Unequal Colleges in the Age of Disparity 9780674982499

Based on quantitative comparisons of colleges since the 1970s, Charles Clotfelter reveals that despite the civil rights

178 9 3MB

English Pages 380 [448] Year 2018

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Contents
Part I. Context
1. Unequal Colleges
2. System, Industry, or Crazy Quilt?
3. Snapshot, Circa 1970
4. Outside Forces
Part II. Supply
5. The Inequality Dividend
6. Zero-Sum Competition
7. Evolution in the Core Business
Part III. Demand
8. Scholastic Segregation
9. Economic Stratification
10. Sorting by Seriousness
11. Sorting by Belief ?
Part IV. Consequences
12. Outcomes
13. Why It Matters
Appendix: Table A.1 Shares by College Category of Total Undergraduate Enrollment in 1,157 Four-Year Institutions
Appendix: Table A.2 The Dwindling Share of Places at Elite Colleges
Notes
References
Acknowledgments
Index
Recommend Papers

Unequal Colleges in the Age of Disparity
 9780674982499

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Unequal Colleges in the Age of Disparity

CHARLES  T. CLOTFELTER

Unequal Colleges in the Age of Disparity

T HE BEL K N A P P RE S S

of

H A RVA RD UNIV ERSIT Y P RE S S

Cambridge, Mas­sa­chu­setts, and London, ­England 

2017

Copyright © 2017 by the President and Fellows of Harvard College ALL RIGHTS RESERVED

Printed in the United States of Amer­i­ca First printing Library of Congress Cataloging-­in-­Publication Data Names: Clotfelter, Charles T., author. Title: Unequal colleges in the age of disparity / Charles T. Clotfelter. Description: Cambridge, Mas­sa­chu­setts : The Belknap Press of Harvard   University Press, [2017] | Includes bibliographical references and index. Identifiers: LCCN 2017016457 | ISBN 9780674975712 (cloth) Subjects: LCSH: Universities and colleges—­United States. | Universities and   colleges—­United States—­Sociological aspects. | Education, Higher—­   Economic aspects—­United States. | Educational equalization—­United States. |   College students—­United States—­Economic conditions. | College students—­   United States—­Social conditions. Classification: LCC LC67.6 .C56 2017 | DDC 338.43378—­dc23   LC rec­ord available at https://­lccn​.­loc​.­gov​/­2017016457 Cover photo courtesy of Thinkstock Cover design by Tim Jones

To my ­mother, Caroline P. Clotfelter

Contents

PART I

Context

 1 Unequal Colleges

3

 2 System, Industry, or Crazy Quilt?

15

 3 Snapshot, Circa 1970

52

 4 Outside Forces

76

P A R T II

Supply

 5 The In­equality Dividend

113

 6 Zero-­Sum Competition

147

 7 Evolution in the Core Business

176

P A R T III

Demand

 8 Scholastic Segregation

207

 9 Economic Stratification

228

10 Sorting by Seriousness

256

11 Sorting by Belief ?

281

P A R T IV

Consequences

12 Outcomes

305

13 Why It M ­ atters

336

Appendix: Table A.1 Shares by College Category of Total Undergraduate Enrollment in 1,157 Four-­Year Institutions 357 Table A.2 The Dwindling Share of Places at Elite Colleges 359 Notes 361 References 409 Acknowl­edgments

425

Index 427

viii C ontents

PART I

Context

1 Unequal Colleges

Although American research universities currently enjoy global preeminence, the same cannot be said for American undergraduate education, which has more often been a cause of anxiety than a source of pride. Concerns about the unevenness of undergraduate education have grown with evidence of falling college completion rates and disappointing results in international comparisons of learning. More prominent have been misgivings about equity and privilege, spurred by evidence that rates of enrollment and completion are strongly correlated to ­family income. As the nation’s income distribution has steadily become more unequal, more than a few observers have asked pointedly ­whether Amer­i­ca’s colleges, rather than acting as a mechanism for equal opportunity, are in fact contributing to more in­equality. This book is intended to provide some much needed perspective, informing discussions inside colleges and policy debates in

Washington and in state capitals. It examines the market for ­baccalaureate education over the last four de­cades, providing an economic and historical portrait of a variegated, complex, and consequential industry. Although this industry is seldom described as such, applying economic concepts like firm, market power, and competition helps to explain how we got to where we are t­ oday. Using the fundamental concepts of supply and demand as organ­ izing themes, the book depicts this market, setting the context by describing the impressive heterogeneity of four-­ year colleges. (Throughout the book I use the term “college” to include universities, not simply for ease of exposition, but also to underscore the book’s focus on undergraduate education.) The book traces society-­w ide changes over the de­cades since 1970 that have affected the college market, one of which was a profound shift in the distribution of income. It examines the supply side of this market, noting in par­tic­ u­lar how the rise of in­equality in the economy paid off handsomely for the already prosperous private selective colleges and how less selective colleges coped with intensifying competition. Shifting the focus to the demand side reveals how students distributed themselves across colleges, beginning with the kind of scholastic sorting that has been linked to the expansion of the market for college from regional to national. The question is w ­ hether students sorted themselves as well by other characteristics, such as their socioeconomic status, studiousness, or po­liti­cal views. Fi­nally, the book examines some outcomes of higher education, among them subsequent gradu­ate training and upward economic mobility, linking ­those outcomes to the colleges where students received their baccalaureate degrees. The book is descriptive. It does not seek to estimate causal models but rather pres­ents a series of quantitative comparisons, both across types of colleges and over time. Some of the categories used to classify colleges are quite detailed, which makes it pos­si­ble to highlight the remarkable heterogeneity in the industry. This heterogeneity turns out to be a star attraction and its explication the basis for the book’s main findings. The numerous graphical depictions address one or both of ­these two questions: How did colleges (and their stu-

4 C onte x t

dents) differ from each other? How did t­hose differences change over time? Many of the graphs make use of the Freshman Survey, an annual questionnaire that posed the same questions to successive cohorts of first-­year students over a period of almost four de­c ades. All of the comparisons over time are based on data for consistently defined categories of colleges and the students who attended them. As many of the figures in the book make clear, some of the differences between colleges and their students are remarkably big. Three main themes run through this book: diversity, competition, and in­equality. Diversity shows itself in many ways, some of which are easily discerned, such as size, age, region, residential surroundings, or the presence of g­ iant football stadiums. Other differences, of equal or greater significance, are less obvious, such as founding princi­ples, academic objectives, or the sorts of courses and majors colleges offer to undergraduates. Public or private, religious or secular, minority-­serving or not—­these distinctions make for a collection of firms—­that is, colleges—­that are remarkably diverse. Competition, the book’s second theme, has taken two forms in the college market. On the supply side, competition resembled the sort found in many other industries, and it was made pos­si­ble not only by the large number of American colleges but also by their in­ de­pen­dence from one another, a feature inherited from centuries of growth among church-­related and other private colleges and the nation’s federal structure. Competition operated on the demand side as well. Over the years the college admissions apparatus became more meritocratic, and the most desirable colleges resisted the urge to ­expand their capacity enough to accommodate the growing demand for places. As a result, students whose hearts w ­ ere set on attending one of ­these desirable colleges, and their parents, found themselves in energetic competition for t­ hese coveted admissions spots. The third theme is in­equality, which inevitably includes both the level of and growth in in­equality. Over the four de­cades covered by this study, the top colleges, which already enjoyed material advantages, further separated themselves from the pack. This pulling away can be observed in colleges’ financial resources and in the kinds of



Unequal Colleges

5

students they enrolled. The major impetus for this increase in in­ equality among colleges was the growth in income in­equality among Americans, an effect further strengthened by a twenty-­year bull market on Wall Street. Another source of growing in­equality among colleges was the competition on the demand side, noted above, for the limited spots at the most selective colleges. Affluent families used a variety of approaches to secure an advantage for their children in the fierce tournament for admission. What Is at Stake

The commodity being supplied and demanded in this market is a service—­baccalaureate education. Not only does it have social significance, it can literally have life-­changing consequences for the student, as is readily illustrated by the experiences of countless individuals, both obscure and famous, who have attended college. Consider the example of Supreme Court Justice Sonia Sotomayor, who at age eigh­teen joined 1,097 other freshmen at Prince­ton University on September 6, 1972, for the opening assembly that marked the beginning of their ­careers as college students. ­These first-­year students heard from the university’s new president, William Bowen, who told them, “Our strength as a university depends, as much as anything, on our attitudes and values, on the personal relationships that we ­ aughter of Puerto Rican–­born parents, are able to develop.”1 The d neither of whom had attended college, Sotomayor had grown up in the Bronx and attended parochial schools close to home. Despite having been the valedictorian at Cardinal Spellman High School, she felt keenly that in g­ oing to college she was entering a wholly unfamiliar world. “­Until I arrived at Prince­ton, I had no idea how circumscribed my life had been, confined to a community that was essentially a village in the shadow of a g­ reat metropolis with so much to offer, of which I’d tasted almost nothing.”2 She had, for example, never heard of Alice in Wonderland or Phi Beta Kappa. But over the next four years she would be profoundly affected by the experience of being a college student. She would have her gram-

6 C onte x t

matical errors highlighted and corrected in one history course. In a psy­chol­ogy lab she would implant electrodes into lab rats. For another class she would write an oral history of her f­ amily, ­going back generations. In a work-­study job, she would gain hands-on knowledge of mainframe computers. She would become an activist, urging the university to hire Hispanic faculty. Spending countless hours in the stacks of Firestone Library, she would write her se­nior thesis on Luis Muñoz Marin, Puerto Rico’s first elected governor. By the time she graduated, she would not only become a member of Phi Beta Kappa but would also win Prince­ton’s top undergraduate prize for academic achievement.3 By any mea­sure, her college experience was life-­changing. In that same fall of 1972, slightly more than one million students entered as freshmen in four-­year baccalaureate degree programs at American colleges and universities.4 Although hers is better known than most from that enormous cohort of freshmen, the story of Sonia Sotomayor serves to illustrate the transformative potentialities that await ­every beginning college student. The economic value of a college education has never been more appreciated than it is ­today, but the value of college goes far beyond any pecuniary return. For many students, as it was for Sotomayor, college can be the gateway to opportunities beyond anything their parents could have ­imagined. Another truth illustrated by Sotomayor is the extraordinary diversity of students’ college experiences. This diversity arises in part from the inherently personal nature of the product. Unlike many goods and ser­v ices that require ­little from the purchaser except money, the ser­vice being bought and sold in the college marketplace requires as one indispensable input attention and exertion by the student herself. Like any paying customer who visits the supermarket or joins a gym, the college student must also be a partner in the production pro­cess. What students get out of college depends on the effort they expend as well as the attitudes they bring with them and the ties they choose to make, as Bowen told the assembly of Prince­ton freshmen. Owing to the marvelous variety among students and the multitude of classes and activities available at most



Unequal Colleges

7

colleges, the ultimate product—­a baccalaureate education—is by its nature an idiosyncratic t­ hing. Beyond t­ hese personal reasons, college experiences differ ­because colleges themselves differ. In addition to easily observed characteristics like location or architecture, they differ in two main re­spects. First, they are not all trying to do the same t­ hing. Although Prince­ton offers undergraduate degrees in arts and sciences and engineering, it lacks many features available at other colleges, simply ­because its objectives are not the same as t­ hose of other colleges. For example, it does not offer an undergraduate degree in business. It has no mandate to educate the state’s public school teachers, so it has no education school. Nor does it have an overarching religious mission. T ­ hese features are the result of Prince­ton’s unique history, and the same can be said for e­ very other college. Besides this variety of objectives, the other reason colleges differ from each other is that they possess unequal capacities for excellence. Prince­ton is one of the most richly endowed colleges in the country, boasting world-­class scholars as well as unsurpassed facilities. In 2013 the value of its endowment was more than $18 billion.5 By contrast, scores of colleges must make do with meager financial under­pinnings, which limit nearly every­thing they can offer their students, from the courses they schedule and the accomplishments of their faculty to the quality of their physics labs and recreation facilities. The in­equality of colleges is also reflected in the wide variation in the talent, training, and predilections of their students and therefore the peers to whom each student is exposed. The other Prince­ton students with whom Sotomayor would share classes, dormitories, meals, and extracurricular activities over the next four years differed markedly—in socioeconomic background and academic preparation, among other ways—­from t­ hose whom she would have encountered at other colleges. Although they all offer four-­year bachelor’s degrees, the institutions that make up the college “industry” differ tremendously in the academic qualifications of their students. This fact is obvious to every­one involved, although it is seldom dwelt on, owing

8 C onte x t

to the constraints of politeness. Beyond the readily observed characteristics, students entering the most selective colleges bring with them habits of studiousness unmatched by t­hose at less selective schools, and this in­equality, also rarely acknowledged, is a central feature of the college market. Thus the differences among colleges arise from the variety of institutional objectives and from the in­equality in resources, both ­financial and h ­ uman. Both of t­ hese sources of difference are of central importance in understanding Amer­i­ca’s college market, and they are the main concerns of this book. In words, diagrams, and numbers, the book describes the remarkable heterogeneity of colleges in this country. It seeks to upset the habit of government statistics and international comparisons that invariably treat all bachelor’s degrees as if they ­were identical, while providing something far more in-­depth than what one encounters in the rankings of U.S. News and in college guides like the Fiske Guide to Colleges. It is surprisingly hard to find studies that examine the broad ­variety of colleges that exist. This is ­because scholars and journalists alike have tended to devote a disproportionate share of their attention to selective, elite institutions, largely ignoring (with the exception of the sports pages) the vast majority of colleges and universities. ­T here is much to be gained from adopting a broader perspective, one that is wide enough to encompass the entire collection of ­four-­year colleges. Most obviously, a broader perspective can be a counterweight to the near obsession with elite colleges, a group that together awards only a tiny fraction of American baccalaureate degrees.  Less obviously, examining the entire set of four-­year colleges makes it plain that the nation’s colleges are profoundly unequal. There has been and continues to be in­equality in financial wherewithal, ­in­equality in the academic rec­ord of entering students, ­in­equality in faculty pay and academic reputations, and in­equality in the social and economic standing of students and their parents. This in­equality is not a new t­ hing. What is new is the mounting concern that the in­equality is getting worse. ­There is alarm in par­tic­ u­lar over what many observers believe to be a growing concentration



Unequal Colleges

9

of wealthy students at the elite colleges, some writers maintaining that the system actually reinforces economic in­equality. This concern finds its way into articles with titles like “As Wealthy Fill Top Colleges, New Efforts to Level the Playing Field,” “The Rich-­Poor Gap Widens for Colleges and Students,” “Amer­i­ca’s Top Colleges Have a Rich-­K id Prob­lem,” “Generation ­Later, Poor Are Still Rare at Elite Colleges,” and “A Bigger Affluence Gap Than Expected at Elite Colleges.”6 The last of ­these reported that ­there ­were thirty-­eight, mostly elite, colleges that enrolled more students whose parents had incomes in the top 1 ­percent of incomes than students with parents in the bottom 60  ­percent of incomes.7 The pressing question for ­those concerned with public policy is ­whether our colleges are offsetting the growth of income in­equality in the country or actually exacerbating it, and if the latter what to do about it. The urgency of addressing this question arises most obviously from the importance it holds for the distribution of income and privilege in this country. Although Americans have been raised to look on their country as a land of opportunity, the truth is that the United States has one of the lowest rates of intergenerational mobility among developed countries. Among thirteen countries in the Organ­ization for Economic Co-­operation and Development (OECD), the United States was third-highest in the degree to which a f­ ather’s earnings predicts his son’s earnings.8 If colleges are at the center of the social machinery that determines how positions of influence and wealth are handed from one generation to the next, t­ hose colleges deserve our serious scrutiny. And it is not just a question of equity. It is also, in the terminology of the economist, a question of resource allocation and efficiency. Does the matching of academically strong students with wealthy colleges represent the best use of resources in the economy? It would if ­those students could get more out of attending an elite college than could less able or less academically prepared students. In that case, some apparent inequity in matching students to colleges might be socially desirable if it leads to a more efficient allocation of resources in higher education, which in turn produces more national income and faster economic growth. 10 C onte x t

If concerns such as ­these are to be aired and policies affecting them are to be debated, it would be useful to consider our system of higher education from a historical perspective, one that views that system in the context of larger developments in society. In the four de­cades ­after 1970 the country underwent several profound changes that had par­tic­u­lar importance for higher education. One was the gradual but unrelenting reshaping of the country’s income distribution, a change that has been accompanied by major shifts in the economy’s industrial makeup and ­labor market. In 1970 the United States was a manufacturing power­house, leading the world, for example, in producing automobiles. Many manufacturing jobs w ­ ere ­unionized, and they offered good pay and generous benefits to workers having just a high school diploma. As ­these well-­paying manufacturing jobs steadily dis­appeared, the earnings of high school gradu­ates stagnated while ­those of college gradu­ates began a steady ascent. More broadly, the country’s income distribution began to change, with t­ hose on the top rungs enjoying growing affluence while many in the ­middle and below saw their fortunes falling. ­These changes in the income distribution affected both demand and supply in the market for college. As the return to college and the incomes of the most affluent families grew, this group’s demand for admission spots in ­those name-­brand elite institutions intensified. Never before, it seemed, had it been so crucial for a young person to be admitted to a prestigious college like Brown, Stanford, or Amherst. Although the supply of coveted admission spots at this handful of colleges did expand somewhat to accommodate the increased demand, it did not expand enough. Admissions standards steadily toughened. One result was a frenzy of activity, by ambitious students and their parents, to find ways to gain admission into one of t­ hese desirable colleges. Not only did thousands of high school se­n iors routinely take SAT test-­prep courses, they also volunteered in soup kitchens, learned new team sports, and spent more time hitting the books. Yet, as all-­consuming as it could appear in suburban communities filled with professional c­ ouples, this kind of frenzied pursuit of admission to an elite college ­really affected only one thin slice of a

Unequal Colleges

11

deeply segmented college system. For the majority of four-­year colleges and universities, admissions standards had never been especially exacting, so students applying to them had no reason to take extraordinary efforts to get in. As college-­age cohorts began to shrink around 1990, admissions at the less selective colleges soon took on the look of a buyer’s market. Indeed, scores of ­these colleges, heavi­ly dependent on the tuition from enrolled students, had already begun living financially precarious lives, driving them to craft financial aid offers designed to strike a balance between attracting tuition-­paying students and students who would raise their school’s academic profile. Many of ­these colleges modified their academic programs, the less prestigious ones tending to add vocational majors to their offerings. They made other changes as well, including spending more on student ser­vices, amenities, and athletics. To understand the importance of decisions like t­ hese, taken at many less competitive colleges, we clearly need to go beyond the fixation on the elite colleges. A Market’s Logic

To understand how Amer­i­ca’s colleges and universities responded to the forces impinging on them in ­these de­cades, it is essential to understand what kind of “industry” the one for baccalaureate education is and what kind of “firm” a college is. A distinguishing mark of this industry, besides the remarkable diversity of its firms, was its decentralization. Unlike Eu­rope, Amer­i­ca has never, except for its military academies, had national universities. History dictated instead that our colleges would be controlled ­either by private governing boards or by state governments. When the federal government did grandly intervene in higher education, during the Civil War and again ­later in the nineteenth ­century, it designated the states, not itself, as the locus of control of public higher education. The result of this choice, on top of two preceding centuries of growth among church-­related and other nonprofit colleges and the nearly absolute tendency of colleges to remain geo­graph­i­cally anchored, has been a

12 C onte x t

vast array of institutions, each seeking to make its own way in the world. As they did so, in­equality among them grew. ­After 1970 the top colleges, which already enjoyed material advantages, further separated themselves from the rest. This can be seen in financial resources. It can also be seen in the students t­ hese colleges enrolled. Even before they arrived on campus for the first time, ­these consumers had sorted themselves thoroughly in terms of academic readiness, as measured by their grades, their test scores, and their study habits. They sorted themselves as well by socioeconomic status. Although the share of top-­ income students enrolling in the most selective colleges did not change much, the incomes of ­these students did, causing average incomes in the vari­ous college categories to spread out over time. The major impetus for this increase in in­equality among colleges was the growth in income in­equality, which boosted demand for elite colleges and stimulated charitable donations to colleges and universities. ­T hese effects ­were bolstered by rising stock prices, which further stimulated donations and in the bargain helped to push endowments in the richest private institutions to unpre­ce­ dented heights. In another contributing development, the college market continued becoming more national, even international, in scope, especially the selective colleges, which exacerbated the winner-­ take-­a ll nature of the competition for admission spots. Aiding t­ hese trends was the rise of the U.S. News rankings, which provided an arena for competition for colleges. One more force for greater in­equality was a decline in the support that state legislatures gave to public colleges and universities. Leaning in the other direction w ­ ere antidiscrimination policies and more meritocratic admissions criteria. Although the practical meaning of affirmative action remained an ongoing, contested arena of debate, the rules and practices governing entry into colleges w ­ ere noticeably altered, narrowing some of the easy paths that affluent and well-­connected families once enjoyed; test scores and Advanced Placement credits mattered more, connections less so. ­Because the most desirable colleges did not expand their capacity enough to



Unequal Colleges

13

accommodate the heightened demand for places, the competition for t­ hese coveted admissions spots intensified. Affluent families found it necessary to look to new strategies for getting their c­ hildren into selective colleges. But adjust they did, using to their advantage merit scholarships, early decision, test-­prep courses and tutoring, retaking the SAT, unpaid internships, athletic skills in nonrevenue sports, private schools, and legacy preferences. One apparent result of ­these efforts was an unexpected bifurcation in reported study time in high school. While average study time in high school declined for most new college students, for ­those attending the most selective colleges study time increased. ­These changes in the market for baccalaureate education did, therefore, have a certain logic to them. Forces of competition ­were at work among the colleges that supplied education as well as the customers who demanded it. Impinging on this competition was a set of unpre­ce­dented economic and social changes that gave comfort to the already well-­endowed, selective colleges, but left many more colleges battling just to maintain the status quo. The result was a spectacularly segmented market that became more so in the 1990s and beyond. Meanwhile the daily business of ­these colleges continued apace, but at dif­fer­ent levels. This business was providing the conditions for students to transform their lives. At the colleges with the richest resources and the best students, some of ­those transformed lives ­were ­those of students who, like Sonia Sotomayor, would become celebrated college gradu­ates. Amer­i­ca’s global preeminence in higher education rested on the quality of institutions such as her alma mater. But the quality of many more colleges did not enjoy such universal acclaim. ­These contrasting reputations are one consequence of the in­equality of American colleges.

14 C onte x t

2 System, Industry, or Crazy Quilt?

If anything about American colleges and universities is true, it is that they ­were not created equal. Nor have they tended to become more equal as they matured. One anonymous observer quipped that the United States might have fifty of the best universities in the world and five hundred of the worst.1 The country’s four-­year colleges differ from each other, sometimes radically, along a host of dimensions, most noticeably by age, location, architecture, and size. G ­ oing beneath the easily observed, they also differ in elusive but more significant ways. One is the oft-­memorialized aspect of “mission.” Does a college have a mandate to provide ser­vices to the citizens or industries of its state? Does the college endorse and follow a set of religious princi­ples? Does its faculty conduct research? Do parties and spectator sports loom large in students’ lives? Do its undergraduate courses teach practical skills that students can use a­ fter graduation to get jobs, or is the cata­log loaded with courses on esoteric subjects?

As we w ­ ill see, the degree to which a college’s courses teach “useful” skills ­w ill be a good predictor of where the college resides along ­a nother elusive dimension of differentiation: prestige. Prestige in turn is correlated with the average SAT scores of students and the ­difficulty of gaining admission. Most of t­ hese differences have roots in history—­each college’s inheritance of founding princi­ples, past achievements, physical structures, location, financial wealth, faculty, and market position. Not to be neglected among t­ hese inherited features is religion. Many if not most private colleges ­were founded by religious bodies or for religious reasons. Although ­these religious ties have weakened over time, as society at large has become more secular, the vestiges of ­these religious origins remain. Among Amer­i­ca’s most famous colleges, some—­like Notre Dame, Brandeis, Brigham Young, or Oral Roberts—­are indelibly marked by their religious origins. Historical roots are evident as well in the most secular of colleges, the country’s public universities. One need only consider the land-­ grant universities established by the first Morrill Act of 1862 to make this point. Compare ­those mostly large universities, with their extension ser­vices and schools of agriculture and engineering, with the myriad public universities originally founded as teachers colleges. Or to the predominantly black colleges covered by the second Morrill Act in 1890, originally designed to serve black students in the states where Jim Crow segregation was the law. Other dimensions of diversity can be seen in universities that began as colleges for w ­ omen or for Native Americans, or as urban commuter colleges or two-­year colleges. Prob­ably the most significant cleavage in the market for college is that which separates for-­profit from the rest, a dividing line so deep that most scholarly studies of higher education—­ including this one—­omit the for-­profit sector altogether. In addition, this book leaves aside two-­year colleges. As broad as their reach is, their purposes and functions are so varied, and generally so distinct from ­those of four-­year institutions, that they can be reasonably excluded. What is left is still vast and variegated—­some 1,600 colleges and universities offering baccalaureate degrees. Of t­ hese, roughly 40 ­percent

16 C onte x t

are creatures of government—­mostly state governments—­a nd the rest are beholden only to their boards of trustees.2 This public-­ private distinction colors every­thing ­these colleges do, although many of the a­ ctual programs they operate can be very similar. Ironically, this diversity coexists with what appears to be a per­sis­ tent urge to standardize. ­These unequal colleges share more than a few common outward features, such as degree names, grading conventions, departmental organ­i zation, and occupational titles. Within this industry we behold a collection of “firms” all of which purport to sell (to customers able to perform certain tasks) one of several seemingly standardized commodities—­degrees—­each beginning with the words “bachelor of.” So standardized are t­ hese degrees that, for many employers and universities and for virtually all statistical purposes, a degree from any college whose quality has been certified by one of the country’s regional accrediting agencies w ­ ill be treated as representing a common level of educational achievement. To be a college gradu­ate means simply having a bachelor’s degree from any accredited college or university. In fact two opposing tendencies are constantly at work among the producers in the market for bachelor’s education. On the one side is a general tendency for organ­i zations to mimic other organ­ izations they see as rivals. With few exceptions, the fear of being dif­fer­ent keeps college leaders from drifting too far away from the pack, keeping in mind that the ­whole universe of colleges is divided into more than one reference group. 3 This tendency t­ oward conformity has been formalized by orga­nizational sociologists as “isomorphism.”4 This line of thought maintains that the tendency ­toward conformity is explained by the need of each organ­ization to maintain its legitimacy in the eyes of the world. 5 This tendency, in other words, arises out of an organ­ization’s most primal instinct—­the urge for survival. This isomorphic tendency can perhaps best be seen in college curricula. Just as colleges abandoned their emphasis on Greek and Latin in the nineteenth ­century and winnowed required courses in ­favor of electives in the twentieth ­century, they have also tended to make other changes en masse.

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

17

Alongside this tendency ­toward conformity, however, exists another per­sis­tent and seemingly countervailing tendency for colleges to try to differentiate themselves from their rivals.6 Often forced to by their own histories, some colleges persist in their distinctiveness. For the most part, however, the impressive diversity in origins across American colleges is tempered by a power­ful tendency ­toward homogenization, animated by the instinct for institutional survival. In the descriptions that follow, t­ hese two opposing forces are on display. Three Colleges

One way to suggest the degree of diversity of American four-­year colleges is to take a close look at several of them, starting with the impressionistic and descriptive rather than the quantitative. This chapter therefore begins with brief portraits of three colleges as they appeared at the beginning of our period of study, around 1970.7 Illinois State University

State governments’ desire for colleges first arose from the urgent need to train teachers for their public schools. The colleges designed to do this—­normal schools—­first appeared before the Civil War, then proliferated as the demand for them grew. Whereas t­ hese normal schools enrolled just 4 ­percent of all college students in 1860, by 1900 they enrolled 30 ­percent of them.8 Illinois State was one of the first. Even the town where it is located ­adopted the name—­Normal. Founded as Illinois State Normal University in 1857, it offered no four-­year degrees for the first fifty years of its existence, and it was not u ­ ntil 1943 that it began training gradu­ate students, by offering a master of science in education. In the 1960s the state’s legislature and higher education governing board approved expanding the university’s official purpose, a shift commemorated by name changes that culminated in the current one, Illinois State University, in 1967. Like other former teachers colleges, Illinois State entered the 1970s as a second-­tier public institution, having to be content to l­abor in

18 C onte x t

the shadow of its state’s flagship research university, the University of Illinois. With slightly more than 15,000 undergraduates in 1970, Illinois State was one of the largest universities in Illinois, though well ­behind the University of Illinois, whose Champaign-­Urbana campus enrolled more than 23,000 undergraduates. Reflecting its tradition of teacher training, 60 ­percent of its undergraduates majored in education, and ­women made up a majority of the students (54 ­percent), far larger than the 38 ­percent female share at Champaign-­Urbana.9 Nearly all of t­hese students, 97  ­percent, ­were from Illinois. Illinois State lagged the state’s flagship in the average per­for­mance of its undergraduates on the ACT standardized test, a requirement for most colleges in the Midwest. Converted to the equivalent SAT score (and expressed in terms of the recentered scale used ­today), Illinois State students averaged 1094, an average that put the college near the 44th percentile among all the colleges and universities in the country in 1970.10 By comparison, students at the state’s flagship university at Champaign-­Urbana had an average SAT score of 1252, slightly above the 90th percentile. Some 82 ­percent of applicants to Illinois State in 1970 w ­ ere accepted for admission. In its 1970 cata­log Illinois State describes itself as a “multi-­purpose liberal arts institution” and states its purpose in ­these words: “Its unique contribution and purpose is to serve the citizens of this state and nation by means of its commitment to knowledge: its creation (research), transmission (teaching), and application (ser­vice).”11 The university’s undergraduate curriculum required students to take courses in En­glish, speech, social sciences, humanities, math and natu­ral sciences, and physical education. Among the twenty-­six majors offered in 1970, six ­were in education (elementary, dance, health and physical, ju­nior high school, special, and business), befitting the school’s origins. Also offered w ­ ere quite a few majors that had a clear connection to practical skills that would be directly applicable in the l­abor market. T ­ hese included business administration, home economics, industrial technology, library science, and speech pathology. Roughly a third of the undergraduate majors could be

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

19

similarly classified.12 In 1970 sociologist Jeanne Binstock described teachers colleges as tax-­supported institutions that fostered Protestant, middle-­class values while they trained middling students for middling jobs in the economy.13 As we w ­ ill see, the emphasis any college places on such practical majors is correlated with its academic standing, its prestige, and the socioeconomic status of its students. Tuition and fees in 1970 / 71 for an Illinois resident was $470 a year.14 The school’s faculty ­were well-­compensated by the standards of the day, with full professors earning an average of $20,200 a year in compensation (salary plus fringe benefits), exactly the same as the average for full professors at the state’s flagship, the University of Illinois at Champaign-­Urbana. On the Illinois State campus t­here ­were seven social fraternities, and 3 ­percent of men did choose to join one. ­There ­were no sororities.15 Based on figures in Cass and Birnbaum’s college guide, about 60  ­percent of undergraduate students lived in dormitories, fraternities, or other university housing.16 Boston College

Our second illustrative college differs in impor­tant ways from Illinois State, in that it is both private and religious in affiliation. Founded in 1863, Boston College was the proj­ect of the Roman Catholic Society of Jesus, better known as the Jesuits, an order that founded other prominent Catholic colleges, among them Fordham, Georgetown, and Holy Cross. Early in the twentieth c­ entury the college purchased a 200-­acre campus in the affluent Chestnut Hill section of Newton, a Boston suburb situated between two arms of the city’s historic Green Line rapid transit network. In terms of its immediate environment, the one offered by Boston College was markedly dif­fer­ent from Normal, Illinois. Despite its explicit Catholic connection, Boston College’s stated purpose, as recorded in its cata­log, lacked any explicit religious references: “The primary objective of Boston College is stated on the official seal of the University: Ever to Excel. Boston College is committed to the conservation, extension, and diffusion of knowledge. Its purpose is to impart an understanding of the unity of knowledge, an appreciation of our 20 C onte x t

cultural heritage, a dedication to the advancement of learning, and a sense of personal and social responsibility, both within the College and University, and beyond, in the community and the world.”17 The undergraduate curriculum differed in several notable re­spects from that of Illinois State. The most striking difference was Boston College’s requirement that students in arts and sciences take at least two courses (out of a total of thirty-­eight) in theology.18 T ­ hese courses did not have to be based on Christian theology, however.19 A newly approved curriculum required Boston College students to take, in addition to the pair of theology courses, two courses each in history, science or math, philosophy, social sciences, and humanities.20 Students who enrolled at Boston College w ­ ere unusually well-­ prepared academically, judging by their average per­for­mance on the SAT. Entering students in 1970 recorded an average score of 1320, ranking Boston College in the top 5 ­percent of all colleges in the country.21 As one indication of the college’s moderate selectivity, 61 ­percent of ­those applying ­were accepted for admission. In sharp contrast to the high share of in-­state students at Illinois State, just 59 ­percent of entering students at Boston College came from New ­England. The college’s religious tradition was evidently a f­actor in who attended: an estimated 90 ­percent of students ­were Catholic.22 Tuition and fees ­were $2,500. The average compensation for full professors was slightly higher than that at Illinois State, $21,000.23 About 22 ­percent of students lived in dormitories located on campus. Adding other housing options operated by the college, slightly more than a third of all undergraduates lived in some type of college-­ controlled housing. The college had no recognized fraternities or sororities. Unlike some private colleges, Boston College offered a number of vocational majors, though it separated them from arts and sciences majors by locating them in distinct schools—of business, nursing, and education. In 1970 the college offered twenty-­t wo majors in the arts and sciences college, plus ­others in ­these three professional schools.24 Some 29 ­percent of students chose to major in some area of business. Most popu­lar among the arts and sciences majors was En­glish, with 10 ­percent choosing it.25 Compared to Illinois State, S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

21

Tennessee Wesleyan

Behavioral science Foreign language Music education

Art Health and physical education Social sciences Arts and science Business education Dance education / physical education French Geography Home economics Industrial technologies Junior high school education Latin Library sciences Russian Special education Speech pathology Theater Illinois State

English Psychology Religion Secondary Business education administration Speech and Classical studies Chemistry theater Economics Fine arts Elementary education Germanic studies History Military science Mathematics Natural science Nursing Philosophy Geology Physics Political science Romance languages and literatures Slavic and eastern languages Sociology

Boston College

Figure 2.1. ​Undergraduate Majors Offered at Three Colleges, 1970 / 71 Source: College cata­logs, 1970 / 71.

Boston College’s list of majors featured more abstract, traditional arts and sciences disciplines. Among the majors it offered that Illinois State did not w ­ ere En­glish, psy­chol­ogy, philosophy, physics, po­liti­cal science, romance languages, and sociology. For its part, Illinois State offered majors that Boston College did not. Most of ­these ­were directly applicable to the professional aims of ­those who would be teachers or would work in schools. Figure 2.1 pres­ents a Venn diagram showing the majors each institution offered in 1970 and the extent to which t­ hose majors overlapped.

22 C onte x t

Tennessee Wesleyan

The last of the trio of illustrative colleges is Tennessee Wesleyan College, a small college located in the town of Athens, Tennessee, sixty miles southwest of Knoxville. As one of more than a dozen eponymous colleges across the country honoring Methodism’s founder John Wesley, it was affiliated from its founding with that denomination, historically the largest Protestant denomination a­ fter the Baptists.26 Founded in 1857, the same year as Illinois State, as Athens Female College, it eventually became coeducational. Over the years it operated successively ­under a total of six dif­fer­ent names. The significance of its Methodist connection went well beyond the financial support the denomination provided. Unlike the objectives associated with the religious under­pinnings of Harvard and most of the earliest colleges, the religious mission of Tennessee Wesleyan was less about training ministers as it was sustaining a community of believers, by providing Methodists a religiously supportive place to send their sons and ­daughters. The college’s 1970 cata­log reflects this religious orientation as it extolls the advantages of its small size, which make pos­si­ble “in-­depth relationship between instructor and student and strong personal identity of the student with the institution. It is designed to prepare gradu­ates ­under the respective majors to be knowledgeable, able to think critically and creatively, and to develop a basis for judgment. We believe ­these to be the ele­ ments by which leadership is produced in the arts, business and industry. The entire program is predicated on the Christian faith which continues to be man’s best hope in his search for Truth.”27 The college required students to participate in a minimum number of religious and cultural events. It also required students to take one course (out of thirty-­three) in religious and philosophical studies. Complementing its small enrollment was a diminutive campus. Its eleven academic buildings included a recently completed science building, a building combining per­for­mance space and dining, a library, an administration building, a gymnasium with bleachers to accommodate as many as 1,500 spectators, a ­house for the president,

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

23

two other ­houses, and three other academic buildings containing classrooms, faculty offices, and an auditorium. Six dormitories plus an infirmary completed the campus. Its 1970 cata­log listed nineteen majors, including business administration and three education majors (elementary, health and physical, and secondary). The most popu­lar majors w ­ ere education (33 ­percent) and business and commerce (24 ­percent).28 Most of its majors ­were also offered by Illinois State or Boston College, or both, as Figure 2.1 shows. The average SAT for entering Tennessee Wesleyan students in 1970 was 1026, slightly below that of Illinois State, placing the college at roughly the 20th percentile of colleges and universities in 1970. Corresponding to this low average score was a relatively unselective admissions pro­cess, with 95 ­percent of t­ hose applying for admission to the college being accepted. Although not as local in geo­g raph­i­c al reach as Illinois State, the college was decidedly regional, with some 87  ­percent of its students coming from the South.29 Two-­thirds of students lived on campus, nearly all of them in one of the six dormitories. ­There ­were three fraternities and four sororities, and slightly more than a third of all students, men and ­women, ­were members of one of them. Contrasting Types of Colleges

As t­hese three thumbnail descriptions suggest, American colleges and universities are a heterogeneous lot. They can differ by o ­ rders of magnitude by size, with enrollments ranging from the hundreds to the tens of thousands. They are strewn all across the country, from well-­ heeled suburbs of booming metropolitan areas to gritty urban neighborhoods well past their primes to sleepy hamlets and middle-­size towns of the heartland whose main employer is that very college. One large subset of them are creatures of state government, bearing some of the marks familiar to government agencies. Among ­those that are not operated by government, many are tied to religious organ­izations, and for some of ­those the connection runs deep. ­Others outside the government sphere are quite secular. Another line that divides American colleges and universities is a dis 24 C onte x t

tinction that exists in no other country—­between ­t hose that do and do not engage in commercial, spectator sports. Quite a large number of colleges and universities compete in such “big-­t ime” sports, some generating sizable revenues and publicity. Colleges differ as well in the academic subjects that students pursue, even at the undergraduate level. While many colleges offer a range of programs that prepare students for employment, some colleges stick to liberal arts courses, leaving it to students to pick up job-­related skills ­later. To t­hese observable dimensions of difference must be added an invisible but supremely impor­tant one: prestige. Long a construct used by sociologists, prestige, or status, can be exhibited and maintained in any number of ways. In the United States, according to legendary sociologist C. Wright Mills, education, along with occupation, became a primary instrument for signaling and maintaining prestige, owing to the country’s lack of an aristocracy.30 Prestige attaches not only to individuals but also to groups and the organ­ izations they join. Thus colleges, like clubs and charities, have their own status hierarchy. Although this prestige resists mea­sure­ment, its imprint is everywhere. Prestigious colleges are emulated and deferred to. Most obviously, their high status finds credible confirmation by way of published college rankings. Indeed, ­there would be wide agreement that the prestigious colleges differ from the pack in terms of academic quality—­a metric approached delicately but one that is unavoidable when assessing an industry that is deeply invested in ­matters of the mind. Clearly, ­there is no shortage of pos­si­ble criteria one could use to arrange the mass of colleges into categories. One study by orga­ nizational sociologists identified no less than a half dozen such schemes.31 To keep ­things as uncluttered as pos­si­ble, the pres­ent study classifies colleges using just three criteria: (1) w ­ hether the college is public or private, (2) its undergraduates’ average SAT score at the beginning of the period, and (3) w ­ hether the college is a historically black college or university (HBCU). The public / private divide provides a natu­ral dividing line for any number of reasons, and it is a familiar division used in statistics of higher education. The HBCU criterion is less standard, but ­these institutions are so distinct in the S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

25

American context that they deserve separate attention. Grouping by student SAT scores is suggested by the work of Caroline Hoxby. 32 She found that the colleges and universities (hereafter colleges) ranking in the top deciles in the 1960s became more selective over time, as mea­sured by the percentile ranking of their students’ SAT scores. In contrast, colleges below the m ­ iddle saw declines over time. According to her analy­sis, ­these trends produced growing heterogeneity (in relative SAT levels, at least) across institutions over time and increasing homogeneity within them. This study likewise classifies each college according to its initial position in the SAT distribution (using 1970 as the starting point) and, for consistency of comparisons over time, each college remains in its original category. ­These three organ­izing criteria lie beneath most of the ­tables and graphs in this book, including all of the figures featuring seventeen dif­fer­ent categories. Although this number of categories might strike some readers as excessive, the resulting fine gradations pay off by revealing impor­tant, and in some cases stunning, differences.33 We ­will see, for example, big differences just among private colleges in the highest 10 ­percent of average student SAT scores. In 2008 / 09 the average per student spending in colleges in the 99th and above category was more than twice the average in private colleges in the 90–95th percentiles. For endowment per student a few years ­later, the corresponding ratio was more than 5:1. 34 ­These categories are central to the method used ­here ­because this approach relies on quantifiable facts, not impressions. And keeping colleges in the same categories is essential for making apples-­to-­apples comparisons over time. This feature is essential, since the purpose is to examine patterns of difference among colleges as well as changes in ­those patterns over the course of four de­cades. Let a Hundred Flowers Bloom: Variety of Aims

To demonstrate their worth in describing the impressive diversity across suppliers in the baccalaureate marketplace, the three dimen-

26 C onte x t

sions can be used to create five illustrative types, or clusters, of colleges. One of the five is the HBCU. The remaining four are (1) low-SAT (0–50th percentile) public colleges, (2) high-SAT (80th + percentile) public colleges, (3) lowest-SAT (0–25th percentile) private colleges, and (4) very-high-SAT (90th + percentile) private colleges. ­These five clusters do not cover all colleges. They leave out colleges with middling SATs. In addition, the gap between the high-­and low-­SAT clusters of private colleges is bigger than that used for the public colleges. 35 The aim ­here is to construct contrasting, illustrative clusters, not to cover all colleges. 36 The following goes beyond the quantitative and offers descriptive material drawn from college cata­logs to provide a fuller picture of t­ hese contrasting types of colleges around 1970. Low SAT Public Colleges

In 1970 ­there ­were 271 public colleges and universities whose average SAT ranked in the bottom half of all the baccalaureate colleges for which data on students’ average SAT scores could be found. Many of ­these schools, like Illinois State, had formerly been normal schools and teachers colleges. For ­these schools, teacher training was not only their original purpose, it also remained a primary focus, although by 1970 they had broadened their course offerings to include other curricular material, including programs in business, other practical pursuits, and the liberal arts. Several referred to themselves as multipurpose universities. For the g­ reat bulk of t­ hese universities (for nearly e­ very one of them had inserted “university” into its name), the statements of purpose that they frequently included in their cata­logs tended to be broad, bland, and unremarkable. Some examples: “Shepherd College is a multiple-­purpose state college whose function is to provide educational programs in teacher education, pre-­professional training, and liberal arts education”; “It is the purpose of Armstrong State College to furnish students with a basic understanding of the intellectual structure of civilized life and to provide some of the knowledge and

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

27

experience necessary to enable them to become responsible contributors to their civilization”; “The objectives of the educational program at Kansas State University are to develop an individual capable of applying an enlightened judgment in his professional, his personal, and his social life.”37 Readers may recognize in such bland generalities a familiar product of deliberative committees whose consensual approach guards against slighting any significant departments, disciplines, or functions being carried out within the university. But the statements themselves—­particularly their conformity—­reflect ­either the fact that many of the schools in this group w ­ ere indeed similar or that they wished not to be seen as dif­fer­ent. Despite this tendency ­toward uniformity, however, ­there ­were a few notable exceptions. One of ­those exceptions is represented by the country’s land-­g rant universities, some of which ended up in this category ­because their students scored in the bottom half of the national SAT distribution in 1970. Chosen by states and financed by the Morrill Acts of 1862 and 1890, t­ hese institutions carry with them an indelible focus on the practical skills associated with agriculture and industry. H ­ ere are many of the country’s most prominent state-­ supported schools of agriculture, engineering, and architecture. Although most of ­these colleges posted average SAT scores in the top half of all colleges in 1970, a few landed in the bottom half, among them the University of Kentucky and Ohio State. (Both w ­ ere 38 close to the 40th percentile in 1970.) In describing its purpose in its cata­log, the University of Kentucky does not mention its land-­grant status, but rather offers this rather generic statement: “Its major function is as a teaching institution, but it also serves as a center for research, as a ser­vice agency, and as a center for the preservation of knowledge.”39 Yet t­hese two universities did stand out, as a list of their majors quickly shows. Among the undergraduate majors ­offered at both of ­these land-­grant universities in 1970 ­were the following applied fields: agricultural economics, agronomy, civil engineering, electrical engineering, entomology, horticulture, mechanical

28 C onte x t

engineering, and plant pathology. To ­these, Ohio State added animal science, dairy technology, metallurgical engineering, poultry science, and welding engineering.40 As one indication of just how distinctive ­these land-­g rant universities w ­ ere from other public institutions, Illinois State, which is not a land-­grant university, did not offer a single one of ­these fourteen majors. Even more distinctive among the low-SAT public institutions, indeed among all four-­year colleges, is the tiny number of military colleges. One of ­these, a state-­supported college whose average SAT in 1970 put it near the 30th percentile, was The Citadel, in Charleston, South Carolina. Describing itself in its 1970 / 71 cata­log as a “liberal arts military college,” The Citadel states an approach to higher education that marks it as dif­fer­ent with a capital D: “Military training teaches the value of a methodical and orderly approach to tasks, of physical and ­mental fitness, and of alertness and self-­ confidence. It teaches how to achieve and maintain morale and discipline. Most impor­tant of all, it instills the conviction that sacrifice is preferable to compromise with princi­ple, and that truth, honor, and integrity are the bases of character.”41 In practice, according to Citadel gradu­ate and author Pat Conroy, ­these princi­ples yielded conditions for first-­year students worthy of a gulag and a dropout rate over 50 ­percent in the first year.42 But how many other colleges would have on its dormitory wall a brass plate with the inscription: “DUTY IS THE SUBLIMEST WORD IN THE ENGLISH LANGUAGE. Robert E. Lee”? ­There can be no better illustration than this college that Amer­i­ca’s four-­year colleges are not simply pale imitations of each other. A significant dimension of difference too often ignored by scholars of higher education is the nonacademic side of college. Despite what economists assume in models of ­human capital development, and as disappointing as it might be to some college professors, college students devote a considerable portion of their attention to nonacademic ­matters.43 Among ­these are romantic relationships, work, socializing, exercise, and friendships (though not, in 1970, enhanced

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

29

by Facebook). Although much in this nonacademic realm cannot be easily quantified, ­there are three mea­sur­able characteristics that do differ from college to college. One is the degree to which a college provides a residential community on campus. Among the low-SAT public colleges whose cata­logs w ­ ere examined, the percentage of students who lived in campus dormitories varied quite widely from colleges that ­were mostly residential in nature, such as Eastern Kentucky (with 63 ­percent of its students living in dormitories) and Appalachian State (58 ­percent) for example, to urban universities where most students commute from elsewhere to attend classes. At Memphis State, just 5 ­percent of students lived in dorms in 1970. Georgia’s Augusta College had no college dormitories at all.44 A difference among the colleges in this group that might influence students’ social lives was in the importance of social fraternities and sororities. Some of t­ hese colleges had no such student organ­izations in 1970, among them North Carolina’s Appalachian State and Mary­land’s Towson State. In contrast, 13 ­percent of students at the University of Southern Mississippi belonged to a fraternity or sorority, and 12 ­percent of Ohio State students did.45 A third nonacademic feature of colleges and universities was commercial spectator sports, an activity unique to American higher education, and one whose reach often extended well beyond the campus walls. Public universities in the low-SAT cluster differed in this re­spect. At one end of the spectrum ­were universities whose football and basketball teams ­were nationally known. Ohio State, which had sponsored football teams for more than a c­ entury, played in the Big Ten conference against other, mostly public flagship, universities in the Midwest. Its football team went undefeated in the 1970 season, losing only to Stanford in the Rose Bowl on New Year’s Day.46 For such universities football games would dominate the school’s campus on many weekends each fall, exerting an outsize effect on students’ college experience. In contrast, schools like Shepherd College and Augusta College never had big-­time college football. As a result, ­those colleges had no comparable focal point for campus life, for better or worse.

30 C onte x t

High-SAT Public Colleges

A relatively small number of public colleges and universities inhab­ ere ited the top rungs of the SAT ladder in 1970.47 Among them w some of the nation’s most famous research universities, such as the University of Washington and Georgia Tech. Georgia Tech, founded by the state as a small school devoted to the study of mechanical engineering, retained its engineering focus a­ fter it emerged as a free-­ standing university.48 Many of the institutions in this high-SAT stratum are the flagship public universities of their respective states, including the University of Washington and the University of Florida, and some of ­these also ­were the state’s land-­grant university. For them, research was a central focus. Like Ohio State and the University of Kentucky, their orga­n izational charts w ­ ere built around separate schools or colleges, each with its own faculty and administration. The University of Florida, for example, had, besides its arts and sciences unit, schools of agriculture, architecture, business, dentistry, education, engineering, forestry, health professions, journalism, law, nursing, pharmacy, and physical education. The physical settings of the universities differed, of course, ranging from the urban campuses of Georgia Tech and the University of Washington to the bucolic environs of UC Davis, originally called the “University Farm.” In a telling reflection of an era marked by urban riots and campus protests, the UC Davis cata­log of 1970 / 71 notes with evident satisfaction, “Our location permits us to be somewhat insulated from the immediate threat of society’s urban prob­lems.”49 A small but distinctive subset of state-­r un institutions ­were the “public Ivies,” state-­run colleges that shared many of the characteristics of elite private institutions. One of t­hese, the College of ­William and Mary, was a public institution that reveled in the appellation of “college” as much as it trumpeted its colonial history and its most famous gradu­ate, Thomas Jefferson. Few American colleges have a backdrop better than the restored Colonial Williamsburg to remind visitors and students alike of this distinctive heritage. Its cata­log places special emphasis on “features that it has long shared

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

31

with liberal arts colleges,” including small classes and “competent, cosmopolitan faculty.”50 Lowest-SAT Private Colleges

Turning to the private nonprofit side of the industry, it is easy to see that religious affiliation is the most prominent marker of differentiation. And this generalization applies with greatest force among ­those private colleges at the low end of the SAT ranking. Within this cluster in 1970 w ­ ere religious colleges that seemed untouched by the steady secularization of twentieth-­century American society, as well as colleges whose self-­description contained no hint of religious influence. An example of the first group is Howard Payne University, in Brownwood, Texas. 51 A Baptist-­a ffiliated college, its 1970 cata­log states its purpose to be “education u ­ nder Christian influence and auspices” and “the advancement of the gospel of our Lord and Savior, Jesus Christ.”52 Another private college with an unabashed religious purpose was Lee University, a college in Cleveland, Tennessee, affiliated with the Church of God, a Pentecostal denomination. Its cata­log states, “A personal commitment to Jesus Christ as Lord and Savior—­ this controlling ­factor is the perspective from which the educational enterprise is undertaken.”53 Most numerous among private colleges with a religious genesis ­were ­those affiliated with the Roman Catholic Church. If their published cata­logs are any reflection, ­these colleges placed less emphasis on religious themes or doctrines than did the colleges with Protestant roots, certainly less than the colleges affiliated with evangelical Protestant denominations. To what extent ­those differences played out in the operation and atmosphere of ­these institutions is not easy to determine four de­cades ­later, but a con­temporary analy­sis based on a close examination of numerous college cata­logs suggests that Catholic colleges resembled public ones much more than Protestant colleges did.54 For Protestant colleges such as Howard Payne and Lee, their formal ac­cep­tance of revealed truth, made plain in their statements of pur-

32 C onte x t

pose, necessarily placed them outside the tradition of ­free inquiry and expression that is so basic to the nation’s most renowned colleges and universities. As a result they stuck to “eminently safe” majors emphasizing occupational skills and “general educational knowledge with a strong emphasis on the past and accumulated knowledge of the west from a classical point of view, rather than a scientific view.”55 In addition to the colleges with religious roots w ­ ere ­others that, from their founding, had less explicit connections to a set of religious beliefs. Their published self-­portraits bespoke both secular humanism and pragmatism. For example, the 1970 cata­log for Atlanta’s Oglethorpe University opines, “Education, as an institution of society, has a social obligation. It cannot neglect e­ither the individual or the community without damage to both.” The college’s approach to this was “to forge the strongest pos­si­ble link between the ‘academic’ and ‘practical,’ between ‘­human understanding’ and ‘know-­how’ . . .”56 Likewise, the University of New Haven, a college founded by the city’s YMCA in 1920, states in its 1970 cata­log that “its primary function” is “ser­vice to the individual student as well as to business, industry and institutions of the New Haven area.”57 A third example of a private, nondenominational, less selective college is York College of Pennsylvania. While asserting an aim that its students would “develop academically” and “grow ethically and morally,” its 1970 cata­log also expresses the hope that they would “satisfy the needs of the nation by providing ever-­increasing numbers of persons for the crafts and the professions.”58 The word “crafts” heralded a socioeconomic orientation wholly dif­fer­ent from that of the prestigious colleges, to which we now turn. Very-High-SAT Private Colleges

Judging only from the words and phrases they used to describe themselves, the private colleges that inhabit the top rungs of the academic pecking order in many re­spects seemed to be indistinguishable from t­ hose at the less-­selective end of the spectrum. Like

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

33

the colleges near the bottom, they paid tribute to good teaching and emphasized rich experiences both in and out of the classroom. T ­ hese highly ranked colleges, as a group, ­were dif­fer­ent, however, in at least three broad ways. The first and most impor­tant way they differed was in their ability and eagerness to claim a level of faculty-­student interaction that colleges in no other group could or did. Thus Bowdoin’s cata­log could claim, “The arrangement of teaching responsibilities of Bowdoin faculty presupposes professional duties not only of original scholarship and creative work but also of supervision of advanced student proj­ects.”59 Caltech could act on its belief that the best way to “train the creative type of scientist or engineer so urgently needed” was through “the contacts of its relatively small group of undergraduate students with the members of its relatively large research staff.”60 And Davidson could aim “to maintain itself as a small community of learners, to emphasize the teaching responsibility of all professors, and to ensure the opportunity for personal relationships between students and teachers.”61 A few colleges—­those with sufficient wealth or confidence in their ability to compete for students—­could afford the luxury of marching to a dif­fer­ent academic drummer. St.  John’s College in Annapolis adhered closely to a curriculum based on the G ­ reat Books, and Eckerd offered small seminars, based on an interdisciplinary core curriculum, taught by interdisciplinary teams of faculty. Sarah Lawrence placed special emphasis on one-­on-­one instruction, while MIT and Cal Tech remained unapologetically technical. And, resisting the strong trend ­toward coeducation, Wellesley College made its status as a ­women’s college a point of distinction and pride. In their unapologetic unorthodoxy t­ hese colleges w ­ ere the exceptions. In contrast, the colleges lower down the selectivity scale seemingly had too much to lose by being too dif­fer­ent. Closely allied to the proud descriptions of learning environments marked by small classes and intellectual engagement is a second generalization about the top-­ranked private colleges: the primacy of the liberal arts. Whereas most of the less selective colleges offered at

34 C onte x t

least some academic programs featuring marketable skills, such as business, education, or health-­related majors, the selective colleges ­were more likely to keep vocational and professional training to a minimum. According to sociologist Steven Brint and his colleagues, this adherence to an abstract, impractical curriculum was a way that high-­status colleges could signal “their association with intellectuality and refinement.”62 A third characteristic of ­these top-­ranked private colleges was that, what­ever their origins, they had become secular institutions. Their words, on the ­whole, ­were more nearly ­free of religious allusions than ­those used by lower-­ranked private colleges. In the small sample of cata­logs for 1970 studied, fewer colleges appeared to be explic­itly religious in fundamental orientation than was the case for the least selective private colleges. T ­ hose that referred to religion typically emphasized their openness to students regardless of belief. For example, the 1970 cata­log for Brandeis University quotes its namesake, Louis Brandeis, in an indirect rebuke to the elite colleges and their history of Jewish quotas: “Brandeis w ­ ill offer its opportunities of learning to all. Neither student body nor faculty ­will ever be chosen on the basis of population proportions, w ­ hether ethnic or re63 ligious or economic.” For two of the selective colleges whose cata­logs ­were reviewed, the de­cades ­a fter 1970 would bring a marked degree of secularization, making the religious themes in their 1970 cata­logs especially striking. In their 1970 cata­logs Davidson College and Wake Forest University both lay considerable stress on their Protestant moorings. One of ­these, Wake Forest, was in 1970 governed by the North Carolina Baptist Convention. Its 1970 cata­log notes that the college had “made marked contributions to Chris­tian­ity.” For Davidson, the statement of Christian foundation could not have been more forthright: “Davidson recognizes God as the source of all truth. As a college committed to the historic Christian faith, it sees Jesus Christ as the central fact of history, giving purpose, order, and value to the ­whole of life.”64

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

35

Historically Black Colleges and Universities

­ here is prob­ably no more distinctive a group of institutions among T ­those that offer bachelor’s degrees than the HBCUs. Few if any of  ­these institutions would ever have existed if the United States had followed a historical path untainted by slavery and Jim Crow segregation. For African Americans in the South, the role of HBCUs went beyond that of educating college students. Like many black high schools in the segregated South, t­ hese colleges became centers for local activities, focal points of community pride, and ave­nues for black leadership. Some of ­these colleges, most notably Tuskegee and the other historically black land-­ grant institutions, emphasized practical, “agricultural and mechanical” skills, consonant with the pragmatism of Booker T. Washington. ­Others, spurred by nineteenth-­ century white missionaries and the elitism associated with W. E. B. Du Bois, viewed themselves as a conduit by which the “best traditions of New ­England” could be made available to blacks in the South. By enforcing strict rules of dress, speech, and be­hav­ior, ­these HBCUs epitomized the values of the black ­middle class.65 And both va­r i­e­ties of HBCUs also came to represent au­then­tic centers of black control, largely ­free of white interference, something that could not be said for the public schools.66 The HBCUs also produced many of the leaders of the civil rights movement, most prominently More­house gradu­ate Martin Luther King Jr. But their share of the country’s black college students declined over time in part ­because of black migration out of the South. This decline began in the de­cade of the 1960s, which witnessed a sea change in the role of t­ hese institutions, as historically white colleges and universities opened their doors to black students. Once the only refuge for black college students in the South, by 1970 the HBCUs w ­ ere having to compete with the likes of Vanderbilt and the University of Georgia for students. Whereas HBCUs in 1954 had enrolled more than four-­fi fths of all black college students in the nation, by 1967 they had just six out of ten. 67

36 C onte x t

In light of HBCUs’ distinctive, racialized histories, and the very large role they played for African Americans in the South, it is ironic that the 1970 sample cata­logs available online hardly acknowledge that role. Perhaps it was a fact so deeply internalized that it did not need to be expressed explic­itly. The educational aims set forth in ­these cata­logs contains no explicit mention of race, only incidental references. The only indication of their historic dedication to the education of African Americans are recitations of t­ hese colleges’ prior names. For example, among the previous names for North Carolina’s Fayetteville State University duly noted in its cata­log are ­these: State Colored Normal School (1877–1916), State Colored Normal and Industrial School (1916–1921), and State Normal School for the Negro Race (1921–1926). Other­wise, the university describes itself in much the same terms Illinois State uses in its cata­log: “Fayetteville State University has become a multi-­purpose institution with Teacher Education its current major quality program.” ­There w ­ ere private HBCUs as well, of course, most of them with strong religious affiliations. Florida Memorial College, a Baptist-­ affiliated college that had moved from St.  Augustine to Miami, notes in its 1970/71 cata­log, “its role as a community of Christian scholars.”68 Oakwood University, in Huntsville, Alabama, founded by Seventh-­day Adventists, used its cata­log to spell out the six aspects of its educational objectives, none of which mentioned race: spiritual: “reflect fully the image of Jesus Christ”; intellectual: “develop . . . ​ attitudes conducive to in­de­pen­dent and creative thinking”; cultural: “develop . . . ​ desirable personalities, refined tastes, and correct usage of the social graces”; personal adjustment: “help the student understand himself”; physical: “engender maximum physical vitality”; and vocational: “train them in practical work.”69 Historical Antecedents

How had the configuration of colleges of 1970 come to be the way it was? Viewing the American higher education landscape in that year, an observer would have beheld a multitude of colleges and

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

37

universities spread across the land, the oldest ones concentrated in the Northeast and the largest ones dotting the most populous states outside that region. It is not hard to explain the geo­graph­i­cal concentration of the oldest colleges. T ­ hese ­were founded where most of the ­people lived in ­those early years. As the country’s population spilled out into the territories and then states of the Midwest, South, and West, ­these early colleges found themselves increasingly distant from the country’s high school gradu­ates. Other aspects are not so easy to explain. Historians of higher education note several re­spects in which the American higher education landscape looked dif­fer­ent from the centralized, state-­run universities that arose in Eu­rope. The United States stood out for its large number of in­de­pen­dent colleges, their autonomy one from another, the large share of them that ­were not creatures of government, and, among t­ hose that w ­ ere run by government, institutions operated by state governments rather than by the nation’s central government in Washington.70 To understand, it is necessary to look back at the development of higher education in the country. The early roots of American higher education ­were private, religious, and privileged. The first colleges w ­ ere established to train 71 ministers and educate gentlemen. In addition to religious instruction, their curricula typically emphasized classical languages, his­ ere tory, and philosophy.72 By most accounts, antebellum colleges w academically weak. In spite of curricula rich in the classics, the instructors ­were often poorly educated, the lectures boring, and the students disinterested. Outside of the classroom, ­those students’ time was marked by high jinx, card-­playing, cursing, drinking, fighting, protesting college rules, and general sophomoric be­hav­ior.73 Resisting t­ hese tendencies ­were the religious denominations, each one resolute in its determination to establish colleges that would “train its own ministers, hold the loyalty of its own young ­people, and convert outsiders.”74 ­These aims, abetted by the proliferation of Protestant denominations, meant that the country, particularly the states east of the Mississippi, would be dotted with scores of small, religiously affiliated colleges.75 Often relying as much on the support

38 C onte x t

of the towns that welcomed them as that of the denominations to which they ­were tied, their localness at once assured them of a reliable source of students and benefactors at the same time that it placed a ceiling on the size which the college might aspire to achieve. By 1890 Ohio had a total of thirty-­seven colleges. Only two ­were public—­Ohio State and the University of Cincinnati—­and only two of the rest had no religious affiliation. Their average enrollment was just over one hundred. Most ­were, in the words of historian Roger Geiger, “small and precarious operations.”76 One consequence of the private, nonprofit status of so many American colleges was the unrelenting need to attract paying customers in order to stay in business. Financial support from a religious denomination was good but seldom sufficient to sustain a college that enrolled just a few students. According to one historian of higher education, “More than merely client oriented, American higher education was consumer oriented, as demonstrated by the exceptional number and diversity of its ser­vices. American universities engaged in an aggressive competition for enrollments.”77 Thus many private colleges turned away from classical subjects, to offer new courses and degree programs on practical skills. Emulating the expanding public institutions, many became “multipurpose,” offering training not only to aspiring teachers but also in such practical fields as nursing and business. Another device colleges found for drawing attention to themselves was to sponsor teams playing the developing American sensation, football. The instinct for survival, and the closely allied imperative to attract students, provide as good an explanation as any for why American higher education, uniquely in the world, incorporated big-­time spectator sports as a central activity.78 Although the earliest colleges w ­ ere mostly private, state governments increasingly entered the college market a­ fter the Civil War, establishing public colleges and universities with more varied purposes and diverse clienteles. ­T hese state universities soon surpassed the private sector in total enrollment. One reason for that was the Morrill Act of 1862, an injection of federal largess into the

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

39

states. This astonishingly far-­sighted law, enacted during the n ­ ation’s most vexing crisis, provided the means for each state to establish a university that would concentrate on the practical skills of agriculture and industry. A second Morrill Act was passed in 1890, providing in part for separate institutions for blacks in the South.79 ­These state universities became vivid examples of the advantages of large scale in higher education. Size allowed departments to proliferate and faculty to specialize, as demanded by advances in knowledge. The lecture format freed faculty to do more research.80 More generally, the de­cades a­ fter the Civil War w ­ ere ones of rapid change in universities, ushering in features now taken for granted, including numbered courses, course credits, and academic departments.81 By 1900 some 725 colleges had been established, but most remained small by t­oday’s standards. Eighty-­five ­percent of them ­were private. The typical private institution (the one attended by the median student) enrolled about five hundred students, and the typical public one had almost eight hundred. Inspired both by the desire to see the benefits of education spread widely across the population and an appreciation of the value of imparting practical knowledge, the state universities, especially ­those in the newer states of the Midwest and West, grew in size. Spurred by rising affluence, increasing demand for college degrees, and federal support, this growth in state-­supported universities continued into the twentieth ­century. By 1924, nine out of ten students in the public sector w ­ ere attending institutions with enrollments of more than a thousand students.82 The share of students attending public institutions increased sharply ­after 1900, rising from one quarter to nearly three quarters by 1980.83 Federal support took other forms in the twentieth ­century, including military-­ related research during World War II; the subsequent G.I. Bill (1944), which provided generous financial support for veterans to attend college; the National Defense Education Act (1957), which supported gradu­ate students intending to become college and university professors; numerous other programs to give financial aid to students; and the direct funding of nondefense spending through agencies

40 C onte x t

such as the National Science Foundation (1950) and the National Institutes of Health. 84 Indirect government aid was also instrumental in the growth of universities, especially for private nonprofit universities. The federal income tax, enacted in 1913, exempted all nonprofit organ­izations from income taxation and, a­ fter 1917, most donations to universities ­were tax-­deductible. The same would be true for the corporate income tax and the estate tax. Private foundations, a noteworthy beneficiary of the tax laws, also provided support to universities. At the local level, universities both public and private ­were exempted from paying most property taxes. Unlike the specialized research institutions in Eu­rope, American state universities served several major objectives: broad-­based undergraduate education, pragmatically oriented professional training, basic research in arts and sciences, and applied research and outreach to industry and farming.85 One reflection of the rising importance of higher education in the United States was a steep rise in postsecondary enrollments. In 1870 college enrollments represented just 1.1 ­percent of the eighteen-­to twenty-­four-­year-­old population. That ratio had doubled by 1900. It doubled again by 1920, and once again by 1940, to 9.1  ­percent. By 1970 college enrollments represented one-­third of this age group, when some 8 million Americans w ­ ere enrolled in college. 86 A direct consequence of this growth was a steady increase in the educational attainment of the population. In 1920 only one in twenty adults aged twenty-­five to twenty-­nine had finished four years of college. By 1970, one in six had achieved this milestone. 87 One last feature that colleges in 1970 had inherited from the past was the complex, troubled, and debated history of preference and discrimination embedded in colleges’ admissions practices. It is difficult to choose between race and religion as the dimension for the first ­great ­battles about equity in college admissions. This is ­because patterns of discrimination by race that had been established long before 1900 ­were, consequently, not part of any active public debate. Thus, it was religion, with an accompanying admixture of national origin, that provided the first twentieth-­century theater of ­battle

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

41

over equity in elite college admission. As documented in Jerome Karabel’s comprehensive history of admissions at Harvard, Yale, and Prince­ton, the percentage of applicants who ­were Jewish increased in the first de­cades of the twentieth c­ entury largely as a function of the massive wave of immigration from Eastern Eu­rope. At Harvard the percentage of enrolled students who ­were Jewish increased from 7 ­percent just before the turn of the ­century to 21 ­percent by 1921.88 Alarmed by this influx of “boys who have come, or whose parents have come, to this country without our background . . . ​[or] the ideas and traditions of our ­people,” Harvard president Lawrence Lowell in 1922 endorsed an informal 15 ­percent quota on Jews enrolled in the college. 89 Yale and Prince­ton also limited the number of Jewish students. Karabel argues that the common mechanism that enabled ­these universities to accomplish this was to adopt a new criterion, in the form of “character” or well-­roundedness, that would give admissions committees leeway in making admissions decisions. By 1970, however, the issue of anti-­Semitism in elite college admission had receded from view, but the use of intangible personal qualities in admissions decisions had not. For the most part, the policies of the elite colleges ­were not in play in the NAACP’s early challenges to racial segregation that predated Brown v. Board of Education or the fights over segregated public university systems.90 Instead, the race issue for the elite colleges centered on affirmative action and racial preferences. An exception to this generalization was a stunning decision in New York that effectively erased the elite status of the City College of New York (CCNY). Notable for its high standards, City College had served as a low-­cost elite college for many first-­and second-­generation New Yorkers. ­Under pressure from black and Puerto Rican groups, New York City’s Board of Education, which operated CCNY, deci­ded in 1969 to drop SATs as a criterion for admission and stop requiring three years of math and foreign language plus four of En­glish. Instead, CCNY would base admission decisions only on students’ high school rank, thus transforming the college’s admissions policy overnight from very selective to virtually open enrollment. The aim was to better reflect the

42 C onte x t

population of the city, and it worked. In the first year ­a fter the new policy’s adoption, the share of students who ­were black increased from 19 to 24 ­percent, and the share of Puerto Ricans also increased. According to one estimate, less than 40  ­percent of the students admitted in the new policy’s first year (1970) would have been admitted ­under the previous admissions standards.91 For the most part, the issue of race in elite college admissions was addressed beginning in the 1960s, as selective colleges across the board increased their enrollments of black students and Jewish students, gradually losing much of their similarity to the exclusive New ­England private schools from which they had drawn so many of their students in past de­cades. The equity issue that would not go away, however, was preferences that colleges gave to the sons and ­d aughters of their alumni. In 2002 the admit rates for legacies at the big three Ivy League colleges ­were roughly ten times the corresponding rates for nonlegacies. Shortly ­a fter 2000 the rates of admission for nonlegacy applicants ­were 4.1 ­percent at Harvard, 3.4 ­percent at Prince­ton, and 2.4 ­percent at Yale; the corresponding admit rates for legacies at t­ hese colleges w ­ ere 39 ­percent, 35 ­percent, and 29 ­percent, respectively.92 In 2014, 20 ­percent of Harvard College’s entering class ­were sons or d ­ aughters 93 of alumni. To be sure, ­t hese ­simple comparisons overstate the ­actual weight given to legacy status, since many of ­these applicants would have been admitted without the extra push of being a legacy. Still, one aspect that made this an impor­tant equity issue was the fact, as one observer put it, at most colleges legacy status “is a near-­ perfect proxy for being white.”94 What Kind of “Market” Is This?

­ hose who are not steeped in the discipline of economics may take T exception to the application of terms like “market,” “supply,” or “demand” in connection with higher education. Many of t­hose who spend their ­careers at a college or university often view their work as a high calling and their places of employment more akin to

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

43

c­ ommunities of comrades than to government agencies or private businesses. Yet to the economist the institutional structures and interactions associated with four-­year colleges fit the standard definition of a market ­because ­there are suppliers who are selling ser­vices to buyers, for a price. The customers are the students and their parents. What is being produced and bought, undergraduate education, is a ser­vice, not a physical commodity (excepting that symbolically significant piece of paper called a diploma). In t­hese re­spects, the market for undergraduate education is ordinary, not unlike the markets for financial advice, health care, or auto repair. As ser­vice industries go, the one for college education is large. Expenditures on it amount to 2.7 ­percent of GDP, counting all levels of higher education, a higher share than the 1.6 ­percent average among the countries of the OECD, but quite a bit lower than the share of GDP the United States devotes to health care.95 But the college market is not without its share of oddities, nine of which are pertinent to the pres­ent study. The first, not unique to colleges, is asymmetric information. Despite the best efforts of numerous college guides, from Barron’s and the Fiske Guide to Colleges to U.S. News and World Report, most customers are quite ignorant about the array of colleges available to them. As any parent of a recent college applicant ­w ill attest, the knowledge that high school se­n iors and their parents possess about the colleges they consider or visit is greatly outweighed by what they do not know. One reason for this imbalance is the amorphous nature of this beast. Perhaps college can best be described as one of t­ hose “experience goods,” one that cannot be appreciated ­until one has become a user of it. Two other peculiarities contribute to this ignorance. The more significant of t­hese, our second feature, is that the production of education requires effort on the part of the consumer, the student. As the example of Sonia Sotomayor illustrates, for students to get very much at all out of college, they must pay attention, read, study, and undertake the intellectual challenges posed by teachers in order to gain from the experience. Other kinds of businesses require consumers to participate actively—­grocery shoppers to fill their own

44 C onte x t

carts or patients to take the medicine prescribed for them, for example—­but few demand so much attention and effort, and on such a sustained basis. The closest analogy is perhaps the gym, which can produce ­little without the customer’s exertion. ­Because students are impor­tant to the success of the endeavor, it is impossible to judge the quality of a college without paying close attention to the attributes and habits of its students. The third distinctive feature of the college market, also an aspect of the educational production function, is the impor­tant role played by other customers. ­These are “peer effects.” Whereas our enjoyment of a restaurant meal may be influenced by the presence and be­hav­ior of other diners, the influence of peers on a student’s learning in schools at all levels can be profound.96 Turning from the consumer to the colleges themselves—­the ­i ndustry—we can observe other aspects that make this market unusual. The fourth distinguishing feature is the degree to which colleges (the firms) are rooted to a par­tic­u ­lar geo­graph­i­cal location. Colleges hardly ever move. Although they may be in the ­human capital business, they are remarkable for the importance of their physical capital. Their buildings are commonly quite central to their identity. Some of their buildings have achieved landmark status. Many are considered iconic. More than a few college campuses rival all but the grandest city parks in the country for the majesty of their monuments and the beauty of their landscaping. One consequence of the nearly absolute immovability of colleges and their stately buildings has been a growing geo­graph­i­cal mismatch over time between colleges and their customers. As a casual glance ­will reveal, among the country’s oldest colleges are an unusually large share of its most prestigious colleges. While the nation’s population was spreading out to the west and south, ­those colleges stayed put, gradually ­exacerbating the mismatch. By 1970 the Northeast was home to 63 ­percent of the spots at the nation’s most selective colleges, but the region produced just 22 ­percent of the nation’s public high school gradu­ates.97 This mismatch meant that, if t­ hose elite colleges wished to remain the destination for top students, the colleges would have

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

45

to attract applicants from more and more distant locations. It also meant that high school se­niors from the Northeast would increasingly face competition from other regions for spots at ­these elite colleges. The fifth notable feature of the college market is the coexistence of markedly dissimilar rival institutions. ­Today, a high school se­ nior hoping to obtain a bachelor’s degree might look to a small liberal arts college where classes are seldom larger than thirty students and faculty do ­little research or a massive public research university enrolling 30,000 or more where some classes are so large that they have to be held in auditoriums. Despite ­these vast differences in circumstances, both institutions offer exactly the same bachelor’s degrees, ostensibly close substitutes for one another. Pricing accounts for two noteworthy aspects of the college market. One is price differentiation, or, more precisely, price discrimination. Like the modern airline industry, it is quite common for customers to pay dif­fer­ent prices, even at the same institution. One reason is the widespread existence of need-­based financial aid, bolstered by federal programs such as Pell Grants and subsidized loans. Another reason is the widespread practice of discounting, whereby colleges offer to reduce the cost of attendance for some applicants for reasons other than financial need. Often dressed up in the form of merit scholarships, t­ hese are no more than a price break.98 The other oddity regarding pricing—­a nd this is a big one—is that most colleges routinely engage in another practice that would be foreign to most businesses. In most competitive markets, prices end up making supply come into alignment with demand. But in the college market it is quite common for price not to play this role. For very selective colleges, the price is usually much too low to clear the market. The result is excess demand, rationed by an admissions office. Symbolized by the rejection letters sent out by admissions offices, this form of nonprice rationing amounts to closing the door on willing customers. To varying degrees, colleges limit who can consume their ser­v ice by reviewing applicants’ academic rec­ords and admitting only ­those applicants they wish to.99

46 C onte x t

Another distinctive feature of the college market is the institution of tenure. By tradition, and backed up by the power­ful American Association of University Professors, most colleges and universities offering bachelor’s degrees make it pos­si­ble for many if not most of its teaching faculty to obtain, ­after a probationary period and conditional on approval, lifetime tenure. This unusual feature, derided by some outside of academia, shows no sign of ­going away, even ­after the disappearance of mandatory retirement in the 1990s. To reduce the burden of a workforce that is virtually immune from firing, colleges have increased the share of teaching duty performed by instructors not eligible for tenure. One last peculiarity of the college market is so-­called positional competition. In several impor­tant ways, other than the obvious case of intercollegiate athletics, the competition that occurs between colleges is of the winner-­take-­a ll variety. Objective assessment of per­for­mance or quality often ­matters less than which institution prevails—­which one successfully recruits the famous scholar or the promising student or lands the big grant. This feature assumed special significance when colleges began to compete in the 1980s for the same set of students. Data on Students at 188 Colleges

A trea­sure trove of data that can be used to mea­sure many of the differences across colleges and track changes over time is the Freshman Survey, an annual survey that has been administered by the Higher Education Research Institute at UCLA since the late 1960s. This survey is uniquely valuable for the proj­ect at hand ­because it has been in operation for a long time, many of its questions have remained unchanged over that period, and quite a few colleges have long rec­ords of participation. Taking advantage of ­these facts, I put together a sample of 188 colleges that participated in the survey at the beginning, m ­ iddle and end of the nearly four-­ decade period, 1972–2009, then divided ­these 188 colleges into categories of similar colleges, the colleges in each category remaining

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

47

unchanged over the entire period of coverage. T ­ hese categories ­were defined using the three characteristics discussed earlier in this chapter: the average SAT of their students around 1970, w ­ hether they w ­ ere public or private institutions, and w ­ hether they are historically black institutions (HBCUs).100 The first of ­these criteria, average SAT, has pre­ce­dent in previous research on higher education in the United States. The other criteria are distinctions that have far-­reaching historical and practical importance in American higher education. The resulting sample comprises a “panel” of groups of similar colleges, meaning that each group of colleges remains the same over the entire period. For example, the public 80th to 90th percentile category was made up of this fixed group of five institutions: Iowa State, SUNY College at Potsdam, UC Santa Barbara, the University of Mas­ sa­ chu­ setts at Amherst, and the University of North Carolina at Chapel Hill. ­Table 2.1 provides some descriptive information for the seventeen categories of colleges into which the sample of 188 colleges is divided. In order to gather enough colleges in the sample to ensure groups of at least five colleges each (a requirement imposed to protect the identities of individual colleges), two pairs of years separated by roughly two decades—1989 / 90 and 2008 / 09—­were selected, in addition to the year 1972, and colleges that participated in the survey in at least one of the two years in each pair ­were identified. Exactly how many categories should be used in dividing up ­these 188 colleges was ultimately a judgment call. More finely differentiated categories promised sharper contrasts, but the required minimum of five colleges per group limited how small the categories could be. In light of this restriction, the 188 colleges w ­ ere divided into seventeen categories, defined according to the three characteristics already noted—­SAT, public or private governance, and HBCU status. The resulting categories of colleges contain between five and twenty-­four colleges each. One feature of t­ hese categories that is immediately evident is that ­there are more categories, with finer gradations in test scores, for the private colleges. The reason for that is simply ­because it was pos­si­ble to make more, ­because a large number

48 C onte x t

­Table 2.1  The Freshman Survey Sample: 17 Categories and 188 Colleges Type and percentile of 1970 SAT

Example

Number of colleges in sample

Number of surveyed students, by survey year 1972

1989 or 1990

2008 or 2009

3,550

3,936

Private colleges and universities 0–25 pct

Iowa Wesleyan College

14

3,167

25–50 pct

Gonzaga University

24

6,052

5,567

9,171

50–60 pct

Ohio Northern University

10

7,065

6,823

8,601

60–70 pct

Texas Christian University

12

7,055

6,053

7,063

70–80 pct

University of the Pacific

19

8,327

9,239

10,517

80–85 pct

Lewis & Clark College

13

5,324

3,970

5,815

85–90 pct

Tulane University

8

3,392

3,733

4,470

90–95 pct

Colorado College

20

8,007

8,782

10,028

95–98 pct

Northwestern University

12

8,121

7,082

8,958

98–99 pct

Wesleyan University

9

4,441

4,889

4,052

99+ pct

Amherst College

9

3,534

3,596

2,544

Public colleges and universities 0–50 pct

University of North Dakota

7

6,752

8,155

9,441

50–80 pct

University of Michigan-­Flint

7

7,689

9,623

10,241

80–90 pct

University of North Carolina, Chapel Hill

5

11,951

12,082

12,052

90+ pct

Stony Brook University

6

6,434

6,820

10,601

Historically black colleges and universities (HBCU) Private

Howard University

8

2,307

3,230

3,077

Public

Winston-­Salem State University

5

1,833

1,947

2,342

188

101,451

105,141

122,909

Total

Source: HERI Freshman Survey, selected institutions. For each institution, data collected for 1972, ­either 1989 or 1990, and ­either 2008 or 2009.

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

49

of private colleges near the top of the score distribution regularly participated in the Freshman Survey. The resulting luxury of finer gradations among private colleges was fortuitous b ­ ecause the colleges in the highest SAT categories are distinctive in impor­tant re­spects, and having small categories composed entirely of ­these colleges makes it pos­si­ble to see ­these distinctions. Thus the 99th and above percentile group of colleges enrolled approximately 1 ­percent of all college students in 1970, ­those attending the colleges with the highest average SAT scores.101 Public institutions ­were less likely to be regulars in the Freshman Survey, so their SAT groupings had to be broader, in order to conform to the minimum size of five. One result is considerable heterogeneity within the categories of public institutions, as evident from the differences among the comparison groups. One anomaly of this sample is the admittedly unrepresentative nature of the highest ­SAT public institutions, ­those at the 90th percentile and above. Among the half dozen colleges in this group are three of the country’s ser­v ice academies—­West Point, the Air Force Acad­emy, and the Coast Guard Acad­emy. ­These academies are grouped with three of the academically strongest state universities—­Stony Brook, Georgia Tech, and the University of California at Santa Cruz.102 All told, the 188 colleges produced more than 100,000 surveys from entering students in each of the three waves of surveys. The first wave consisted of surveys completed by students entering one of ­these 188 colleges in the fall of 1972. The second wave combined students entering in 1989 or 1990, denoted hereafter as 1989 / 90, and the third wave covered students entering in 2008 or 2009, denoted 2008 / 09. In an explanation he inserted at the front of Huckleberry Finn, Mark Twain advised readers that in writing the book he had taken the trou­ble of crafting distinct accents for certain characters, so that readers should not “suppose that all t­ hese characters ­were trying to talk alike and not succeeding.” Likewise, one reason colleges differ from each other is not b ­ ecause of variations in their ability to achieve a common objective, but b ­ ecause they have genuinely dif­

50 C onte x t

fer­ent aims. A starting point for reflecting t­hese differences is to focus on how colleges describe their institutional aims. Taken together, the expressed aims yield a sort of a constrained pluralism, one that allowed for a marked variety of ultimate goals within the confines of producing a rather standardized service—­a baccalaureate education. ­These colleges are, ­a fter all, firms in a market, and they are in princi­ple competing for the business of customers. To be sure, the market is highly segmented—by region, religion, prestige, and academic selectivity, among other ­t hings—­wherein most students applying to college seriously consider no more than a handful or two. In an effort to reflect some of the variation across colleges, the empirical analy­sis used ­here makes distinctions along three dimensions: public vs. private (not-­for-­profit); status as a historically black college or university (HBCU), and the average standardized test scores achieved by their entering students in 1970. Using t­ hese three criteria, the 1970 colleges are divided into seventeen mutually exclusive categories, categories for which consistent, detailed information about students was available from the Freshman Survey. ­Every comparison between groups of colleges in the book utilizes ­these same seventeen categories or combinations of them.

S y s t e m , I n d u s t r y , o r C r a z y Qui l t ?

51

3 Snapshot, Circa 1970

In the first four days of May 1970, dozens of colleges across the nation witnessed or­ga­nized protests following President Nixon’s announcement on April 30 that he had ordered U.S. soldiers to cross the border from Vietnam into Cambodia. One of ­those colleges was Kent State University, a comprehensive public university located thirty miles south of Cleveland, Ohio. Protests t­ here had turned ugly. One day a­ fter a late-­n ight disturbance that left dozens of broken win­dows in stores downtown, the university’s wood-­frame ROTC building had been set ablaze. Governor James Rhodes ordered some 750 soldiers from the Ohio National Guard, already called up ­because of a threatened Teamster’s strike, to the scene.1 Shortly ­after noon on May 4, a crowd of demonstrators gathered in a grassy campus area called the Commons. ­There they w ­ ere confronted by troops armed with tear gas equipment and loaded M-1 r­ ifles. Ordering the demonstrators to disperse, the soldiers advanced and then began

lobbing tear gas canisters t­oward the crowd. Demonstrators, in turn, threw rocks and canisters back, chanting “Pigs off campus.” Suddenly, soldiers opened fire on the crowd, killing four students and wounding nine.2 This tragedy, still shocking more than forty years l­ ater, serves as both a reminder of the turbulent period that preceded it and a beginning point for a new period in American higher education, one brimming with unfamiliar challenges. To set the stage for a consideration of changes over four de­cades in the college market, this chapter offers a snapshot of the college scene as it was at the beginning of that period, around the year 1970. Most visibly, college leaders feared that peaceful operations might never be restored. They also worried out loud about ominous trends in their institutions’ finances. Beneath the headlines lay issues of more lasting significance, including large gaps among colleges in students’ academic ability and socioeconomic status and deep ­disparities in resources. Another demarcation that remained impor­ tant but largely unmentioned was athletics and its two distinct manifestations. Embattled Colleges

As professors, deans, and presidents across the country readied themselves for the start of the 1970 fall term, the August 31, 1970, issue of the Chronicle of Higher Education arriving in college mailboxes across the country carried four front-­page stories, nicely encapsulating the two worries that weighed on colleges and their leaders. Two of the headlines embodied worries that flowed directly from Kent State and the years of upheaval that preceded it: “Presidents Prepare for Disorders, Seek Ways to Keep Campuses Open” and “Some Students ­Will Take Up Guns, Panel on Campus Unrest Is Told.”3 Guns and unrest—­indeed this was a singular moment in the history of American higher education. The other two front-­page headlines in this issue of the Chronicle revealed anx­i­eties of another sort: “Financial Crisis Worsens for Colleges; Some Close, Many Show Deficits” and “Small Colleges Face Shortage of Students.”4 As

S n a p s h o t , C i r c a 1 9 7 0

53

illustrated by t­ hese front-­page articles, fear and gloom ­were the order of the day. The anxiety over campus unrest seemed fully justified, in light of the preceding de­cade of po­liti­cal turmoil. Nor ­were the financial worries simply a figment of overactive imaginations. Protest at the National Level

The climate that existed on college campuses in the fall of 1970 can be understood only by appreciating the profound challenges that had shaken the country’s existing social and po­liti­cal order. Beginning in the late 1950s, when a young Baptist minister in Montgomery, Alabama, led a boycott of racially segregated city buses, numerous groups ­rose to challenge both the accepted social order and national policies. Coordinated protests, backed by rulings from federal judges, succeeded in ­doing away with much of the racial segregation in schools and public accommodations that had been an accepted fact of life in the South and other parts of the country. Congress passed historic civil rights legislation—­the 1964 Civil Rights Act and the 1965 Voting Rights Act. Hundreds of colleges whose student bodies had been entirely white, or nearly so, in 1960 had enough nonwhite students by 1970 to be called integrated. Emerging to operate on a parallel track was a nationwide protest against the country’s heavy involvement in an ongoing war in Vietnam. In 1968 ­these movements encountered a year of tragedy and reversal, with the assassinations of Martin Luther King  Jr. and Robert Kennedy and a ­fracas-­scarred Demo­cratic Convention in Chicago. The forces that drove t­hese challenges to “the establishment” did not stop at the campus gates. Campus Unrest

With a few exceptions, such as the sit-in movement launched in Greensboro in 1960 by a handful of students from North Carolina A&T, the nation’s protest movements did not hit college campuses ­until the self-­styled F ­ ree Speech Movement rocked the University

54 C onte x t

of California’s Berkeley campus in 1964–1965. Soon thereafter campus protests revved up during the late 1960s at campuses mainly in the Northeast and Midwest. In 1966 and 1967, students protested at the University of Wisconsin over defense-­related research, with authorities resorting to tear gas in April  1967. Also in 1967, a car carry­ing Secretary of Defense Robert McNamara was blocked at Harvard by hundreds of students protesting the war in Vietnam. 5 In 1968, three weeks ­a fter the King assassination, more than seven hundred students at Columbia took over a major administration building, demanding an end to defense-­related research and an end to a construction proj­ect in nearby Morningside Heights.6 In 1969, students at Duke in February and Harvard in April occupied the respective central administration buildings, the latter receiving widespread attention.7 Buildings ­were also taken over by students at other campuses in the Northeast, including the University of Pennsylvania, City College of New York, Prince­ton, Rutgers, Swarthmore, and Wesleyan.8 At Cornell, black students occupied Willard Straight Hall, emerging ­a fter thirty-­six hours with leaders toting ­rifles, one bedecked with an ammunition ­belt over his shoulder.9 If ­these scenes ­were not chilling enough, then came the Kent State shootings in May, followed by another firebombing at the University of ­Wisconsin of a building housing research for the U.S. Army in ­August 1970, this bomb killing a researcher working t­ here.10 The Aftermath

The first effects of the Kent State killings ­were immediate and widespread. Spurred by demands by students and urged by faculty, colleges across the country suspended regular operations, some ending classes altogether and excusing students from taking final exams. At Ohio State, for example, ­a fter protesters pushed into the administration building and the president’s home, then w ­ ere driven back by national guardsmen, the university’s president ended the term precipitously.11 Within a week ­a fter the Kent State shootings, some 450 college campuses had ­either been subjected to some kind

S n a p s h o t , C i r c a 1 9 7 0

55

of strike or had been closed down entirely.12 The breadth and alacrity of this shutdown was stunning. T ­ hese events revealed, in part, the extent to which years of protest had altered structures of power on the country’s college campuses. Thus the summer of 1970 became, for campus leaders, a time of sober assessment, spurring that Chronicle front-­page story, “Presidents Prepare for Disorders, Seek Ways to Keep Campuses Open.” In meetings over the summer, university presidents resolved not to repeat the widespread shutdown of May, many making it clear they would use police if necessary to keep their campuses operating. The president of the University of Arizona spoke for the majority at one of ­these meetings of presidents of land-­grant universities, in f­ avor of “a clear-­cut statement that the institution ­will stay open.”13 Such statements appeared to be a direct response to legislatures unhappy with the widespread shutdown of campuses in May. It was no easy time to be a college president. Yale’s Kingman Brewster observed, “You have the mob within whose impatience takes the form of wanting to make the university serve their po­liti­cal purposes. And you have the mob without who want to withdraw support ­unless the university is conforming to their po­liti­cal preferences.”14 A report commissioned by President Nixon, released in late September, strongly affirmed the law-­and-­order theme. It declared, “­There can be no more ‘trashing,’ no more rock-­throwing, no more arson, no more bombing by protestors.”15 The longer term effects of the de­cade punctuated by Kent State ­were harder to pinpoint. Some observers cited grade inflation as one consequence, a symptom of the decline in the authority of faculty, or a reluctance of faculty to make judgments. Between 1960 and 1969 the average GPA at colleges across the country increased from 2.4 to 2.56.16 In the de­cade 1963 to 1973 the percentage of “A” grades given among a large number of colleges doubled. This apparent temerity was especially acute when low grades could force male students to flunk out, leaving them vulnerable to the military draft.17 A more impor­tant outcome of the period, in the view of some ­observers, was the demise of the princi­ple of in loco parentis. In concert with

56 C onte x t

the general dissolution of their authoritarian apparatus, colleges relaxed or did away with rules governing undergraduate housing and social life.18 One consequence of this weakening of in loco parentis might have been to allow faculty to respond more readily than they other­wise would have to growing pressures at many colleges and universities to do more publishable research. If this turning away took the form of easier courses, it could explain subsequent declines in students’ study time. A “Frightening” Economic Outlook

Besides their fears that student protests might keep spinning out of control, the other anxiety shared by ­those who led colleges and universities in 1970 was financial. Deficits—­actual or anticipated—­were the specter that haunted t­hese leaders. The period from the mid1950s to mid-1960s had been something of a golden age for American higher education, as generous support from state and federal government helped build research universities and expand the capacity of other colleges, private and public.19 By 1970, however, revenue growth was slowing and expenditures w ­ ere accelerating. T ­ hese threats ­were spelled out in that “Financial Crisis Worsens” front-­page story in the Chronicle on the eve of the 1970 school year. It began, “The financial crisis that ­people in higher education have been talking about for years may fi­nally have arrived.” Several small private colleges had closed down, and an “alarming number” of colleges and universities w ­ ere reporting bud­get deficits. One university president called the situation a “deep financial crisis.” The president of a private liberal arts college waxed apocalyptic, declaring, “The Day of Judgment is upon us.” Columbia University had run deficits for five years, and Prince­ton’s deficit had doubled from the year before. In his annual report to the American Association of University Professors (AAUP) in the summer of 1970, economist Peter Steiner of Michigan cited the mounting costs and lagging revenues facing private institutions as well as a disastrous job market for gradu­ate students, pronouncing the outlook “a frightening picture for the

S n a p s h o t , C i r c a 1 9 7 0

57

f­ uture.” He wrote, “That ­there is a ‘Crisis in Higher Education’ can hardly come as a surprise to anyone who reads this report.”20 A report for the AAUP by William Baumol and Maryse Eymonerie stated: In the next de­cade higher education in general and the public institutions in par­tic­u­lar face a set of serious economic threats. . . . ​The private institutions find their costs mounting more quickly than their sources of support are expanding. The costs of private universities have been trebling ­every de­cade in the postwar period and any extrapolation of this trend produces a frightening picture for the ­future. The public institutions, which have borne the brunt of the increased demand for higher education, face legislatures which may show increasing re­sis­tance to their financial requests and a federal government which at pres­ent seems l­ ittle inclined to take up the slack.21

Another report put together by the Association of American Colleges summarizes the findings from a survey of more than five hundred private colleges and universities. It painted a grim picture. Buffeted by rising costs, decreases in federal aid, and the prospect of sustained declines in applications, many institutions had been ­running deficits since the 1968 / 69 school year. The average college surveyed, which had run a surplus in 1967 / 68, ran a $20,000 deficit in 1968 / 69, which grew to $103,000 and $115,000 in the next two years. For institutions with enrollments over 4,000, the average deficit was five times this amount.22 The most vulnerable appeared to be the small private colleges. A report three years before had warned that many of them ­were facing “financial crisis.” To cover rising costs, ­these colleges ­were raising tuition charges at a rapid pace, all the more so to offset dwindling endowment income.23 Adding to the woes of the private colleges, public universities w ­ ere competing more and more to enroll students. The Chronicle’s front-­page article gave examples of small colleges struggling to enroll enough students to balance their bud­gets. One of

58 C onte x t

them accepted 90 ­percent of its applicants. Another undertook unpre­ ce­dented recruiting efforts. Despite ­t hese efforts, many of them opened their school years with dozens of empty dorm rooms, and deficits. College leaders applied pressure on their admissions officers, first to generate more applications and, second to turn ­those applications into “warm bodies.”24 One report found that, of 554 private institutions surveyed in 1970, 47 ­percent ­were operating at a deficit.25 A report issued in 1971 by the Association of American Colleges stated that more than a hundred private colleges and universities had exhausted their liquid assets “and are hovering on the brink of financial disaster.” The author of the report gave this interpretation: “­Behind ­these mounds of grotesque deficits lie the broken remains of curtailed operations, of abbreviated departments, of decimated academic programs, of faltering plans and languishing aspirations, of innovation untried and of creativity curbed.”26 Still another study, ominously entitled The New Depression in Higher Education, presented detailed case studies of forty-­one institutions. Of ­these, twenty-­nine, or more than two-­ thirds, ­were judged to be “headed for trou­ble or already in difficulty.” An institution was deemed to be in “difficulty” if existing financial circumstances had caused it to cut programs or ser­v ices. According to the author, the prob­lem was that revenues w ­ ere not 27 growing as fast as expenditures. Stratification

In 1970 the market for college was already marked by deep divisions and striking contrasts, most importantly in terms of academic readiness of entering students, their economic backgrounds, and the resources of the colleges themselves. Academic Readiness

For all the upheaval that was taking place at the beginning of the 1970s, some ­things about the vast college industry changed no

S n a p s h o t , C i r c a 1 9 7 0

59

faster than the proverbial glacier. One of ­these was academic standing. ­There existed a well-­defined academic pecking order, and that hierarchy had proven to be quite durable over the years. As stolid as the ­great academic buildings that graced Harvard Yard, the Lawn at Mr. Jefferson’s college in Charlottesville, or Berkeley’s Sproul Plaza, the reputations of colleges had proven equally sturdy, nearly impossible to budge, ­either up or down. A corollary to this stability is the real­ity that the gulfs that separate the academic rigor and the resources across the spectrum of colleges can be enormous. ­Here, barely beneath the surface, was a dimension of differentiation that, perhaps ­because it can be a delicate subject, is often left unstated, or only referred to obliquely. Rather than adjectives such as “mediocre” or “undistinguished,” college guides employed more tactful terms in differentiating academic quality. For example, Cass and Birnbaum’s Comparative Guide to American Colleges employs five gradations of the term “selective” in rating the nation’s top several hundred colleges: selective, very selective, very(+) selective, highly selective, and among the most selective in the country. To illustrate how ­these ­were applied in its 1972 edition to some illustrative colleges: the Texas Christian and the College of Charleston ­were selective, Catholic University and Iowa State ­were very selective, Boston College and Georgia Tech w ­ ere very(+) selective, Grinnell and UC Santa Cruz w ­ ere highly selective, and Prince­ton was among the most selective in the country.28 The differences among colleges along ­these academic dimensions could be dramatic. Consider the academic preparation, or aptitude, of the students who ­were entering vari­ous colleges. The grouping of colleges used in the pres­ent study, based as they are on the average SAT and ACT scores of their entering students, are well suited to highlight such academic distinctions between colleges, or at least the students attending them. To illustrate how wide the gulf was between colleges at the extremes of SAT rankings, Figure 3.1 shows the proportion of students enrolling in each of the seventeen categories of colleges who had been top performers in high school. Using the first-­year students’ own reports of their average grades in high school, the figure displays the percentage of students by college cat-

60 C onte x t

Private D

C

C+

Public B−

B

B+

A−

Public

Private

90+

80–90

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0–25

Proportion of first-year students

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

HBCU

A or A+

Figure 3.1. ​High School Grades, by College Category, 1972 Source: Freshman Survey, students from 188 colleges, 1972. N=101,451. Note: The figure shows the proportion of students in the category who reported having a high school grade average in each of the indicated ranges. The survey question was, “What was your average grade in high school?”

egory who said their grades averaged A or A+, A−, B+, and so forth. Granting that ­these self-­reports are imprecise and that grading rigor surely differed across high schools, the differences between college categories are nevertheless simply enormous. Consider the percentage of students reporting A or A+ averages. At one end we have the top private colleges, a group that includes the likes of Amherst, Harvey Mudd, Cal Tech, and Swarthmore. Among the students who ­were entering one of t­ hese colleges, about 38 ­percent reported having achieved A or A+ averages in high school. At the lower end of the private colleges—­those whose SAT averages ­were in the bottom half of the national distribution—­the percentage of A or A+ students was only slightly more than 5 ­percent. And the HBCUs had even fewer of ­these top students. At the public HBCUs, the percentage of entering students reporting a high school grade average of A or A+ was just 2  ­percent, while nearly 40  ­percent of their students reported high school averages of C or D. In the four college categories where students had the weakest high school rec­ords—­private colleges in the bottom half of the SAT distribution and the HBCUs—­more than a

S n a p s h o t , C i r c a 1 9 7 0

61

fifth of the entering students reported high school grade averages of C or D. The academic gulf revealed by ­these differences among students entering colleges in ­t hese vari­ous categories was gigantic. Given the size of t­ hese gaps, it is not hard to imagine that t­ here could also have existed very large differences in such hard-­to-­measure aspects of their high school experiences as rigor of homework assignments and tests, although documenting such differences might be next to impossible. Surely it was no accident that the academic hierarchy of colleges had remained so stable over time, why the same set of colleges continued, year ­a fter year, to receive applications for admission from the nation’s brightest high school se­niors. Year ­a fter year, top high school se­n iors applied, disproportionately, to colleges like Stanford, Amherst, Michigan, and Columbia, not to former teachers colleges or low-­ranked private colleges. Information about colleges was crucial in sustaining this stability, and it was available—­from high school counselors, parents, and published college guides. Yet one source of information had come to assume a position of paramount importance—­scores on standardized tests, the SAT, and its ­sister test the ACT. The origins of the SAT go back to 1900, when a dozen selective colleges in the Northeast, including Columbia, Prince­ton, and Bryn Mawr, joined forces to form the College Board.29 Their ostensible aim was to streamline the pro­cess of selecting students, which had up u ­ ntil then entailed exams devised and graded by the colleges themselves. ­Under the new consortium, applicants would take a common essay exam. The first examination produced by the College Board, administered in 1901, had nearly a thousand test-­takers. But it was not ­until 1926 that the College Board administered a test that resembles ­today’s SAT. Named the Scholastic Aptitude Test (a name long ago dropped in ­favor of the capitalized letters), it was modeled on psychological intelligence tests such as the Army Alpha, a test used to classify soldiers in World War I. In the succeeding de­cades, u ­ nder the development of the Educational Testing Ser­vice (ETS), the SAT continued to be a test of intel-

62 C onte x t

lectual aptitude, a first cousin to an IQ test. Among the ­things it did not become ­were a test of subject ­matter, a more general test covering other ­human traits arguably as impor­tant as intellectual aptitude in determining readiness for college work, or a test that would correct for the power­ful effects of social background. ­T hose in charge of developing and marketing the SAT briefly flirted with ­these and other kinds of tests, but always returned to the aptitude test. The SAT as it evolved was attractive b ­ ecause it was reliable (a student’s score seldom changed much from one taking to the next), and it was cheap to administer, thanks to computer technology. 30 The test’s coverage remained concentrated in the Northeast, however, with three-­fourths of ­those taking it applying to Harvard, Yale, Prince­ton, Smith, Wellesley, or Vassar. Over time, its reach expanded, to cover much of the country, ceding only the Midwest to the ACT. By 1960 the two tests ­were being taken by 533,000 high school se­niors a year. By 1970, a­ fter a de­cade of phenomenal growth, the total had reached nearly 4 million.31 By 1970 a vigorous debate had developed as to what effect this new testing regime might have on the college marketplace. One view was that the SAT would aid in solidifying the hierarchy of colleges that had already developed.32 A contrary viewpoint was that standardized tests would weaken the influence of Eastern prep schools, as bright students from the provinces could easily show their academic potential by turning in a high score. Indeed, this latter possibility had been the motivation for a pioneering outreach effort by Harvard president James Conant in 1934 to enroll more young men from the hinterlands of six Midwestern states. One of the early recipients was a young man from Champaign, Illinois, named James Tobin, who would gradu­ate in 1939 summa cum laude and go on to win a Nobel Prize in economics in 1981. The g­ reat value of the SAT was the information it provided, not only to colleges but to the students who took it. It allowed high school students, who in the past might only have had their classmates for reference, to perceive their own abilities in quantitative, nationally comparable terms. Not only did the SAT provide a student with this information, it also became a new

S n a p s h o t , C i r c a 1 9 7 0

63

source of esteem. In the words of one historian, it became “a single national prestige ranking.”33 The College Board’s president noted, “I learned from hearing my own ­children’s conversation that SAT scores have now become one of the peer group mea­sur­ing devices. . . . ​ The bright, steady student is appreciated with his high 600, and the unsuspected genius with his 700 is held in awe.”34 In the pro­cess the spread of the SAT and ACT could not fail to make the market for colleges more national in scope. It allowed a college to consider seriously an applicant from a distant state, attending an unfamiliar high school, b ­ ecause of the standardized mea­sure of academic merit available. And the possession of a high SAT score could embolden the student from an unknown high school to consider that distant college. Economist Caroline Hoxby argues that a result of this improved information, for firm and customer alike, would be an increase in academic stratification. Chapter 8 examines this possibility. Socioeconomic Status

To finish up our sketch of the state of the college market as it was as the 1970s began, another kind of stratification demands attention—­ socioeconomic. ­Because the United States had often been portrayed as a land of opportunity and ­because education was widely viewed as the principal ave­nue for climbing the ladder of economic betterment, the issue of economic stratification always lurked like an unwelcome guest in debates about the social role of higher ­education. Although it usually remains beneath the surface in media coverage, everyday conversation, and official pronouncements by colleges, this issue has certainly not been neglected by scholars of higher education. Historians of higher education have taken pains, for example, to point out the degree to which college has always tended to serve the m ­ iddle and upper classes.35 What early social science evidence that exists certainly does reveal class bias in patterns of enrollment. A survey at the University of Michigan showed that, even at a public university, students came

64 C onte x t

disproportionately from the families of professionals and homeowners. Of 500 Michigan undergraduates surveyed in 1924, fewer than 15 ­percent had ­fathers whose occupations could be described as blue collar.36 And the parents of t­ hese students ­were much more likely to be homeowners than citizens of the state at large.37 Data collected by the census showed that ­children of more educated f­ athers had a better chance of attending college than c­hildren of less educated ones. Over the de­cade from 1915 to 1925, 8 ­percent of t­ hose whose ­fathers had less than eight years of school enrolled in college, compared to 47 ­percent of ­those whose f­ athers had at least some college education. Thirty years ­later, the corresponding rates of college attendance w ­ ere 14 ­percent and 78 ­percent, respectively. 38 The gap had widened considerably, from 39 to 64 percentage points. Not only ­were college students as a group distinct from their noncollege peers, ­ there ­ were systematic socioeconomic differences across colleges. Historian O. Edgar Reynolds reported in a 1927 book on surveys done of the incomes of parents of college students. Whereas the parents of 17 ­percent of the students at Indiana’s ­Wabash College had incomes over $10,000, the comparable percentage was 28 ­percent at Mills College (California), 35 ­percent at Washington and Lee, 47 ­percent at Yale, and 68 ­percent at Williams.39 Students’ ­family incomes tended to be lower at public institutions than at private selective colleges. The 1924 median f­ amily income for students attending a sample of state universities was $3,172, compared to $4,889 and $5,140 for sampled private men’s and w ­ omen’s colleges, respectively. The more diverse and less selective collection of coeducational private colleges w ­ ere closer to the state universities, with a median ­family income of $2,964.40 ­After World War II, a commission appointed by President Truman highlighted disparities in educational attainment.41 Closer in time to 1970, evidence of differences in college attendance by socioeconomic status had been documented in recently published studies based on Proj­ect Talent, a national survey undertaken of the high school class of 1961. Of high school gradu­ates whose ­fathers ­were professional or semiprofessional, 67 ­percent enrolled in

S n a p s h o t , C i r c a 1 9 7 0

65

college and 40 ­percent completed college. By contrast, for ­children of factory workers, craftsmen, and unskilled workers, only 26 ­percent enrolled and 15 ­percent graduated.42 Even when students’ achievement levels ­were controlled for, socioeconomic differences persisted. For men who scored in the top quartile of mea­sured achievement, ­those in the top socioeconomic quartile had a 92 ­percent chance of attending college, but ­those in the bottom quartile had just a 61 ­percent chance. Such socioeconomic differences ­were evident at ­every achievement level.43 ­There was less evidence about socioeconomic differences within the ranks of college-­going students. An exception was a study published in 1969 examining the distributional effects of California’s differentiated system of higher education, a system that featured large subsidies at the campuses of the world-­renowned University of California, compared to smaller subsidies at California’s state colleges and ones smaller still at the state’s numerous community colleges. The study made clear that students who qualified for and attended the University of California system tended to come from higher-­income families than t­ hose attending the other systems. The authors estimated that the median income of families with ­children attending a campus of the university was $12,000, compared to $10,000 for ­those with ­children at one of the state colleges and $8,800 for ­those with ­children at a community college.44 That college enrollment rates had a strong socioeconomic bias could not be denied. As the 1960s wound down this real­ity remained, uncomfortably at odds with the American ideology of equal opportunity. Some social commentators viewed colleges as an instrument by which the ruling class perpetuated its advantages by using educational disparities to legitimate economic ones.45 For other commentators, this view was too ­simple. In a 1967 book, Christopher Jencks and David Riesman offer a more nuanced view. Without denying the advantages enjoyed by ­middle and upper class, they posit that, to prevent decay, the economic system needed to leave room for some bright and ambitious individuals to rise above their parents’ economic status to inject innovation and energy.46

66 C onte x t

Not only was ­there a socioeconomic distinction between ­those who went to college and t­ hose who did not, t­ here w ­ ere distinctions within the ranks of college-­going youth. Taking advantage of the rich information available for first-­year students entering our sample of 188 colleges, we can examine two familiar mea­sures of socioeconomic status: parental education and ­family income. Figures 3.2 and 3.3 show how ­these two indicators differ across the seventeen types of colleges in the fall of 1972. The stories t­ hese graphs tell are much the same. Figure 3.2 shows the percentage of students whose ­fathers ­were college gradu­ates. Three unmistakable conclusions can be drawn. First, the students ­going to private colleges ­were more likely than t­ hose attending the public ones to have college-­educated ­fathers. Second, within each of ­those subsets, parental education tended to rise with the SAT ranking of the college. Among private colleges, the lowest share of college-­educated ­fathers was in the 0 to 25th percentile category, with a third of the students ­there having 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

Private

Public

Pub

Priv

90+

80–90

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0–25

0.0

HBCU

Figure 3.2. ​Proportion of Students Whose F ­ ather was a College Gradu­ate, 1972 Source: Freshman Survey, students from 188 colleges, 1972. N=101,451. Note: The figure shows the proportion of students in the category who answered about their father, “college degree,” “some gradu­ate school,” or “gradu­ate degree.” The survey question was, “What is the highest level of formal education obtained by your parents?”

S n a p s h o t , C i r c a 1 9 7 0

67

$35,000 $30,000 $25,000 $20,000 $15,000 $10,000 $5,000

Private

Public

Pub

Priv

90+

80–90

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0–25

$0

HBCU

Figure 3.3. ​Average ­Family Income of First-­Year Students, by College Category, 1972 Source: Freshman Survey, students from 188 colleges, 1972. N=101,451. Note: The survey question was, “What is your best estimate of the total income last year of your parents (not your own f­ amily, if you are married)? Consider annual income from all sources before taxes.” Categories ranged from less than $3,000 to $50,000. For the calculation of average ­family income, values w ­ ere assigned at midpoints of indicated categories. For the top category, income was estimated as the average adjusted gross income for all taxpayers in 1971 with adjusted gross incomes of $50,000 or more.

f­ athers who ­were college gradu­ates. The highest rates ­were in the top four categories of private colleges, ranging from 60 to 73  ­percent. ­There was also a correlation within the public colleges, with the percentage of college-­educated f­ athers ranging from 28 ­percent for the 0 to 50th percentile category to 47 ­percent for the top category. The third generalization applies to the HBCUs. Their students ­were least likely to have college-­educated ­fathers, especially the students in the public HBCUs, for whom the rate was just 10 ­percent. A very similar picture emerges if income is used as the mea­sure of socioeconomic status. Students reported ­family incomes (expressed in 1971 dollars) that tended to rise with the SAT ranking of their college, with ­those attending private institutions exceeding t­hose ­going to public ones. As with f­ather’s education, the college group

68 C onte x t

showing the highest value was the next-to-highest ranked group of private colleges, t­ hose in the 98th to 99th percentiles. Next highest ­were t­ hose in the 90th to 95th percentiles. As in the previous figure, the lowest value by far was in the public HBCUs. The average reported income of ­these students was only about one-­fourth that of the students in the highest income group. As a reminder, that top college group includes such colleges as Carleton, Mount Holyoke, Johns Hopkins, and Prince­ton, while the group of public HBCUs includes Alabama A&M, Morgan State, and Winston-­Salem State. The socioeconomic gap between ­these two extreme categories is indeed vast. Resources

Besides the large disparities that existed among colleges in the academic and economic standing of their students, ­there ­were obvious gaps in resources to which students w ­ ere exposed. One needed only visit a campus to be struck by the markers of solid wealth at the prestigious colleges, compared to the modest and utilitarian look of the teachers colleges and small, nonselective private colleges. Such gaps might show up in indicators like the number of books in the college library or the percentage of instructors with PhDs. An early attempt to quantify the differences in resources across colleges was a survey of financial data made in 1909. Although this sample size makes it too small to be definitive, the patterns in t­ hese data indicate differences in revenue per student that are both large and consistent with the selectivity rankings for 1970 that are used in the pres­ent book. The five private colleges that would l­ater rank in the highest 1 ­percent of average SATs had by far the highest revenue per student, and the private colleges that would l­ ater rank just below had the next highest.47 Among the public institutions, the four that would l­ater rank in the top 20  ­percent of SATs had considerably higher revenue per student than the public institutions at lower SAT levels. To an extent unmatched among the firms of most industries, the firms in the college market stubbornly held onto their historic positions in the quality / prestige pecking order.

S n a p s h o t , C i r c a 1 9 7 0

69

Such gaps are evident in a comparison of 1970 faculty salaries. This data reflects the total compensation of full professors for a random sample of 160 colleges and universities for the 1970 / 71 academic year. The data was published by the Association of University Professors for twenty randomly selected colleges in each of eight groups, defined by average SAT score and public / private status. Expressed in current 1970 / 71 dollars, the comparison reveals rather large proportional differences, as shown in Figure 3.4. Between the top-­and bottom-­ranked private colleges, the gap is an impressive 70  ­percent. The figure shows that public colleges tended to pay more than comparably ranked private ones. For ­those below the ­middle of the SAT ranking, the public colleges paid 20 ­percent more,

$25,000

$20,000

$15,000

$10,000

Private

90–100

50–90

0–50 and HBCUs

98–100

95–98

90–95

50–90

$0

0–50 and HBCUs

$5,000

Public

Figure 3.4 ​Average Faculty Compensation in 1970 / 71, by Broad College Group Source: AAUP Bulletin, Vol. 57, No. 2 (June 1971), total compensation for full professors. Note: Sample consists of stratified random sample of 160 institutions, drawn from 1,157 institutions with data on SAT circa 1970. For each of the eight groups of institutions specified in the figure, data on average compensation for a random sample of twenty institutions was drawn, based on SAT percentiles for 1970.

70 C onte x t

on average, and between the 50th and 90th percentiles the publics ­were about 5 ­percent higher. College Athletics

Although it may appear incongruent, even unwelcome, in a volume about the market for college, the topic of college athletics deserves attention. The reason, quite simply, is ­because it is impor­tant, much more so than would be suggested by the lack of attention given to it by most scholars of higher education. T ­ here is no shortage of attention to college sports outside of scholarly research, of course. It is a familiar subject in newspapers’ sports sections, popu­lar books about sports, and college games broadcast on tele­v i­sion and radio. But scholars of higher education, with some notable exceptions, have ­until recently paid ­little attention to college athletics as an integral aspect of colleges. 48 In approaching this subject, it is crucial to begin by making an impor­tant distinction between the two, almost mutually exclusive, meanings of the term “college athletics.” One of its meanings refers to an activity played by students, mostly for reasons of their own, not for the appeal their matches might have for spectators outside their friends and families, nor for their contribution to ­future success in professional sports. For ­these students, being on a college team was an integral part of being a college student. This definition of college athletics decidedly conforms to the “amateur model.” The vast majority of college athletes play on teams that not only draw few spectators but bring in no revenue to speak of and are lucky to be mentioned in the student newspaper. It is a time-­honored aspect of American colleges to field athletic teams that compete in intercollegiate leagues. That colleges do establish and finance ­these teams suggests that many potential students want them, or want to go to a college that has them. Although it might be a stretch to argue that colleges view participation in athletics as primarily educational, most ­people who take curricular issues seriously at the college level would view participation in athletics as not only a healthful activity but one wholly consistent

S n a p s h o t , C i r c a 1 9 7 0

71

with the aims of baccalaureate education. Many gradu­ates who ­were athletes in college believe the experience taught them impor­tant lessons, ones not readily gleaned from formal academic c­ oursework. Consider writer Pat Conroy’s perspective on his time as a basketball player at The Citadel: Basketball provided the nearest approach I’ve ever made to the realms of ecstasy. The sport consumed the best part of my dreaming self, and I found myself in reverie ­a fter reverie moving swiftly in the flow and anarchy of games. My ­whole philosophy of life was caught up with what I believed ­were the responsibilities of a point guard—­the importance of outhustling your opponent, watching for the unexpected, moving teammates to their proper spots on the floor, barking ­orders and calling the plays, exhorting and inspiring your team, and never quitting ­until the buzz­er has sounded.49

­ hese college teams are or­ga­nized first and foremost for students, T perhaps in large part to attract them, certainly as an optional part of their education. Although some of their games and matches do attract spectators, the principal reason for organ­izing t­ hese teams is to provide an or­ga­nized setting for students to play together and compete, not to entertain onlookers. In its second meaning, the term “college athletics” refers to a commercial enterprise whose educational component is small relative to its commercial component. Much has been written about this commercial enterprise u ­ nder the term “big-­time college sports,” although it has been almost wholly ignored by scholars of higher education. But it should not be ignored in any comprehensive analy­sis of American higher education ­because it is simply too impor­tant.50 Mea­sured in terms of the attention it receives from the general public or by the weight it is given by university leaders, commercial sports operation is in truth one of the core functions in a hundred or so universities in the country. Among t­ hese are some of the best universities in the world.

72 C onte x t

Such commercial sports operations are just one of several apparently unrelated university functions that may influence baccalaureate education, among them gradu­ate and professional training, research, hospitals, and other public ser­vice. The first of t­ hese may affect undergraduate education naturally, by way of teaching assistants or participation in research proj­ects. The goals and functions of ­these vari­ous core functions simply have too many interrelationships for ­there not to be some effect. Two of t­ hese functions are unabashedly commercial—­hospitals and big-­time sports. But all of ­these apparently unrelated functions except big-­time sports are fully consistent with the institutions’ teaching-­research-­service objectives, as made explicit in the scores of mission statements that universities proudly display on their websites. This lack of connection to the educational mission may be the reason that scholars of higher education have, by and large, skirted the issue of spectator sports. But this is a ­mistake if the aim of research is to analyze universities as they are, rather than as one might wish them to be. The rec­ord reveals that, for universities with big-­ time sports operations, not only do they value the activity, they value it for its own sake as well as for any benefits it might bring the educational mission. ­There does indeed appear to be real marketing value in having a prominent and successful football or basketball team. This marketing value appears to have motivated colleges to or­ga­nize football teams in the nineteenth ­century, and it continues to be a reason given for playing football and basketball at the highest, and priciest, level of competition. Thus it is that involvement in the commercial form of college athletics constitutes another dimension of variety across American colleges. Let us see what a snapshot of participation in 1970 looks like. Figure 3.5 shows the extent to which colleges in the previously defined categories participated in big-­t ime college sports, as indicated by the designation as NCAA Division I-­A , which means having a football team playing at the highest and most expensive level, the one that allows colleges to offer the maximum number of scholarships. As

S n a p s h o t , C i r c a 1 9 7 0

73

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

Private

Public

Pub

Priv

90+

80–90

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0–25

0.0

HBCU

Figure 3.5. ​Share of Colleges in Division I-­A , by College Category, 1970 Source: Soren Sorensen, “College Football 1970, Ranking of Teams within College Conferences,” University of Tennessee, Department of Physics & Astronomy, http://​ www​.­phys​.­utk​.­edu​/­sorensen​/­cfr​/­cfr​/­output​/­1970​/­CF​_­1970​_­Conferences​.­html. Note: Bars show the percentage of students in each college category attending colleges that w ­ ere in NCAA Division I-­A institutions. Shares are weighted by enrollment.

the figure shows, big-­time college sports was primarily but not exclusively an activity undertaken by public institutions in 1970, although a fair number of private institutions, ranked in the top half by SAT, fielded Division I-­A football teams as well. Among the top-­ ranked private colleges that ­were in Division I-­A in 1970, the eight that made up the Ivy League deserve an asterisk. Although officially part of this least-­restricted, most-­commercial division in the NCAA, that league has departed from the pack by forbidding its members from giving athletic scholarships, in effect not participating fully in big-­time, commercial college sports. 51 According to a recent article, conferences such as the Ivy League perform an entirely non-­athletic function—to proclaim to the wider world the kinship of similar colleges, thereby creating exclusive clubs that establish and maintain status for its members. 52

74 C onte x t

The View from 1970

The starting point for this four-­decade numerical chronicle of the evolving market for college is, roughly, 1970, a choice dictated mostly by the existence of a trove of detailed information on college students, the Freshman Survey, which was just getting ­under way about then. In hindsight, the year 1970 also represents something of an inflection point for the college market in Amer­i­ca. For one t­ hing, that year marked the end of most serious student protests and of the biggest advances in the numbers of black students attending predominantly white colleges, especially in the South. The year also augured a period of growing anxiety about the ability of private colleges to pay their bills and maintain their quality. Ironically, some of the colleges most worried about the ­future would find the next four de­cades to be a heady time for advancement and expansion rather than retrenchment. Two seemingly unrelated features of the college market at this juncture deserve par­t ic­u ­lar attention. One is the feature best expressed by the so­cio­log­i­cal term “stratification,” the sharp differentiation within the college market, showing up principally in differences across colleges in the academic ability of enrolled students and in their social and economic status. ­There ­were large gaps as well in resource levels across colleges. Before tracing changes in patterns of stratification, it is impor­tant to begin with a snapshot of ­things as they w ­ ere at the period’s outset. The second feature highlighted in this chapter is college athletics. Although rarely included in sober assessments of higher education, the tendency to shun the topic is wrong-­headed. Athletics in both its meanings are worth taking seriously in describing this market. As an activity that students take part in, it is both a marketing tool and an au­then­tic component of the educational experience. In its commercialized form, its role in baccalaureate education is altogether dif­fer­ent, and the effects on competition and the efficiency of outcomes may be larger than from athletics of the first sort.

S n a p s h o t , C i r c a 1 9 7 0

75

4 Outside Forces

“The real world” is a term that college students have been known to use to refer to every­thing beyond the bound­aries of their campus. Across the four de­cades covered by this study, a handful of forces in that real world w ­ ere having lasting effects on the market for ­college— on demand and on supply, all the way down to the details of how learning occurred. For several of ­these outside influences the connection to higher education seems obvious, but for the o ­ thers the effects have become discernible only in the rearview mirror, and then only gradually. Of ­t hese outside forces, two ­were so gradual that they might be easy to overlook. Over a period of de­c ades, though, especially with the help of a copy of the Statistical Abstract of the United States, even t­ hose can be readily discerned. The first of t­ hese two stealth trends was the steady reshaping of the nation’s income distribution. Beginning in the late 1970s, incomes of t­ hose at the top began to increase rapidly while incomes

for ­those in the bottom half stagnated. ­Running in parallel with this reshaping of income distribution w ­ ere a steady decline in the importance of manufacturing and an equally steady increase in the use of computers across businesses of all types. Both of t­ hese developments boosted the economy’s demand for skilled ­ labor, pushing up the earnings of college gradu­ates relative to t­ hose with just a high school education. The other stealth trend was the metamorphosis of the nation’s racial and ethnic composition. Whereas income in­equality is often invisible, a demographic shift is ­t here for all to see, leaving a vis­i­ble imprint on workplaces, communities, and colleges across the land. The third force, technology, had  influence both gradual and dramatic. The remaining two influences—­government support and college rankings—­had more obvious connections to the college market. This much applies to all five of ­these forces: to understand how the college market changed over this period requires being aware of t­ hese aspects of “the real world.” Growing In­equality

Beginning sometime near the end of the 1970s American income distribution began what would become a sustained pro­cess of reshaping itself. One aspect of the change was a turnaround in the gap between the earnings of college gradu­ates and t­ hose with only a high school diploma. During the 1970s, this gap narrowed, reflecting both the continued strong demand for blue collar workers and the manufactured goods they produced and the plentiful supply of college gradu­ates, spurred in part by Vietnam War deferments.1 ­These trends moved ­labor economist Richard Freeman in 1976 to write a book entitled The Overeducated American.2 But events would quickly make that book, at least its provocative title, outdated. As the country began to lose its dominant role in manufacturing and the power of ­unions faded, the demand for high school gradu­ates began to recede.3 Si­mul­ta­neously the growing demand for skilled ­labor began to outpace supply, and the average earnings for college gradu­ates

Ou t s i d e F o r c e s

77

began a long ascent. The result of ­these two trends was a rapid increase in the “college earnings advantage.” Simply put, the income in­equality increased ­because the incomes of ­people at the top of the distribution, especially ­those at the very top, began to grow much faster than ­those of every­one ­else. This divergence can be seen by tracing how the average income for ­those in the ­middle one-­fifth of the income distribution changed over time, as compared to the fortunes of families in the top 5 ­percent and the top 1 ­percent of the income distribution. Between 1981 and the eve of the G ­ reat Recession, incomes of the m ­ iddle fifth increased, ­after inflation, by 21 ­percent. Over the same period incomes of families in the top 5 ­percent more than doubled. Mea­sured from 1981 to 2012, the top 5 ­percent enjoyed a 96 ­percent increase in average income, compared to just 14 ­percent for the middle-­income h ­ ouse­holds.4 The growing income in­equality evident in this comparison played out across the entire income distribution. Between 1981 and 2012 families in the next-­to-­lowest one-­fi fth saw their incomes increase just 7 ­percent in real terms, and for t­ hose in the bottom fifth, average income actually declined by 6 ­percent.5 At the very top of the distribution, by contrast, incomes shot up at astronomical rates. The top 1  ­percent of ­house­holds, made famous during the 2012 presidential election, saw their average, inflation-­adjusted income nearly triple between 1981 and 2015. 6 ­T hese divergent trends ­were just one of many markers of increasing income in­equality. Another was the Gini coefficient, a widely used metric of a country’s income in­equality. The Gini coefficient calculated for ­house­hold income ­rose from 0.406 in 1981 to 0.477 in 2012. Accompanying ­these trends was a decline in intergenerational income mobility, that is, how closely ­children’s economic standing was tied to that of their parents.7 In­equality in ­house­hold wealth also increased over ­these de­cades. Driven by rapid increases in net worth by t­hose at the top of the wealth distribution, wealth in­equality ­rose during the 1980s, remained steady ­until the ­Great Recession, and then jumped sharply ­after 2007. Between 1979 and 2012, the share of ­house­hold wealth held by the richest 0.1  ­percent of h ­ ouse­holds increased from

78 C onte x t

7 ­percent to 22 ­percent. 8 Driving that post-2007 increase w ­ ere widespread declines in the net worth of middle-­class ­house­holds, caused in large part by the precipitous plunge in h ­ ouse prices and the accompanying wave of foreclosures. Black and Hispanic h ­ ouse­holds ­were particularly hard hit. Reflecting ­these developments, the Gini coefficient for ­house­hold net worth, ­after remaining roughly steady for nearly two de­cades, jumped ­after 2007, from 0.834 to 0.870 in 2010.9 ­These profound shifts in the nation’s income and wealth distributions could not but influence the market for college. Granting that the causal chains undoubtedly run in both directions between the distribution of income and the nation’s colleges and universities, the changes in income distribution might have affected the market for college in several ways. Prob­ably the most obvious way is by putting more money into some customers’ bank accounts and less into ­others’. Affluent parents of college-­age c­ hildren, for example, enjoyed a very good run for three de­cades, making the high tuition rates of many private colleges a feasible outlay even without financial aid. The tuition list price—­the price before any financial aid—­increased faster than the rate of prices or incomes in general.10 All this is to make the rather obvious point that access to money makes most consumer purchases easier, including paying for college. And this logic extends to expenditures besides tuition and fees. For example, colleges typically charge not inconsequential fees—­$80 was typical in 2014—­just to pro­cess an application, fees that can quickly add up for high school se­niors applying to multiple colleges.11 To be sure, ­these obvious considerations are complicated by the widespread availability of need-­based financial aid, assistance that can extend even to application fees. The point is that families with higher incomes almost always have an advantage compared to the less affluent when it comes to having the financial wherewithal to send their ­children to college. At the top end of Amer­i­ca’s income distribution, therefore, rising college costs are less intimidating than they are likely to be for every­one ­else. A second reason growing income in­equality has likely affected the college market is that it highlights the value of a college degree

Ou t s i d e F o r c e s

79

as a financial investment. In­equality in wages, not disparities in income from capital, has been the chief source of overall income ­in­equality, and wages are highly correlated with educational attainment. Although it borders on the simplistic, the idea that a college education is a sound financial investment seems readily apparent, even obvious, if one just compares the average incomes of college gradu­ates to ­those with just a high school diploma. A common calculation is based on comparing median incomes of full-­time workers in ­those two categories of educational attainment. ­A fter falling through the late 1960s and most of the 1970s, the percentage gap between t­ hese two incomes—­dubbed the “college earnings advantage”—­ reversed direction at the end of the 1970s, growing steadily ­after that despite the constant outpouring of new college gradu­ates. Figure 4.1 traces this remarkable turnaround. It graphs the proportional gap in median earnings of men and ­women aged twenty-­ five and above between ­those with just a high school diploma and ­those with just a college degree. In 1979 men who had college degrees but no more education earned 24  ­percent more than ­those who ­were high school gradu­ates. By 1982, however, the gap had risen to 31 ­percent, and by 1992 it was 51 ­percent. By 2014 it had risen to 67 ­percent. For ­women, the story was almost exactly the same. This gap made itself evident in dozens of ways that w ­ ere impossible to ignore. The reliable manufacturing jobs that high school gradu­ates had used to build middle-­class lives gradually dis­appeared in many parts of the country. The economic distress went well beyond the union-­dominated Rustbelt into the textile and furniture towns of the nonunionized South as many industries moved their plants to China, Mexico, and elsewhere. ­These new realities emphatically underlined in the public mind the growing importance of having a college degree. The demand for baccalaureate education intensified, in other words, as its product became less a luxury and more a necessity.12 As intuitive and easily calculated as this percentage gap is, it is by no means a complete or even altogether accurate mea­sure of the financial payoff from attending college. For one ­thing, it ignores the costs of attendance as well as the complex issue of assigning causal

80 C onte x t

0.70 0.65 0.60

Proportional difference

0.55 0.50 0.45 0.40 0.35 0.30 0.25

1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013

0.20

Year Men

Women

Figure 4.1. ​The College Earnings Advantage: Proportionate Difference in Median Earnings between College and High School Gradu­ates, Selected Years, 1967–2013 Source: 1967–1999: U.S. Bureau of the Census, Current Population Reports, Series P-60, Money Income of House­holds, Families, and Persons in the United States (Washington, DC: U.S. Government Printing Office); 2000–2003: U.S. Census Bureau, Money Income in the United States: 2000, ­Table PINC-03. http://­w ww​.­census​.­gov​/­hhes​/­w ww​/­cpstables​/­macro​/­032001​ /­perinc​/­toc​.h ­ tm; 2004–2014: U.S. Census Bureau, Income, Poverty and Health Insurance in the United States: 2004, T ­ able PINC-03. http://­w ww​.­census​.­gov​/­hhes​/­w ww​/­cpstables​ /­macro​/0 ­ 32005​/­perinc​/t­ oc​.­htm. Note: Lines show three-­year rolling average differences in median earnings for full-­t ime, year-­round workers, age 25 and older. High school gradu­ates include t­ hose with four years of high school or a high school diploma only. College gradu­ates include t­ hose with four years of college or a bachelor’s diploma only.

Ou t s i d e F o r c e s

81

significance to observed differences in earnings. Second, it is based on averages across the population. In fact, some college gradu­ates saw high returns while o ­ thers had only modest gains. Moreover, ­there is evidence that that the average earnings of college gradu­ates may have stopped growing a­ fter 2000, with recent research suggesting a general weakening of the demand for cognitive skills. College gradu­ ates with weaker skills, who might have aspired to relatively undemanding occupations, saw routine functions progressively taken over by computers. Some of ­t hese college gradu­ates took jobs formerly filled by less educated workers.13 Notwithstanding t­ hese complications, ­there can be ­little doubt that the “college earnings advantage” is an intuitive and persuasive piece of evidence that getting a college degree is a smart investment of time and money, thus fueling the demand for college. A third reason why the changing income distribution could have an impor­tant effect on the market for college is the strong likelihood that f­amily income—­quite apart from ­family characteristics that are merely correlated with it—­influences a young person’s chance of getting into a competitive college. In a host of ways, affluence can enhance the education of ­children and their academic readiness for college. F ­ amily income is correlated with any number of potentially consequential variables, among them well-­fi nanced public schools, neighborhood amenities, travel and other summer enrichment experiences, and private school attendance. The economic disparities in spending and teacher quality in public schools are among the nation’s most per­sis­tent policy challenges. But the advantages of income extend beyond the school day. In 1972 and 1973, ­house­holds in the top income quintile spent an average of $3,536 (in 2008 ­dollars) on child-­enrichment goods and ser­vices such as m ­ usic lessons and summer camp while ­house­holds in the bottom quintile averaged just $835. Three de­cades ­later, in 2005 and 2006, the corresponding average expenditures ­were $8,872 for top-­quintile ­house­holds, compared to just $1,315 for t­hose in the bottom quintile.14 Thus, over this period of diverging incomes, spending on ­children’s enrichment by top-­quintile ­house­holds, which already dwarfed the low-­income

82 C onte x t

average, increased twice as fast (2.8  ­percent versus 1.4  ­percent a year). In addition to ways that affluent parents might help their ­children’s admissions prospects by enhancing their ­children’s educational development, money can buy better prospects through expenditures that are more strategic than educational, as discussed in Chapter 8. A fourth pos­si­ble connection between the nation’s income distribution and the market for college comes not by way of prospective students but rather by way of donors. Among the subsectors of the vast array of organ­izations receiving charitable contributions, universities rank near the top. It is a well-­established fact that income is positively correlated with charitable giving, and that high-­income donors tend to devote a disproportionate share of their giving to col­ hether t­ hese tendenleges and universities.15 The question remains w cies have played out over the de­cades of growing income in­equality to change the pattern of donations to colleges and universities and, if so, which institutions have benefited most within the higher education domain.16 Chapter 5 takes up this question. Changes in the College-­Age Population

Demography may not be destiny, but demographic trends surely have had a potent influence on the baccalaureate market. How Many Eighteen-­Year-­Olds

Although college students can be younger or older, students traditionally attend college immediately a­ fter high school, when students are eigh­teen years old or thereabouts. So the projected number of eighteen-­year-­olds is of no small interest to colleges and their planning officers. Over the four de­cades covered ­here, that number ­rose for the first de­cade (by almost 20 ­percent), fell for the next de­cade and a half (by about 25 ­percent), ­a fter which it gradually increased again, undoing all of the post-1970s decline. T ­hese movements  are  displayed in Figure  4.2, which shows the number of

Ou t s i d e F o r c e s

83

200

180

160

140

120

100

Number 18–19-yr-olds

2011

2009

2005

2007

2001

2003

1999

1997

1995

1993

1991

1989

1985

1987

1981

1983

1979

1977

1975

1973

1971

80

Enrollment, public 4-yr

Enrollment, nonprofit 4-yr

Figure 4.2. ​Four-­Year College Enrollment and Number of Eighteen-­and Nineteen-­Year-­Olds, 1970–2011 Source: U.S. Department of Education, Digest of Education Statistics 2012, ­Tables 19 and 223. Enrollments are total fall enrollment in degree-­g ranting institutions. Totals expressed relative to 1970 levels.

e­ ighteen-­year-­olds relative to the 1970 level. Among other ­things, the graph shows the relative size of the college-­age population at the three dates used in the current study for making most comparisons—1970, 1990, and 2010. Notice that the eighteen-­year-­old population was about the same size in the first two of ­these years, meaning that students observed at t­hese two dates faced roughly similar competition for college enrollment places. By 2010, however, the numbers had grown by roughly a fifth, making that a considerably more competitive year from the perspective of aspiring students. ­These ups and downs tend to have predictable effects on the market, creating buyers’ markets when cohorts are small and sellers’ markets, marked by reduced ac­cep­tance rates, when cohorts are large.17

84 C onte x t

Regional Shifts

Another demographic trend was the long-­term geographic shift away from the country’s Northeast, the location of many of the country’s most selective private colleges. When the American nation first took hold on the Atlantic shore and its first cities arose in the Northeast, so too did its earliest colleges. Given the nearly immutable rule that established colleges rarely move, the strong regional footprint left by three centuries of American higher education history meant that many of the nation’s most prestigious colleges w ­ ere and still remain located in the Northeast. All eight Ivy League institutions, for example, are located ­there.18 This concentration in the Northeast is reflected in Figure 4.3. In 1970 the nine states of the Northeast contained private colleges and universities that enrolled 41  ­percent of all students in private colleges. ­Those states contained an even larger share of the country’s top-­ranked colleges. Of the American colleges, public and private, whose 1970-­era SAT scores put them in the top 5  ­percent of all four-­year institutions, 63 ­percent of their students w ­ ere in colleges located in the nine Northeastern states. Yet in that same year, 1970, ­those same states accounted for just 22  ­percent of all public high school gradu­ates in the country. And this regional mismatch only grew worse over time. By 2010 t­hose nine Northeastern states accounted for only 18 ­percent of all high school gradu­ates. This shift in population away from the region containing so many of the storied Northeastern colleges can be seen by comparing the bottom two bars in Figure 4.3. Over the four de­cades from 1970 to 2010, the Sunbelt grew rapidly while the industrial states of the Northeast and Midwest stagnated. For example, the shares of public high school gradu­ates living in the Northeast and Midwest fell (from 52 to 41 ­percent), while t­ hose of the South and West increased (from 48 to 59 ­percent). One consequence of this shift was to move students farther away, on average, from private colleges, many of which ­were located where population had previously been concentrated. This geographic shift meant that, if the colleges traditionally in the top

Ou t s i d e F o r c e s

85

Enrollment in top 5% SAT colleges, 1970

Enrollment in private colleges, 1970

Number of public HS graduates, 1970

Number of public HS graduates, 2010 0%

20%

40%

60%

80%

100%

Regional share Northeast

Midwest

South

West

Figure 4.3. ​The Regional Distribution of Top Colleges, Private Colleges, and Public High School Gradu­ates Source: First bar: author’s calculations; Second bar: U.S. Office of Education, Digest of Education Statistics 1971, ­Table 82. Third bar: Digest of Education Statistics 1971, ­Table 67. Fourth bar: Digest of Education Statistics 2012, ­Table 123. Note: The first bar shows the distribution by region of students enrolled in one of the sixty-­six colleges in the enrollment-­weighted highest 5 ­percent of SAT scores in 1970, and the second bar shows the comparable distribution for students enrolled in any private college in 1970. Regions are defined as follows: Northeast: Connecticut, Maine, Mas­sa­chu­setts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont; Midwest: Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, South Dakota, Ohio, Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Mary­land, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, V ­ irginia, West ­Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, Wyoming.

selectivity echelon wanted to retain their share of the nation’s top students, they would have to recruit from increasingly distant locations. By the same token, ­these shifts worked to the advantage of colleges in the South and West, which would find more and more potential matriculants close at hand.

86 C onte x t

Racial and Ethnic Diversity

An altogether separate set of demographic forces reshaped the demand for college in a dif­fer­ent way, literally transforming the face of American higher education. Whereas the fluctuations in numbers and the geographic trends described above w ­ ere largely invisible, the changes in composition by race, ethnicity, and national origin over ­these four de­cades ­were plainly vis­i­ble. Both demography and policy ­were at work. Demographic forces featured waves of immigrants and differing rates of population growth, spurring marked shifts in the racial and ethnic makeup of the college-­going population. The major policy change was the dismantling of state-­sanctioned racial segregation of the South and an end to much of the explicit racial discrimination in college admissions nationwide. The first of ­these changes, in the racial composition of college students, is easily observed, though comparisons over time are complicated ­because categories used in surveys have evolved over time. One way of appreciating the changes that have occurred over time in the racial composition of potential college students, while using up-­to-­date classifications, is to compare the racial / ethnic composition of dif­fer­ent age cohorts using the same con­temporary set of categories. This approach is imperfect, of course, to the extent that the composition of older cohorts naturally has been changed over time by immigration and mortality. Bearing ­these complications in mind, Figure 4.4 shows the racial and ethnic breakdown of three cohorts of Americans, using ­today’s classifications. ­These are the cohorts corresponding to each of the three waves of first-­year students covered by the Freshman Survey sample, but the source of data is dif­fer­ent. Reflecting the changes that occurred over time, t­hese shifting distributions represent the demographic under­pinnings of the rising share of nonwhites in t­ hese cohorts of college students.19 The figure’s loudest message is that the percentage of non-­Hispanic whites (the left-­out category) has declined over time. Whereas non-­ Hispanic whites accounted for 74 ­percent of the cohort of Americans

Ou t s i d e F o r c e s

87

45

Share of cohort in 2009 (percent)

40 35 30 25 20 15 10 5 0 Two or more races American Indian alone Asian or Hawaiian, PI alone Hispanic origin Black alone

1972 0.9 0.9 4.3 9.5 11.3

1989 1.2 1.0 6.0 17.2 13.2

2008 2.3 1.3 4.1 18.1 16.0

Figure 4.4. ​Racial and Ethnic Makeup of Three Cohorts Source: U.S. Census Bureau, Statistical Abstract of the United States: 2012–13, ­Table 11, 14–15. Hispanic origin is considered an ethnicity, not a race. Components may sum to more than total nonwhite ­because nonwhite Hispanic origin is counted twice.

who turned eigh­teen in 1972, that group made up only 60 ­percent of the cohort that turned eigh­teen in 2008. Among the minority groups that ­were part of this displacement of non-­Hispanic whites, the fastest growing ­were Hispanics / Latinos, whose share almost doubled between the 1972 cohort and the 2008 cohort, from 9.5  ­percent to 18.1 ­percent. Over this period, Hispanic overtook African American as the largest minority group, though the percentage of African Americans did increase across ­these cohorts, from 11.3  ­percent to 16.0 ­percent. Asians remained in all the cohorts the third-largest minority group, rising and then falling in share.

88 C onte x t

Besides demography, the other force affecting racial patterns in colleges was the demise of racial segregation in all, or nearly all, of American higher education. Owing to the strict racial segregation laws and practices in the former states of the Confederacy, virtually no colleges in that region had more than a slightest hint of interracial enrollment at the undergraduate level u ­ ntil the late 1960s. Before the 1954 Brown v. Board of Education decision, colleges in the South and border states w ­ ere strictly segregated, except where a few court cases had opened small cracks in Jim Crow segregation in gradu­ate programs.20 ­A fter Brown, the border states moved to desegregate their public university systems, but the states of the former Confederacy mostly stalled ­until the mid-1960s.21 By 1976 white students in state-­supported colleges and universities in half of the states in the South ­were on campuses where the percentage of nonwhites was at least half what it would have been if students had been distributed evenly across all campuses. By and large, private colleges and universities in the South followed suit. Yet one component of racial segregation stubbornly persisted, in the form of the historically black colleges and universities. Created to serve black students in the era of state-­sanctioned segregation, they continued, and still continue, by and large, to serve a predominantly black clientele, with predictable results for overall levels of racial segregation in higher education, both public and private. To see how t­ hese changes in racial mix and segregation policies affected the diversity of the country’s four-­year colleges, the series of bars in Figure 4.5 represents the shares for four groups of race / ethnicity groups of American students, along with the share of foreign students. To show the divisions more clearly, HBCUs are not included in the figure. The most striking aspect of the figure is the marked increase in e­ very category of colleges in the share of students who ­were nonwhite or international.22 In 1972 not one of the categories of colleges averaged as much as 15 ­percent nonwhite and international. By 2008 / 09 no group had less than 15 ­percent. Each of the 15 groups of predominantly white colleges saw increases over the nearly four-­decade period, with the increases ranging from

Ou t s i d e F o r c e s

89

25–50

0–25

72 90 09

50–60

72 90 09

72 90 09

70–80 Private

80–85

72 90 09

72 90 09

90–95

95–98

Hispanic (not international)

85–90

72 90 09

Other/American Indian/2 or more races (non-international)

Black (not international)

60–70

72 90 09

72 90 09

72 90 09

99+

0–50

72 90 09

International (non-citizen)

Asian (not international)

98–99

72 90 09

80–90 Public

50–80

72 90 09

90+

Source: Freshman Survey, 1972, 1989, 1990, 2008, and 2009 for selected colleges. Note: Bars show the percentage of first-­year students who identified themselves, respectively, as black or African American, Hispanic, Asian, other race, or not a citizen or permanent resident. For each college category, the three bars refer to the three survey waves: 1972, 1989/90, and 2008/09.

Figure 4.5. ​Racial, Ethnic, and National Makeup of First-­Year Students

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

72 90 09

0.45

72 90 09

0.50

72 90 09

about 9 percentage points (among the lowest-­ranked public institutions) to more than 30  percentage points (among the top-­ranked private and public institutions). Among the racial groups contributing to this rise in nonwhite percentage, none has greater historical importance that African Americans. By 1972 most predominantly white colleges and universities across the country, private as well as public, had black students. Actually, by 1972 most of the increase in the share of black students in four-­year institutions that would occur had already been achieved, with surprisingly ­little increase taking place ­after 1972. The enrollment-­weighted average p ­ ercent black for all students in four-­year colleges using figures for the 188 colleges in the sample, excluding international students, ­rose from 7.1  ­percent in 1972 to 9.2 ­percent in 1989 / 90 and then declined to 7.8 ­percent in 2008 / 09. Excluding HBCUs, the percentage black was 4.0  ­percent in 1972, 6.1 ­percent in 1989 / 90, and 5.0 ­percent in 2008 / 09.23 In contrast to African Americans, students in two other r­ acial / ethnic categories markedly increased their shares among college students, and this growth is evident in our sample. Hispanic students grew as a share of all four-­year college students. Whereas they made up less than 1 ­percent of students in 1972, their share increased over the next two de­cades, more than doubling in most categories of colleges. In our sample, the increases ­were especially large for three categories: the ten private institutions in the 50th to 60th percentile range (this group included two in Texas and another in California), the five public universities in the 80th to 90th percentile group that included UC Santa Barbara, and the highest-­ranking public institutions (which included three ser­vice academies and UC Santa Cruz). As remarkable as ­these increases certainly are, the shares of black and Hispanic students in four-­year colleges did lag b ­ ehind the shares of the corresponding population cohorts who w ­ ere black or Hispanic, reflecting their comparatively lower rates of college enrollment.24 In 2011, for example, when 43 ­percent of whites aged eigh­teen to twenty-­four ­were enrolled in college (two-­or four-­year), the corresponding percentages ­were 38  ­percent for blacks and 32  ­percent

Ou t s i d e F o r c e s

91

for Hispanics. For Asians, by contrast, it was 64 ­percent.25 Among the highlighted groups, Asian students increased at remarkable rates. In 1972 they ­were few and far between among the first-­year students in our sample, accounting for less than 1 ­percent of all students in all but one group. Two de­cades ­later, ­these shares had increased across the board, reaching 8 ­percent in the top-­SAT private and public institutions. And the shares increased again over the next two de­cades. By 2008 / 09 Asian students made up 15 ­percent of the students at the private and public institutions in the highest SAT categories, as defined by 1970-­era average SATs. Another noteworthy pattern is the growing importance of the fourth category, which includes two or more races. By 2008 / 09 students identifying themselves in this category ­were outnumbering ­those identifying themselves only as black. One impor­tant driver of diversity in several of the college categories was an increase in the share of students from other countries. Although for most categories of colleges foreign students began and remained a small share of the total, especially in the bulk of public institutions, among private colleges inhabiting the top decile of 1970 scores foreign students accounted for a growing share, rising to more than 8 ­percent in 2008 / 09 for private colleges at the 98th percentile and above. In public institutions, by contrast, the shares of foreign students remained low. A dimension of diversity imperfectly explained by changes in categories such as ­these is the degree to which En­glish was not students’ native language. Based on our Freshman Survey data, En­glish was not the native language for some 4  ­percent of first-­year students in 1998 / 99. In 2008 / 09 that share had increased to 5 ­percent. But this share differed widely across college types, as Figure  4.6 makes plain.26 In 1989 / 90 public institutions below the 80th percentile and HBCUs had below-­ average shares of such students, whereas private colleges above the 95th and public ones above the 90th all exceeded the national average by 4 percentage points or more. By 2008 / 09 t­hese gaps had widened considerably, with the top-­ranked public and private colleges having shares that exceeded the U.S. average (4.8  ­percent) by 10  percentage points or more.

92 C onte x t

0.08 0.06 0.04 0.02 0.00 −0.02

0.05

0.06

0.07

0.05

0.07

0.10

0.11

Priv

90+

80–90

Pub −0.03 −0.01

0.03

0.00

−0.03 −0.02

0.01

0.01

0.07

0.04

0.03

0.11

0.01

0.03

0.01

0.01

0.04

HBCU

0.04

0.03

0.00

−0.02 −0.02

0.01

2008/09

−0.02 −0.01

1989/90

0.00

Public

−0.01

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

−0.04

0–25

Group mean minus U.S. average

0.12 0.10

Figure 4.6. ​En­glish Not Their Native Language Source: Freshman Survey, 1989, 1990, 2008, and 2009 for selected colleges. Note: The survey question was, “Is En­glish your native language?” The figure shows the difference between the proportion of students in the category who answered no and the enrollment-­weighted mean for all first-­year students. The enrollment-­weighted mean for all first-­year students in four-­year colleges w ­ ere 0.039 for 1989 / 90 and 0.048 for 2008 / 09.

­ hese high shares of students who did not grow up in English-­ T speaking families are only partly a function of the rise in international students, in that the shares of non-­native En­glish speakers exceed that of international students. For example, in private colleges in the 50th to 60th SAT percentiles, some 8.7 ­percent of first-­ year students said En­glish was not their native language, yet only 2  ­percent of students ­were foreign. The biggest difference between ­these two shares was in the 90th percentile public colleges, where 15.3 ­percent spoke another language as a child, but only 3 ­percent ­were foreign students. We see in t­ hese numbers the high share of immigrants who have gravitated to highly ranked public universities that offer strong STEM programs. Taken together, t­ hese shifts in race / ethnicity produced a sea-­ change in diversity among private colleges and the top public ones.

Ou t s i d e F o r c e s

93

By 2008 / 09 the percentage nonwhite (including foreign nationals) exceeded 30 ­percent in the top three groups of private colleges, the top group of publics, and, as one would expect, in the HBCUs, which began and remained almost entirely nonwhite. Whites remain a majority, but their numerical predominance has waned. This growing diversity, driven by a historic civil rights movement and high rates of immigration, suggests opened doors, wider access, and new opportunities. Yet that vision of a new meritocracy is, at best, too simplistic and, at worst, a mirage, for it must be reconciled with seemingly contradictory facts. Coeducation and the Rise of W ­ omen

The other notable demographic change affecting the demand for college is the stunning rise of ­women from minority to majority among college students and gradu­ates. Historically underrepresented in colleges, ­women made up just 19  ­percent of ­those receiving bachelor’s degrees in 1900. Their share grew steadily through the ­century, though, rising to 24 ­percent in 1950 and to 43 ­percent in 1970. In 1982, w ­ omen surpassed men for the first time, and their share r­ ose to 57 ­percent by the end of the ­century and remained ­there through the first de­cade of the twenty-­fi rst c­ entury. 27 By 2011, 44.9 ­percent of ­women aged eigh­teen to twenty-­four w ­ ere enrolled in college, compared to only 39.1 ­percent for men.28 In this rise to predominance in college enrollment, American ­women are similar to ­those in most advanced nations. Although the reasons b ­ ehind this thoroughgoing change are still a ­matter of debate, it has been noted that such a reversal of a longstanding social disparity is a rare event.29 A related, longstanding trend was the decline in single-­sex colleges. During the 1960s the number of all-­male colleges fell by a third and the number of all-­female colleges fell by a quarter. 30 The trend continued into the early 1970s. Between 1969 and 1972 such storied institutions as Prince­ton, Yale, ­Virginia, Williams, Brown, Dartmouth, and Davidson went coed. 31 ­T hese trends continued ­after 1972, especially among private colleges in the highest SAT cat-

94 C onte x t

egories, several of which had been all-­male in the 1960s and two that ­were all-­male in 1972, Amherst and Haverford. Between 1972 and 1989 / 90, for example, the share of ­women in the most selective private category among the sample of 188 colleges doubled, from 27 ­percent to 54 ­percent.32 Technological Advances

In the 1973 movie American Graffiti, actor Richard Dreyfuss plays Curt, a recent high school gradu­ate living in the San Joaquin Valley of California in the early 1960s. In the movie’s final minutes he boards an airplane en route to the East Coast, where he w ­ ill attend college. For most American high school students during the ­couple of de­cades ­after World War II, the prospect of attending a college 3,000 miles away from home would have been unthinkable. But, as propellers gave way to jet engines and expensive long-­distance calls ­were replaced by cheap long distance and Skype, the barriers of distance receded over time. The fictional Curt represents one stage in a longer progression that has witnessed a steady expansion of the geographic reach of colleges, especially for the rich and influential ones. For the generation before Curt’s, this widening reach could be represented by the bright, nonfictional high school se­nior from Illinois, James Tobin, who would attend Harvard College and go on to win the Nobel Prize in economics. Although it is easy in the twenty-­first c­ entury to take them for granted, developments in transportation and revolutionary ones in communications constitute a third consequential trend affecting the market for college. On the demand side, ­these developments had the effect of diminishing the significance of distance. Paving the way for this expanding geo­graph­i­cal reach ­were forces of public policy, technology, and economics. The Interstate Highway System, authorized in 1956, was nearly complete by 1980, totaling some 40,000 miles all together, making automobile travel faster and safer. 33 Spurred by t­ hese highways, the number of passenger miles in automobiles increased by an astounding 53 ­percent in just one de­cade,

Ou t s i d e F o r c e s

95

between 1960 and 1970. Over that de­cade the miles traveled by the average American by car, train, and airplane increased from 6,628 to 9,145.34 And the speed of commercial air travel increased, as jet engines replaced propellers, cutting the flight time from New York to San Francisco, for example, from eleven and a half to six and a half hours.35 Technological change came as well to communications, with a vengeance. As difficult as it might be to imagine t­ oday, it was not uncommon in 1970 for students away at college to write letters, with stamps, to communicate with parents and sweethearts back home. In ­those days a ten-­minute phone call in the lower forty-­eight states could cost as much as $29 (in 2010 dollars). But technological change and deregulation combined to drive down the cost of long-­ distance phone calls. Adjusting for inflation, the cost of coast-­to-­ coast calls fell by more than 85 ­percent between 1970 and 1990. By 2010, the cost was just a tenth of what it had been forty years earlier.36 Even more dramatic was the effect of computers and email on the ease and cost of long-­distance communication. By way of ­these developments in transportation and communications, the power of distance to influence the college choices of students and their families inevitably receded. As automobiles expanded the market areas of retail grocers, allowing supermarkets to replace corner stores, ­these late twentieth-­century technological advances also allowed colleges to expand their geographic reach. With the drag of distance slowly receding and information more readily available, students like Curt could seriously consider colleges once thought to be too far away from home. The result was a gradual reshaping of the nation’s market for college. Students with the best academic credentials could look beyond their state and region in hopes of attending one of the nation’s most selective colleges. And, on the strength of that kind of interest, ­those selective colleges could in turn become pickier in their admissions decisions. As less selective colleges lost their hold over top students living nearby, ­t hese colleges increasingly had to replace them with students a notch or two down the academic ladder.

96 C onte x t

Government Support

The fourth outside force on the college market was substantial financial support from government, both federal and state. The point has been made that nearly all of American colleges and universities came into being and grew up outside the ambit of the national government. But that does not mean the federal government was uninvolved. Besides the nineteenth-­century Morrill Acts, which established the land-­g rant universities, still ­today a prominent part of American higher education, the post–­World War II G.I. Bill sent hundreds of thousands of veterans to college who surely would not have attended other­wise. Add to that the post-­Sputnik National Defense Education Act of 1958 and the Higher Education Act of 1965, legislation that provided for grants to colleges and universities as well as guaranteed loans to students, and it adds up to a rather muscular federal role.37 As impor­tant as ­these federal programs ­were, state support of public colleges and universities was prob­a bly even more significant in opening up opportunities to attend college. In return for generous support from state legislatures, public colleges and universities enticed in-­state students with low tuitions. Together, ­these forms of government support combined to push rates of college attendance and degree attainment to the top of the international charts.38 In the 1970s the Congress loudly announced that it intended to be a potent force in helping students find the money to pay for college. It did this first by passing in 1972 amendments to the Higher Education Act of 1965, new legislation that established a financial aid program in the form of a portable voucher that low-­income students could use at almost any college where they chose to enroll. 39 In 1978 Congress made t­ hese scholarships more generous and broadened their coverage to include students with higher incomes. This new legislation also lifted the income cap on families eligible to receive subsidized federal guaranteed loans.40 As a ­whole, federal aid to college students grew rapidly, doubling over the de­cade of the 1970s.41 Significantly, loans became a progressively bigger part, and grants a smaller part, of this federal aid. In 1975 federal grants w ­ ere

Ou t s i d e F o r c e s

97

four times as large as federal funds for loans. By 1981 loans w ­ ere just as large as grants. And by 1997 grants had shrunk to less than a third of the amount devoted to loans.42 At the heart of the student financial aid system, funded in part by federal programs, lay a methodology for determining need that may have seemed to f­ avor private colleges. Financial need was defined, roughly, as the difference between the cost of attending the college chosen by the student and his or her ­family’s “expected contribution,” a precisely calculated amount resembling a steeply progressive tax on income and wealth.43 Since private colleges invariably had higher tuitions than public ones, the average amount of aid tended to be higher for students attending them. In 1987 Congress mandated the method by which colleges would make ­these calculations in order to be eligible to receive federal funds.44 In the 1990s the federal government added tax credits to its arsenal of student financial aid policies, with two tax credits enacted in the mid-1990s. ­Because they ­were not refundable, they w ­ ere generally useful only to taxpayers with middle-­class incomes and above.45 Meanwhile Congress allowed the maximum Pell Grant award, which had not been indexed for inflation, to dwindle in real value, falling by more than a third between 1977 and 1993. But the legislators then reversed course and increased the top award, pushing it by 2015 nearly to its 1977 inflation-­adjusted value.46 By 2015 federal support for student aid was chiefly in the form of loans, which together accounted for almost 60 ­percent of federal spending. About 28 ­percent of federal aid was in the form of grants, most of it Pell Grants, another 11 ­percent was from vari­ous tax benefits, and work-­study expenditures weighed in at 1 ­percent of the total.47 Student aid was by no means the only form of federal support for colleges and universities, of course, though it was prob­ably the most impor­tant form of support affecting undergraduate education. In addition to the student aid programs, the federal government supported the research done at universities through a variety of ave­nues, providing in total about 60 ­percent of the research and development funds for universities in 2008.48

98 C onte x t

As impor­tant as the federal government had been in supporting college attendance, states ­were arguably more impor­tant. The states had established large systems of public higher education beginning in the nineteenth c­ entury. State support mostly took the form of unrestricted, general appropriations, to be spent by institutions ­under the watchful eyes of the vari­ous state governing boards. Like the federal government’s student aid programs, state support had the effect of making college more affordable for students, but the mechanism for achieving that was quite dif­fer­ent. Whereas the federal programs helped students and their families by way of explicit subsidies to defray the cost of attendance, state support mostly operated invisibly. What amounted to a social contract was at work. State legislatures showered their colleges and universities with appropriations. In return ­those institutions would allow in-­state students to attend at bargain-­basement prices. At the University of California, in-­state students who w ­ ere admitted could attend the world-­class Berkeley campus for $960 a year in tuition and fees in 1970 ($5,860 in 2015 dollars).49 It was even cheaper to attend the University of North Carolina at Chapel Hill, where in-­state students paid only $398 ($2,430 in 2015 dollars) in tuition and fees. In the case of North Carolina, the low-­cost part of the social contract was literally enshrined in the state’s constitution, which stated that its state colleges should, “as far as practicable,” be f­ ree. 50 Alongside this tradition of low tuition was a reluctance to offer need-­based financial aid, perhaps based on the reasoning that the cut-­rate tuition was assistance enough. In comparison to the vari­ous federal student aid programs, state policies ­were markedly less redistributional. The low-­tuition approach taken by states, despite its egalitarian appearance, did ­little to aid low-­income families, whose ­children ­were often academically ineligible to attend the heavi­ly subsidized flagship universities, a point made by Hansen and Weisbrod in 1969. 51 What­ever its distributional effects, this social contract appeared to fray in the late 1980s, as state support began to falter. Combined state appropriations for higher education, which amounted to about $8.30 per $1,000 of personal income in 1986, had fallen to $6.60 per

Ou t s i d e F o r c e s

99

$10,000 $9,000 $8,000 $7,000 $6,000 $5,000 $4,000 $3,000 $2,000 $1,000 $0

19 7 19 1 7 19 3 75 19 7 19 7 7 19 9 81 19 8 19 3 85 19 8 19 7 89 19 9 19 1 9 19 3 95 19 9 19 7 99 20 0 20 1 0 20 3 05 20 0 20 7 09 20 11

Revenue per FTE in 2013 constant dollars

$1,000 by 1998. The likeliest explanation for this decline in support was the growing cost of other demands on state bud­gets, especially Medicaid. Between 1988 and 1995, Medicaid expenditures increased as a percentage of gross state product from about 4.5 ­percent to 8 ­percent. 52 To compensate for the stagnation in state appropriations, public colleges and universities began to increase their tuition rates. As shown in Figure 4.7, the average inflation-­adjusted tuition charged by all public colleges and universities began a steady ascent beginning around 1980. Over the ensuing three de­cades, this average tuition tripled. For most of this period of tuition increase, at least two de­cades ­after 1980, state appropriations per student ­were on a roller-­coaster but one with no long-­term trend e­ ither up or down. ­A fter 2001, however, ­these appropriations began to fall. Between

Academic year All public tuition and fees

All public state appropriations

4-yr public tuition and fees

4-yr public state appropriations

Figure 4.7. ​State Appropriations vs. Tuition at Public Colleges and Universities, All Public Institutions, Four-­Year Public Institutions, 1971–2012 (2013 dollars per full-­t ime equivalent [FTE] enrollment) Source: U.S. Office of Education, Digest of Education Statistics, 1970–71, ­Table 121; 1987, T ­ able 270; 1995, ­Table 319; 1996, ­Table 322; 2000, ­Table 328; 2004, ­Table 332; 2005, T ­ able 330; 2009, ­Table 352; 2013, ­Table 333.10. Note: Data are for academic years denoted by ending year; that is, 2012 is 2011 / 12.

100 C onte x t

2001 and 2004, average appropriations per FTE student fell by more than 15  ­percent, holding steady at this new, lower level ­u ntil the ­Great Recession. But ­a fter 2008, a second decline began. By 2012, appropriations per student at all public institutions had fallen by an additional 23 ­percent, and by 25 ­percent at four-­year ones. By 2011 tuition had become a bigger source of revenue for public four-­year institutions than state appropriations, and by 2012 this was true for all public institutions taken together. For all public institutions, state appropriations in 1971 had been three times as impor­tant as tuition as a source of revenue. By 2012 appropriations had become less impor­tant than tuition. Although most states stuck with the practice of avoiding programs of explicit financial aid, some states did institute new scholarship programs. A few of t­ hese, such as Chapel Hill’s Carolina Covenant and ­Virginia’s Access UVa, ­were aimed at low-­income, first-­generation college students. 53 But by far the larger tendency was for states to institute state-­financed merit (no-­need) scholarship programs. The granddaddy of them all was Georgia’s HOPE scholarships, launched in 1993. It offered to students in the state who achieved a B average or better in high school f­ ree tuition to any state institution for as long as the student maintained a B average in college. The program was not only popu­lar with voters, it was an extraordinary success in boosting rates of college enrollment for Georgians and increasing the number of high-­achieving students choosing to attend the University of Georgia and Georgia Tech. 54 By 2012 fifteen states had ­a dopted merit scholarship programs along the lines of Georgia’s HOPE scholarship. ­T hese state merit scholarship programs appeared largely to benefit middle-­class ­children, ­those g­ oing to college and especially t­hose attending the most heavi­ly subsidized ones. In the case of Georgia, the lack of redistributional effect was exacerbated when, in the program’s first two years, the original income cap of $80,000 for eligibility was increased and then simply eliminated. To make m ­ atters even less pro-­poor, the program was paid for by earmarked revenues from the state’s lottery, certainly one of the most regressive sources of state revenue.55

Ou t s i d e F o r c e s

101

Over the four de­cades covered by this study, government remained a potent force on the market for college. At the federal level, an ever-­ changing collection of programs amounted to what the Office of Management and Bud­get (OMB) described in 1985 as “a shotgun approach that has indiscriminately sprayed assistance at students regardless of income.”56 Loans assumed an ever-­increasing share of all aid, and tax subsidies became a new fixture in the total effort. Through it all, the one policy aimed at giving low-­income students access to college, the Pell Grant, held its own but did not expand. For the public colleges and universities run by the states, appropriations from the respective legislatures became less impor­tant while tuition became more so. But ­these tuition increases failed to compensate fully for dwindling state appropriations. The predictable consequence was to weaken public institutions at a time when leading private institutions ­were thriving. And, in apparent emulation of the many private colleges that had expanded their merit aid, in part to attract and retain good students and t­ hose who could afford the tuition, a number of states instituted their own merit aid programs, as we have seen. A few states introduced new need-­based scholarships, as noted above, but t­hose w ­ ere small relative to the new no-­need scholarships. The Rankings Phenomenon

A fifth outside force on Amer­i­ca’s college marketplace burst forth from the staid and unassuming confines of the country’s third-­ ranked national weekly news magazine, U.S. News and World Report. In 1983, that magazine deci­ded to devote a portion of one of its weekly issues to a ranking of U.S. colleges. Few decisions in modern magazine journalism have proven to be more successful or lucrative. Following a format that had occasionally worked in other applications for magazines in the past, they deci­ded to build this feature around a comprehensive numerical ranking of the nation’s four-­year colleges and universities, reasoning that this format would appeal to readers who, in other contexts, evidently liked to see how college

102 C onte x t

football teams ranked in the opinions of sports writers or coaches (on newspaper sports pages and Sports Illustrated), how dif­fer­ent brands of vacuum cleaners performed (on the pages of Consumer Reports), or how presidential candidates stacked up in terms of likely electoral vote totals (in the pages of their own weekly magazine). ­Unless the magazine’s leadership could have seen into the ­future, “success beyond their wildest dreams” is the only way to describe the outcome for U.S. News. The first rankings, published in 1983 and 1985, w ­ ere reputational, relying entirely on responses from college officials across the country surveyed by the magazine. In 1987, ­under the direction of newly hired Robert Morse, the magazine began adding quantitative information and reducing the weight given to its reputational surveys to 25 ­percent. In 1990 the feature became its own free-­standing issue, immodestly entitled, Amer­i­ca’s Best Colleges. It became an instant bestseller. The rankings business soon outshone the magazine. By 2010 the stately old weekly magazine had folded, victim of the financial pressures besetting print journalism, including competition from cable tele­vi­sion and the internet, leaving the college ranking issue as a standalone business. 57 Surely few ­people in 1983, let alone 1970, could have predicted how impor­tant this surprisingly good business decision would become in the market for college in the United States What began as a moderately expensive experiment designed to boost sales by a struggling national news magazine quickly morphed not only into a healthy profit center but also an annual news story, feverishly anticipated at hundreds of colleges, energetically discussed at bridge clubs and church socials, and eagerly scrutinized by high school se­niors and their parents. To explain this nearly overnight rise to national prominence, one must point to at least four ­factors. First, the U.S. News ranking was the first organ­ization to provide numeric assessments in place of the rich but messy collection of facts and impressions jammed into the pages of existing college guides like Lovejoy’s College Guide and Cass and Birnbaum’s Comparative Guide to American Colleges.58 ­Because ­these new rankings ­were numeric—­the magazine took the trou­ble to

Ou t s i d e F o r c e s

103

be transparent about their calculation—­the rankings could reasonably be judged to be closer to the “objective” side of the subjectivity scale than the traditional college guides. A second ­factor that explains the ranking’s success, is, ironically, its changeability. A third is that consumers paid attention. As the popularity of the annual issue makes clear, students and their parents bought this publication and took it to heart. Fourth, the ranking just happened to jibe with the sort of positional competition that underlay the rivalries that existing in the vari­ous pockets of the nation’s college market. At a nuts-­and-­bolts level, the U.S. News ranking is nothing more than a weighted average of vari­ous quantifiable items. All that is necessary to turn all this into a ranking is a list of items that are widely understood and easily collected, all expressed as numbers. To boil ­these dif­fer­ent mea­sures to one single number—­the necessary ingredient for the one-­dimensional ranking—­requires just one more heroic but arbitrary decision: how much weight to assign to each of t­ hose mea­sures. To illustrate how U.S. News has done this, consider the methodology they used for the 1997 edition, which was, as it was in ­every year, clearly spelled out for anyone to see who had an inclination to go into the details. In that year the ranking was based on slightly more than a dozen dif­fer­ent mea­sures. 59 Among t­ hese, the mea­sure to which the designers assigned the biggest weight was, interestingly, a subjective evaluation, based on a compilation of several thousand survey forms sent out to presidents, deans, and admissions officers, each of whom w ­ ere asked to rate a group of colleges deemed similar to their own. To gather the remaining data needed to calculate the ranking, U.S. News sent out a survey to e­ very college, complete with detailed instructions. Since, by 1990, most colleges had officers or offices whose job it was to collect such data and fill out such surveys, regularly filling out forms for government agencies, accrediting agencies, and other regulatory bodies, the U.S. News operation took advantage of easily gathered information, collected for f­ ree. Over the years U.S. News devised a number of checks on the accuracy of the numbers the institutions submitted.60 Other items collected included mea­sures of selectivity (ac­cep­tance rate,

104 C onte x t

yield rate, and the SAT scores and high school grades of entering students), information on faculty pay and degrees held, institutional expenditures, student retention, and alumni giving.61 ­After objectivity, the second ingredient in the ranking’s success has been, ironically, its changeability. Due almost entirely to tweaks in its methodology from one year to the next, each year’s ranking inevitably looked a bit dif­fer­ent from the year before. A weight is adjusted; a mea­sure is slightly redefined; the result is a set of mostly minor changes in the ranks of individual colleges. Sometimes ­these tweaks occurred a­ fter colleges complained that this or that aspect was unfair or could be improved, but the pro­cess by which the U.S. News editors adjusted the formula was by no means as transparent as the resulting formula itself is. What­ever the mechanism or motivation ­behind them, ­these subtle changes had the effect of rendering last year’s ranking outdated, much as the edits a textbook author makes to differentiate the newest edition from the previous one. In addition to the changes themselves, the prospect of shifts in rank added drama to the anticipation leading up to each year’s unveiling of the newest set of rankings. As arbitrary as the choice of mea­sures and certainly the mathematical weights ultimately are, year-­to-­year changes in them seem doubly so, or at least they highlight the almost whimsical aspect of numeric rankings. Seemingly small adjustments in the formula could lead to surprisingly large readjustments in rank that bore no apparent connection to a­ ctual changes in a college’s quality. For example, in 1994 Berea College, a college in the mountains of Kentucky known for its long history of racial openness and its requirement that students cover school costs through paid employment, found itself moved to a dif­fer­ent category in the U.S. News ranking. Whereas it had been rated the third-­best regional university in the South, it became the next year an unranked, third-­ t ier national university, simply ­because the Car­ne­g ie Foundation, the ultimate arbiter of institutional status, had reclassified the college. 62 In another instance, Georgetown dropped from seventeenth to twenty-­fifth place in the nation in just one year, between 1993 and 1994, all ­because U.S. News

Ou t s i d e F o r c e s

105

changed the mea­sure it used to reflect the level of faculty compensation. What had previously been average compensation for full professors became average compensation for all ranks, and t­ hese figures ­were adjusted for cost of living, which together dropped Georgetown fifty-­four places in the ranking for “faculty resources.”63 The third reason the U.S. News ranking assumed importance is perhaps the most obvious. As illustrated by the issue’s healthy sales volume, ­those in the market for college ­were paying attention. Two statistical studies provide direct evidence strongly suggesting that the ranking’s influence went beyond making applicants aware of colleges, but that the rank assigned to a college m ­ atters. One, using data for a single college, showed that students ­were more likely to apply to colleges that had been ranked highly. 64 Another showed that colleges moved onto the first page of the U.S. News issue enjoyed a significant bump in applications.65 Indirect evidence of an effect is provided by a study by James Monks and Ronald Ehrenberg, who found that colleges that had experienced a drop in rank ended up with higher ac­cep­tance rates and lower yield rates.66 The fourth, and possibly the most impor­tant, explanation for the U.S. News formula was that it fit right in with colleges’ instincts to compete for the attention of potential students by demonstrating that they are better than their rivals, and to do this in as public a way as pos­si­ble. Some scholars have theorized that intercollegiate athletics is a proxy for carry­ing out this kind of head-­to-­head competition. In college rankings colleges found themselves in an equally unambiguous contest, one having a clear winner ­every time except for the occasional tie. This competition was based on academic, not athletic, merit. But college leaders w ­ ere no more willing to embrace a high U.S. News ranking as an accurate mea­sure of their quality than they would be to use its football team’s win-­loss rec­ord for the same purpose.67 Indeed, it became de rigueur for presidents and provosts of colleges with lofty U.S. News ranks to pooh-­pooh the entire rankings enterprise as unworthy of their attention. Yet most of ­those same administrators no doubt retired to the privacy of their offices or conference rooms, filled with a certain quiet satisfaction.

106 C onte x t

In contrast to this official disdain from leaders at highly ranked colleges, many of the striving colleges across the land seemed all too willing to boast about good rankings. One need only look at a few of their websites for proof. Perhaps the most stunning aspect of the U.S. News phenomenon is the rapidity with which rankings, and particularly this one, became a force in the college marketplace. Their popularity among consumers was one ­thing. What was most remarkable was how the rankings became for college leaders both reviled and irresistible. As it had been with the loyalty oaths of the 1950s, leaders came to detest and denounce the U.S. News rankings, but most of them could not afford to go so far as to refuse to participate. Several colleges did in fact try to do just that but in the end they had to relent. One telling case was Sarah Lawrence, a highly selective former Seven ­Sisters college in Westchester County, New York, just an hour by train from ­Grand Central Terminal. Known for the time and attention its faculty devoted to students’ writing, the college deci­ded in 2006 not only to stop requiring applicants to submit SAT scores—­a step already taken by a number of colleges—­but also to stop looking at the scores at all.68 They literally tossed the score reports into the paper shredder. ­A fter it became clear to U.S. News that Sarah Lawrence was no longer submitting SAT scores as a part of its annual survey, the magazine dealt with this prob­lem by devising an arbitrary replacement for the missing data. It deci­ded to calculate the college’s rank by assigning it an SAT score set some 200 points lower than the college’s erstwhile peers.69 Viewing this response as unfairly punitive, the college’s president complained on the op-ed page of the Washington Post, “The real­ity is that the magazine’s rankings issue has a large circulation and that parents and students rely on ­these rankings to make a college choice that has enormous educational and financial implications. This gives the magazine the power to keep colleges playing the game it sets and controls.”70 But, ­a fter a brief standoff, Sarah Lawrence was the first one to blink. Following two years of double-­digit percentage drops in numbers of applications, the college relented, resuming the practice of accepting, and reporting,

Ou t s i d e F o r c e s

107

SAT scores.71 Among the outside forces shaping the evolution of the college marketplace, few had a more vis­i­ble effect on college practices, as discussed in Chapter 6. Although U.S. News retained its place of prominence from its debut in the 1980s, it did not go unchallenged. In 2005 Washington Monthly launched a competing product. Rejecting the U.S. News approach as biased ­toward wealthy private colleges, it offered a formula that was radically dif­fer­ent. Eschewing mea­sures based on inputs—­l ike the SAT scores of entering students, the salaries of faculty, and the faculty-­student ratio—­t hey devised an approach that attempted to assess outcomes, broadly defined. This approach recognized the share of students who w ­ ere not from affluent families, the share who participated in ROTC or who went on to serve in the Peace Corps or get PhDs. Using a statistical model to adjust for the income level of entering students, the ranking also mea­sured graduation rates and the earnings of gradu­ates. It also added mea­sures reflecting the research done by faculty. The result was a ranking of national universities much friendlier to public institutions. Whereas the 2016 U.S. News top-­t wenty list of national universities included only one university that could be classified as public (Cornell), the Washington Monthly list contained eleven. On top of its ranking of liberal arts colleges was not Amherst or Williams (though both did very well), but Berea, the college in Kentucky famous for educating needy students who help pay the bills by working on campus.72 Belying their cloistered courtyards and hushed reading rooms, the nation’s colleges w ­ ere by no means insulated from changes that ­were occurring all around them. No market as vast as the one for baccalaureate education could be. The biggest force influencing the college market was the massive, if glacial, shift in the nation’s distribution of income. Not only did it ratchet up the financial payoff to getting a college degree, it also moved money around in ways that would come to have far-­reaching effects. A second set of tectonic shifts, barely discernible in real time, w ­ ere demographic. They helped to create a buyers’ market for college places, especially in the Northeast. They also contributed to college classrooms that w ­ ere, like the

108 C onte x t

nation as a w ­ hole, becoming more and more racially and ethnically diverse. Technological change was a third major outside force, drawing the nation closer together and its market for college along with it, at least the segment of the market that matched the best applicants with the most selective colleges. Government, both federal and state, was the fourth major outside force, affecting the college market mainly by way of financial aid to undergraduates. From the federal government this meant some need-­based grant aid and a growing amount of loans. From the states it came largely in the form of low tuition for ­those high school gradu­ates who could qualify for public colleges and, in recent de­cades, merit scholarships. The last of the highlighted pressures for change was the meteoric rise of college rankings, particularly U.S. News, as a force to be reckoned with, by consumers seeking to chart their course ­toward ac­cep­t ance and by colleges seeking to position themselves in a highly competitive environment.

Ou t s i d e F o r c e s

109

P A R T II

Supply

5 The In­equality Dividend

A parable in the gospel of Matthew concludes with this verse: “For unto e­ very one that hath s­hall be given, and he s­hall have abundance: but from him that hath not ­shall be taken away even that which he hath.”1 In a 1968 article in Science, sociologist Robert Merton recast this normative teaching into an empirical hypothesis in social science, to explain why eminent scientists tended to receive more than their share of credit for research conducted jointly with ­others. Merton called this tendency the “Matthew effect,” and the term has been used since then to describe the tendency for the rich to get richer.2 Merton’s term aptly describes the fortunes of Amer­i­ ca’s four-­year colleges in the era of growing economic disparity. The richest colleges and universities, most of them private, grew richer; another swath of institutions did all right; but most of the rest strug­gled to maintain their quality or even to stay afloat.

The backdrop for all this was growing income in­equality. As described in Chapter 4, the most affluent ­house­holds saw their incomes increase while most p ­ eople’s earnings stagnated. The share of income received by the most affluent 10 ­percent of ­house­holds in the United States, which had averaged about 33 ­percent between 1942 and 1978, began to rise around 1980. Between the mid-1980s and the late 2000s, incomes in this top 10 ­percent increased at a rate of 1.9 ­percent a year. By contrast, incomes in the bottom 10 ­percent increased at a rate of only 0.5 ­percent a year.3 The concentration at the very top was even more striking: the share of total income claimed by the top 1 ­percent of ­house­holds more than doubled between 1976 and 2011, rising from about 9 ­percent to 20 ­percent.4 Wealth also became much more concentrated at the top, with one estimate, noted in Chapter 4, showing the share of wealth held by the wealthiest 0.1 ­percent of ­house­holds tripling between 1979 and 2012.5 Even as their professors ­were pointing with alarm at ­these trends,6 the country’s wealthiest colleges and universities became inadvertent beneficiaries of this growing in­equality. ­These wealthy, largely private, institutions benefited in two main ways. First, they gained ­because rising incomes at the top of the income distribution boosted the demand for the t­ hing they w ­ ere selling. But this ­t hing called a college education was no garden variety “normal good,” defined by economists as one that ­people want more of when their incomes rise. To an extent that went beyond what was true for such goods as real estate or automobiles, admission to elite colleges became a must-­have prestige good. The prestige emanated in part from the limited number of places available at the most desirable colleges. Although the elite colleges rarely charged what the market would have borne, this condition of unremitting excess demand made it pos­si­ble for ­these colleges to raise tuition at rates consistently higher than the inflation of prices overall. A second way the wealthy colleges benefited was by way of a boom in donations, fueled by t­hese rising incomes at the top and a sustained bull market in stocks. Investors evidently approved of companies that paid gigantic CEO pay packages while skimping on their

114 S upply

blue-collar workers, b ­ ecause such firms made money. T ­ hese profits boosted stock values. Combined with their rising personal incomes, ­these rising stock values put more and more well-­off individuals in a position to make sizable charitable contributions. Like generations of well-­off donors before them, ­t hese high-­dollar philanthropists looked to higher education as a favorite category of nonprofits for bestowing donations. ­These donations, plus extraordinary rates of return, caused college endowments to grow to unpre­ce­dented levels. For the wealthiest colleges, this happy confluence of trends made it pos­si­ble to raise faculty salaries, add new programs, and, in effect, gold-­plate their facilities. Although no one at t­hese colleges spoke out in support of growing in­equality, the colleges themselves w ­ ere being made wealthier by the year as a result of it. Meanwhile, private colleges in the bottom half of the selectivity distribution soldiered on, making do with modest infusions of money. And over on the public side of baccalaureate education, conditions became gradually more dire. State legislatures in e­ very region, faced with increases spending for health care, prisons, and pensions and gripped by growing re­sis­tance to tax increases of any kind, put the brakes on increases in appropriations for higher education. ­These trends served to accentuate the in­equality that was already ­there. As made plain in Chapter  3, any comparison of resources across colleges and universities offering the baccalaureate degree quickly reveals a distinct, if not perfectly defined, hierarchy, stretching from the well-­heeled affluence of a handful of very selective institutions to the relative penury of a vast number of lesser-­ known colleges. Without putting too fine a point on it, the players in this multitiered industry run a wide gamut, all the way from elite institutions that look like country clubs to scrawny colleges facing existential questions. Thanks to the economic forces that played out ­after 1970, ­these gaps widened. For the colleges at the top, this was the in­equality dividend. This chapter examines the effect of the growing income in­ equality on the large, vis­i­ble aspects of the college market, leaving for phi­los­o­phers and psychologists to consider the more subtle,

Th e I n ­e q u a l i t y Divi d e n d

115

perhaps more pernicious consequences. To make sense of the supply-­ side effects, it is necessary to take a closer look at the firms ­themselves—­the colleges—­and what makes them tick. It w ­ ill be especially impor­tant in this context to pay as much attention to the hundreds of nonselective colleges as we are accustomed to devoting to the several dozen elite ones. As we w ­ ill see, the forces unleashed by growing in­equality looked quite dif­fer­ent to the little-­known, poorly funded institutions, both private and public, than they did to the famous colleges that occupied lofty spots in the U.S. News national rankings of colleges. To understand t­ hese disparate effects on the colleges, it is essential to ask a basic question: What drives t­ hese colleges to do what they do? Next we can apply that work­horse model of microeconomics, supply and demand, to explain some of the unusual aspects we observe in this market. We can then turn to data to document the extent of the in­equality dividend as it was enjoyed by colleges at the top, with less favorable consequences to be seen elsewhere. Church or Car Dealership?

To explain the be­hav­ior of colleges, economist Gordon Winston invokes a pair of dueling meta­phors, calling colleges part church and part car dealership. Like the car dealership, ­every college ­faces the pragmatic necessity of generating revenue. It must find customers and sell them something of value. But colleges are not just commercial businesses, he asserts, ­because they are, at heart, also au­then­tic charities, at least the ones that are not legally for-­profit companies. ­These colleges routinely do t­hings that reduce, not increase, their net income. In this re­spect, he argues, colleges are like churches and other charities. They give price breaks to needy students, something no car dealership would countenance. The colleges whose product is most in demand do not charge as high a price as the market would bear, thereby leaving money on the ­table and unrequited customers barred from purchasing their ser­v ices at all. At the most sought-­ after colleges—­Winston no doubt has his beloved Williams in mind—­ students in fact typically receive sizable subsidies, owing to the fact

116 S upply

that they pay less in tuition (net of financial aid) than the cost of the educational ser­vices they receive. Anticipating the wide differences across colleges that are highlighted in this book, Winston created ­tables showing that t­hese subsidies w ­ ere quite generous at the nation’s elite colleges. ­These subsidies are, in his view, the tangible sign of charitable intent. Although Winston’s whimsical hybrid charity model is a reasonable depiction of the aims of some colleges, it is less useful for understanding the motivations driving most colleges and universities. The vast majority of public and other nonprofit colleges in the United States can more usefully be divided into three loosely defined types. The first of t­ hese is in fact Winston’s church-­like college, one willing to behave in ways that do not necessarily make good business sense. This criterion fits a good number of religious colleges, some of them Evangelical Protestant and some of them Roman Catholic. It may also apply to colleges with no religious link but with a mission focus that goes beyond merely appealing to a market niche. By banning alcohol from dormitories or requiring students to take courses in religion, to attend chapel ser­vices, or to have jobs on campus, for example, colleges in this category necessarily limit their attractiveness to large numbers of high school students who have grown up in the highly secular and permissive environments typical of many American communities. Examples of t­ hese colleges in Chapter 2, include The Citadel and Howard Payne University. T ­ hese colleges might more accurately be called mission-­driven, suggesting a primacy of aims having ­little to do with ­either revenue or prestige. As described, this group of colleges is considerably less numerous than what Winston has in mind. For Winston, the practice of giving need-­based financial aid reveals colleges’ aim to promote “income re­distribution and equality of opportunity.”7 Another sign of their charitable motive, he argues, is colleges’ willingness to charge students less than the full cost of education, instead subsidizing them by drawing on endowments, the happy residue of past donations.8 But a more realistic interpretation of such apparent altruism is to see financial aid not as charity but as a tool that colleges use to protect

Th e I n ­e q u a l i t y Divi d e n d

117

their academic reputations. By this reasoning, the seemingly charitable policies cited by Winston mainly serve to bolster a college’s reputation for high academic standards and especially to discredit any notion that admission is open only to ­children of affluent parents. In truth, it is sometimes difficult to discern very much that is charitable about most American colleges and universities other than their government-­certified tax-­e xempt status and the associated privilege of receiving tax-­deductible contributions. That tax deduction, ­after all, was explic­itly granted for gifts to qualifying educational and scientific organ­izations in addition to charitable ones.9 Donations to colleges are rightfully deductible, but it is ­because of the donee institutions’ educational, not charitable, nature. If not charity, then what does the college or university strive for? This question has two reasonable answers, and the answer a college gives ­w ill necessarily depend on its financial condition. Economist Philip Wicksteed wrote, “A man can be neither a saint, nor a lover, nor a poet, u ­ nless he has comparatively recently had something to eat. The ­things that money commands are strictly necessary to the realisation on earth of any programme whatsoever.”10 As poetry is a luxury requiring food, so ­grand academic schemes are a luxury requiring financial security. Largely ignored when scholars study wealthy institutions like Williams is the fact that, like most organ­ izations of any stripe, colleges have a primal urge to survive. With rare exceptions, no college president wants to be the president who has to close the institution’s doors for good. ­These are not the concerns, of course, of t­ hose who lead colleges blessed with reliable sources of revenue. But for the myriad tuition-­dependent colleges that routinely operate near the edge of bud­get deficits, the specter that lurks right ­behind deficits is the threat of closure. For many public institutions desperate for revenue, the specter may not be closure per se, but rather the ignominious shedding of programs or consolidation into some other unit in a state system. Some readers may dismiss as obvious survival as an aim of colleges, especially since so few colleges actually do go out of business. But this survival imperative drives much of the be­hav­ior observed

118 S upply

among large numbers of private colleges, and a few public ones as well. Thus a second type of college embraces all ­those for whom financial viability is a recurring, year-­to-­year concern. T ­ hese are the tuition-­dependent colleges, the colleges of quiet desperation. The only reason survival itself, and the colleges whose most pressing concerns are focused on it, have not been discussed more in the academic lit­er­a­ture on higher education is ­because of scholars’ tendency to concentrate their attention on well-­endowed institutions, t­hose where, happily, the possibility of closure cannot even be ­imagined. This third type of college, whose financial circumstances allow them to look beyond the next payroll, are the name-­brand colleges. ­Here are the universities, both public and private, on whom scholars have lavished the bulk of their attention, universities like the ones where most of ­those scholars studied and where many of them ­were or are employed. For ­these favored colleges, formally expressed institutional aims and major initiatives alike arise out of complex intramural strug­gles for influence among competing interests. The rhe­toric that is used to summarize the resulting consensus often expresses a nebulous devotion to excellence, which in practice cannot easily be distinguished from maximizing prestige, or ranking. Writing in the 1970s, economist Howard Bowen and sociologist James Coleman expressed the view that universities ­really have no objective at all, other than a generalized devotion to excellence. For Bowen, the goals w ­ ere “educational excellence, prestige, and influ11 ence.” Indeed, a simpler view, not inconsistent with the somewhat nihilistic views of Bowen and Coleman, is that colleges simply seek prestige.12 For sociologist Steven Brint, Duke University’s mission statement offers an unvarnished exemplar of this general objective. It states: “Our overriding goal therefore is to be among the small number of institutions that define what is the best in American higher education.”13 Coleman attributed this lack of specific aims to the loose orga­n izational structure of universities, one better characterized by the competing constituencies of a congressional district than by the top-­down hierarchy of the modern corporation.14

Th e I n ­e q u a l i t y Divi d e n d

119

Bowen’s theory is worth spelling out in more detail, in part b ­ ecause it anticipates con­temporary notions of “positional” competition and winner-­take-­a ll markets. In such competition, winning is the only goal, since ­there is other­wise no objective mea­sure of output or accomplishment. Best seen in sports, this kind of competition has one ­simple objective: coming out ahead of one’s rivals. Competition between such entities devolves to an arms race. Since, u ­ nder t­ hese circumstances, ­t here is no logical limit to potential expenditures that might serve to advance the aim of winning, expenditures have no natu­ral check except the availability of funds. In this kind of competition, when position is all that m ­ atters, the college becomes an acquisitive enterprise, always looking for money, always clutching a list of worthy proj­ects needing ­doing, needing funding. Elegant in its simplicity, Bowen stated his famous law in 1970: “Institutions raise as much money as they can get and spend it all. Cost per student is therefore determined primarily by the amount of money that can be raised.”15 A de­cade l­ ater, he elaborated, writing, “The incentives inherent in the goals of excellence, prestige, and influence are not counteracted within the higher educational system by incentives leading to parsimony or efficiency. The question of what ­ought higher education to cost—­what is the minimal amount needed to provide ser­v ices of acceptable quality—­does not enter the pro­cess except as it is imposed from the outside. . . . ​The duty of setting limits thus falls, by default, upon ­those who provide the money, mostly legislators and students and their families.”16 H ­ ere, then, is a succinct and ser­viceable list of the institution’s post-­survival aims: excellence, prestige, and influence. Not a bad list, to be sure, but surely one no more actionable than “to be the best.” In practice, ­these ill-­ defined aspirations inspire a generalized effort to be better than one’s rivals, to rise in the rankings. Such competition is termed “positional” b ­ ecause the only objective is to come out ahead of one’s rival. It is not hard to see how this competition, waged in terms of excellence, prestige, and influence, would be a fertile field for someone to come along with a way to keep score. Enter the U.S. News ranking of colleges.

120 S upply

Supply and Demand in the Segmented College Market

In describing the market for undergraduate education, historian Roger Geiger noted the remarkable stability in the number of students entering four-­year colleges and universities. For the two and a half de­cades following the mid-1970s, although the number of high school gradu­ates had seen swings as large as 30 ­percent (from 3.2 to 2.3 million and back to 2.8 million), he stated, the number of first-­ time students remained remarkably steady, varying within a narrow range between 1.1 and 1.2 million.17 He took this as evidence that baccalaureate enrollment is determined largely by the supply of places in four-­year colleges and universities, a conclusion he found difficult to reconcile with the apparent unused capacity at many colleges, both public and private. With the luxury of the hindsight afforded by an additional de­ cade of history, it appears that this apparent constancy highlighted by Geiger was not so much proof of some immutable collegiate version of Say’s Law but rather a historical coincidence, resulting from two opposing trends. The first was a sustained decline, over most of the period Geiger was observing, in the number of high school gradu­ates. Between 1975 and 1991 that number fell by more than a quarter. But a second trend over the same period went in the other direction: the rate at which ­t hese high school gradu­ates ­were enrolling in four-­year colleges was steadily increasing, rising by more than 25 ­percent itself. Supporting this opposing trend’s explanation is the fact that, as the demographic decline ended, first-­time enrollments in four-­year colleges, far from remaining constant, began rising rapidly. ­A fter 1993 new enrollments in four-­year institutions increased in ­every single year through 2009. The driving force for this growth was a general increase in four-­year college enrollment rates, which r­ ose from an average of 45.5 ­percent during the 1990s to more than 50 ­percent ­after 2000.18 As shown in Figure 5.1, first-­ time enrollments in four-­year colleges, a­ fter remaining more or less steady ­until 1995, increased steadily ­after that, outpacing the growth in high school gradu­ates. This growth occurred in private as well as

Th e I n ­e q u a l i t y Divi d e n d

121

1970 = 100, private rel to total in 1970

180 160 140 120 100 80 60 40 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011

20

Year High school graduates In private colleges

First-yr students in all 4-yr colleges

Figure 5.1. ​High School Gradu­ates, First-­Year College Students at Four-­Year Colleges, 1964–2012 Source: U.S. Department of Education, Digest of Education Statistics 2013, ­Tables 302.10, 305.10. Note: Figure shows high school gradu­ates relative to the total in 1970 and first-­year enrollment in all four-­year colleges and in private four-­year colleges, both relative to total first-­year enrollment in all four-­year colleges in 1970.

public colleges, as indicated by the rise in the bottom line, which shows private first-­year enrollments relative to the 1970 total of all first-­year enrollments. In fact the private share of new enrollments, which had remained about 36 ­percent since the 1970s, hit 37 ­percent in 2000 and did not fall below that share afterward. Although “supply-­determined” may not accurately characterize enrollment for the entire market for four-­year colleges, that term comes much closer to the truth for at least one segment of the market. This segment is the handful of “elite” colleges and universities, widely admired and highly desired, that regularly sit atop the U.S. News annual ranking of American colleges. Exemplified by such familiar names as Stanford, Harvard, Yale, and Prince­ton, ­these colleges are famous for the very high SAT scores of their students and

122 S upply

the lavish resources they have at their disposal. They are also distinguished by the difficulty of gaining ac­cep­tance to them. In 2014 Stanford led the way in the selectivity competition, accepting a scant 5 ­percent of its applicants for admission.19 For many families, a son or ­d aughter’s ac­cep­t ance to a college in this rarified group ranks among a lifetime’s most prized trophies. Unlike other sought-­after goods, such as exclusive real estate or luxury cars, however, letters of ac­cep­tance to the likes of Stanford or Yale are not for sale, at least not in any conventional sense. In the sort of stylized markets described in economics textbooks, an increase in the demand for a good or ser­vice ­will normally induce firms to produce more, to raise their prices, or both, depending on the exact nature of supply. In the special case where the number of available units is more or less fixed—­for example, Impressionist paintings or h ­ ouses in a fash­ion­able neighborhood—an increase in demand in a freely operating market ­will inevitably cause the market price to go up. But selective colleges have quite deliberately chosen not to allow their prices to rise as high as the market might other­wise push them. Instead of relying on prices to equilibrate the market, t­ hese prized admissions spots are rationed on the basis of carefully developed criteria that look nothing like price, criteria that are embodied in the dreaded letter of rejection, received by thousands of hopeful applicants ­every spring. Just how unresponsive the supply is in this corner of the baccalaureate market can be seen by considering the share of the college market made up by just fifteen of the most selective colleges, a group that includes such institutions as Amherst, Columbia, Harvard, Stanford, and Yale.20 In 1972 t­ hese fifteen colleges together enrolled about 1 ­percent of all four-­year college students in the country. By 2010, however, their share had fallen to 0.6 ­percent. Although they had, as a group, increased their enrollment during the thirty-­eight-­ year period (by a fifth), this expansion failed to keep pace with the total number of full-­time students in all four-­year colleges, which had increased by more than 75  ­percent.21 A similar story of unresponsive supply applies to the somewhat more inclusive group of fifty-­eight private colleges whose average SAT levels in 1970 put

Th e I n ­e q u a l i t y Divi d e n d

123

them in the highest 5  ­percent of U.S. undergraduate institutions. This larger group of still very selective colleges saw their collective share of all undergraduates decline, from 3.6 ­percent in 1972 to 2.5 ­percent in 2010.22 One direct result of the unwillingness of selective institutions to expand in proportion to rising demand has been to make it harder and harder for students to be admitted. Stanford is only one of the handful of extraordinarily stingy colleges. In 2014, Harvard, Yale, Columbia, MIT, and Chicago also had ac­cep­tance rates of 8 ­percent or less.23 Not only do con­temporary cohorts of high school gradu­ ates find themselves in a more challenging competition for admissions spots than their older pre­de­ces­sors faced, the burden of this unresponsive supply fell especially hard on students living in ­those parts of the country where most of the selective colleges are located, the Northeast and California.24 To be sure, as they engage in this nonprice rationing, the colleges most in demand cannot fail to notice that they could, if they so desired, raise their tuition—­significantly—­and still be able to fill all their available spots. Although they did not raise tuition as much as they could have, the selective colleges did regularly raise it faster than the rate of inflation. Between 1980 and 2000, for example, Harvard’s tuition and fees went up an average of 6.5 ­percent a year, compared to 3.7  ­percent a year for inflation. Other selective colleges ­increased their prices at similar rates—­Wellesley at 7.3 ­percent a year, Swarthmore at 8.2 ­percent, and Duke at 8.8 ­percent, for example.25 In fact, since around 1980 published (sticker-­price) tuition rates nationwide have increased faster than inflation. This tendency is portrayed in Figure 5.2, which compares the average annual increase in tuition at public and private four-­year colleges and universities to the rate of inflation as mea­sured by the Consumer Price Index. During the rapid and unpredictable inflation of the 1970s, tuition increases in turn fell ­behind and then caught up with increases in general prices, making up lost ground in the mid-1980s. Since that time tuition increases have averaged roughly 2 percentage points above inflation.

124 S upply

0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00

19

7 19 0 7 19 2 74 19 7 19 6 7 19 8 80 19 8 19 2 8 19 4 86 19 8 19 8 9 19 0 9 19 2 94 19 9 19 6 98 20 0 20 0 0 20 2 04 20 0 20 6 0 20 8 10 20 12

−0.02

Year 4-yr public tuition

CPI

4-yr private tuition

Figure 5.2. ​Tuition Increases and Inflation, Private and Public Institutions, 1971–2003 Note: Before 2001, private tuition covers all private four-­year colleges; beginning in 2001, it covers only nonprofit, private four-­year institutions. Source: U.S. Department of Education, Digest of Education Statistics 2013, ­Table 330.10.

Although this generalization applies to tuition increases in both the public and private sectors, the explanation for the increases are dif­fer­ent. For the private colleges and universities, the sustained increases over inflation began with the highly selective institutions, ­those experiencing significant excess demand. T ­ hese above-­inflation increases ­were spurred by quality competition and made pos­si­ble by three ­things: the bandwagon nature of the increases, which meant that no one college’s price ­rose much relative to the o ­ thers; a need-­ based financial aid system that protected low-­income students from the effects of higher tuition; and the excess demand itself, the fact that many families w ­ ere willing and able to pay the higher tab, requiring l­ ittle or no financial assistance to do so. On the public side of the market, tuition levels started out much lower than they did on the private side. It was the growing reluctance by states to make appropriations that forced the public institutions to raise tuition. In

Th e I n ­e q u a l i t y Divi d e n d

125

percentage terms, the increases turned out to be virtually the same as ­those in the private sector. The Matthew Effect at Work

As described in Chapter  3, the economic outlook for colleges and universities around 1970 looked decidedly bleak. Costs ­were outpacing revenues, and the prospects for a return to good times seemed unlikely. The title of one report, The New Depression in Higher Education, summed up the dire outlook facing many institutions. A de­ cade ­later, a university as seemingly secure as Duke was considering radical surgery in order to get its spending ­under control and put itself on firmer financial footing. Duke’s chancellor, Kenneth Pye, proposed eliminating the university’s education department and undergraduate nursing program, two programs with significant enrollments. Arguing that Duke’s financial position compared unfavorably with ­those of public universities such as the University of North Carolina, which had been enjoying real increases in state appropriations, he argued that retrenchment was the only way Duke could maintain its quality and remain competitive.26 But much of this pessimism ultimately proved to be unwarranted, especially for the colleges and universities at the top of the prestige pecking order. Harvard and Stanford, both designated by the New Depression report as being in serious financial trou­ble, soon soared in resources and prestige, beyond their already lofty positions. Duke did much the same, without resorting to many of the cuts discussed by its chancellor. Indeed, most of the colleges and universities at the top, especially the private ones, soon entered a period of unpre­ce­dented success and prosperity. In studies that traced the fortunes of colleges over time, economists showed that expenditures per student increased faster in the more selective colleges than in the less selective ones. One study examining changes in the college market between 1972 and 1992 found that inflation-­adjusted expenditures per student increased by 80 ­percent in highly selective private institutions, compared to 23  ­percent among the top fifty

126 S upply

publics, and 11 ­percent among less selective private colleges. Among the less-­selective public institutions, expenditures per student actually fell slightly.27 ­Here was the Matthew effect at work in the college market. Although other explanations have been offered, ­there can be ­little doubt that the driving force ­behind this phenomenon as it played out in the American college marketplace was the profound change in the income distribution that began during the late 1970s.28 This was the one indisputable smoking gun, though it was helped along by a stock market that ­rose to unpre­ce­dented heights, leaving many investors with valuable assets that they could contribute at comparatively l­ittle cost.29 The most vis­i­ble consequence of t­ hese developments was a surge in donations to universities. For several years, the online magazine Slate published lists of ­people who had contributed the most money in the previous year. 30 In 2011 it listed the top sixty donors. Of the top ten of ­these philanthropists, five made gifts to universities in that year: $119.5 million to the University of California at San Diego, $117.2 million to the University of Oxford, $101 million to Oklahoma State, $100.5 million to UCLA, and $100 million to the University of California at San Francisco. The next dozen individuals on the list accounted for another six gifts to universities, all $50 million or more. ­Because American higher education, to an extent unmatched in most other developed countries, depends on private support, contributions such as ­these are not only welcome but, for ­those institutions that have come to expect them, vital. Gifts of such magnitude, however, require donors with enormous wealth at their disposal. One significant consequence of the changes in the country’s income and wealth distribution, loudly decried by many scholars, was to boost the wherewithal of potential donors to make large gifts to the very colleges and universities that employed many of t­ hose same scholars. It is a long-­established fact that ­those with higher incomes tend to contribute more than t­hose who make less. 31 This rule ­applies with special force to donations made to colleges and universities, since the most affluent donors have historically devoted a

Th e I n ­e q u a l i t y Divi d e n d

127

disproportionately large share of their largess to colleges and universities (in addition to health care and cultural organ­i zations). An analy­sis of an earlier year’s Slate 60 showed what a large share of the largest gifts goes to higher education. Of the gifts of $5 million or more recorded in 1996, 56 ­percent went to universities, over and above the 9 ­percent that went to university-­affiliated medical centers.32 Besides their extraordinary magnitude, t­ hese charitable gifts are also marked by their unevenness. They tend to be concentrated among a relatively small group of universities. Of the $34 billion in donations made by individuals, corporations, and foundations to colleges and universities in 2013 recorded by the Council for Aid to Education, roughly a sixth of the total went to just ten universities. T ­ hese ten are a Who’s-­W ho of U.S. research universities: Stanford, Harvard, University of Southern California, Columbia, Johns Hopkins, Penn, Cornell, NYU, Yale, and Duke. 33 With impressive regularity, prominent American research universities have in recent years announced, and then met or exceeded, increasingly ambitious capital campaigns. In 2012 and 2013, for example, university fundraising campaigns raised $3.9 billion for Yale, more than $3 billion for Cornell, more than $4 billion for Columbia, and a rec­ord $6.2 billion for Stanford. In September 2013 Harvard announced a goal of $6.5 billion for its upcoming campaign.34 Reflecting on that university’s already huge endowment, one commentator complained that the rich in higher education ­were becoming too rich: “Harvard’s timing is impeccable. The wealthiest Americans have recovered all the money they lost during the G ­ reat Recession and then some, while legions of potentates and businessmen worldwide are e­ ager to buy a piece of the elite American dream for their kids. Over the last de­cade, private universities have separated from their public competitors, ramping up spending and poaching faculty members and students. Now they can run up the score.”35 Thus a profound and lasting effect of income in­equality was that colleges became more unequal, thanks in part to the dynamics of tax-­subsidized charitable giving. Given the strong statistical correlation between a person’s income and the amount of his or her char-

128 S upply

itable giving, and the documented preferences that affluent donors have exhibited ­toward colleges and universities as objects of their generosity in the past, it is not hard to imagine that the rising fortunes of families at the top of the income distribution ­a fter 1980 might have resulted in rising donations to colleges. To analyze exactly what did occur, we need to address two empirical questions. First, did the changes in the income distribution boost contributions to higher education? Second, did contributions such as ­these go disproportionately to the best-­endowed universities in the country, thus increasing in­equality in higher education? To address t­ hese two questions, two sources of data on donations received by colleges and universities can be analyzed. The first consists of annual survey data collected by the Council for Aid to Education (CAE), and the second is financial reports, covering more institutions but a shorter period of time. The CAE annual surveys used h ­ ere covered voluntary gifts and grants received by colleges and universities over the period 1969 to 2011 for a small but significant subset of all U.S. colleges and universities. Of the roughly 1,150 four-­year colleges and universities offering general bachelor’s degrees in 1970, the CAE survey covered a l­ ittle more than 200 of them in 2010. ­These institutions likely accounted for the lion’s share of all contributions to U.S. colleges and universities. Omitting universities with excessive gaps in reporting further reduces the sample, but contributions for ­those omitted ­were small relative to ­those for the included institutions. For 2010, for example, the 121 colleges and universities in the more restricted sample used below accounted for $13.5 billion in donations (in 2010 dollars), while 100 institutions sampled by CAE in 2010 but not included among the 121 accounted for just a third of that amount, $4.3 billion. The CAE survey covered donations from six sources: foundations, alumni, other individuals, corporations, and religious and other organ­izations. One feature of the data that stands out is the stability over time at the top end of the distribution. This stability is illustrated in ­Table  5.1, which is based on the sample with reasonably complete data. The ­table lists the top dozen recipient universities in 1970 and 2010. Of

Th e I n ­e q u a l i t y Divi d e n d

129

Table 5.1  Universities with the Most Donations, 1970 and 2010 (2010 dollars) Year

Rank

1970

2010

Total Donations

Institution Name

State

1

Harvard University

MA

$302,026,976

2

New York University

NY

$234,766,608

3

University of Chicago

IL

$221,546,176

4

Yale University

CT

$189,449,728

5

Columbia University

NY

$183,690,192

6

Stanford University

CA

$175,649,536

7

University of Pennsylvania

PA

$151,060,576

8

Northwestern University

IL

$144,429,424

9

Cornell University

NY

$141,159,408

10

Dartmouth College

NH

$125,882,824

11

Washington University

MO

$114,859,408

12

Mas­sa­chu­setts Institute of Technology

MA

$113,748,312

1

Stanford University

CA

$617,794,432

2

Harvard University

MA

$615,806,336

3

Johns Hopkins University

MD

$441,090,400

4

University of Southern California

CA

$439,463,648

5

Columbia University

NY

$415,057,088

6

University of Pennsylvania

PA

$393,636,640

7

Yale University

CT

$392,926,688

8

New York University

NY

$360,236,992

9

Duke University

NC

$356,372,832

10

Indiana University

IN

$353,639,232

11

University of California, Los Angeles

CA

$351,151,776

12

Cornell University

NY

$317,948,480

Source: Council for Aid to Education, Voluntary Support of Education, data on voluntary support of education, total contributions that universities received from foundations, alumni, other individuals, corporations, and religious and other organ­i zations. Sample includes 121 universities with complete or nearly complete data. See text.

130 S upply

Donations in millions (2011 dollars), quartile average

350 300 250 200 150 100 50

19 6 19 9 71 19 7 19 3 7 19 5 77 19 7 19 9 81 19 8 19 3 8 19 5 87 19 8 19 9 91 19 9 19 3 9 19 5 97 19 9 20 9 0 20 1 0 20 3 0 20 5 07 20 09

0

Lowest quartile

2nd quartile

3rd quartile

Highest quartile

Figure 5.3. ​Average Donations by Quartile of 1969 / 70 Donations Source: Council for Aid to Education, Voluntary Support of Education, data on voluntary support of education, total contributions that universities received from foundations, alumni, other individuals, corporations, and religious and other organ­i zations. Sample includes 121 universities with complete or nearly complete data for the period. Donations are expressed in constant 2010 dollars. See text.

t­ hose that constituted the list in 1970, seven ­were also on the list in 2010.36 Data for 121 institutions that had complete or nearly complete rec­ ords over the entire period ­were used to trace trends in donations to colleges and universities over time.37 For the few years where gaps did occur, total support was estimated by linear interpolation. 38 The primary aim was to compare the fortunes of the haves with the have-­nots. Since most of the institutions covered in the CAE sample are established, with histories of receiving some donations, the notion of “have-­nots” must be a relative one. Thus Figure 5.3 pres­ents average inflation-­adjusted donations by quartiles based on their donations at the beginning of the period (the average of 1969 and 1970) and then traces how each of the four groups fared over the next

Th e I n ­e q u a l i t y Divi d e n d

131

Average donations per student, 2009 dollars

$30,000 $25,000 $20,000 $15,000 $10,000 $5,000

Pub

Priv

90+

264 286

1,596 3,277

HBCU 2,970 3,376

80–90 2,388 2,112

1,274 1,553

840 746

29,492 17,618

22,121 9,980

15,434 5,983

Public 9,939 5,191

5,713 2,366

3,783 2,034

5,143 2,187

2,500 1,388

3,730 1,859

2009

2,556 1,825

1990

2,217 1,988

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0–25

$0

Figure 5.4. ​Donations per Student, by College Category, 1990 and 2009 Source: Author’s calculations based on data from the Delta Cost Proj­ect for 1,035 of the 1,157 institutions in the full sample. For definitions of the sample, see text.

four de­cades. Although donations in the lowest quartile grew the most rapidly in percentage terms, the inequities of the starting point grew in absolute terms. Whereas the average top quartile university received $107 million more in donations than the average bottom quartile university in 1970, by 2010 the difference had more than doubled, to $266 million. To see how ­these diverging growth paths differed by type of institution, we turn to the second source of data, one that covers most of the four-­year colleges in the country. ­These are the annual reports that colleges and universities fill out for the federal government. Figure 5.4 displays per-­student donations received by colleges two de­ cades apart, in 1990 and 2009, based on data for more than 1,000 colleges and universities. For this figure, institutions ­were divided into the more detailed seventeen groups used elsewhere in this book.39 The figure reveals yawning disparities in donations. In 1990, for example, the private colleges in the highest 1970-­era SAT category received an average of $17,610 per student (expressed in 2009

132 S upply

dollars), a number that dwarfed virtually all other groups. The nextclosest group was the 98th to 99th percentile private colleges, which averaged an impressive $9,980 per student. Only two other college categories received as much as $5,000 per student in donations in 1990. Public institutions below the 90th percentile mostly held steady, but those above the 90th percentile fell by half. In sum, for the top private colleges the growth in donations was remarkable. Beginning at already high levels in 1990, donations to t­ hese top private institutions ­rose dramatically. In percentage terms, donations to ­these schools grew at rates comparable to ­those for public universities and several of the categories of private colleges. But in absolute terms, they w ­ ere able to increase their advantage over all other groups of colleges.40 In sum, donations received by American colleges and universities grew impressively over this forty-­year period, spurred in part by rapid increases in wealth among the affluent. Many institutions, especially the state universities, which had received only modest voluntary support, saw their total donations grow rapidly in percentage terms. In fact, growth rates in donations among institutions that began the de­cade of the 1970s with modest levels of giving tended to be high, and the share of total donations g­ oing to a very few institutions fell. But the institutions that began the period with an advantage saw that advantage, in dollar terms, increase, propelling the richest ones further ahead in absolute terms than they had been in the early 1970s.41 Another development that played an impor­tant role intensifying the in­equality of colleges was an extraordinary bull market in stocks that persisted for some two de­c ades. A ­ fter losing half of their value between 1970 and the early 1980s, stocks entered a period of unpre­ce­dented and sustained price increases. Between 1980 and 1990 the Dow Jones Industrial Average increased, in inflation-­ adjusted terms, at an annual rate of 7.4 ­percent, an impressive rate of return by any standard. But in the following de­cade the Dow ­rose even faster—­a spectacular 10.1 ­percent a year.42 ­These sustained increases in stock prices had the potential to help colleges in two ways.

Th e I n ­e q u a l i t y Divi d e n d

133

First, ­because the higher stock prices added to the wealth of potential donors, colleges stood to collect more in charitable contributions, all the more so since gifts of appreciated stock enjoyed extra tax advantages over gifts of cash.43 Second, college endowments would grow, helping most the colleges with large endowments with heavy reliance on stocks rather than bonds. As we have seen, donations tended to flow to the richest institutions. As for the asset mix of college portfolios, we must turn to additional information. This tendency for the rich to get richer over this period was amplified by the manner in which the donations w ­ ere invested. The universities that collected the most also made it grow the fastest. Estimates by Thomas Piketty suggest that, from 1980 to 2010, universities with the biggest endowments enjoyed higher average returns than ­those with smaller ones. Whereas t­ hose in the highest endowment category (over $1 billion) earned an average of 8.8 ­percent a year ­after inflation, ­those in the lowest (less than $100 million) earned just 6.2 ­percent.44 Other studies covering more recent periods confirm this conclusion. For example, one study of university endowments showed that the eight Ivy League universities consistently outperformed other private universities as well as public ones in their average rate of return.45 Another showed that, for the de­cade 2003 to 2013, the total investment returns ­were highest for universities with the biggest endowments and lowest for ­those with the smallest. Universities with endowments over $1 billion earned an average 8.3 ­percent a year (net of management costs), compared to just 6.3 ­percent for ­those with endowments u ­ nder $25 million.46 That the rich colleges would not only receive more in donations but also enjoy higher rates of return makes this an especially apt application of the Matthew effect. What explains this dollop of additional return for the rich colleges? One ready economic explanation is economies of scale, wherein the relatively fixed administrative costs can be spread over more assets, thus lowering average costs for bigger endowments. It seems unlikely, however, that differences in scale alone explain the gaps in rates of return and such large differences in portfolio composition. A more impor­t ant difference be-

134 S upply

tween the large and small endowments is the composition of investment portfolios. Universities with the biggest endowments have tended to put less money into safe investments, such as bonds, and more into riskier assets with higher average yields, such as stocks and real estate. In recent de­cades, the richest universities have significantly boosted their reliance on “alternative strategies,” such as private equity, hedge funds, venture capital, noncampus real estate, natu­ral resources, and distressed debt.47 In fiscal year 2013, universities with endowments over $1 billion held 59  ­percent of their portfolios in such assets, compared to just 11 ­percent for universities in the under-­ $25-­million category.48 One study of university endowment per­for­ mance proposes two additional explanations, sophistication and connections: “While the SAT scores of a university’s students are unlikely to affect investment per­for­mance directly, average SAT score may be a proxy for the skill of the university administration, the wealth and connections of the alumni network, and the prestige of the university brand, all of which could affect endowment management over and above the role of endowment size.”49 Another ­ factor in the extraordinary success of the richest ­universities may be in rates of spending out of endowments, but the evidence is mixed. One analy­sis of seventy-­eight private research universities over a thirty-­year period found that ­those which had the highest endowments per student devoted a higher share to reinvestment than o ­ thers, giving them lower spending rates. 50 But surveys done by the National Association of College and University Business Officers (NACUBO) revealed that universities with the largest absolute endowments had higher rates in some years, but not in ­others. 51 One ready mea­sure of financial strength is the value of endowments. Figure 5.5 compares per-­student endowment by college category in 1970, 1990, and 2013. 52 ­Here the yawning gaps between college categories that we saw in donations are further accentuated. Compare, for example, private colleges in the 50th to 60th percentiles of SAT scores with ­those in the 99th and above category. In 1970 the average endowment levels (in 2013 constant dollars) w ­ ere

Th e I n ­e q u a l i t y Divi d e n d

135

about $7,000 per student in the former group and about $224,000 in the latter one. Quite a difference! Over the following two de­cades, ­these per-­student amounts r­ose to $15,000 and $311,000, respectively. The difference between the amounts increased a bit, but not by a ­great deal. But ­after 1990, the gap increased dramatically—­$34,000 versus $1,010,000. The gap between t­ hese two college categories had quadrupled in size from where it stood in 1970. The colleges at the top of the selectivity distribution pulled away from the rest in the last two de­cades, further widening the gulf between them and the rest. If one judges equality in terms of ratios rather than dollar differences, 2013 actually looks more equal, since the ratio of the higher to the lower amounts fell, from 32:1 to 30:1, but ­these ratios do not reflect differences in what can be purchased per student with income from ­these endowments as the gaps in per-­student endowments do. 53 The message that Figure 5.5 sends should be loud and clear: the rich colleges got richer, much richer. Widening Resource Gaps

A natu­ral consequence of the vast differences in donations regularly bestowed on colleges and universities ­were growing gaps in both expenditures and assets. It was pos­si­ble to see evidence of t­ hese widening gaps in painted win­dow sills and manicured lawns as well as new recreation centers or high faculty salaries. Referring to physical capital, one study concluded that “the capital stock used in U.S. higher education is simply very large, very impor­tant, and very unevenly distributed among colleges and universities.”54 In 1993 the richest 277 institutions owned 30 ­percent of the capital stock.55 Their capital stock per student ($116,760) was more than twice that of the next decile ($52,848). 56 The focus ­here is not only on differences such as ­these across institutions, but also on how t­ hose differences have grown over time. This approach necessitates finding mea­sures of resources that can be compared over time, among a fixed group of colleges and universities. Although comparable mea­sures of general expenditures and as-

136 S upply

$800,000 $600,000 $400,000 $200,000

Pub

Priv

90+

559

18,677 33,769

790

908 1,135 40,350

19,534

1,662 1,287 13,631

4,338

80–90

3,808

70,286 80,756 256,616

4,127

44,093 84,828 188,246

17,234

12,819 23,425 52,920

943

16,944 28,111 69,299

1,238

14,233 24,115 56,403

HBCU

8,348

8,061 16,229 29,004

1,010,097 310,593 224,033

7,001 14,515 33,852

522,628 189,732 163,963

6,332 11,674 21,194

2013

9,171

1990

10,811

1970

Public

18,745

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

$0

0–25

Average endowment per student, 2013 dollars

$1,000,000

Figure 5.5. ​Endowment per Student, 1970, 1990, and 2013, by College Category (2013 dollars) Source: Based on calculations for 705 colleges and universities with data for 1970, 1990, and 2013. 1970: U.S. Office of Education, National Center for Education Statistics, Higher Education General Information Survey (HEGIS), 1969–1970: Financial Statistics, ICPSR02095-­v1. Ann Arbor, MI: Inter-­u niversity Consortium for Po­l iti­cal and Social Research [distributor], 1998; http://­doi​.­org​ /­10​. 3 ­ 886 ​/­ ICPSR02095​.­v1. 1990: ­Because data for endowments for June 30, 1990, are not available, the ending values for the 1988 / 89 and 1990 / 91 years w ­ ere averaged. Integrated Postsecondary Education Data System (IPEDS), National Center for Education Statistics, accessed October 27, 2015, http://­nces​.­ed​.g­ ov​/­ipeds​/­d atacenter​/D ­ efault​.­a spx; and 2013: Integrated Postsecondary Education Data System (IPEDS), National Center for Education Statistics, accessed December 12, 2015, http://­nces​.­ed​.g­ ov​/­ipeds​/­d atacenter​/D ­ efault​.­a spx. Figures are aggregate value of endowments at the end of the fiscal year in each college category divided by the total number of undergraduate and gradu­ate students in the category. For colleges for which 1990 was missing, the average for 1989 and 1991 was used.

sets are available for only half the period ­under study, one excellent indicator of resource levels that can be gathered over the entire four-­ decade span of time is available—­faculty compensation. The AAUP collects and publishes data by rank for most college and universities in the land. Figure  5.6 summarizes, for eight groups of colleges

Th e I n ­e q u a l i t y Divi d e n d

137

Average compensation, 2011/12 dollars

$180,000 $160,000 $140,000 $120,000 $100,000 $80,000 $60,000 $40,000 $20,000 $0

0–50 50–90 and HBCUs

90–95

95–98 98–100

Private Universities

0–50 50–90 90–100 and HBCUs Public Universities

1970/71 86,409.9 104,666 110,972 124,470 129,715 94,950.6 114,468 128,473 1990/91 90,222.8 113,937 119,720 141,693 148,851 102,486 130,002 142,134 2011/12

100,660 118,385 137,535 164,110 176,135 113,820 140,230 165,466

Figure 5.6. ​Average Compensation for Full Professors, Eight Groups of Institutions, Selected Years (2011 / 12 dollars) Source: For each of the eight groups of institutions specified in the figure, data on average compensation for a random sample of twenty institutions was drawn, based on SAT percentiles for 1970. For this panel of institutions, data w ­ ere taken from AAUP published figures for each of three years, as follows: 1970 / 71: AAUP Bulletin 57 2 (June 1971); 1990 / 91: Academe 77 2 (March–­April 1991); 2011/12: “Appendices to the 2011–12 Annual Report on the Economic Status of the Profession,” accessed May 29, 2014, http://­w ww​.a­ aup​.o ­ rg​/­reports​-p ­ ublications​/­2011​-­12salarysurveyappendices. Dollar figures ­were converted to 2011 / 12 using the CPI.

defined by sector and percentile of 1970-­era SAT, the inflation-­ adjusted level of compensation for full professors, the rank containing the bulk of faulty. For each of ­these categories, a random sample of twenty institutions was drawn to make the calculations, and compensation data ­were collected for each of the sampled colleges for all three years. 57 The graph reveals the same unmistakable hierarchy by SAT level in each of the three years seen for 1970 in Chapter  3. In the 2011 / 12 academic year, for example, professors’ compensation ranged across the categories of private institutions from $101,700 for the lowest-­ranked colleges to $176,100 in the highest. A similar hierarchy was evident for the public universities, with average compensation rising from $113,800 for the lowest-­ ranked institutions to $165,500 for the top ones. Comparing the pri-

138 S upply

vate and public sectors, professors at public universities earned more in both the 0–50th and 50th–90th categories, but comparisons cannot be made for institutions at the highest SAT level, as the categories are not comparable. The hierarchy in faculty pay that existed in 1970 was still quite apparent forty years l­ater, but the gaps had grown larger, both relatively and absolutely. Over this period compensation for full professors grew by 36 ­percent in the highest-­ranked private institutions; next-fastest growth was in the 95 to 98th private category, with an increase of 32 ­percent. In no other group did compensation grow by more than 30  ­percent. ­These public-­private comparisons may be misleading, however, due to the prevalence of small colleges in the categories of private institutions. 58 One unsettling dimension of this growing in­equality, only barely vis­i­ble in Figure 5.6, was a growing divide among public and private research universities in faculty compensation. During the 1980s, the small edge that top private research universities enjoyed over their public counter­parts grew larger, and this trend has continued to the pres­ent day. Consider the case of two research universities in Illinois—­ Northwestern, the private university located in the Chicago suburb of Evanston, and the University of Illinois at Champaign-­Urbana, the state’s flagship university. In 1970 / 71 the average compensation for full professors at Illinois, $22,200, was 86 ­percent of the corresponding average at Northwestern, a gap of 14 ­percent. In 1980 / 81 the gap had narrowed to 7 ­percent. But the fortunes of ­these two universities, like ­those of public and private research universities across the country, began to diverge during the 1980s. By 1990 / 91 the gap in average full professor’s compensation had risen to 20  ­percent. ­A fter holding at that level for a de­cade, the gap r­ ose to 22 ­percent by 2015 / 16. Much the same was true for the leading research universities in North Carolina. In 1970 / 71 average compensation for full professors at the flagship Chapel Hill campus was 94 ­percent that of its neighbor Duke, for a gap of 6 ­percent. By 2015 / 16 the gap in compensation had mushroomed to 27 ­percent, making it very hard indeed for Chapel Hill to recruit and retain the most sought-­a fter full

Th e I n ­e q u a l i t y Divi d e n d

139

­Table 5.2 Average Salaries for Full Professors, Research I Universities (2015 dollars) Year

Public

Private

Public / ­Private

1971

$120,131

$127,120

0.95

1980

$96,491

$101,796

0.95

1990

$114,427

$129,787

0.88

2000

$123,811

$149,459

0.83

2015

$141,205

$186,582

0.76

Source: Courant and Turner (2016), ­Table 1. Figures are based on surveys of universities, the Higher Education General Information Survey (HEGIS), and the Integrated Postsecondary Education Data System (IPEDS).

professors in a competitive academic job market. That ­these two examples illustrate a broader trend can be seen in calculations two economists made for faculty salaries at public and private Research I ­ able 5.2, the comparisons for full profesuniversities.59 Shown in T sors at ­these research universities reveal a 5 ­percent advantage for the private universities in 1971 that held steady for a de­cade. By 1990 the advantage had grown to 12  ­percent, then 17  ­percent by 2000, then increasing all the way to 24 ­percent by 2015, when the average full professor at the public Research I universities ­were being paid just 76  ­percent of the average at comparable private universities.60 Much of this deterioration in the position of the public universities can no doubt be attributed to the growing reluctance of state legislatures to support higher education, dealing as they ­were with other demands and slowing revenues. But for a country whose exalted reputation for ­great research universities rests on universities supported and operated by state governments, this dimension of the growing in­equality in the market for college constitutes a legitimate worry. For the most recent two de­cades, it is pos­si­ble to go beyond faculty compensation to examine other kinds of comparable data on resources. Figure  5.7 shows one very general mea­sure of expenditures, the so-­called education and related expenditures per stu-

140 S upply

Expenditures per student, 2012/13 dollars, group median

$60,000 $50,000 $40,000 $30,000 $20,000 $10,000

Pub

Priv

90+

17,039 14,279

26,681 22,774

HBCU 18,164 14,214

80–90 20,352 19,259

16,433 14,711

14,030 11,912

60,664 39,607

53,583 32,115

45,553 27,219

Public 37,062 24,923

31,120 23,523

28,587 20,122

25,667 19,354

24,239 17,688

24,003 17,392

2012/13

21,796 16,239

1989/90

19,302 14,147

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0–25

$0

Figure 5.7. ​Education and Related Expenditures per Student, by College Category, 1989 / 90 and 2012 / 13 Source: Based on calculations for 1,114 colleges and universities with data for 1990 and 2013. IPEDS, Delta Cost Proj­ect. Medians by group. Note: Education and related expenditures are defined in terms of expenditures on the following: instruction +  student ser­v ices  + (education share) × (academic support + institutional support + operation and maintenance), where the education share = (instruction + student ser­v ices) / (instruction +  student ser­v ices  + research + public ser­v ice). The median for education and related expenditures was divided by the median of FTE enrollment for each category. The enrollment-­weighted average of class medians was $15,762 for 1989 / 90 and $19,465 for 2012 / 13.

dent, ­here shown separately for the more detailed seventeen college categories used elsewhere. ­Here again, the hierarchy suggested by grouping institutions according to their 1970-­era SATs turns out to track expenditure levels with remarkable fidelity, rising in 1989 / 90 from $14,100 in the lowest-­ranked private group to $39,600 in the highest and from $11,900 in the lowest public to $22,800 in the  highest. What is most startling is the increase in per-­student spending among the private institutions over the next two de­c ades,

Th e I n ­e q u a l i t y Divi d e n d

141

an increase not matched in the public sector. Especially impressive is the 50  ­percent increase in spending per student in the 99th + private institutions. While expenditures at the private colleges w ­ ere increasing rapidly, spending in the public sector increased only modestly. By 2012 / 13, per-­student spending in the most selective private colleges was more than four times that in the vast number of public colleges below the 50th percentile, $60,700 versus $14,000. One way this public-­private spending gap manifested itself was in the ratio of students to faculty. A calculation for research universities compared student-­faculty ratios for public and private universities in 1971 and 1997. For public universities the ratio increased, from 21.1 to 21.5. For private universities the ratio fell, from 17.3 to 15.7, offering one more indication of the growing in­equality in resources between public and private universities.61 A point that economists (and college administrators) sometimes like to make is that college students rarely pay enough to cover the full cost of their education, even students at the most expensive colleges. In effect, colleges subsidize students’ educational costs by drawing on other sources of income, including donations, endowment earnings, and, in the case of public colleges, appropriations from state legislatures. The average subsidy a college offers its students is calculated simply as the difference between what it costs to educate the average student and the amount of money the average student ­actually pays.62 Actually, the biggest subsidies occur in the priciest of colleges, as Figure 5.8 illustrates. Grouped again into the seventeen college categories, the bars reveal rather modest subsidies for students at lower-­ranked private colleges, but the subsidies rise to impressive heights for t­ hose attending the highest-­ranked private colleges. At the top, where Harvard and Cal Tech reside, the average subsidy reached a stunning $40,200 per student in 2012 / 13. Over in the public sector, subsidies at most institutions w ­ ere modest, except for the most selective institutions. Especially noteworthy w ­ ere the changes in subsidy rates between 1989 / 90 and 2012 / 13, with subsidy rates in the most selective private colleges (above the 95th per-

142 S upply

Subsidy per student, 2012/13 dollars, group median

$40,000 $35,000 $30,000 $25,000 $20,000 $15,000 $10,000 $5,000

Pub

90+

8,926 11,182

80–90

Priv 7,416

17,234

6,495

5,724

11,943

HBCU 11,182 16,175

7,504

6,472

9,484 14,671

5,365

5,848

9,759

4,042

6,000

6,032

4,383

4,396

8,251

3,259

4,342

5,769

2,903

3,719

40,194 21,641

3,560

2012/13

22,740 10,998

1989/90

3,960

Public

3,109

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0–25

$0

Figure 5.8. ​Subsidy per Student, by College Category, 1989 / 90 and 2012 / 13 Source: Based on calculations for 1,114 colleges and universities with data for 1990 and 2013. IPEDS, Delta Cost Proj­ect. Note: Subsidy is defined as education and related expenditures minus net tuition, where net tuition is sticker price tuition minus grant aid. The enrollment-­weighted average was $8,556 in 1989 / 90 and $6,861 in 2012 / 13.

centile) ­going up spectacularly, rising more modestly in all but the least selective private colleges, and everywhere ­else ­going down. ­These two de­cades saw legislatures across the country cutting their support of state universities, necessitating significant hikes in tuition in the public sector. The result was a general decline in average subsidies across the public sector. ­There is no more dramatic illustration of the in­equality in resources across the colleges and universities offering baccalaureate degrees than a comparison of net assets per student. Encompassing land, buildings, and equipment as well as financial assets, net assets per student ­were manifestly unequal in their distribution. Data on net assets for 2012 / 13, summarized in Figure 5.9, reveal almost

Th e I n ­e q u a l i t y Divi d e n d

143

$1,000,000 $800,000 $600,000 $400,000 $200,000

Private

Public

Pub

Priv

90+

80–90

50–80

99+

0–50

98–99

90–95

95–98

85–90

80–85

70–80

60–70

50–60

25–50

$0 0–25

Net assets per student, category median

$1,200,000

HBCU

Figure 5.9. ​Net Assets per Student, by College Category, 2012 / 13 Source: Based on calculations for 1,102 colleges and universities with data for 2013. IPEDS, Delta Cost Proj­ect.

ludicrously large differences. The wealthiest institutions ­were the private colleges in the top three SAT categories. ­Those in the 95th to 98th percentile group had median assets worth $334,000 per student; t­ hose in the 98th to 99th, $492,000; and t­ hose in the top category, a staggering $1.2 million per student. In no other category ­were net assets greater than $200,000 per student. In fact, the colleges at the bottom, the least selective public colleges and the private HBCUs, had assets amounting to less than $36,000 per student. Between ­these least wealthy colleges and ­those at the most affluent category, in other words, was a gap in net assets of more than a million dollars per student. A graphic illustration of the relative penury of the colleges at the bottom was a row of four large clocks that adorned one wall in the lobby of Howard University’s Mordecai Johnson Administration Building. Evidently designed to give the time in four cities across the globe, they no longer did so when observed by the author on a July morning in 2013. Just three of the clocks w ­ ere labeled (Washington, DC; Cape Town; and Rio de Janeiro), and no two of them had minute hands that matched. Equally telling was the

144 S upply

fact that, three and a half years ­later, ­these clocks had not been repaired, remaining unsynchronized and incompletely labeled. Abundance for the Few

Forces of in­equality—­the growing in­equality of incomes, a red-­hot stock market, and widely divergent investment returns—­produced three sets of effects, all of which increased in­equality across the colleges and universities in the country. The first effect was intensified demand for scarce places at the nation’s most selective institutions. As a group, ­t hese institutions chose not to expand enough to accommodate the demand. One consequence of the resulting inelastic supply was to brighten the glitter on what had already become one of the ultimate prestige items in con­temporary Amer­i­ca. The other consequence was to make it relatively easy for private colleges to raise their tuition. Indeed, price had become for many consumers an indicator of quality. So, for a college to lag too far ­behind one’s competitors could easily send the wrong signal to applicants and their parents. Thus the selective colleges could continue to raise tuition faster than inflation, secure in the knowledge that t­ here was safety in numbers and that low-­income students would be protected by compensating increases in need-­based financial aid. For the less selective private colleges, tuition increases ­were more likely to be a two-­edged sword, potentially driving away the paying students they would need to enroll. How ­these colleges would cope with this dilemma is a topic for Chapter 6. Suffice it to say, the advantages arising from excess demand accrued largely to the haves, not to the have-­nots. The second effect of the forces of in­equality was to boost contributions to colleges and universities. The bulk of ­these donations landed on campuses already lustrous and favored. As shown in Figure 5.4, between 1990 and 2009 the gap between the most selective private colleges and the below-­median public institutions in average donations doubled, rising from $12,300 to $26,400 per student. The third effect, the increase in endowments, likewise helped ­those colleges seemingly least in need of it. The richest colleges

Th e I n ­e q u a l i t y Divi d e n d

145

tended to have the highest shares of equities in their portfolios, ­u nless they had invested in even more lucrative unconventional ­a ssets. The result was a rapidly expanding divergence in endowments. The already hefty gap of $300,000 per student between the most selective private colleges and the below-­median public ones in 1990 had become an astounding $1 million per student by 2013 (Figure 5.5). ­These growing disparities in financial wherewithal translated easily into disparities in the nuts and bolts that go into producing a college education, as illustrated by the doubling of the gap in spending for core educational functions between the most selective of the private colleges and the vast number of public institutions in the bottom half of selectivity (Figure 5.7). It was, in short, a textbook case of the Matthew effect. Even as scholars decried the country’s growing in­equality, the selective colleges and universities where many of them worked ­were enjoying a financial golden age, while the ­great bulk of meagerly resourced institutions strug­gled to make ends meet. An industry already marked by considerable in­equality before 1970 emerged in the twenty-­first ­century as one even more unequal.

146 S upply

6 Zero-­Sum Competition

Soon a­ fter 1970 an ominous fact that demographers had known for some time began to be written about and then worried about. That fact, which had emerged from well-­documented demographic projections, was this: beginning around 1980 the number of eighteen-­ year-­olds in the United States would decline precipitously, falling by a million, from 4.2 million (in 1980) to 3.2 million by the early 1990s.1 During the 1960s and much of the 1970s ­there had been a surge in college applications fueled by the coming of age of the postwar baby boom, assuring most colleges of a dependable flow of high school gradu­ates e­ ager to enroll. This demand had enabled many colleges to become more selective. With justification, they had come to view their admissions office as a gatekeeper, charged with the task of choosing the best applicants from a slew of possibilities. But that golden age would quickly recede with ­these projected declines in the number of potential freshmen. It did not require a slide

rule to perceive the implied threat to the colleges that inhabited the less selective lower tiers, whose livelihoods depended on a dependable stream of paying customers. What had been a sellers’ market quickly turned upside-­down, for all but a small number of selective colleges. Sometime around 1980 the market for baccalaureate education became a buyers’ market. How did competition in the college market in this new environment play out? Answering that question uncovers an industry responding to the new circumstances in ways that seemed to make business sense. Some of the steps ­these tuition-­dependent colleges took to remain afloat w ­ ere based on solid princi­ples of marketing. Other actions that colleges used to improve their prospects w ­ ere simply strategic, while some of them went over the line of acceptable be­hav­ior. In the background of t­ hese actions lay traditions of preference for certain categories of applicants, including the ­children of alumni. This competition was zero-­sum: success required no more nor less than coming out ahead of your rivals. The Plight of the Tuition-­Dependent College

The demographic downturn that began around 1980 was frightening to most but not all colleges and universities. It was most feared by colleges for which two ­things w ­ ere true: they could not be certain of receiving an abundance of qualified applications each fall, and they depended for most of their income on tuition-­paying students. At ­t hese colleges the mood was no doubt grim; it may have been closer to quiet desperation. T ­ hese colleges awaited each year’s batch of ac­cep­tance letters with trepidation, and for good reason, for the revenue that matriculating students would bring with them would make up the bulk of the college’s working funds. In truth, most colleges ­were not very selective. One group of three colleges in Ohio studied by Elizabeth Duffy and Idana Goldberg had already suffered a decline in selectivity by 1980. In the fall of 1965 they had as a group managed to attract a large number of high-­ achieving students: 64 ­percent of their entering students in that year

148 S upply

had ranked in the top 20 ­percent of their high school class. By 1980, however, only 31  ­percent of their entering freshmen had ranked that high.2 More generally, ­these authors found that yield rates (the percentage of admitted students who chose to attend), which had been falling since 1965 due to an increase in the number of schools to which students ­were applying, dropped rapidly ­after 1980.3 A survey of more than a thousand colleges in 1985 revealed that just half had ac­cep­tance rates of less than 80 ­percent. If “selectivity” w ­ ere defined as accepting less than half of the students who applied to your college, only 9  ­percent of private colleges and 8  ­percent of public ones would have satisfied the criterion in that year.4 To be sure, selectivity cannot be summarized by any one metric, but t­ hese rates would suggest that the portion of the colleges that could be thought of as selective was relatively small. It is not difficult to imagine that such stark differences in selectivity would shape the attitude a college’s leaders bring to the annual task of selecting the next year’s entering class. An honest answer to a single question posed to the officer overseeing a college’s admissions and financial aid operation would be as good a way as any to classify colleges: “What is your primary function in enrolling a new class, choosing or recruiting?” Undoubtedly the typical image of college admissions, bolstered by the occasional feature article describing the inter-­workings of a college admissions office, culminating in tense sessions where final decisions are made, would suggest the former. Indeed, admissions officers are often viewed as gatekeepers who control entry into the inner courtyard of opportunity and privilege associated with college. In real­ity, that image has never accurately described more than a handful of colleges, and over the period covered by this study the number of colleges to which it ­applied actually declined. To describe how colleges responded to  a changing set of external circumstances, therefore, we must begin by separating the relatively few colleges for which the gatekeeper label continued to apply from the much larger collection of colleges that enjoyed ­l ittle if any of the market power possessed by t­ hose few.

Z e r o - ­Su m C o m p e t i t i o n

149

What made the dwindling number of eighteen-­year-­olds so frightening to private colleges was that most of them depended heavi­ly on tuition to finance their operations. As a general rule, all colleges face the same basic bud­get constraint: the tuition they receive ­after all the scholarships have been paid must be sufficient to cover expenses that are not covered by nontuition revenue. (The major sources of nontuition revenue are state appropriations, gifts, and income from endowment.) Where “sticker price” refers to the advertised tuition rate, this constraint can be written more formally as [sticker-­price tuition]−[scholarships]+[nontuition income] ≥ [expenditures]

For most private colleges and universities, nontuition revenue was remarkably sparse, leaving them heavi­ly dependent on tuition, net of scholarships. The extent to which private colleges ­were dependent on tuition can be seen in Figure  6.1. All but the most selective among them (­t hose with 1970-­era average SATs below the 90th percentile) depended heavi­ly on tuition as a source of revenue. T ­ hese colleges relied on tuition for more than 70 ­percent of their revenue, based on data for 1990. Among private colleges, it was only the most selective sliver of the spectrum—­those few colleges in the 99th percentile—­for whom tuition did not constitute at least half of their required revenue. T ­ hese few institutions ­were the elite among the name-­brand colleges. For them, the gatekeeper meta­phor remained the most accurate way to characterize their admissions office. In contrast to the high tuition dependence of private colleges, public colleges and universities looked to tuition for only a fraction of their revenue, owing to sizable appropriations from state governments.5 For public institutions below the 80th percentile of 1970-­era SATs, tuition amounted to about 30 ­percent of educational and related spending, and it was just 20 ­percent for the most selective publics. Among the HBCUs t­ here was a corresponding contrast between the private ones, whose tuition covered about half of spending in 1990, and the public ones, whose tuition covered above a fifth. In

150 S upply

Medians, by college category

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

Private

Public

Pub

Priv

90+

80–90

0–50

50–80

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0–25

0.0

HBCU

1989/90 0.72 0.77 0.82 0.81 0.77 0.77 0.77 0.68 0.77 0.63 0.48 0.29 0.32 0.22 0.22 0.53 0.22 2012/13 0.84 0.82 0.85 0.80 0.77 0.78 0.81 0.68 0.63 0.55 0.40 0.59 0.62 0.49 0.49 0.62 0.43

Figure 6.1. ​Share of Education and Related Expenditures Covered by Net Tuition Source: Based on calculations for 1,110 colleges and universities with data for 1990 and 2013. IPEDS, Delta Cost Proj­ect. Note: Bars show the median ratio of net tuition to education and related expenditures. The enrollment-­weighted average of class medians ­were 0.413 for 1989 / 90 and 0.631 for 2012 / 13.

1990, in summary, tuition-­dependence was largely a condition of private colleges—­a ll except the most selective ones. But tuition revenue was to become a much more impor­tant source of revenue for public institutions when state appropriations dwindled ­after the G ­ reat Recession. The Specter of Bidding Wars

Staring the demographic facts in the eye, the multitude of tuition-­ dependent colleges faced the grim real­ity that they would need to depart from business as usual in order to survive in anything close to the manner they had previously operated. In front of them lay four alternatives: slash their spending, do something to continue to attract students of accustomed quality, lower standards, or face

Z e r o - ­Su m C o m p e t i t i o n

151

financial ruin. The first promised no more than marginal savings, but at the expense of abundant push-­back, the third was unpalatable, and the last was unthinkable. They ­were left with just one acceptable option, to boost demand for the ser­vice they had to sell. To accomplish this, colleges responded in the way one would predict if one assumed, as is reasonable, that chief among their aims is survival. The colleges looked to their business officers, who w ­ ere steeped in the princi­ples taught in e­ very MBA program. This newly found business orientation, itself a perspective alien to most faculty, meant seeing the still sizable masses of high school se­niors as potential customers, to be contacted, to be persuaded to apply, and to be convinced to enroll. To a considerable extent, t­hese tasks boiled down to garden-­ variety promotion. A 1979 article by reporter-­turned-­college guide author Edward Fiske described how faithfully the colleges’ new techniques came to mimic familiar business practices.6 Colleges upped their spending on promotion, sending out glossy view books ­featuring campus scenes with happy and attractive students.7 To increase the share of admitted students who would ultimately choose to enroll (the so-­called yield rate), they or­ga­nized “yield parties” in vari­ous cities hosted by local alumni or invited ­these prospective students to campus for choreographed visitation days. To t­ hese promotional efforts they added market research, such as “image surveys” and questionnaires sent to admitted students who had chosen not to attend. This research even spurred some colleges to add new academic programs. Possibly of greatest importance, colleges turned proactive. They began routinely to send printed brochures and other promotional material to students who had not requested anything.8 What made ­these unsolicited mailings more than a shot in the dark was a ser­ vice offered beginning in 1970 by the College Board, the same com­ pany that administered the SAT. For a price, the College Board (through its “Student Search Ser­vice”) would provide lists of names and addresses of high school se­niors who had taken the test. Colleges could order up ­these lists by gender, location, and score. This

152 S upply

ser­vice allowed Wellesley, for example, to obtain a list in 1977 of the top female science students and then to send them an article about its Science Center.9 Another selective college, Williams, used this ser­vice to bolster its attempts to attract more students from beyond the Northeast. In 1983 the college sent material to 5,000 students whose names they had obtained from the College Board. From that group, 265 students applied, of whom 204 ­were accepted and 67 enrolled, a number of entering freshmen that represented 13 ­percent of their entering class that year.10 Just how extensive this form of marketing had become is revealed by the number of names the College Board’s Student Search Ser­vice was selling. In that same year, 1983, the College Board sold an astonishing total of 30 million names to colleges, although just 1.4 million students had actually taken the test.11 In addition to promoting their product, colleges turned to a second strategy to ensure that a sufficient number of applicants, with acceptable credentials, would enroll. This was to offer financial inducements in the form of scholarships. T ­ hese w ­ ere merit scholarships, ones not based on financial need. Although many colleges had long offered such scholarships, it did not require an overactive imagination to see how they might lead to an expensive arms race to compete for the same dwindling group of attractive students. Wrote one observer, “­There are only so many good kids to go around and too many schools who want them. This circumstance has forced colleges to arm themselves with new and unsettling weapons—­ aggressive marketing techniques, hard-­sell persuasion, even offers of large scholarships to students who ­don’t have money worries.”12 The director of admissions at Wellesley wrote, “Unfortunately, ­there are signs that point to the real beginnings of cut-­throat competition. College admissions and financial aid conventions, honored since the early fifties, are breaking down. T ­ here is outright buying or bribing of top candidates, preferential packaging in financial aid awards, disregard of the candidates’ reply date, requirement of large deposits to retain a place—­such practices are becoming more and more prevalent.”13

Z e r o - ­Su m C o m p e t i t i o n

153

Indeed, the use of merit scholarships did increase. Among a group of thirteen liberal arts colleges in Ohio and Mas­sa­chu­setts, for example, the amount of money devoted to institutionally funded scholarships more than doubled over the two de­cades from 1970 to 1990. The median college’s ratio of institutional scholarships to net tuition revenue ­rose from 10 ­percent in the years between 1955 and 1970 to 25 ­percent in 1990.14 Enrollment Management

Techniques such as t­hese to boost demand soon became incorporated into a larger, rather comprehensive set of functions that came to be called “enrollment management,” a way of thinking that would be eagerly embraced by hundreds of tuition-­hungry colleges. This application of marketing princi­ples to a college’s challenge of attracting and keeping tuition-­paying students meant ­going well beyond using the admissions office as a gatekeeper. One version of the genesis of the approach is that it was developed by a physicist-­ turned-­administrator at Boston College.15 The concept was applied first in private colleges—­for whom the importance of tuition as a source of revenue ranges from very impor­tant to all-­important—­ that ­were run by business-­oriented administrators. The approach then spread to middling public universities that found themselves pressed to keep their enrollment numbers up. One early adopter of enrollment management was the University of Miami, a college whose location and prominent football team had proved to be mixed blessings in the university’s attempts to attract an academically strong applicant pool. The city’s reputation for crime vied with its warm winters, and the football team brought attention, not all of it good.16 Shocked into action when it missed its bud­geted enrollment target by 1,000 students in 1981, the university put together a comprehensive planning operation to oversee all aspects of enrollment, beginning with first contacts with prospects and extending all the way to graduation. In 1986 the university hired a consulting com­pany that specialized in enrollment manage-

154 S upply

ment, and the next year it combined admissions and financial aid offices to form an “enrollment management team.” By 1989 it had developed its first linear programming enrollment model. A central component in Miami’s approach was purchasing from the College Board lists of students taking the SAT and PSAT.17 Already the biggest purchaser of names, Miami increased the number of names it purchased in 1994, from 170,000 to 300,000. This jump produced more student inquiries and subsequently led to a higher average SAT among its incoming students, a success that encouraged Miami to keep increasing ­these purchases. Meanwhile, the university cut back on other recruiting efforts that w ­ ere judged not worth the cost, including admissions trips to some high schools and visits by groups of administrators and faculty to cities for recruiting events. To increase retention, the university a­ dopted a model to identify first-­year students in danger of dropping out, ­later adding “geodemographic data ele­ments.”18 It was a broad range of functions that fell ­u nder the heading of enrollment management.19 One consulting firm published case studies highlighting the recommendations it gave to several of its clients. One client was the University of Alabama in Huntsville, for example, whose aim was to increase the size, diversity, “academic profile,” and share of its students who ­were full-­time. ­A fter studying the university’s enrollment office, recruiting tactics, and marketing documents, the firm recommended a series of recruiting campaigns to “create a more identifiable and unique niche in the crowded university marketplace,” with the aim of increasing its appeal to students beyond the north Alabama region from which it traditionally drew students.20 Many colleges and universities brought the vari­ous enrollment management components together ­under one roof. At Purdue, for example, recruitment and retention w ­ ere placed u ­ nder the command of an associate vice provost of enrollment management, who also served as dean of admissions. This office reported in 2011 / 12 that it had a­ dopted the Common Application in order to make it easier for high school se­niors to apply to Purdue and that its staff had

Z e r o - ­Su m C o m p e t i t i o n

155

visited 922 college fairs and 772 high schools. It cited an increase of 7.3 ­percent in the number of visits of prospective students to campus (reaching the staggering total of 39,229) as a sign that “interest in Purdue remains high and that off-­campus recruitment programs and effective communication are working.” The report focused as well on retention, noting that its one-­and two-­year retention rates had reached all-­time highs.21 Of the marketing steps taken by colleges u ­ nder the banner of enrollment management, none was more vis­i­ble, lasting, or controversial than the fine-­tuning they applied to their price structures. The featured attraction was the merit scholarship, awarded to prospective students without regard to financial need. By awarding ­t hese scholarships to applicants they deemed to be especially desirable, a college effectively offered a discount to some customers but not to ­others. Far from being the desperate tactic it might have seemed in the early 1980s, the idea of offering dif­fer­ent discounts to dif­fer­ent customers was actually supported by a well-­established economic model called “price discrimination.” Its basic idea is that u ­ nder certain circumstances, a seller can generate more revenue by charging dif­fer­ent prices to dif­fer­ent groups of customers. The idea works to increase a firm’s revenues by offering the lowest prices (the biggest scholarships) to the most price-­sensitive students, making fewer efforts to entice applicants who are judged to be likely to enroll in any case. In the airline industry it has meant charging more for tickets purchased the day before a flight than for tickets bought months in advance. For colleges, one straightforward application of this princi­ple has been to offer a merit scholarship, or a price discount, to academically strong applicants while insisting that mediocre applicants pay full price. ­Doing so would boost the college’s U.S. News ranking. This price discrimination would be likely to work since the strong applicants are desired by rival colleges while mediocre applicants are not. This logic also inspired some colleges to base their scholarship offers on other bits of information, by being less generous to applicants who showed signs of a fervent wish to attend the college, for example by visiting the campus or listing the college first among colleges

156 S upply

being considered. But exploiting affection in this way appears unseemly if not downright unfair. Johns Hopkins became the poster child for such tactics ­after the appearance of a front-­page Wall Street Journal article in 1996 that spelled out in some detail that college’s efforts to fine tune its offers of merit aid. Applying the princi­ple of price discrimination, their approach suggested being less generous to students who had made a campus visit, since they ­were likely to attend without much financial inducement.22 This logic is at work in the statistical models, produced by the small industry of for-­profit consulting firms that sprang up to advise colleges, designed to predict yield rates for dif­fer­ent categories of applicants. Sometimes t­ hese advisors pres­ent their college clients with options expressed in matrix form showing applicants arrayed according to their SAT scores and predicted f­amily income. Applicants who are high on both scales are the toughest to get and thus require the biggest discounts (that is, scholarships). W ­ hether the analy­sis is done in-­house or contracted out, such business thinking became standard practice for hundreds of colleges facing the challenge of balancing their bud­gets.23 A report prepared for Indiana State University by an enrollment management consulting firm illustrates how ­these concepts could be applied. That report recommended raising scholarships from institutional funds for students whose high school GPAs ­were in the range 2.8 to 3.0. Noting that the average award to Indiana residents in this group had been less 15 ­percent of the average award for applicants at the next-higher GPA level, 3.0–3.3, the report argued that larger awards could attract more of this 2.8–3.0 group. It argued, “As ­these are desirable students (B-­ish students on average), some kind of discounting through institutional aid may very well help drive increased enrollment as t­ hese students may respond enthusiastically to some sort of award (or ‘reward’) for their solid (if not stellar) academic standing.”24 Basic to all of this analy­sis are models that predict w ­ hether or not an applicant ­will enroll. To see what kind of ­factors are considered, consider the same report written for Indiana State. It pres­ents and

Z e r o - ­Su m C o m p e t i t i o n

157

then interprets statistically estimated equations that predict enrollment. One model was estimated for students living sixty miles or more from the campus, evidently using data from the FAFSA form filled out by applicants. The most impor­tant predictor of enrollment was a stated preference to attend Indiana State; applicants who ranked Indiana State first ­were more than seven times as likely to attend than ­those who did not. Other strong predictors ­were ­whether students had made a campus visit or, with less force, made telephone or email contact.25 Some colleges have been criticized for their use of the college preference rankings that students entered when filling out their financial aid forms. To improve their yield and lower their ac­cep­tance rates, some colleges chose to reject some students who had ranked them far down the list, reasoning that ­these students would most likely have turned the college down if accepted. 26 Tactics such as ­these have been criticized as an unsavory symptom of an increasing tendency of colleges to mimic business. Wrote one critic, “Bottom-­feeding private colleges desperate to fill classroom seats hired companies to run their entire recruiting and admissions operation.”27 To be sure, the practice of discounting tuition has been around for a long time, as discounting is at the heart of all college scholarships, both need-­based and merit. But the practice has increased over time.28 Based on surveys of nearly 1,000 four-­year colleges, the percentage of them offering no-­need scholarships increased rapidly between 1979 and 1985, from 60 ­percent to 86 ­percent among public colleges and from 61 ­percent to 77 ­percent among private ones. From ­those levels, the percentages fell over the next seven years. By 2000 the percentage of colleges giving no-­need scholarships ­were 76  ­percent and 73 ­percent in the public and private sectors, respectively.29Among first year students at all four-­year colleges the share who received any merit aid ­rose from 6 ­percent in 1995 / 96 to 14 ­percent in 2007 / 08. The percentages receiving merit aid ­were highest in private colleges near the ­middle of the selectivity distribution. The corresponding percentages for ­these colleges ­were 34 ­percent and 56 ­percent, respectively. Among public colleges, price discounting was by far the most

158 S upply

in the Southeast.30 A more instructive mea­sure of discounting is the discount rate itself, the share of sticker-­price tuition that is forgiven by colleges in the form of college-­funded scholarships. An annual survey of business officers of private four-­year colleges has traced a sizable increase in this rate over time, from 27  ­percent in 1990 to 38 ­percent in the early 2000s to 42 ­percent in 2008.31 The ­factor that drove up discount rates was merit aid, not need-­based aid, of course, which became more and more common, as indicated by the long upward march of the average discount rate. The results of this price discounting can be seen in Figure  6.2, which shows the average discount rate by college category in 1990, 2000, and 2012. It was a phenomenon more pervasive in the private sector. Note that it was not confined to the most tuition-­dependent colleges, but actually r­ ose with SAT rank. At the upper end of the private sector, it was need-­based aid, not merit aid, which drove ­these discount rates up. The public universities, which historically have not used much need-­based aid, or indeed, much merit aid, had low average discount rates in 1989 / 90. This changed in the early 1990s, however, with the advent of state merit aid programs such as Georgia’s HOPE Scholarship. One study compared the percentage of students receiving need-­based and merit scholarships in 1995 / 96 and 2007 / 08. At private colleges, the percentage getting merit aid ­rose from 24  ­percent to 44  ­percent while the percentage getting need-­based aid barely changed. For public institutions, the percentage of students getting merit awards r­ ose over that period from 8 ­percent to 18 ­percent.32 Alongside ­these increases in discount rates, it is worth noting changes in tuition-­dependence. As shown in Figure 6.1 this dependence tended to increase a bit in the two de­cades between 1990 and 2010 among all the private colleges except t­ hose in the most selective categories. ­These increases ­were sizable only in the least selective private colleges (tuition’s share of revenue ­rose from 72  ­percent to 84 ­percent in colleges in the 0–25th percentile private category), underscoring t­hese colleges’ tenuous financial position. Among the public institutions, however, the years ­after 1990 brought significant

Z e r o - ­Su m C o m p e t i t i o n

159

0.45 Median rate by college category

0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05

Private

Public

Pub

Priv

90+

80–90

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0–25

0.00

HBCU

1989/90 0.15 0.18 0.16 0.19 0.18 0.20 0.21 0.22 0.21 0.21 0.25 0.10 0.11 0.12 0.19 0.15 0.10 1999/00 0.24 0.28 0.25 0.30 0.34 0.33 0.32 0.31 0.30 0.29 0.29 0.14 0.14 0.18 0.20 0.19 0.14 2012/13 0.32 0.37 0.35 0.37 0.41 0.40 0.37 0.43 0.35 0.43 0.41 0.16 0.17 0.25 0.25 0.28 0.19

Figure 6.2. ​Median Discount Rate by College Category Source: Based on calculations for 644 colleges and universities with data for 1990, 2000, and 2013. IPEDS, Delta Cost Proj­ect, http://­w ww​.­deltacostproject​.­org​/­delta​-­cost​-­project​-­d atabase. Medians by college category. Note: The discount rate is defined as the ratio (grants to students from restricted and unrestricted funds) divided by (net tuition plus grants to students from restricted and unrestricted funds). For colleges with missing data for 1989 / 90, data for 1988 / 89 w ­ ere used if available or 1990 / 91 if available, other­w ise 1990 / 91 if available. Similar replacements ­were used to replace missing data for the other two points in time. A total of 644 colleges had data usable for all three time periods.

change in the degree of tuition dependence. Owing to the declines in state funding for public higher education, public institutions turned to tuition to make up the difference, and the result was a spectacular doubling of their dependence on tuition as a source of revenue. For public institutions in the bottom half of the SAT distribution, for example, tuition’s share of revenue ­rose from 29 ­percent in 1990 to 59 ­percent in 2009. For the top public colleges, the rate more than doubled, from 22 ­percent to 49 ­percent. HBCUs, both private and public, also saw increases in this dependence on tuition.

160 S upply

The clear exception to the general trend for American colleges to become more dependent on tuition over ­these two de­cades was in the most selective private colleges, t­hose at the 95th percentile and above. By 2012 / 13 the private colleges in the super-­selective 99th percentile depended on tuition to cover just 40 ­percent of their spending. ­Here, then, was another manifestation of the Matthew effect. For public colleges and universities, the jolting rise in their dependence on tuition ­after 1990 seen in ­these figures was a signpost pointing to a new era of in­de­pen­dence, one forced on them by legislatures too preoccupied or other­wise reluctant to support them in the manner to which they had become accustomed. For many of ­these public institutions, the search for new sources of tuition revenue took an in­ter­est­ing and controversial turn. ­Because tuition rates for out-­of-­state students invariably exceeded ­those for in-­state students, universities that could do it had a strong incentive to substitute nonresidents for residents. For the University of California in 2014 / 15, for example, the differential was mammoth, the out-­of-­ state rate of $37,000 a year being three times the in-­state rate. A scathing report from the state auditor noted that enrollment by nonresidents had increased by 18,000 over the previous four years while enrollment by residents had decreased by 2,200.33 Not surprisingly, state legislatures tended to take a dim view of this kind of substitution, as suggested by the fact that many states had long limited the share of out-­of-­state students in their public institutions. Perhaps in recognition of the financial straits its parsimony had caused, the Wisconsin legislature in 2015 removed its ceiling on nonresidents altogether.34 One par­tic­u­lar source of nonresident tuition revenue was foreign students. Although by no means restricted to public institutions, public universities, particularly public research universities, pivoted rapidly to enroll more foreign students in the wake of stagnant support from state legislators. In states where appropriations grew the slowest, enrollment of foreign students increased the fastest. 35 Among HBCUs, rates of tuition dependence increased markedly over the two de­c ades. The public HBCUs, which had enjoyed

Z e r o - ­Su m C o m p e t i t i o n

161

significant nontuition funding, showed a 21  percentage point increase in this de­pen­dency rate. Inevitably, the logic of price discrimination model worked against need-­based financial aid. Purdue’s enrollment management report implied as much: “To make better use of limited institutional funds, the dollars ­were concentrated on high merit students.”36 ­Because most colleges offer need-­based financial aid as well as any merit scholarships they award, the needy student typically generates fewer dollars in net revenue than does the wealthy one. So as long as the college is being strategic in offering merit scholarships anyway, the practice of price discrimination means acknowledging this advantage by looking kindly on t­ hose applicants who appear able to enroll without asking for need-­based aid.37 This conclusion is bolstered by a study of thirteen private colleges that began offering non-­need-­ based aid a­ fter 1987. The share of students at ­these colleges receiving Pell Grants declined within a de­cade of starting their merit aid programs. They also saw a decline in the enrollment of low-­income students and minorities.38 Over all, colleges, both public and private, shifted their limited financial aid dollars away from low-­income, t­oward relatively affluent students. Consider the findings of one detailed analy­sis of financial aid to students classified by ­family income. The study used nationally representative sample of college students covering the 1992 / 93 and 2003 / 04 school years. A mea­sure of the relative generosity of colleges t­ oward affluent students is the average amount of aid given to a student whose f­ amily was in the richest one fifth of the income distribution as a percentage of the average aid given to a student from the bottom one fifth. B ­ ecause a significant amount of all financial aid is need-­based, it would not be surprising if this percentage is less than 100 ­percent, indicating that low-­income students receive more aid on average than affluent students. Indeed, in 1992 / 93, among students at public doctoral universities, this percentage was 62 ­percent. By 2003 / 04 the corresponding percentage was 71 ­percent, indicating that aid given to affluent students, while still less than the average given to low-­income students, had grown

162 S upply

faster. This trend was found in e­ very group of colleges studied. In public nondoctoral institutions, average aid to the affluent as a percentage of that given to low-­income students increased from 74 ­percent to 85 ­percent. In private doctoral universities, it increased sharply, from 36 ­percent to 69 ­percent. The most dramatic change occurred in private nondoctoral institutions, which included the vast number of private, tuition-­dependent colleges. For ­those colleges, the ratio increased from 56 ­percent in 1992 / 93 to 112 ­percent in 2003 / 04, indicating that the average affluent student actually received more than the average low-­income student in ­these colleges. ­There is no better demonstration than ­these changes to illustrate the drift away from supporting low-­income students in the wake of the rise in merit aid. One study also linked the growing reliance on nonresident students to declines in the share of students from low-­income families.39 In conclusion, the practice of price discrimination was widespread, since virtually all colleges gave scholarships to many of their students to defray the cost of attendance. When such scholarships ­were calculated as a function of financial need, their use could arise out of ­either self-­interest or charity, but in ­either case ­there could be ­little objection to the practice. However, when such price discrimination took the form of merit scholarships strategically calibrated to maximize net income, t­ here is ­little that is inherently virtuous in it. Rather, that practice simply highlights colleges’ desire to maximize net revenue, an honest expression of a fundamental instinct for survival, but a policy that is hard to justify on the basis of high princi­ple. The Name-­Brand Colleges

In contrast to the ­great majority of colleges that accepted virtually all students who apply for admission, t­ here remained a small but distinctive set of selective colleges—­t he name-­brand colleges. T ­ hese ­were the colleges whose wealth made it pos­si­ble and credible to proclaim lofty aims, such as t­ hose cited in Chapter 2 for Bowdoin,

Z e r o - ­Su m C o m p e t i t i o n

163

Cal Tech, and Davidson or t­hese, by Amherst: “Amherst College educates men and ­women of exceptional potential from all backgrounds so that they may seek, value, and advance knowledge, engage the world around them, and lead principled lives of consequence. Amherst brings together the most promising students, what­ever their ­financial need, in order to promote diversity of experience and ideas within a purposefully small residential community. . . . ​ Amherst College is committed to learning through close colloquy and to expanding the realm of knowledge through scholarly research and artistic creation at the highest level.”40 At this end of the spectrum, the colleges’ financial planners w ­ ere prudent but confident. ­There would never be enough money to do every­thing that could be desired, of course, but t­ hese enterprises remained on solid footing, and t­here would be more than enough ­eager applicants to allow the admissions office to be very picky in accepting applicants, not to mention to ensure a steady stream of tuition revenue. ­These name-­brand colleges boldly illustrate the profound in­equality that marks the material well-­being across Amer­ i­ca’s vast array of colleges. Most colleges in the United States simply do not have the financial wherewithal to deliver what Amherst College promises. Whereas the bulk of unselective colleges faced the bleak prospects of a buyers’ market, ­those lucky enough to be among the ranks of selective colleges could retain the mindset of gatekeepers choosing among legions of e­ ager applicants. For ­t hese colleges the strug­gle was not about staying in the black but rather about achieving and maintaining a high rank. This was so-­c alled positional competition, in which a college aimed simply to outdo its rivals. By the 1980s, thanks to U.S. News and its imitators, ­these colleges would have a ready metric for mea­sur­ing their success in this zero-­sum competition. Indeed, ­these elite colleges w ­ ere proud of their consistently high ranking by U.S. News, although they rarely boasted, ­preferring to remain above the fray and politely dismiss all such rankings as simplistic. An exception was Amherst’s early exuberance over its top rank among liberal arts colleges in U.S. News’

164 S upply

first reputational survey, published in 1983. The college’s admissions office purchased 25,000 copies of the report and sent them to applicants around the country.41 ­There ­were gradations within this group, to be sure. Among the “Best National Universities” as ranked by U.S. News in its 2014 edition, degrees of selectivity differed markedly. At the very top in that year was Harvard, which accepted just 6 ­percent of t­ hose who applied for admission. Another seven colleges had ac­cep­tance rates of 10 ­percent or less: Prince­ton, Yale, Columbia, Stanford, MIT, Dartmouth, and Brown. Moving down the U.S. News list, this indicator of selectively rather quickly rises, showing that “selectivity” is an inherently relative concept. ­After ­these eight usual suspects, another sixteen colleges posted ac­cep­tance rates of 25 ­percent or less. By the time you get to colleges ranked around sixty, most ac­cep­tance rates ­were 50 ­percent or more. Tied at sixty in 2014 ­were Southern Methodist and the University of Georgia, whose ac­cep­tance rates w ­ ere 42 54 ­percent and 56 ­percent, respectively. The speed at which t­ hese ac­cep­tance rates increase serves to illustrate how small a share of American colleges are highly selective. It is just one small segment of the college market that could be deemed a sellers’ market. ­There was a continuum of selectivity, to be sure, but it was clear that the gatekeeping function, so necessary at the top-­ranked schools, very quickly gave way to recruitment as a college’s major admission challenge. And this point bears repeating: it was not an astronomical price that brought demand for admission slots in this submarket into line with the number available but rather each college’s office of admissions. But that price of admission could be adjusted, even by some of ­these selective, name-­brand colleges. One way the discounting issue pres­ents itself in this category is the bright-­l ine distinction highlighted by a very small group of colleges that make it a policy to base all their financial aid offers on financial need, that is, to offer no merit scholarships whatsoever. This exclusive group includes the Ivy League, but excludes such selective colleges as Duke, Rochester, and Tulane, which do offer scholarships not based on need.43

Z e r o - ­Su m C o m p e t i t i o n

165

The discounting issue arises in a second, more subtle form in connection with the admissions policies that have the potential to treat applicants in dif­fer­ent ways, depending on their presumed ability to afford the cost of tuition without financial aid. Most impor­ tant among ­these policies is “need-­blind” admissions, the pledge that a few dozen of the richest, most selective colleges make. ­These colleges promise to make admissions decisions without regard to an applicant’s financial need, a pledge that rules out price discrimination based on financial circumstances—­assuming the financial aid office meets “full need” and does so in a way that does not differentiate by ­family income. Some quite prominent colleges abandoned the practice, saying it was simply too expensive. Brown, for example, dropped it in 1990.44 Some colleges v­ iolated that latter condition, however, by varying the attractiveness of financial aid packages, offering less desirable applicants more in loans and more desirable applicants more in grant aid. Meanwhile, the tuition rates posted by t­ hese name-­brand colleges increased at a remarkable rate. During the 1980s and 1990s average tuition increased roughly 2 to 3 ­percent a year faster than the rate of inflation.45 Such increases allowed selective colleges and universities to build on their quality. They hired more faculty, allowing teaching loads to fall.46 They raised professors’ salaries, pushing the pay for superstars to previously unimagined heights. ­These ­were the institutions, the ones focused on by many scholars of higher education, whose objectives ­were summarized by Bowen’s law and its vision of worthy proj­ects waiting for the right donor. ­These institutions found it both easy and rewarding to increase their tuition rates. ­A fter all, the rising incomes of the already ­a ffluent kept boosting the demand for places at their colleges; as long as competing colleges w ­ ere ­doing the same t­ hing ­there would be safety in numbers, and need-­based aid remained a safety net to assure that worthy low-­income applicants could attend.47 And one more force might have made t­ hese tuition increases easier to pull off. According to some close observers of the admissions scene in the name-­brand college segment of the market, a Veblenesque price

166 S upply

effect was at work. Referred to by terms like “the Mercedes syndrome” and “the Chevas Regal phenomenon,” consumers ­were said to take tuition itself as a gauge of quality.48 If applicants and their families ­were ­going to look at a lower tuition rate as a signal of low quality, ­there would be no reason for a college not to join with their competitors and raise tuition, as long as that increase was in line with o ­ thers in the pack. Thus in the rarified group of selective colleges, t­ here was competition, to be sure, but it was waged on quality, or perceived quality, not on price. One ready scorecard for this competition was the U.S. News ranking of colleges, a scorecard of special interest to the name-­brand colleges, their public statements notwithstanding. Gaming the System

Given the almost overnight emergence in the mid-1980s of U.S. News as a force in college admissions, it should come as no surprise that college officials would begin to think about what they could do to move up in the rankings. In fact the magazine opened the door to strategic be­hav­ior in 1987, when it overhauled its rankings methodology, reducing the weight given to its semiannual reputational survey from 100 ­percent to 20 ­percent. For the remaining 80 ­percent the magazine used hard numbers and then weighted the results using a highly transparent formula, albeit a formula that they tweaked on a regular basis. Except for the reputational component, therefore, the college itself could control or at least influence virtually ­every other number g­ oing into its ranking. With the aid of a spreadsheet program if not a handheld calculator, an administrator would have ­l ittle trou­ble asking questions like, “If we increased from X p ­ ercent to Y ­percent the share of our classes having fewer than twenty students, how would that affect our rank?” Posing a counterfactual question like this one is an example of reverse engineering. What one calls a­ ctual changes made in order to rise in the rankings is less clear, however, but one uncharitable possibility would be gaming the system.

Z e r o - ­Su m C o m p e t i t i o n

167

Evidence abounds that colleges did take actions in apparent response to the U.S. News phenomenon. The evidence is circumstantial, it is mostly in the form of newspaper and magazine articles, and it almost exclusively focuses on selective colleges. Despite t­ hese limitations, this evidence leaves open the strong possibility that t­ hese actions have indeed been a reaction to U.S. News and that actions like them have been carried out by unselective colleges as well as the selective colleges that are featured in national publications. Perhaps the most impor­tant tactic in this kind of strategic be­hav­ior involves the use of early decision, but other tactics deserve mention as well. Well before 1979 many selective colleges offered prospective applicants an alternative to the conventional method of applying for admission. Instead of sending applications to numerous colleges in January of their se­nior year, students who ­were sure they wanted to attend Smith College, for example, could instead apply in November just to Smith. Forsaking all other colleges, any student taking this option would be committed to attending Smith if admitted. From the standpoint of the college, admitting students by this early decision route had the advantage of improving two of the statistics that went into the U.S. News calculation of rank: the percentage of applicants who ­were admitted (the ac­cep­t ance rate) and the share of ­those admitted students who ultimately enrolled (the yield rate). The requirement that applicants g­ oing this route ­were committed to attend the college guaranteed virtually a 100 ­percent yield rate for this group and for many colleges nailed down a quarter or more of all the available slots. This early decision mechanism allowed the admissions office to reject more applicants out of the regular pool, thus reducing the college’s overall ac­cep­tance rate.49 ­Because ac­cep­ tance rates tend to be considerably higher in the early decision round, counselors report that many high school se­niors feel pressured to take this route, effectively cutting off the opportunity to make a more considered choice among alternate colleges l­ ater on. 50 Beyond early decision, colleges used a host of tactics, some of them of questionable social value, in an apparent effort to jigger the metrics that would be sent to U.S. News. The bulk of ­these tactics worked

168 S upply

on the same two selectivity mea­ sures most amenable to being ­manipulated by the college—­admit rates and yield. For all of ­these tactics, as with early decision, the logic of price discrimination provided the rationale: other ­ things equal, direct your ac­ cep­ tance letters to the applicants most likely to accept your offer of admission. Not only ­will this approach improve your yield rate, it ­will also allow you to accept fewer of the rest, thus pushing down the ac­cep­ tance rate, in the direction of greater selectivity. One tactic, employed by some colleges viewed by high school se­niors as a “safety” school, was to reject applicants with the strongest rec­ords if they had not done something to indicate a strong desire to attend the college, like visiting the campus. In 2000 Franklin and Marshall College chose not to accept 140 applicants with rec­ords at the top of their applicant pool. T ­ hese applicants, who had neither visited campus nor interviewed with a college representative, ­were judged to be unlikely to enroll. Instead of accepting them, the college offered them places on the waiting list, an offer that almost all of them passed up.51 Another piece of evidence that colleges used as a signal of sincerity was ­whether an applicant listed the college as their first choice, ­either on the application or the informational form filled out with the SAT. 52 One other way to assess an applicant’s likelihood of accepting an offer of admission was to place the applicant on the waiting list, figuring that applicants whose hearts ­were elsewhere ­were unlikely to stick around. 53 A college could look beyond the admissions-­determined metrics of ac­cep­tance rates and yield in the effort to improve its U.S. News numbers, of course, but t­ hese actions could run into real money. Of all the reports of colleges that took deliberate steps to move up in the rankings, none was a more spectacular success than the all-­out effort undertaken by Boston’s Northeastern University. Led by a president who made no secret of his aim to reverse engineer the rankings and put money where it would have the biggest payoff, Northeastern ascended an astonishing forty-­t wo places on the U.S. News ranking, rising from number 162 in 1996 to number 49 in 2013. The college lowered the enrollment caps on many courses to nineteen, thus

Z e r o - ­Su m C o m p e t i t i o n

169

increasing the share of classes considered small. They built new dorms in order to increase retention and graduation rates. They hired more faculty to reduce average class size. They began accepting the Common Application, making it pos­si­ble for students to apply with minimal effort, thus increasing applications. The president made a special effort to make contact with presidents of top-100 colleges, whose opinions would be reflected in the reputational component of the ranking. The president even met with the U.S. News leadership, convincing them to exclude Northeastern’s co-op students from the calculations.54 Unfortunately, a few colleges pursued tactics that went over the line. Actions taken in this small group included consistently misreporting their incoming students’ SAT scores and high school GPAs, overstating graduation rates and alumni giving. One college offered admitted students $300 to retake the SAT and a bonus if their score ­rose by more than fifty points. One college admitted that it gave rival institutions low reputational marks across the board. Among the offenders w ­ ere colleges known far and wide. 55 Nonscholastic Favoritism in Admissions

As the foregoing makes plain, colleges ­were engaged in vigorous competition to enroll talented high school se­niors, or ones whose families could afford to pay most or all of the sticker-­price tuition, or ones who offered both attributes. This competition took place, of course, in a certain historical context, one so familiar to con­ temporary observers that it tends to be taken for granted and thus not remarked on. It is worth highlighting one aspect of that historical context—­the preferences that colleges routinely accorded to certain groups of applicants. Three main groups enjoyed favored treatment: ­children of alumni, athletes, and underrepresented minorities. The preferences for each of ­these groups have been controversial in their own ways, although neither of the first two has been debated so vigorously, or in such lofty policy forums, as has the issue of preferences regarding race and ethnicity. H ­ ere, briefly, is a

170 S upply

description of t­hese preferences in terms of s­imple magnitudes, leaving the deeper normative issues for other to discuss. A practice that dates back at least a ­century and a half is that of showing favoritism in admissions ­toward the sons and ­daughters of alumni. 56 Although such legacies rarely w ­ ere actually guaranteed admission, the thumb placed on the scale has at times been substantial. The existence of such a preference can be observed, for example, in the high rates of admission for legacies, as compared to ­those who are not ­children of alumni. Historically, ­these rates have been especially high, or at least prominently cited, at the nation’s most prestigious colleges. For example, for t­hose applying to be in Harvard’s class of 1984, the sons and d ­ aughters of Harvard alumni had a 39.7 ­percent chance of being accepted, while t­ hose applicants without a Harvard connection had just a 12.6 ­percent chance. 57 At Yale, four years ­later, legacies had a 42.5  ­percent chance of being accepted, compared to 19.4 ­percent for the applicant pool as a ­whole. For Yale’s class of 1994, the comparable probabilities ­were 45.4 ­percent and 19.7  ­percent.58 By 2002 the admit rates for legacies at the Ivy League’s most famous three ­were roughly ten times t­ hose for nonlegacy applicants. In comparison to the vanishingly small admit rates for nonlegacies of 4.1 ­percent at Harvard, 3.4 ­percent at Prince­ton, and 2.4 ­percent at Yale, the corresponding admit rates for legacies w ­ ere 39 ­percent, 35 ­percent, and 29 ­percent, respectively.59 One result of ­these dramatically dif­fer­ent rates of ac­cep­tance was an unusually high share of each class at t­ hese colleges made up by legacies, especially among white students. In 2014 a fifth of all the white students in Harvard College’s entering class ­were legacies.60 Of course not all of the difference in admission rates between legacies and nonlegacies can be written off as rank favoritism, since many of t­hese ­children of alumni entered the admissions contest with strong academic rec­ords. To the extent that ­children of alumni pres­ent stronger academic credentials than the average applicant, any ­simple comparison of ac­cep­tance rates overstates the pure advantage of being a legacy. To assess how much ­these observed differences can be attributed to favoritism, some researchers have

Z e r o - ­Su m C o m p e t i t i o n

171

used multiple regression analy­sis to control, statistically, for differences attributable to academic qualifications. In a study of admissions at one selective university spanning three entering classes from 1976 to 1999, James Shulman and William Bowen estimated that, purely by virtue of being a legacy, an applicant had roughly a 20 to 25 ­percent higher chance of ac­cep­tance than another applicant with the same SAT score who was not a legacy.61 Another study, based on data over a similar period for three selective institutions, yielded almost identical estimates.62 ­These legacy applicants ­were distinctive in several re­spects, and consequently they affected the composition of colleges once they enrolled. A study at Duke compared legacies to two groups of nonlegacy students: t­ hose who had a parent with an advanced degree and ­those whose parents had at least one college degree between them but no advanced degree. Compared to ­these two groups, legacy students ­were more likely to be white or Protestant and more likely to have attended a private high school. Eighty-­four ­percent of legacies ­were white, compared to 67 ­percent and 64 ­percent for the two nonlegacy groups, while only 7 ­percent of legacies ­were Asian, compared to 17 ­percent and 15 ­percent for the nonlegacy groups, respectively. Fifty-­t wo ­percent of legacies ­were Protestant, versus 26 ­percent and 42 ­percent for comparison groups, and for private schools 39 ­percent of legacies w ­ ere Protestant, versus 35  ­percent and 28  ­percent. Legacy students had lower SATs on average—1393 versus 1433 and 1405. The legacy students also ­were distinctive in terms of their ­career plans, being less likely than the nonlegacies to aspire to be a medical doctor, a professor or scientist, or an engineer.63 A second category of applicant that has enjoyed wildly favorable treatment over the years is athletes who w ­ ill play on a college’s intercollegiate teams. Unlike the two other categories of students discussed ­here, this form of favoritism in admissions has received very ­little attention, perhaps ­because it is a preference that many ­people accept as a cost of ­doing business. In order to field competitive teams, the argument goes, it is simply necessary to apply dif­fer­ent admissions standards when evaluating exceptional athletes for ad-

172 S upply

mission. In the case of revenue sports—­football and men’s basketball at Division I-­A universities—­the existence of admissions double standards is quite plain. One ready, though imperfect, indicator is markedly lower graduation rates for recruited athletes, especially football and basketball players. Among fifty-­eight public and private universities in Division I-­A, whose average graduation rate was 72 ­percent in 2010, the average graduation rate for football players was 56 ­percent. For basketball players it was 42 ­percent. For uni­ ere even larger. versities highly ranked by U.S. News, the gaps w Compared to an overall average graduation rate of 88 ­percent for fourteen Division I-­A universities ranked in the top thirty-­five in 2010, the rates ­were 68 ­percent for football players and 50 ­percent for basketball players.64 But the preferences applied to recruited athletes go well beyond ­these two spectator sports played at their most commercial levels. In their 2001 book Shulman and Bowen showed that admissions preferences w ­ ere quite substantial even in some of the nation’s most selective colleges. They computed the average SAT for students applying to one of four prominent Division I-­A public universities in 1989.65 Whereas the average SAT score for all students at the four universities was 1154, it was 137 points less for t­ hose recruited to play football, basketball, or hockey and 94 points less for other recruited athletes. For the nine private Division I-­A schools, where the average student entered with an SAT score of 1287, athletes playing one of the three high-­profile sports averaged 284 points below, and ­those recruited to play other sports w ­ ere 122 points below the mean.66 Shulman and Bowen discovered, moreover, that ­t hese gaps had grown wider over time.67 The third form of favoritism, directed t­oward members of some racial and ethnic groups, has been considerably more contentious than ­those for legacies and athletes. Where discrimination or outright prohibition had once existed, ­there sprang up in the 1970s a widespread tendency for traditionally white colleges to show preferences ­toward applicants from underrepresented minority groups. In due time ­these policies met ­legal challenges in such cases as Bakke,

Z e r o - ­Su m C o m p e t i t i o n

173

Grutter, and Hopwood, as well as popu­lar referenda renouncing affirmative action. Surveys of college admissions policies document that such preferences, once common, declined ­ after the early 1990s. In 1979, 45 ­percent of public institutions surveyed reported that they made exceptions to formal academic admissions requirements for racial and ethnic minorities; for private institutions the comparable percentage was 39 ­percent. By 2000 t­ hese percentages had dropped to 20 ­percent and 14 ­percent, respectively.68 To mea­sure the practical effect of such policies, as noted above, researchers have used multiple regression analy­sis, in an effort to control for the effects of other variables. One such analy­sis examined thirteen highly selective colleges and universities. The authors estimated that underrepresented minority applicants received a 28 percentage point boost in the chances of being admitted to one of t­ hese colleges. That is, the chance of admission for an applicant who is an underrepresented minority would be 28  percentage points higher than for a nonminority applicant with the same SAT score. For ­these thirteen institutions, this substantial advantage is slightly smaller than that enjoyed by recruited athletes (30  ­percent), but larger than that for legacies (20 ­percent).69 Another study, based on three selective colleges, found similar advantages. For the 1997 entering class, the authors of the study found that black applicants received a boost of 33 points and Hispanic applicants one of 21 points. Both of ­these ­were down from previous years. In the 1980s the comparable preferences ­were 42 and 40 points for t­hese two groups.70 Over this same period, when preferences for ­these two minority groups had fallen, the boost for athletes had grown, from 25 to 35 points.71 More impor­tant than discounting, the gatekeeping issues that are paramount for the selective colleges tend to be more fundamentally about equity in deciding who gets admitted. ­Because a letter of ac­cep­tance to one of ­these favored institutions is seen, with some justification, as the ticket to a life other­wise inaccessible, the admissions decisions made by several dozen elite colleges have outsized im-

174 S upply

plications for the degree to which the country lives up to the princi­ple of equality of opportunity. This equity issue underlies debates and policies regarding discrimination in admissions, affirmative action, and the preferences historically given to certain groups, especially the c­ hildren of alumni. Competition on Two Fronts

Competition was evident in the market for baccalaureate education on two quite dif­fer­ent fronts. Among colleges t­ here was competition to attract students, the lifeblood of financial security. But not just any students. Ideally, ­those students should not only have strong academic qualifications, but they should also have the financial wherewithal to pay for all or most of the cost of attending. Made more desperate by a decline in the number of high school gradu­ates ­after 1980 and made more efficient by the flood of comparative information on colleges furnished by the U.S. News rankings, this competitive pressure pushed colleges to adopt corporate marketing strategies. Among ­those strategies was to move ­wholesale into the business of offering tuition discounts, reasoning that this form of price discrimination could preserve tuition as a major source of revenue. Such discounting was a strategy mainly used by private colleges, although a number of states did much the same ­thing when they enacted statewide merit scholarship programs. The other front for competition was among students and their parents, made all the more intense b ­ ecause admissions spots at selective colleges grew much more slowly than the demand for ­those spots. This form of competition is addressed in Chapter 9.

Z e r o - ­Su m C o m p e t i t i o n

175

7 Evolution in the Core Business

In 2014 New York City’s College of Mount Saint Vincent faced a choice. Should it add ice hockey to the dozen or so intercollegiate athletic teams it already had? This was one of the options on the t­ able as it sought to make its way in an increasingly competitive market for undergraduate students. Despite its scenic perch on a steep bluff in the Bronx overlooking the Hudson River, the college had strug­gled to attract students. With the assistance of a prominent enrollment management consulting firm, the college had been successful in expanding its applicant pool, in part by buying more names of SAT test-­takers who lived in the Bronx. But marketing could do only so much. Like scores of other colleges facing the rigors of competitive pressure, the College of Mount Saint Vincent considered the advisability of adding programs or facilities, such as an ice hockey team or a new major. ­Whether such potential changes ­were thought of as tools for attracting more students or as means of improving quality

is unknowable. What­ever the ultimate motivation, colleges did make substantive changes as they faced the competitive pressures of a buyers’ market. How did Amer­i­ca’s colleges respond to the market pressures they faced? In this chapter we turn to some of the ways the colleges modified their stated aims, such as the courses and sports teams offered or how they operated. ­T here is ­little doubt that they did make changes, and some of ­these changes ­were roundly criticized. To see how ­these changes played out across the spectrum of colleges we examine adjustments in the colleges’ academic programs and in how the colleges described themselves, paying special attention to their statement of purpose and the majors they offered to undergraduates. We also look at shifts in the use of an impor­tant input, namely, faculty. Trends in the reliance on tenure-­track faculty are calculated for dif­fer­ent clusters of colleges. We examine some information, again by college category, on one indicator of attention being paid to students’ happiness and well-­being, expenditures on student ­ser­vices. Fi­nally, we turn to one of the biggest optional activities a college can undertake, commercial sports. Just Another Business?

In the textbook rendering of competition, market forces are applauded for forcing firms to produce what consumers want, at minimum cost, or ­else. In the market for college, though, anything that looks like slavish obedience to consumers is typically viewed with disdain. Numerous critics of American higher education have taken exception to the ways that American colleges and universities have modified undergraduate education in their efforts to respond to market forces. Reflecting what is surely a widely held suspicion, if not antipathy, among many academic observers ­toward markets in general, colleges have been criticized for being too responsive to market demand, in par­tic­u­lar to student demand. Applying a term that became an epithet when used by critics, Clark Kerr, the long-­ time leader of the University of California through most of the

Ev o l u t i o n i n t h e C o r e B u s i n e s s

177

1950s and 1960s, decried the “shift from academic merit to student consumerism” as a guiding princi­ple for university leaders.1 Critics worried that colleges had fallen ­under the spell of the market model, with the wishes of students exerting undue influence on university decision making. 2 ­Under pressure to recruit tuition-­paying students, the critique went, colleges began to think of themselves as businesses. In the pro­cess they ­adopted the vocabulary that comes with that mercantile outlook—­terms like customer, stakeholder, revenue center, marketing, enrollment management, and, of course, branding.3 One con­temporary critic delivered the ultimate insult, warning that higher education might simply degenerate into “just another business.”4 ­These commercial tendencies ­were thought to have at least four consequences. The first was to diminish the role of the liberal arts. Students worried about job prospects in an uncertain economy, increasingly demanding courses that would readily lead to good jobs. 5 And colleges ­were listening, as illustrated by this statement on Mount Saint Vincent website: “A College of Mount Saint Vincent education prepares students for ­careers, not just jobs.”6 The second consequence of a market orientation, say critics, had been a general decline in standards and a rise in grades. Driven in part by an excessive focus on research and in part by a reluctance to judge students harshly, an implicit “nonaggression pact” had been instituted, whereby faculty demanded less work from students and students demanded less attention from faculty.7 A third general effect of consumerism, according to critics, had been a rush by colleges to add amenities and student ser­v ices that would make students happier. Some observers saw an amenity arms race, in which colleges competed with each other to attract students, using as weapons nicer dorms, up-­to-­date recreational facilities, and fancy dining options that included turning student cafeterias into food courts with gourmet options.8 Along the same lines, observers noted an increase in the provision of student ser­vices, such as counseling.9 One critic likened the newly added student ser­vices staff to camp counselors.10 Another sneered that the undergraduate experience had become “a four year frolic.”11 The fourth consequence that critics saw

178 S upply

arising from colleges’ excessive attention to market forces was the growing use of adjuncts and other nonregular instructors in place of tenure-­track faculty to carry the burden of teaching, done in the name of reducing costs and increasing flexibility.12 Criticisms such as ­these ­were typically laid at the feet of higher education in general, without distinguishing among clusters of colleges or singling out exceptions. Such generalizing is quite consistent with the isomorphism idea—­that colleges tend to mimic each other. Yet ­there w ­ ere exceptions, as sociologist David Riesman argued back in 1980. He pointed to evangelical colleges as one exception to the general tendency to conform. While most of higher education was following society in loosening its ties to or­ga­nized religion and throwing out the rules devised during the era of in loco parentis, ­these evangelical colleges retained much of the old structure. And students continued to enroll at ­those colleges, implicitly accepting that authority.13 The Academic Program: Stated Aims

Four de­cades is a long time, and it would not be surprising to discover that colleges made changes in their academic programs over that period. What is less predictable are what the changes w ­ ere and ­whether colleges of dif­fer­ent types changed in systematically dif­ fer­ent ways. It is instructive to begin by noting how colleges describe themselves—­their purposes and priorities. Looking at a sampling of 1970-­era cata­logs, we can compare how certain colleges describe their aims ­today with how they described them back in 1970. The other sort of change is in each college’s array of majors it makes available to undergraduates. Although lists of majors provided in cata­logs or websites do not indicate the number of students who selected each of them, the lists themselves are an illuminating indication of a school’s priorities and collective expertise. Using the same illustrative groupings of colleges employed in Chapter  2, we can compare the 1970 statements of purpose with their present-­day counter­parts.

Ev o l u t i o n i n t h e C o r e B u s i n e s s

179

Low-­SAT Public Colleges14

The self-­descriptions offered on con­temporary websites by ­these institutions are impressive in their breadth, inclusivity, and generality. In their statements ­these unselective colleges seem to strive to mean all ­things to all potential students. Brimming with corporate buzzwords, they offer mission statements, vision statements, core values, and strategic goals. They pay tribute to transparency, accountability, diversity, shared governance, communication, quality, collaboration, engagement, accessibility, community, integrity, excellence, and teamwork. To an extent unmatched by any other group of self-­portraits examined h ­ ere, the statements of ­these low-­SAT public colleges reflect generalizations couched in the corporate idiom. If one is looking for examples of business lingo, they are ­here. Eastern Michigan states its commitment to “continuous improvement” and “providing greater customer satisfaction.” Towson University states its aim is to “cultivate a campus-­w ide culture of excellent customer ser­vice.” Among the four priorities stated by the University of Southern Mississippi on its website is “image development,” which “promotes the internal and external reputation of the university.” Among the “supporting mea­sures” it intends to use in evaluating its success along ­these lines are scores in the U.S. News and Forbes university rankings, positive national media coverage, and the Gallup Branding Index.15 If it signifies nothing ­else, such ­wholesale adoption of the business vernacular suggests a remarkable attentiveness to the demands of the corporate world. High-­SAT Public Colleges

The websites for the several public universities with high 1970-­era SAT scores are considerably less corporate in style, although the term “mission” certainly does appear. If ­there is a pattern to them, it is to follow the tripartite exposition that characterizes research universities, both public and private, by listing the three ­g reat functions—­ research, teaching, and ser­vice. But some distinctive features remain. William and Mary stresses its history, Georgia Tech its technological

180 S upply

expertise, UC Davis its STEM fields and health-­related research, and the University of Washington its programs in health and aeronautics. ­These public colleges distinguish themselves from lower-­ranked public colleges in their espousal of the liberal arts. William and Mary’s website, for example, states that its educational goal is to provide students with “a superb liberal arts education and the breadth of knowledge that characterizes an educated person.”16 But similar statements can be found on the websites of other highly ranked public universities known as leaders in STEM fields. One of the University of California San Diego’s colleges, Revelle College, “emphasizes intellectual challenge and preparation for competition in a complex world, featuring a well-­rounded, structured liberal arts curriculum.”17 Another public university known for its science, Georgia Tech, similarly stresses the liberal arts: “Our mission is to merge the liberal arts into the larger science and technology-­centered context of the Institute and the world beyond.”18 ­These appeals to the liberal arts theme set ­these selective public colleges off from ­those ranked lower in average SAT, and prestige. Fi­nally, as perhaps an antidote to the sameness of so many of the public colleges’ mission statements, the University of Florida offers this singular declaration: “We are The Gator Nation, a diverse community dedicated to excellence in education and research and shaping a better ­future for Florida, the nation and the world.”19 Lowest-­SAT Private Colleges

Schools in this unselective category include a disproportionate number of religious colleges. Prob­ably the most noteworthy observation that arises from comparing contemporaneous statements with ­those of 1970 is the surprising number of colleges that show no sign of secularization whatsoever. Writing in 1980 sociologist David Riesman noted the steady adherence by some evangelical colleges to their founding belief.20 Not only did ­these colleges stick to their princi­ples, they evidently survived, by attracting students who ­were willing to accept the authority of a college that stood by

Ev o l u t i o n i n t h e C o r e B u s i n e s s

181

t­ hose princi­ples, in an age when most other colleges ­were content to look the other way. Among the schools in our sample seemingly untouched by the steady secularization of American society was the Baptist college in Brownwood, Texas, Howard Payne University. Its website in 2014 revealed no weakening of its 1970 self-­description as a “Christ-­centered academic community,” stating, “Howard Payne University remains dedicated to honoring Christ and serving His Church by providing an educational experience that integrates faith and learning.” Tennessee’s Southern Adventist University (formerly Southern Missionary College) revealed, in its 2014 mission statement, a similar adherence to its longstanding aims: “Southern Adventist University as a learning community nurtures Christ-­likeness and encourages the pursuit of truth, ­wholeness, and a life of ser­vice.” Two of the evangelical colleges in the sample, Lee University and Trevecca Nazaerene University, both in Tennessee, provided statements in their 2014 websites that ­were nearly identical to t­hose used some forty-­four years before.21 A fifth illustration is William Carey University, a ­college whose current self-­portrait reveals, if anything, a stronger religious emphasis ­today than in 1970. By 2015 its website was considerable more fulsome in its references to belief, referring to itself as “a Christian university which embraces its Baptist heritage.” Among its listed core values ­were “We value the truth of God’s Word and its ground values for life” and “We value ser­vice as the fulfillment of the commission of Jesus.”22 Other colleges with religious connections, by contrast, evinced a shrinkage in religious identification. Elon University, which had in 1970 noted in passing its denominational connection, by 2014 had omitted any mention of the connection in its mission statement other than the oblique reference that it “embraces its found­ers’ vision of an academic community that transforms mind, body, and spirit and encourages freedom of thought and liberty of conscience.”23 Another telling sign of its secular turn was its decision to replace its sports nickname, “Fighting Christians,” with “The Phoenix.” In similar

182 S upply

ways, the colleges in the group affiliated with the Roman Catholic Church now tend to place ­little stress on their religious roots. For example, Mount St. Mary’s College of California, in Los Angeles, describes itself on its con­temporary website as a Catholic college. Its program and educational philosophy seem quite similar to nonsectarian colleges, with a statement that includes “We are committed to social justice, re­spect for all persons, and the power of education to transform lives.”24 Similarly, Saint Leo University, located in the Florida city of the same name, offers ­little of an explic­itly religious nature, stating, “We are Catholic, but we are for every­one.”25 Among the least selective private colleges w ­ ere t­ hose that had no connection with any religious organ­ization. ­Today Atlanta’s Oglethorpe University describes its core curriculum thus: “­Every student at Oglethorpe experiences the Core together, examining life’s most vexing questions and building an intellectual foundation for a life, not just a job.”26 The University of New Haven continues the same pragmatic theme of its 1970 cata­log in 2015: “Our mission is to prepare our students to lead purposeful and fulfilling lives in a global society by providing the highest-­quality education through experiential, collaborative and discovery-­based learning.”27 Likewise, York College of Pennsylvania offers a vision in 2015 that is even more explic­itly vocation-­related: “At York College we prepare students as professionals in what­ever ­career they pursue.”28 To what extent statements such as t­ hese translate into course offerings is a question discussed below. This emphasis on pragmatic education could be seen as well in  ­those colleges with religious roots. Elon, which in 1970 had stressed the liberal arts, t­ oday highlights its three professional schools offering undergraduate majors or degrees, citing a “curriculum grounded in the traditional liberal arts and sciences complemented by distinctive professional and gradu­ate programs.”29 Howard Payne University boasts having “100 majors, minors and pre-­ professional programs in six schools,” including schools of business and education.30

Ev o l u t i o n i n t h e C o r e B u s i n e s s

183

Very-­High-­SAT Private Colleges

Neither religion nor vocational training is prominent in con­ temporary self-­portraits offered by the most selective private colleges. The secularism of t­ hese top-­ranked private colleges is perhaps best illustrated by two schools whose 1970 cata­logs made explicit reference to religious values. In their 1970 cata­logs, Davidson College and Wake Forest University both laid considerable stress on their Protestant moorings, but all of this was gone by 2015. Davidson’s con­temporary website mentions character, integrity, “humane instincts,” and “moral courage” in addition to the college’s honor code. This shift parallels changes in governance at the college. In 2005 the college dropped its previous requirement that all members of the governing board be Presbyterians.31 Similarly, Wake Forest, which had still been governed by the North Carolina Baptist Convention in 1970, reached an accommodation in 1979 that ended its financial support in exchange for virtual autonomy from the increasingly conservative denomination.32 Its 2015 website reveals almost no evidence of once having been a Baptist college, nor mention of religion. The closest it comes is this expression of the brand of humanism favored by the most selective colleges: “It’s about us, and the greater meaning [of] what it means to be ­human. Our motto, Pro Humanitate (For Humanity), is a calling to use our knowledge, talents and compassion to better the lives of ­others.”33 A second generalization concerning top-­ranked private colleges is the emphasis they place in their self-­portraits on the liberal arts. As much as anything, this theme takes the form of advocacy for an approach being threatened by the rise in anxiety over employability, especially on the websites of liberal arts colleges. For example, Swarthmore College’s website proclaims, “A liberal arts education fuels lives of purpose and creativity. Its variety and vitality empower students to excel in a rapidly changing world, fostering civic and social engagement, personal growth, and happiness.”34 Brandeis states, “We see a liberal arts education as a pro­cess of self-­ scrutiny and self-­transformation, developing each student’s percep-

184 S upply

tion, reasoning and oral and written expression.”35 Mount Holyoke states, “A liberal arts education consistently emphasizes critical reading and thinking, clear writing and speaking, and leadership.”36 Lehigh writes, “A strong liberal arts education allows our students to think critically and creatively. We work across disciplines to solve ­today’s most challenging issues.”37 Bucknell justifies its liberal arts emphasis by stating, “We believe that a liberal arts education is the best choice for t­ hose who want to think sharply and broadly, communicate well, serve their communities, work ethically and keep learning and growing—­a lways.”38 Bates asserts, “An education which teaches one to critically analyze situations w ­ ill help develop skills sought ­after in most disciplines. Liberal arts gradu­ates develop the ability to learn.”39 Kalamazoo College explains its approach with similar words: “We believe the liberal arts undergraduate learning experience is the best preparation for a life of ser­vice and fulfillment—­a life with the highest pos­si­ble standard of being.”40 Goucher College opines, “The purpose of liberal education traditionally has been to educate ­people who are thoughtful and engaged and who can see the world broadly. They are critical thinkers; they are strong communicators, and they are adaptable to the world’s changing landscape.”41 And Agnes Scott College states, “Liberal arts colleges provide students with the tools to think deeply and critically and to lead lives of purpose and consequence informed by that thinking.”42 When compared with the practical, utilitarian, and vocational themes stressed by the lower-­ranked public and private colleges, t­ hese schools have genuine differences in mission. HBCUs

The five HBCUs with comparable descriptions include both public and private institutions. What the public HBCUs choose to put on their websites is very similar to what less selective, historically white public colleges emphasize. ­These HBCUs offer mission statements and appeal to the same general princi­ples, such as shared governance, diversity, and excellence, mentioning only in passing their

Ev o l u t i o n i n t h e C o r e B u s i n e s s

185

identity as HBCUs. While noting its African American legacy, for example, Savannah State employs terms wholly indistinguishable from ­those of historically white public universities: “The university ­will create an efficient, student-­centered culture, responsive to the needs of its stakeholders, supportive of ideals and ethical verities, and loyal to its rich legacy and heritage.”43 The two private HBCUs in the sample are both church-­related. Of t­ hese, Florida Memorial University hardly mentions its Baptist connection, but does emphasize its HBCU role, naming as one of its core aims “the transmission and preservation of African American history and heritage.” As for its academic mission, it gives equal weight to liberal arts and professional education.44 In contrast, Oakwood University’s statement is thoroughly permeated by religious references. In its statement of values, the first is, “We value the Bible as the primary reference for life’s direction and qualities.”45 The Academic Program: Majors Offered

In 1970 Madonna College, a Catholic college in Livonia, Michigan, offered to its 400 undergraduates a total of fifteen dif­fer­ent majors from which to choose. Alongside conventional liberal arts fields like biology, chemistry, En­glish, history, and sociology, it also made available a few majors that contained a heavy practical, occupational component, among them medical technology, nursing, and m ­ usic 46 education. By 2014 the college’s list of majors had grown to forty­one, a sizable increase, but a smaller percentage than the growth of its enrollment, which more than qua­dru­pled. What most of the new majors had in common was a strong occupational or professional ele­ment, being majors that might reasonably help students find employment in the not-­too-­d istant ­future. Among the thirty-­six majors that Madonna College had added over the four-­decade period ­were the following practical ones: accounting, business administration, computer technology, dietetics, e-­commerce, emergency management, fire science, graphic design, and hospitality and tourism management. Meanwhile, the college had dropped some majors,

186 S upply

thirteen in all, including home economics, sociology, French, Spanish, ­music education, and religious education. Changes such as t­ hese in the menu of academic majors offered by a college provide a telling perspective on its fundamental business. To some extent the waxing and waning of majors inevitably reflect broad trends in the academic world. A trend that has been evident for some time is a decline in the share of students taking courses in the humanities. A study that examined more than five hundred colleges and universities found that the percentage of students with degrees in the humanities dropped from 28  ­percent in 1972 to 17 ­percent in 2009.47 Another study, based on analy­sis of the cata­ logs of some sixty-­three colleges between 1970 and 2000, identified some fields with noticeable declines. Among them ­were Romance and Germanic studies, history, and sociology.48 Another loser was home economics, a major commonly offered in 1970 (even at Wellesley), which had nearly dis­appeared by 2014. But most majors persisted as v iable fields throughout the four-­ ­ decade period, the exception being majors that had barely come into being as fields in 1970. Another study of college cata­logs for the period from 1980 to 2000 identified some of t­ hese rapid growth fields. They included computer engineering, w ­ omen’s studies, and cognitive sciences.49 With ­these national trends as background, individual colleges still had to decide how to shift their own menu of majors, decisions that inevitably posed hard choices. B ­ ecause new majors typically must go through a college-­level approval pro­cess, if not hiring new faculty as well, their appearance and disappearance at a given ­college is unlikely to be accomplished merely as a result of one department’s whim. Call it strategic or call it a response to student demand. What­ever the motivation, ­these changes in available majors represent significant choices for colleges. Invest in a new department or program and keep up with major shifts in research or pay to maintain quality in an old field with lagging enrollments? If colleges ­were anxious to maintain their enrollments, or improve the quality of the students they enrolled, they might have tried to make themselves more attractive by adding subject areas that students could

Ev o l u t i o n i n t h e C o r e B u s i n e s s

187

major in, in something of the way that product variations have proliferated in consumer products like cookies, candy, and soda. Another pos­si­ble response colleges could make, particularly with growing anxiety following the ­Great Recession, would be to offer more majors that promise a potential employment payoff. To explore ­these two possibilities, we can examine information in 1970 college cata­logs and on 2014 websites, looking at undergraduate majors offered at fifty colleges—­a ll the colleges for which the 1970-­ era cata­log could be located online and which fell into one of four identifiable and differentiable groups of colleges, as shown in ­Table 7.1. As this t­ able suggests, how the menus of majors changed depended on the academic selectivity of the college. The number of majors offered at t­hese colleges increased across the board, in some cases by a g­ reat deal. The t­ able shows the median number of distinct majors offered by the colleges in each ­Table 7.1  Number of Undergraduate Majors Offered and Percentage Vocational, 1970 and 2014 Median share of majors vocational

Median number of majors offered

1970

2014

1970

2014

Number of Colleges

Low-­SAT public (0–50th)

33

61

30%

44%

21

High-­SAT public (80th +)

23

99

10%

30%

5

Lowest-­SAT private (0–25th)

20

33

33%

42%

16

Very-­high-­SAT private (90th +)

28

35

7%

3%

8

Category of colleges

Source: 1970-­era college cata­logs and 2014 college websites. For classification as vocational majors, see the online appendix. Note: Sample includes colleges in stated cluster for which 1970-­era cata­logs could be obtained.

188 S upply

cluster. 50 Whereas the numbers ­rose in ­every category, the increases are considerably bigger in the public universities. For the five highly ranked public universities, the median number of majors offered skyrocketed, increasing more than fourfold. Occupation-­related majors increased in share, rising in all but one of the featured clusters of colleges. For the private colleges in the bottom quartile and public ones in the bottom half, the increases are similar, the median occupational share rising from about one-­ third to slightly more than 40 ­percent. It is illuminating to list some of the newly added majors in this occupational category. Except for the least-­selective private colleges, the most commonly added major considered to be occupational is computer science, an area of study that literally came into being over ­these intervening de­cades. Beyond that, however, the new occupational majors differ between the top-­ and bottom-­ranked colleges. In the low-­SAT public institutions, the most commonly added new occupational majors other than computer science are nursing, ­music education, several business majors (management, accounting, finance, and marketing), several sports and recreation majors (athletic training; sports management; and ­hotel, restaurant, and tourism management), and social work. 51 Among the lowest-­SAT private colleges, the most common new practical, occupational majors are nursing, several business majors (accounting, management, and marketing), and exercise science.52 For the high-­SAT schools, the lists look quite dif­fer­ent. Among the top-­ranked public universities sampled, the most common new practical majors besides computer science are computer engineering, health science, business administration, landscape architecture, and several engineering majors (computer, biomedical, biological systems, and materials). 53 The top-­ranked private colleges, as would be suggested by the preponderance of liberal arts majors, have only a handful of new occupational majors besides computer science. They include bioengineering, business economics, and architecture. 54 ­These findings mirror another analy­sis, based on hundreds of college cata­logs, which found that well-­endowed private colleges w ­ ere least likely to drop majors in economics and some STEM subjects55

Ev o l u t i o n i n t h e C o r e B u s i n e s s

189

and most likely to drop education. Notably, the high-­SAT private cluster of colleges, already lowest in the percentage of occupational majors in 1970, saw a decline in that share while the other three experienced increases. 56 One in­ter­est­ing difference emerges along the selectivity dimension: whereas “exercise” appears in majors in the less-­selective institutions, it is “kinesiology” that appears among the selective ones, if at all. ­These findings bolster the notion that the selective private colleges have both the ­will and the way to maintain their longstanding practice of embracing academic arts and sciences majors.57 Not only do the more selective colleges tend to offer a greater share of academic majors, ­there is evidence that, within colleges, a hierarchy of majors has developed, with the best-­prepared students gravitating t­ oward the academic arts and sciences majors.58 Who Did the Teaching

­ here can be l­ ittle doubt that the chief resource at the disposal of a T college as it pursues its aims is its teaching staff. Not only does the teaching faculty play a pervasive, perhaps dominant, role in shaping the learning environment for students, their numbers and credentials inf luenced colleges’ scores in such rankings as U.S. News.59 Although the institution of tenure surely limits how nimbly deans and provosts can respond to shifts in student interests, it is certainly pos­si­ble for t­hese administrators to increase or reshape the teaching force to some extent. For colleges whose main worry is maintaining an accustomed level of academic quality, or, for t­ hose in more desperate straits, simply covering expenses, holding onto successful teachers is a necessity. For colleges aiming to rise in the rankings, holding on is not good enough. In a book published in 1996, I looked in detail at four institutions, all financially strong, that ­were competing hard to attract top students (Harvard, Chicago, Duke, and Carleton College). In addition, the three research universities in the group w ­ ere also competing to excel at research. This meant more than just paying good salaries. It also meant meeting the evolving expectations among fac-

190 S upply

ulty regarding teaching loads, expectations that w ­ ere constantly being revised through the workings of the academic ­labor market, as professors weighed competing offers and colleges learned what their competitors ­were offering. Over the fifteen-­year period covered by the study, 1977 to 1992, average teaching loads in each of three representative departments declined at ­every one of the institutions. The average teaching loads w ­ ere highest at the liberal arts college, Carleton, but teaching loads ­there declined as well, in fact even more rapidly than at the three research universities. Over the period the unweighted average of courses per full-­time equivalent faculty member at t­ hose four institutions fell by 26 ­percent in their chemistry departments, 12 ­percent in their history departments, and 28  ­percent in their economics departments.60 ­These declines w ­ ere not the result of falling enrollments; rather, they ­were the direct result of deliberate policies to increase the number of faculty. ­Whether declines in teaching such as t­ hese ­were occurring outside the orbit of elite colleges is less clear. Certainly a decline was not uncovered in successive surveys of faculty time use over the same period. Between 1975 and 1989, according to successive surveys, the median number of hours faculty spent teaching undergraduates did not decline.61 A related trend t­ hese findings touch on is an apparent increase in the expectations that liberal arts colleges had about how much research would be carried out by their faculty. A survey conducted in 1987 found that faculty who taught at liberal arts colleges spent 65 ­percent of their time in teaching, compared to just 30 ­percent for t­ hose at research universities.62 ­W hether the heavy emphasis on teaching has been reduced remains to be seen. In looking at how colleges have reacted to the changing college market, one big question, then, is w ­ hether they chose to increase the size of their teaching staff, ­either to reduce class size or to attract better faculty by reducing course loads. To be sure, expanding the faculty for e­ ither reason is an expensive proposition, but ­there could also be the prospect of a payoff in terms of stronger student demand or higher rankings or both.

Ev o l u t i o n i n t h e C o r e B u s i n e s s

191

A related question is w ­ hether colleges replaced tenure-­track faculty with adjuncts or other faculty not eligible for tenure. Over the years commentators have raised concerns about the apparent growing reliance on nontenure-­track faculty. Many colleges and universities increased their reliance on part-­time and nontenure-­track faculty.63 Nationwide, the percentage of faculty who are full-­time fell from almost four-­fi fths in 1970 to half in 2007. 64 The percentage of faculty not on the tenure track increased from 19 ­percent in 1975 to 37 ­percent in 2007.65 Apart from concerns that nonregular faculty positions are tenuous and low-­paying, observers worry about relying on instructors who may be less invested in the success of the college and its students.66 Perhaps reflecting ­these shifts, expenditures on instruction have grown more slowly over the past two de­cades than ­those on student ser­vices, research, and other support. Between 1987 and 2008, the real annual rate of growth in public and private two-­and four-­year institutions was 1.1 ­percent, compared to 1.6 ­percent for academic support and institutional support, 2.2 ­percent for student ser­vices, and 2.6 ­percent for research.67 To see what changes colleges have made over time in their teaching faculty since 1970 we can examine detailed rec­ords showing the number of listed faculty, along with their ranks, for specific colleges at the beginning and ending points of the approximately four-­decade period. For 1970 we have lists of teaching faculty by department in published college cata­logs and for con­temporary figures, lists of teaching faculty found on college websites in 2014. Using three disciplines for which most colleges have departments—­chemistry, history, and economics—­instead of disciplines on the rise or on the decline (neuroscience or geography, respectively, for example), allows us to mea­sure like against like but may not be representative of the ­whole. As it is, a number of colleges did not have at least one of t­ hese three departments in one or both of the sample years. To see ­whether patterns of change differed between public and private colleges, or colleges at the top and bottom of the SAT distribution, only findings for the same four illustrative clusters of colleges

192 S upply

used above—­low-­SAT public, high-­SAT public, lowest-­SAT private, and very-­high-­SAT private—­are compared.68 In e­ very comparison, data are drawn only for colleges for which information is available in both years. The ratio of instructors to students provides a rough-­and-­ready indicator of resources devoted to teaching. T ­ able 7.2 shows, for each of the three representative departments and each of the four clusters of colleges, the number of instructors in a given department per 1,000 undergraduates in the college. Since this mea­sure does not reflect enrollments in the classes taught by each department, the ratios cannot be used to infer average class size. The ratios are simply general indicators of resources devoted to three traditional departments that would be central to any liberal arts curriculum. Except for the very selective private colleges, the predominant tendency was for enrollments to grow faster than the teaching staff in ­these departments. Among the illustrative public institutions, undergraduate enrollments generally ­rose faster than the teaching faculty in ­these departments. Relative to undergraduate enrollments, faculty in the public institutions w ­ ere smaller than t­hose in the corresponding private ones, and they tended to be sparsest in the low-­ ranked institutions. With the exception of chemistry in the low-­SAT publics, enrollment growth outpaced faculty growth in ­these departments. The most emphatic example of student growth outpacing that of instructors was in the lowest-­SAT private colleges, where the number of instructors per 1,000 students fell by a third in chemistry, by half in history, and by more than 40 ­percent in economics. ­These are sizable reductions, indeed. They could be explained by bigger class size or the shift away from the liberal arts that was suggested by the new emphasis on occupational majors in ­these colleges. What­ever the under­lying explanation, the large declines in the instructor-­student ratio suggest significant changes in aims or operations over the four-­decade period in t­ hese colleges. Similar but smaller changes occurred in the sampled public colleges. With the exception of chemistry in the low-­SAT public institutions, noted above, enrollments grew faster than instructional staff. Economics saw the largest decline in instructor-­student ratios,

Ev o l u t i o n i n t h e C o r e B u s i n e s s

193

­Table 7.2  Faculty Size in Three Representative Departments, Relative to Undergraduate Enrollment, 1970 and 2014, Colleges with 1970 Cata­logs Available Low-­SAT Public (0–50th)

High-­SAT Public (80th+)

Lowest­SAT Private (0–25th)

Very-­ high-­SAT Private (90th+)

CHEMISTRY Average number of instructors in department per 1,000 undergraduates 1970

1.6

2.5

3.7

5.1

2014

2.1

2.0

2.5

5.3

P ­ ercent change

31%

−19%

−32%

3%

Percentage not tenure track 1970

7%

11%

16%

5%

2014

21%

18%

20%

14%

20

5

13

10

N HISTORY

Average number of instructors in department per 1,000 undergraduates 1970

2.2

2.6

5.2

6.2

2014

2.1

2.0

2.6

6.3

P ­ ercent change

−2%

−21%

−50%

2%

1970

9%

13%

22%

13%

2014

23%

21%

9%

5%

21

5

Percentage not tenure track

N

14

10

ECONOMICS Average number of instructors in department per 1,000 undergraduates 1970

1.5

1.9

3.8

4.7

2014

1.2

1.5

2.2

5.3

P ­ ercent change

−17%

−22%

−42%

14%

1970

15%

13%

13%

11%

2014

15%

19%

13%

2%

19

5

9

10

Percentage not tenure track

N

Source: 1970-­era college cata­logs and 2014 college websites. Note: Sample includes colleges in stated cluster for which 1970-­era cata­logs could be obtained and with the indicated department in both academic years.

with declines of 17 ­percent in the low-­SAT publics and 22 ­percent in the high-­SAT publics. The five high-­SAT publics in the sample had declines also in chemistry and history, each ­going down by about a fifth. The very-­high-­SAT private colleges pres­ent a sharply contrasting pattern. In all three of the representative departments, faculty size kept pace with the growth in undergraduates, and in economics the growth in faculty was slightly more rapid, allowing the ratio of faculty to undergrads, already higher than colleges in the comparison clusters, to increase over the period. Simply put, ­these prestigious colleges took a dif­fer­ent path when it came to staffing t­ hese three traditional liberal arts departments. While public and low-­SAT private colleges w ­ ere economizing in chemists, historians, and economists, the high-­SAT colleges ­were continuing to invest in them, increasing their numbers at least enough to keep pace with the rise in overall enrollments. Turning to the use of nontenure-­track faculty, it is pos­si­ble to calculate the percentage of faculty who are not on the tenure track by recording the ranks of individuals in listings of teaching faculty.69 This information produces T ­ able 7.2’s calculations by college cluster and for each of the three departments. No overall trend is apparent in two of the representative fields. Only in chemistry is t­ here a common tendency, and that is increases in the share of faculty who are off the tenure track. ­There is a similar increase in the history departments, but only in the public colleges. In the two illustrative clusters of private colleges, history courses are increasingly being taught by tenure-­ track instructors. In economics, the changes are minimal, with the highly ranked private colleges showing the largest change—­a near disappearance of nontenure-­track faculty. By 2014 the history and economics departments at the ten representative very-­high-­SAT private colleges ­were relying almost exclusively on tenure-­track faculty, just one more indication of the charmed life ­these colleges ­were leading. Based on information taken from cata­logs and websites for about fifty colleges, T ­ ables 7.1 and 7.2 reveal a widening gulf between prestigious, private, high-­SAT colleges and three other clusters of colleges.

Ev o l u t i o n i n t h e C o r e B u s i n e s s

195

Whereas the colleges in the other three groups showed increasing attention to practical, occupational majors, the high-­SAT private colleges stuck to their liberal arts roots. T ­ hese private, prestigious colleges also managed to match the increases in students they enrolled with increases in teaching faculty, at least in the three sample departments examined h ­ ere. Judging from t­ hese illustrative colleges, it would appear that some of the contrasts that ­were evident in 1970 remained, and became more pronounced, over the next four de­cades. “A G ­ reat Place to Be You”

Judging from the kinds of models that economists routinely employ to explain students’ decisions to enroll in college, chock full of risky investments, expected returns, and opportunity cost, a reader might suppose t­ here was no such ­thing as a fraternity party or a football game or a college romance. Alas, in the real world inhabited by eighteen-­to twenty-­one-­year-­olds spread across hundreds of college campuses, such considerations—of recreation, friendship, and affection—do ­matter. And, if the historians are to be believed, it has been thus from the earliest days of American higher education.70 So in order to attract and retain paying customers has meant that colleges and universities have, with re­spect to their undergraduate customers, always had to balance the faculty’s inclination ­toward academic rigor against the bursar’s expectation of tuition payments. Not surprisingly, spokesmen for the former have occasionally denounced what they viewed as administrative pandering to “student consumerism,” but the predominant reaction to any accommodation to the party life has been silence. When we consider how colleges responded to the economic challenges set before them, no one should be surprised to discover that this fine balancing act between rigorous adherence to academic rectitude on the one side and a pragmatic appreciation of the value of tuition revenue on the other would not abate, but rather loom larger than ever. For the hundreds of colleges for which net tuition

196 S upply

was the lifeblood of continued survival, enrollment management and strategic discounting of tuition could take you only so far. It became increasingly clear that, to make sure enough of ­those admitted students enrolled, and remained in school, ­those students needed to be courted, or at least more attention needed to be paid to their care and feeding. At Elon, for example, this attention took the form of a weekly donut break with faculty and administrators at the center of campus and a “turning 21” ceremony for e­ very student reaching that milestone.71 The care and feeding imperative took several forms, such as spending on student ser­v ices or building consumer-­oriented facilities. The first of ­these, raising expenditures on student ser­v ices, meant spending more on such items as counseling and subsidies for student clubs. An administrator at the University of Miami reported that between 1995 and 2014 the number of sports clubs sponsored by the university more than doubled and the total number of student organ­izations had grown by 42 ­percent.72 Between 1990 and 2013 the percentage of general educational spending by colleges and universities increased from 9.7 ­percent to 12.6 ­percent. As shown in Figure 7.1, the increases w ­ ere largest for the lowest-­ranked private colleges, t­ hose most heavi­ly dependent on tuition revenue. For the private colleges in the least selective cluster, for example, the share increased from 14 ­percent to 19 ­percent over the two-­decade period. This share increased at the least selective public institutions, but not by as much. A second form this accommodation took was in enhanced dormitories, recreational facilities, and dining options. In the place of cafeterias of the 1960s, colleges offered a dizzying array of dining alternatives, often mimicking the food courts of shopping centers, all designed to appeal to the increasingly sophisticated palate of the modern student. Remarking on the newly opened Au Bon Pain on campus, a Duke sophomore praised the restaurant’s “eclectic se­ lection.”73 Colorado State’s plan to build a $226 million football stadium on campus was justified in part by its hoped-­for appeal to students.74 The University of Missouri built a dorm containing an indoor ­water park, with river and waterfall.75

Ev o l u t i o n i n t h e C o r e B u s i n e s s

197

0.18 0.16 0.14 0.12 0.10 0.08 0.06 0.04

Private

Public

Pub

Priv

90+

80–90

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

60–70

70–80

50–60

25–50

0.02 0.00

0–25

Proportion by college category

0.20

HBCU

1989/90 0.13 0.13 0.11 0.13 0.12 0.13 0.12 0.12 0.10 0.11 0.09 0.08 0.07 0.08 0.07 0.11 0.10 2012/13 0.19 0.18 0.18 0.18 0.18 0.17 0.17 0.15 0.12 0.11 0.12 0.11 0.09 0.09 0.08 0.14 0.10

Figure 7.1. ​Student Ser­v ices as Proportion of Total Expenditures, 1989 / 90 and 2009 / 10, by College Category Source: Based on calculations for 1,112 colleges and universities with data for 1990 and 2013. IPEDS. The share was calculated by dividing expenditures on student ser­v ices for undergraduate and gradu­ate students by total educational and related expenditures for undergraduates and gradu­ate students.

In the effort to provide amenities for undergraduates, few colleges could outdo North Carolina’s High Point University. Soon ­a fter it named as president a former motivational speaker, the college launched a building campaign financed in large part by borrowing. Among the dozen or so buildings constructed over a five-­year span was a 277,000-­square-­foot combined student center and dormitory. This fa­cil­i­ty featured four restaurants, a movie theater, and an arcade outfitted with video games, carnival amusements, and a fun-­house mirror. One of the restaurants, which offered at no additional cost steak dinners and waiter ser­v ice, was, according to the college’s website, “a learning lab where students learn the art of fine dining and proper dining etiquette.”76 The theater showed movies, many of them recent releases, almost ­every night of the week, along with complimentary popcorn, drinks, and candy. Prominently located near the entrance to the campus, situated between two dormitories,

198 S upply

was a large swimming pool, surrounded by deck furniture and umbrellas. A putting green adorned the lawn in front of one residence hall. A promenade through campus was studded with more than a dozen bronze statues of famous historical figures, with piped-in m ­ usic audible to pedestrians. Among the details in the vari­ous housing options ­were pool ­tables, plasma screen tele­v i­sions, and granite­top vanities in bathrooms. The buildings constructed by the university included a dozen Georgian-­style ­houses for fraternities and sororities.77 More in­ter­est­ing than such expenditures designed explic­itly to make college more fun ­were accommodations having no obvious bud­getary impact. Perhaps the most impor­tant w ­ ere the accommodations colleges made for alcohol consumption and the collegiate party scene. Usually this meant supporting a muscular Greek system while dealing with underage drinking by looking the other way. This was the case at the large Midwestern University (MU) that was the subject of a 2013 ethnographic study by Elizabeth Armstrong and Laura Hamilton entitled Paying for the Party: How College Maintains In­equality. They write: MU’s campus is dotted with over forty large, stately mansions housing Greek organ­izations. White columns, massive wooden doors, large porches, and meticulously groomed grounds make them one of the more picturesque features of the campus. . . . ​On game days, the street comes alive as Greek students head out in force to tailgate: drinking, partying, and other­wise creating MU school spirit. Thursday through Saturday nights in the fall and late spring, throngs of scantily clad ­women—­d ressed in their party best and stumbling drunkenly in high heels—­can be seen making their way across fraternity lawns.78

As we saw in the descriptions of some of the colleges featured in Chapter 2, fraternities and sororities exist at many colleges, although their members usually account for only a minority of students. Yet ­these organ­izations can play an outsized role in the social life of a

Ev o l u t i o n i n t h e C o r e B u s i n e s s

199

campus. At MU, where just 17  ­percent of students w ­ ere members, the Greek system had disproportional influence, in part b ­ ecause of university policies that granted favored treatment to fraternities. One such policy was to punish underage drinking strictly when it occurred in the dormitories but to allow the Interfraternity Council to police parties that occurred in fraternities.79 In addition, students had few opportunities to socialize outside this Greek system of parties, since “sterile dormitories are structured in ways that reduce coed group interaction.”80 The authors concluded, “MU’s fraternity ­houses thus control some of the most valuable resources on campus: space to congregate socially, a large supply of alcohol, and the promise of ­legal impunity.” At MU, the influence of the Greek system was bolstered by administrative support, institutionalized in the form of an advisory committee of Greek organ­izations that reported directly to the dean of the undergraduate dean. 81 In their marketing, many colleges made a point to emphasize the fun aspects of their colleges. Samplings from college websites included ­these: “Wright State University is like a fun, safe, medium-­sized city with a 24-­hour police department, comfortable living areas, and, of course, a Jamba Juice.” “Bloomsburg University is a ­great place to be you!” “Our 100-­acre, Georgian style campus features plenty of places to study, enjoy a quick bite or just unwind. And our new, state-­of-­the-­a rt residence halls boast some serious creature comforts” (West V ­ irginia Wesleyan). “­There is literally something for every­one who wants to be a part of the William Carey University f­ amily experience. The more involved you are in campus life, the more likely you are to succeed in college and to feel at home.” “At Tennessee Wesleyan College, we believe in hard work and serious fun.” “With sunny Florida skies, palm trees and plenty of green space across our 186 acres, life on the Saint Leo University campus is pure

200 S upply

bliss. Join the diverse group of students who work, study, serve and play together.” “From clubs to concerts to camping or cross country skiing, t­ here is something for every­one. What’s more, thanks to SUU’s residential campus that keeps more than 90 ­percent of students within walking distance, ­there’s always a friend nearby to join in the fun” (Southern Utah State College). 82

Evidence on the effect of amenities on student demand is hard to come by, but one study by Brian Jacob, Brian McCall, and Kevin Stange examines students’ choices of colleges, finding that students ­were more likely, other t­ hings being equal, to choose colleges with higher expenditures on student ser­vices and auxiliary entertainment (including fraternities and Division I sports). 83 Even at colleges that compete at the Division III level, where athletic scholarships are not allowed, many leaders evidently believe that athletic teams attract students. Almost two-­thirds of the college presidents surveyed in 2014 reported that they w ­ ere planning to increase spending on athletics, while just 10 ­percent said they planned to reduce spending.84 If colleges sought to entice students by enhancing recreation and the social scene, it would have been logical to consider as well ways to make academic rigors less forbidding. While no administrator would admit to watering down requirements or dumbing down courses, two facts do suggest relaxed academic expectations. One of ­these is the secular decline in the amount of time undergraduates in American colleges reported spending studying or in class. Whereas the average college student surveyed in 1962 as part of Proj­ect Talent spent 40 hours a week attending class or studying outside of class, the average had fallen to 27 hours a week by the early 2000s. 85 The second piece of evidence consistent with a general easting of academic rigor is the long-­term increase in average grades earned by college students, a trend that afflicted virtually e­ very college whose grades ­were examined. 86 That ­these two trends might be connected is supported by a study that examined course evaluations from twelve quarters at the University of California, San Diego. Using

Ev o l u t i o n i n t h e C o r e B u s i n e s s

201

information provided on ­these evaluations, researchers found that the time students spent studying was lowest for courses for which the predicted average grade would be highest. Spectator Sports

Another realm where competitive pressures might induce colleges to modify how they operated is spectator sports. As described in Chapter 3, this is the form of college athletics whose most numerous beneficiaries are spectators rather than players. It is marked by large stadiums, well-­paid coaches, and membership in the NCAA’s Division I-­A . 87 This form of college athletics has only the faintest connection to the educational mission of the universities where it is practiced (the players must be enrolled students, and many of the spectators are students). By enveloping u ­ nder one roof both an e­ ntirely commercial enterprise and the traditional functions of teaching, research, and ser­vice, universities that engage in revenue-­generating spectator sports in Division I-­A are akin to conglomerates in the modern business world—­corporations that combine seemingly unrelated businesses u ­ nder one corporate roof. Although many scholars of higher education would deny that this commercial enterprise is part of higher education, or wish that it ­were not, the fact is that spectator sports have long been a core function of many American colleges and universities. As discussed in Chapter 3, its prac­ti­tion­ers include most of the largest and all of the “flagship” public universities, a majority of private universities in the highest selectivity clusters, and a smattering of ­others. Competition in this realm is intense, both on and off the field. To hire and retain the most successful coaches, universities respond to market forces, allowing the pay for football coaches, for example, to escalate at rates far exceeding ­those of other university employees. As an illustration, a comparison using data for twenty-­one public universities for the period from 1986 to 2010 reveals that, ­after inflation, while average compensation for full professors increased by 32 ­percent and that for presidents increased by 90 ­percent, the com-

202 S upply

pensation for head football coaches ­rose by 650 ­percent. 88 The universities in Division I-­A also moved aggressively a­ fter 2010, by way of a dizzying series of conference mergers, to consolidate into five large conferences of universities the financially strongest universities in an effort to secure the most favorable tele­vi­sion contracts. 89 Perhaps the best evidence of the value that some colleges and universities place on spectator sports is the decision of a few institutions to launch major sports programs in the face of daunting financial barriers to entry. Despite the rarity of universities that are able to pay for all their sports teams with the money made from revenue sports, ­there is evidence that success in big-­time college sports can attract applicants.90 Revenue sports may also help with fundraising and building po­liti­cal capital for public universities. What is clear is that governing boards appear to believe that the benefits exceed the costs. Colleges responded to changing conditions in an ongoing effort to compete for students and resources. How they responded depended on where they ­were initially. As a result of ­these reactions, the gaps between the prestigious colleges and the rest appeared to widen. The prestigious colleges generally stuck with the liberal arts and did not hold back from increasing their tuition faster than inflation. Other colleges appealed to practical needs of getting a job. In resources the gaps grew as well. Based on a small sample comparing staffing in three departments over four de­cades, the faculty-­student ratio held steady or increased in high-­SAT colleges, but not in general elsewhere. All colleges, but especially the less selective ones, upped their game when it came to dining halls and other student amenities. With few exceptions, it was the rich colleges that became richer.

Ev o l u t i o n i n t h e C o r e B u s i n e s s

203

P A R T III

Demand

8 Scholastic Segregation

Amer­i­ca’s colleges are sharply differentiated by the academic qualifications of their students. ­Whether the mea­sure is test scores, high school grades, or the share of students who earned Advanced Placement credit or won National Merit Scholarships, the variation across colleges can be dramatic. The resulting hierarchy is as sturdy as it is widely recognized. Few with any exposure to college w ­ ill fail to recognize that Stanford and MIT have more star students than comprehensive or open-­enrollment colleges of similar size. As spelled out in Chapter 2, the top private colleges are as richly stocked with high school standouts as the bottom-­ranked colleges are lacking them. This is the face of scholastic segregation. Star students rarely attend colleges alongside mediocre students. This segregation and the academic hierarchy that engenders it are so well known that their existence is entirely unremarkable. They are simply accepted as features of con­temporary life.

This ac­cep­tance is noteworthy in that it contrasts sharply with unease at the K-12 level that surrounds inequities and segregation of any kind. This discomfort at the K-12 level is well illustrated by debates over academic tracking—­the practice of grouping students by aptitude or achievement. It has been a longstanding practice in many American high schools to divide students between college-­prep, general, and vocational tracks. ­Middle schools, sometimes elementary ones, often feature honors classes. Critics have charged that such practices perpetuate socioeconomic distinctions and penalize students assigned to less challenging tracks. On the other side, proponents claim that teaching can be more effective when students with similar aptitude or achievement are grouped together. As controversial as academic tracking has been in K-12 schools, its corollary in baccalaureate education seems to be widely acknowledged and readily accepted. T ­ here are no calls for dismantling an academic hierarchy that features colleges sharply differentiated on the basis of students’ rec­ords of courses, grades, and test scores, one made all the more public, thanks to the establishment and widespread ac­cep­tance of the SAT, and its ­sister test the ACT, as well as to the meteoric rise of the U.S. News ranking of colleges The main question posed in this chapter is ­whether the academic stratification in the college market, so clear in the evidence presented in Chapter  3 for 1970, increased over the following four de­cades. Some previous research suggests that it has done so; t­ hese findings are worth checking using dif­fer­ent data. Before discussing this new evidence, it w ­ ill be helpful to review the connection between this scholastic segregation and the broad developments in the national market for college. We then examine new evidence on scholastic segregation, focusing on changes across the vari­ous categories of colleges over the four de­cades beginning in 1970. The Demand for Quality

A central, indispensable ele­ment in the college market is a set of widely shared beliefs regarding quality differences among colleges. In a

208 D emand

host of ways, aspiring college students and their parents regularly signal their preference for colleges that are deemed better as against ­those of lesser quality. This preference shows up both in detailed statistical studies as well as the choices made by thousands of applicants ­every year as they decide where, among the colleges to which they have been admitted, they ­will actually enroll. Mea­sur­ing quality directly is challenging, of course, for reasons already touched on, not the least of which is the importance of the student’s own effort. Even the most astute scholars have difficulty in mea­sur­ing college quality. Notwithstanding such challenges, a preference for high quality makes perfect sense to most economists who study the demand for college. In the standard economic model the principal effect of education is higher productivity, which normally translates to higher earnings.1 In fact, two economic theories vie for ac­cep­tance, but both end up being consistent with the broad empirical regularity that getting a college degree pays off in economic terms.2 Some readers w ­ ill surely object that such a wholly materialistic explanation, like other models from economics, is too narrow to explain the clear hierarchy of preferences observed in student demand. Sociologists, for example, might argue that students and their parents are driven in their choice of a college by a desire for prestige or status. What­ever ultimately motivates them, students and their parents do make a collective statement about the relative desirability of dif­ fer­ent colleges, and the implied hierarchy is highly correlated to some observable characteristics, including the percentage of faculty holding PhDs and the student-­faculty ratio.3 That t­here exists a mea­sure of college quality to which consumers seem so attuned carries an impor­tant implication for interpreting what has happened to the college market in the last twenty-­five years. Rough notions of relative quality across colleges had long been shared among high school counselors and sophisticated parents of applicants, and t­ hese common understandings ­were perpetuated if not promulgated in part by published college guides, such as Cass and Birnbaum’s Comparative Guide to American Colleges or the Fiske Guide to Colleges.

Sch o l a s t ic S e g r e g a t i o n

209

But it was not u ­ ntil the appearance and rise of the U.S. News operation that the rankings themselves became so well known, right down to minute differences in numerical position. Coming to prominence in the mid-1980s, this ranking ser­vice quickly became a ­factor of the first order in the market for college, as Chapter 4 describes in some detail. Soon a­ fter its debut, it came to be based on a mechanical aggregation of numerical mea­sures. Frequent changes in methodology made for shuffling in ranks, with annual rounds of newspaper articles chronicling ­those changes. Not surprisingly, students and their parents pay attention to all this, and statistical studies have shown that their enrollment decisions are influenced by ­these changes in colleges’ rankings, especially choices involving higher-­ranked colleges.4 This sensitivity means that any college able to break into the top fifty national universities, typically displayed on the first page of the featured article in the annual U.S. News issue, can expect a healthy jump in applications.5 Over time, as demand by aff luent and well educated parents for high-­ quality colleges increased, and the supply of available places in selective colleges failed to keep pace, the perceived quality differences among colleges rapidly took on a heightened importance. And, thanks to their national dissemination, t­ hese rankings soon became national news, thus opening the door to the geo­graph­i­cal widening of the market, at least for the top colleges. Owing to the power of the U.S. News rankings, and colleges’ recognition of this power, t­here developed the potential for a self-­ perpetuating feedback loop. A study by economists James Monks and Ronald Ehrenberg found that a rise in a college’s ranking had three consequences: a fall in the ac­cep­tance rate, an increase in the yield ratio, and an increase in the average SAT of entering students. Since each of t­hese changes improves a college’s ranking, the ele­ ments of a self-­perpetuating cycle are pres­ent. Academic quality is not the only t­ hing students care about, of course. Many of them look to their college years as ones of plea­sure as well as learning. Brian Jacob, Brian McCall, and Kevin Stange captured this pleasure-­seeking dimension in a statistical model of

210 D emand

the ­actual choices made by college applicants. The authors estimated a model of discrete choice to see if students’ ­actual choices are influenced, for example, by expenditures on student ser­vices or the presence of sports teams. The findings imply that students across the board appeared to be influenced, but that only students with strong academic credentials w ­ ere swayed in their decisions by the academic standing of the college.6 A National Market for College?

A subtle change affecting the market for college over time, highlighted and discussed by Caroline Hoxby, is its geo­graph­i­cal expansion. In the not-­too-­distant past, argues Hoxby, the geographic reach of most colleges was local, or regional at most. Most students chose colleges close to home. But, in what might be called the national market hypothesis, two sets of forces helped to transform what had been a collection of mostly statewide or regional markets for colleges into a network that began to look national in scope.7 First, the advent of standardized tests used in college admissions, starting in the 1940s, provided college admissions offices with cheap and comparable information on the academic readiness of students outside their localities or personal networks. The scores, augmented by the development and ac­cep­tance of a standardized method of assessing the financial need of applicants, also gave applicants more reliable information by which to assess their own prospects for admission. 8 Second, the steady decline in the costs of transportation and communication had the effect of diminishing the practical and psychological barriers of remote places, as it was ­doing with travel and immigration. If this story is accurate, ­there would be two verifiable effects. First, the distance that students go to attend college would have increased over time, particularly for students at the top of the college readiness / SAT pile. Second, successive cohorts of students would begin to re-­sort themselves, in terms of mea­sured aptitude, increasing the degree of scholastic segregation across colleges. In terms

Sch o l a s t ic S e g r e g a t i o n

211

of the average academic credentials of their students, colleges would tend to drift apart, as the top colleges garnered more of the best students, leaving the less prepared students for the remaining colleges. This increasing segregation would also produce, as a corollary effect, student bodies more homogeneous in academic terms. An implication of the hypothesis noted by Hoxby is that the correlation between colleges’ resources and the aptitude of their students would increase, since the re-­sorting pro­cess puts the strongest students into the best-­ endowed colleges. She argues, however, that this sorting should be on the basis of aptitude, not f­amily income. Accordingly, Hoxby argues, the correlation between resources and students’ f­ amily incomes would decrease. The national market hypothesis implies that college students over time ­will choose colleges that are, on average, farther from their homes and sort themselves across colleges so that they increasingly have peers of similar aptitude. Having collected and analyzed archival data for hundreds of colleges, Hoxby pres­ents evidence that ­these changes did in fact take place over a forty-­five-­year period. Using data for a panel of 1,551 four-­year colleges, she calculated the percentage of students who attended college within their home state. Among students at public colleges, the share fell from 96 ­percent in 1949 to 84 ­percent in 1994. For ­those attending private colleges, the decline was considerably steeper, with the in-­state share falling from 80 ­percent in 1949 to 55 ­percent in 1994.9 To test the second implication, of increased scholastic segregation, she pres­ents data on differences in SAT scores between and within colleges for a twenty-­five-­year period. To show that differences in student aptitude between colleges have increased, she analyzed data from 1,121 colleges for years from 1966 to 1991, comparing collegewide average SAT levels. She found that the gaps did increase, using a variety of mea­sures of dispersion. For example, the gap between the average SAT score at the college at the 25th percentile and that for the college at the 75th percentile increased, rising from 69 points in 1966 to 180 points in 1991. This growing dispersion occurred among private colleges and public colleges alike. Notably, it was al-

212 D emand

most entirely the result of falling SAT levels at lower-­ranked colleges.10 To show that colleges became internally more homogeneous over time, in terms of student aptitude, Hoxby used data on the distribution of scores by college to calculate standard deviations of scores for the 1,121 colleges in her sample for the same twenty-­five-­ year period, summarizing them as enrollment-­weighted averages, separated into public and private and size category. ­Here again, her calculations are consistent with the hypothesis, with the average standard deviations falling monotonically for ­every one of the categories, the average decline being about 30 ­percent.11 Parents, Meritocracy, and the Application Frenzy

An instinct that could date back to the caveman, but certainly from the earliest day of colleges, is the parental desire to help one’s ­children in all ways pos­si­ble, including in the college application pro­cess. ­Today it seems natu­ral to assume that most parents ­will go all-­out to give their c­ hildren ­every pos­si­ble opportunity to make the best of their potential, although at least one widely cited study argues that it was not ­until the Progressive Era that most parents could afford to stop viewing ­children as extra workers for the ­family but instead as “priceless” trea­sures to be nurtured.12 What­ever its history, the tendency for parents to help their ­children in the college application pro­cess appears to be quite widespread, at least within the social milieu inhabited by social scientists and university faculty.13 According to one sociologist who studies families very closely, Annette Lareau, when it comes to how parents approach their ­children’s school experience, t­ here is a characteristically middle-­class approach that differs sharply from that taken by working-­class parents. Whereas the latter tend to take a hands-­off approach to school, leaving it to principal and teacher to instruct and punish, as need be, for middle-­ class parents the school is just one more of life’s bureaucracies to be negotiated with and conquered.14 This orientation is illustrated by an in-­depth study of an affluent public high school by Peter Demerath. Parents of students at this school not only pushed their

Sch o l a s t ic S e g r e g a t i o n

213

c­ hildren to excel academically, they also stood ready if necessary to intervene with school authorities on their c­hildren’s behalf to achieve desired outcomes.15 If such interventionist instincts operate at the K-12 level, it should come as no surprise that they would extend to the college application pro­cess as well. Before parents could become he­li­cop­ter parents, they would first need to see to it that their ­children got into college. How best to wage this campaign in partnership with their ­children came to be a major worry for parents across the country, and t­here is no better rendering than the fictional Jane Austen in Scarsdale. In this parody, a college admissions counselor advises a roomful of parents anxious about their ­children’s prospects in the upcoming college admissions cycle: “The cornerstones of any successful college application are the two P’s: ‘Push’ and ‘Package.’ “ ‘Push’ means you do what it takes to make your kid crack the books. You nag, you bully, you threaten. Enough with the ‘Sweetie, ­wouldn’t it be nice if you did your homework?’ You gotta say: ‘Do your homework, buster, or ­you’re never ­going to get season tickets to the play-­offs or own a Porsche.’ “And d ­ on’t be afraid to raise your voice!” instructed the speaker sternly. “­These kids ­can’t hear the normal speaking voice. You gotta be forceful, or t­ hey’ll turn you off the way they always do.”16

­ here is ­little doubt that this parental instinct to run interference T for one’s c­ hildren delivers a bigger punch in the top half of the income distribution than it does in the bottom half. Indeed, the c­ hildren of affluent and highly educated parents consistently receive advantages along the way, many of which ­will improve their chances when the year comes to apply to college. Although the Found­ers asserted that all men are created equal, ­there is considerable evidence that the ­children of the affluent get more than their share of breaks.17 To what­ever extent the reader grants that this is an imperative widespread in the population, ­there can be no doubt that the chal-

214 D emand

lenges and opportunities for parents to help their c­ hildren changed markedly over the four de­cades ­under study, especially for parents whose college aspirations included ac­cep­tance to a selective or other­ wise prestigious college. One challenge that grew out of the fact that the supply of admission spots at selective colleges did not keep pace with the growing number of students entering four-­year colleges, making it harder to get into one of t­ hose highly desired colleges. The situation was especially daunting for students living in California and the Northeast. Stocked with more than their share of prestigious colleges, students in t­ hese areas had tended to enroll in t­ hese schools partly due to propinquity. But as the market for selective colleges became more national in scope, ­these students soon found themselves competing with applicants from the Midwest and South.18 The other trend that challenged parents’ efforts to help their ­children navigate the seas of college admissions was the inexorable advance of meritocratic criteria in the college admissions pro­cess. This meant more objective or documentable items on an application form, at the expense of subjective assessments. As the geographic scope of the market for selective colleges gradually expanded, college admissions officers’ local knowledge of high schools came to have less value, forcing ­those officers to rely more on metrics that ­were reliable no ­matter where the applicant lived. This meant the SAT, ACT, and Advanced Placement tests. In the era of affirmative action, the values of other objective indicators of merit ­were enhanced as well, including evidence of awards or participation in worthwhile extracurricular activities. In the face of t­ hese challenges, parents and their ­children engaged in a variety of activities to boost their chances in the college application pro­cess. ­There ­were four kinds of approaches. One was to take all pos­si­ble steps to build a high school rec­ord that could be turned into an impressive college application. This included, obviously, taking a rigorous set of courses, one that included calculus and courses designed for AP testing. Not only did students competing for spots at selective colleges take more of ­these courses, the rates at which they did so increased over time. Based on data from successive national

Sch o l a s t ic S e g r e g a t i o n

215

surveys, researchers estimated that among high school se­niors applying to college the share who took calculus in 1982 was one-­fifth. Among ­those applying to selective private colleges, the share was more than double that, 44 ­percent. By 2004 the share taking calculus among all applicants to four-­year colleges had risen a bit, to 23 ­percent, but the share of t­ hose applying to ­those selective private colleges had increased by 8 percentage points, to 52 ­percent. Similarly, the share of students who took at least one AP exam increased over this twenty-­ two-­year period. For all applicants the share ­rose from 28 ­percent to 45 ­percent. For ­those applying to selective private colleges, the jump was from 60 ­percent to 78 ­percent.19 ­There remained a concern that socioeconomic bias might lurk in the use of AP tests, however. A study published in 2008 reported that affluent students ­were more likely than their low-­SES (socioeconomic status) counter­parts (88 ­percent vs. 81 ­percent) to attend a school with an AP program.20 A second way to increase the prospects for getting into a selective college was perhaps the most straightforward: do something to increase one’s score on the SAT (or ACT). It is no secret that an entire test-­preparation industry has grown up to meet this demand. In 1970 this industry was no more than a microscopic boutique tutoring business in Brooklyn run by an entrepreneur named Stanley Kaplan, whose success depended in large part on recording and using ­actual test questions gathered from test-­takers. When New York state enacted a truth-­in-­testing law in 1979, however, getting old questions became infinitely easier, as copies of past tests ­were required to be made public, thus spurring the industry to unimagined heights.21 That the share of students taking test-­prep courses has grown over time is well documented. Based on the same national samples cited above, of the high school se­niors who applied to four-­ year colleges the percentage who took a private class or had private tutoring to prepare for the test increased from 18  ­percent to 23 ­percent in just the twelve years between 1992 and 2004. For ­those applying to private selective colleges, the percentages ­were considerably higher, with one-­third of such applicants having a private course or tutoring in 1992 and 36 ­percent ­doing so in 2004.22 But t­ hese rates

216 D emand

do not apply evenly across the income distribution. For high school se­niors in 1992 the percentage who paid for private classes or a private tutor was considerably higher among the most affluent than even the next-highest income group.23 For the highest income group the percentage was some 26 ­percent, compared to about 18 ­percent for the next-lower income group.24 It should prob­ably not be surprising that ­there would be an income gradient for engaging private tutors, even if ­those tutors did not all charge the $300 per hour paid by some families who spend their summers in affluent communities on the eastern edge of Long Island.25 A course offered by one of the national chains can be pricy as well, ­running $1,000 or more.26 Another, simpler approach to raising SAT scores was to take them more than once.27 This was often a successful way to boost scores. But ­because following it not only requires another test-­taking fee but uses an additional day that might be spent working for pay, this was a tactic that turned out to be easier for t­ hose with higher incomes. Both the SAT and ACT test allowed a prospective test-­taker to apply for extra time to take the tests on the basis of a condition that makes it difficult to concentrate, such as dyslexia or ADHD. Between 1988 and 2003 the percentage of SAT takers nationwide who requested and received special accommodation for the test tripled, providing one bit of suggestive evidence that parents might be using this accommodation to gain an advantage for their ­children.28 Another piece of circumstantial evidence came ­after 2003, when the SAT ended its practice of indicating to colleges which students had received an accommodation. In the year following this change, the average verbal score achieved by t­ hose with accommodations saw an 8-­point increase, the largest one-­year jump ever. The increase in the math score for ­those with special accommodations increased almost as much. In Washington, DC, a state-­level jurisdiction with an unusually large share of affluent test-­takers, the increase for ­those with accommodations was especially large—30 points in verbal and 29 in math. Other evidence from California suggested that accommodations w ­ ere being utilized disproportionately by advantaged students. Whereas 68  ­percent of the state’s regular SAT-­takers in

Sch o l a s t ic S e g r e g a t i o n

217

2000 ­were white, 84  ­percent of ­those taking it with accommodations ­were.29 One explanation for such a difference may lie in the cost of having a child tested privately, which can exceed $1,000. 30 A third strategy ­adopted by students and their parents was to acquire any other skills or recognition that might improve the chances of gaining admission to a desirable college. One striking bit of evidence that families ­were pursuing this strategy was an increase in the mid-1990s in the amount of time that parents ­were devoting to childcare, particularly for older ­children. Relying on a succession of time-­use surveys spanning more than four de­cades, Garey and Valerie Ramey demonstrated that the increase was roughly twice as big for college-­educated parents as for parents with less education. Among the activities with the largest differences by education ­were “organ­izing and attending activities” and “chauffeuring.”31 Employing the ­minivan, team roster, and orange slices, therefore, affluent parents could assist their c­ hildren in learning the violin, or soccer. Especially helpful was learning to play a sport, since most selective colleges give preferences to applicants who can play for one of their teams. Statistically speaking the best sports for high school students to learn for this purpose are ones that are relatively rare in high school but that are played by selective colleges, such as lacrosse, crew, squash, even rugby.32 More generally, ­children who aspired to be strong applicants ­were well advised to acquire activities or recognition that might impress admissions officers. At the extreme, this effort to load dossiers led to “hypercredentialing.” In his study of a suburban high school, Demerath observed that many parents pushed their ­children to make high grades, helped them with homework, and encourage them to fill their time with activities. Students for their part strived to receive extra credit when pos­si­ble. In 2003 the school named forty-­seven valedictorians, all of whom had GPAs of 4.0 or more. 33 A final approach in this admissions frenzy is to be strategic about the application pro­cess itself. For parents with means, ­there are private college admissions counselors, even admissions boot camps, where rising high school se­niors learn tips for writing the admissions essay, presenting their accomplishments, preparing their ap-

218 D emand

plications, and making the most of interviews with admissions officers. 34 Among purely strategic actions, choosing to apply in the early decision round may be the most consequential. As outlined above, this practice was developed by colleges to improve the numbers they report to U.S. News, but it relies on applicants and their parents to be willing participants. New Evidence on the National Market Hypothesis

Using the Freshman Survey data described in Chapter 2, we can explore the national market hypothesis over a more recent period than Hoxby did, using more fine-­grained categories of colleges. The data include average SAT scores for more than 1,100 colleges and survey results covering students from three survey waves who attended one of 188 colleges. Organ­i zing the Freshman Survey data by college groups creates a panel that is similar in structure to the sample of college-­level data analyzed by Hoxby. The result is another panel of colleges, smaller in number and l­ ater in time. For 1970 and 2010 we have scores for each college. In addition, for the survey waves 1989 / 90 and 2008 / 09, we have self-­reported SAT scores (or ACT scores, suitably converted, or their weighted average) provided by individual students, each of whom is identified in the data with a group of similar colleges, not a par­tic­u­lar institution. The first implication of Hoxby’s model of the college market is that the geo­graph­i­cal reach of colleges, especially selective ones, ­will expand over time. That is, as the costs associated with distance recede, the distance that students travel to go to college w ­ ill tend to increase over time. First-­year students w ­ ere asked in the Freshman Survey to indicate how many miles their college was from their home. Figure 8.1 shows the percentage of students who lived more than 500 miles from their college for each of the seventeen college categories repeated for each of the three survey waves.35 Between 1972 and 2008 / 09, the share of students traveling more than 500 miles increased in nearly ­every category of college. The increases ­were largest in the private colleges in the 95th to 98th and 99th and above

Sch o l a s t ic S e g r e g a t i o n

219

Proportion of students

0.5 0.4 0.3 0.2 0.1

Private 1972

Public

Pub

Priv

90+

80-90

50-80

0-50

99+

98-99

95-98

90-95

85-90

80-85

70-80

60-70

50-60

25-50

0-25

0.0

HBCU

0.12 0.30 0.10 0.13 0.16 0.19 0.26 0.39 0.20 0.32 0.33 0.06 0.06 0.06 0.46 0.34 0.07

1989/90 0.18 0.20 0.11 0.13 0.22 0.22 0.35 0.43 0.30 0.40 0.43 0.06 0.07 0.06 0.41 0.44 0.18 2008/09 0.17 0.22 0.12 0.13 0.22 0.25 0.33 0.46 0.35 0.41 0.48 0.08 0.12 0.07 0.28 0.42 0.16

Figure 8.1. ​College Is More Than 500 Miles from Home, 1972, 1989 / 90 and 2008 / 09 Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). The question was, “How many miles is this college from your permanent home?” Note: The nationwide enrollment-­weighted average percentages w ­ ere 0.12 for 1972, 0.13 for 1989 / 90, and 0.15 for 2008 / 09.

percentiles, categories that included colleges such as Davidson, Northwestern, Cal Tech, and Swarthmore. Even ­after the period examined by Hoxby, therefore, the tendency she hypothesized and documented continued to operate. 36 A feature of the graph at least as significant as the time trend is the sizable differences across the groups. In 2008 / 09, for example, just 8 ­percent of students at below-­average public institutions (which ­were, ­after all, populated mainly by in-­state students) traveled more than 500 miles, whereas almost half the students at private colleges did. Distances for students attending private colleges tended to be greater than t­hose attending comparably ranked public colleges (with the exception, again, of the anomalous 90+ public group). One noteworthy feature of this graph is the long distance traveled by stu-

220 D emand

dents in the private HBCUs. In 2008 / 09, more than 40 ­percent of first-­year students attending one of ­those HBCUs traveled more than 500 miles to go to college, a share considerably higher than for any of the groups of private colleges below the 90th percentile. It would appear the market for private HBCUs has as good a claim as any to be called “national.” It is worth noting some findings from other research that links distance to other characteristics of students. A study by Griffith and Rothstein (2009) revealed that students who attended selective colleges had a proximity advantage over ­those who did not. Comparing students’ homes to the closest selective college (not necessarily the college they attended), t­ hose who did attend a selective college lived closer to one than ­those who did not. Moreover, students living in the Northeast, where many of the established, elite colleges are located, tended to live closer to a selective college than students in other regions.37 ­There was also an income bias to this proximity: for students in the highest income quartile, the closest selective college was eighty-­seven miles away, whereas for ­those in the bottom quartile, the closest one was ninety-­five miles away. The second implication of the national market hypothesis is scholastic segregation. This means that the most selective colleges ­will see their successive cohorts of matriculants getting academically stronger over time, while the academic qualifications of students in the less selective colleges ­will trend in the other direction. To see if this kind of scholastic segregation in fact increased in the four de­ cades ­after 1970, we can examine data on standardized test scores and high school grades. Employing a variety of data sources, 1,085 colleges with published data on enrolled students’ SAT and ACT scores ­were identified for both 1970 and 2010. ACT scores and 1970 SAT scores ­were converted to the recentered SAT scores used in 2010, and enrollment-­weighted averages w ­ ere calculated for each college where both tests ­were taken. For ease of explication, the resulting averages are simply referred to as the SAT score. Figure 8.2 shows ­these collegewide average SATs for 1970 and 2010. 38 T ­ hese

Sch o l a s t ic S e g r e g a t i o n

221

200 100 0 −100

Private

Public

Pub

Priv

90+

80–90

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−300

25–50

−200

0–25

Average SAT minus mean for all colleges

400 300

HBCU

1970 −107 −29

9

33

67

98

123 169 229 266 319 −78

36

110 186 −296 −285

2010 −65 −30

22

22

52

50

139 189 258 305 359 −58

7

91

159 −212 −228

Figure 8.2. ​Average College-­L evel Mean SAT, Deviation from Mean, 1970 and 2010 Source: 1970: college guides; 2010: IPEDS. Collegewide average SAT scores for 1,085 colleges with information on scores and enrollment in both years. ACT scores converted to equivalent SAT scores; 1970 scores are recentered. 2010 scores based on mean of 25th and 75th percentile scores. Note: Averages are shown as deviations from the enrollment-­weighted means for all first-­year students, which ­were 1104 for 1970 and 1081 for 2010.

averages are shown separately for each of the same seventeenth categories of colleges described in Chapter 2, each category’s average shown in comparison to the national average, e­ ither above or below. Recall that each pair of bars is based on data from precisely the same set of colleges. Over the four-­decade period, average SAT scores for the nation fell slightly. Notably, they fell relative to the national average in the top three categories of predominantly white public institutions, as they did in private colleges between the 60th and 85th percentiles of the 1970-­era SAT distribution. In the top-­ranked private colleges, t­ hose above the 85th percentile in 1970, however, the averages did go up. This pulling away at the top echoes Hoxby’s findings for collegewide mean SATs for an earlier period. This finding is not supported, however, by trends in SATs based on students’ reports of their own scores. In 1989 / 90 and 2008 / 09, but not in 1972, the Freshman Survey asked students about their scores

222 D emand

200

100

0

−100

Private

Public

Pub

Priv

90+

80–90

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−300

25–50

−200

0–25

Average SAT minus mean for all first-year students

300

HBCU

1989/90

−28 −57 −23 −16

62

46

90

123 166 233 236 −60

52

59

116 −95 −227

2008/09

−24 −75

49

5

93

143 168 205 230 −25

11

56

73 −131 −248

30

−49

Figure 8.3. ​Average SAT, Deviation from Mean, by Category, 1989 / 90 and 2008 / 09 Source: Freshman Survey, students from 188 colleges, two waves (1989 or 1990 and 2008 or 2009). Note: The figure shows the difference between the average SAT in the college category minus the enrollment-­weighted mean for all first-­year students. The enrollment-­weighted means for all first-­year students in four-­year colleges w ­ ere 1123 for 1989 / 90 and 1197 for 2008 / 09.

on the SAT and ACT.39 Figure  8.3 shows the mean SAT scores by group for the two waves of surveys that asked about scores, again expressed as differences from the average score at each time. The figure reveals the same marked differences by college category seen in Figure 8.2, indicating the strong tendency for colleges not to stray far from their past ranking. However, the average scores converged rather than diverged over the two-­decade period from wave 2 to wave 3. To be sure, t­ hese two figures are not necessarily contradictory, since they cover dif­fer­ent periods. But this second set of means does not indicate scholastic segregation.

Sch o l a s t ic S e g r e g a t i o n

223

The Freshman Survey offers another mea­sure of academic readiness for college—­g rades in high school. In 1972 the first-­year students attending the colleges with the highest average SAT scores in 1970 also tended to be ­those most likely to report that their average grade had been A or A+, as shown in Figure 3.1. In successive years, this correlation held up, but with one unmistakable addition—­grade inflation. The percentage of all first-­year students in the nation reporting that their high school average had been A or A+, based on the enrollment-­weighted average from Freshman Survey responses, ­rose from 9.6 ­percent in 1972 to 14.0 ­percent in 1989 / 90 to 24.5 ­percent in 2008 / 09, a stunning increase. To see through this grade inflation and determine ­whether disparities between colleges actually increased over time, Figure 8.4 shows the percentage of first-­year students reporting that their high school grades ­were A or A+, again as deviations from each year’s average for all four-­year college students.40 This figure makes it clear that t­ hese disparities did increase, as the shares of students with reported A or A+ averages in high school increased for the groups of initially highly ranked private colleges. For the highest-­ranked private colleges, the jump in reported grades in 2008 / 09 is remarkable. By 2008/09, whereas just 20 ­percent of students attending public institutions in the bottom half of the SAT distribution reported having A or A+ averages in high school, the share of ­those in the top private colleges was a whopping 63 ­percent. The only negative trends of note w ­ ere for private colleges in the 80–85th percentiles, public colleges in the 90th and above, and the public HBCUs. Indeed, the decline in the academic fortunes of the HBCUs, reflected also in the reported SAT scores in Figure 8.3, has been a source of concern, as their historic role has shifted as traditionally white colleges have attracted many high-­achieving black students.41 In sum, although ­these figures are based on self-­reports, they produce as convincing a piece of evidence as any that the pro­cess of scholastic segregation across colleges did indeed continue unabated a­ fter 1990.

224 D emand

0.30 0.20 0.10 0.00

Pub

Priv

90+

−0.17 −0.11 −0.08

0.14 0.18 0.09

−0.08 −0.05 −0.06

80–90 0.04 0.05

0.00

0.02

0.01

HBCU

0.01

0.28 0.38

−0.05 −0.06 −0.03

0.24 0.27

0.29

0.19 0.22

0.21

0.12 0.18

0.15

0.05 0.07

0.08

0.06 −0.01

0.04

0.02 0.06 0.03

0.04

0.00 −0.01 0.02

0.00 0.02

2008/09

Public

0.03

1989/90

−0.05 −0.03 −0.04

1972

−0.02 −0.02 −0.04

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−0.20

25–50

−0.10

0–25

Proportion A or A+ for college category minus U.S. average

0.40

Figure 8.4. ​Share of First-­Year Students with High School Grade Averages A or A+, Deviations from Mean Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). The figure shows the difference between the proportion of students in the category who reported having a high school grade average of A or A+ minus the enrollment-­ weighted mean for all first-­year students. (The question was, “What was your average grade in high school?”) Note: The enrollment-­weighted means for all first-­year students in four-­year colleges w ­ ere 0.096 for 1972, 0.140 for 1989 / 90, and 0.245 for 2008 / 09.

Other Mechanisms Influencing Demand

The hypothesis analyzed above focuses largely on distance and its declining importance to students. But t­ here is another, unspoken desire for quality. In addition, price, or tuition net of financial aid, has been found to be a significant influence on enrollment decisions. Consider financial aid, an ever-­present and growing f­ actor in students’ decisions. ­Because of its growth, the effect of tuition differences has dissipated over time.42 In making its offers of admission, a college

Sch o l a s t ic S e g r e g a t i o n

225

can customize ­these costs by applicant, a technique that is part of the science known as “enrollment management,” described in Chapter 6. For reasons tied up with the growing importance of college rankings, many colleges have made it a practice to offer merit scholarships to students with strong academic credentials. As to the effect of aid on student choices, economic studies of demand for college have long supported the standard prediction that students w ­ ill be attracted by lower costs of attendance and repelled by higher costs. But research in the last de­cade suggests that rationality does not always prevail. One study of high-­achieving high school se­niors revealed that many of them made decisions that could not be explained in terms of dollars and cents. For example, many students ­were influenced as much by a dollar of loans as by a dollar of grants, although the two are not equivalent. Students ­were more likely to respond favorably if a grant was in the form of a scholarship with a name. In addition they appeared to care about the percentage of costs covered by an award more than about the amount of the award.43 Is This Anything New?

Based on every­thing we know, colleges in the United States have always been quite stratified by the academic aptitude of their students. In a pro­cess described by Caroline Hoxby, colleges became more stratified a­ fter World War II, as improvements in transportation and communications made it more feasible for students to attend colleges far removed from their homes. Judging from the distance students traveled to go to college, this pro­cess continued during both the first two de­cades of our period of study for most groups of colleges and the second two de­cades for the most selective colleges. Another indicator of stratification is seen in the high school grades of entering students. By this metric, the most selective colleges continued to distance themselves from the rest. Unlike other markets whose geo­graph­i­cal limits ­were spreading, the consumers in this market, especially in its upper echelons, ­were

226 D emand

natu­ral rivals, ­because some suppliers, the selective colleges, had chosen to limit the supply of admissions spots. At the same time, to an extent that went beyond the high degree that had already occurred, admission to elite colleges became a must-­have prestige good. Unlike most other items bought and sold in markets, however, this one could not be purchased simply by bidding up the price. For affluent professional ­couples, some who had grown accustomed to hefty annual bonuses and most whose life experience had been replete with success in achieving goals and abundance in consumption, this exception to the usual market rules and the prospect of rejection at this vital gateway presented a challenge that was as unwelcome as it was unfamiliar. It was a challenge that could be overcome, however, but only with the full cooperation of the ­children who would become applicants. Not only did colleges compete with each other, so did consumers, or rather, aspiring consumers. This competition came in the form of extraordinary efforts to achieve the goal of admission to a desirable college, using any means at hand.

Sch o l a s t ic S e g r e g a t i o n

227

9 Economic Stratification

Many reports and op-ed pieces in recent years have decried the inequities in American higher education and warned that disparities are increasing. Elite colleges have been called “bastions of privilege,” and in their admissions pro­cess, according to an article in the Atlantic, “students from well-­off families have a big advantage.”1 One study was subtitled, “how colleges compete for wealthy students and leave the low-­income ­behind.”2 According to one observer, “higher education has become a power­ful force for reinforcing advantage and passing it on through generations.”3 ­These denunciations echo a longstanding theme among sociologists of education, that the systems of education in advanced countries are designed to be unequal.4 More than a few observers have opined that in­equality is intensifying. 5 As Chapter 8 makes clear, the entire system of four-­year colleges and universities is indelibly marked by large differences in both resources and student ability, and has been for a long time.

The more pressing question is ­whether ­these disparities have actually been increasing. Much editorializing seems to take this as an established fact, viewing the extent of current disparities between colleges in resources and socioeconomic status as something new. But it is by no means obvious that disparities are any bigger ­today than they w ­ ere four de­cades ago. The aim of this chapter is to examine evidence bearing on this question. ­Whether developments in the market for college have resulted in more pronounced economic segregation is a consequential question to explore, for educational as well as social reasons. The rising economic rewards for college attainment, noted previously, have heightened the importance of access and diversity, traditional aims of U.S. postsecondary education policy.6 The stakes are further raised ­because the characteristics of a student’s peers ­matter in impor­tant ways as a component in the pro­cess of learning. A growing body of research has shown that peer effects are a power­ful force during the college years in influencing students’ academic achievement, their attitudes, and their be­hav­ior.7 As a social mechanism, the composition of college student bodies is also impor­tant, for example, in the pro­cess of finding marriage partners.8 ­Whether enrollment patterns are increasingly producing socially more homogeneous student bodies, as they appear to be producing more academically homogeneous colleges, is a question of real significance, b ­ ecause the byproduct of more homogeneous colleges is, obviously, more separation between students who are dif­fer­ent from each other. It seems self-­evident that such stratification could easily aggravate disparities by income, social class, race, or other dimensions of difference. B ­ ecause changes of this sort in enrollment patterns, if they occur at all, prob­ably occur gradually over time and would be expected to affect dif­fer­ent kinds of colleges in dif­fer­ent ways, it is valuable to be able to observe changes over several de­cades, something that can be done using the survey data on first-­year college students employed in this study. ­A fter reviewing some of the prior research on socioeconomic disparities in U.S. higher education, we can return to the Freshman Survey, looking for evidence that socioeconomic gaps have actually

Ec o n o m ic S t r a t ific a t i o n

229

widened and seeking to reconcile the findings regarding socioeconomic status (SES) with ­those related to racial diversity. As throughout this book, we pay special attention to differences within the vast universe of four-­year colleges, contrasting colleges of dif­fer­ent types. ­Here we can look specifically at HBCUs and ask how their existence influences socioeconomic stratification across four-­year colleges. This chapter ends by reconsidering the notion of meritocracy in the context of the baccalaureate market.9 Socioeconomic Gaps in College Enrollment

That the education system in the United States is riven with social and economic in­equality is beyond dispute. Disparities appear at early ages. ­Children growing up in low-­income ­house­holds, on average, hear fewer words and have less intellectual stimulation than ­children in affluent ­house­holds. One study showed that c­ hildren in poor ­house­holds ­were half as likely as ­those in nonpoor ­house­holds to have summer enrichment activities, such as clubs, or­ga­nized sports, or m ­ usic lessons.10 Parents at higher socioeconomic status levels spent more than ­those at low levels on enrichment activities such as summer camp, and the differences have increased over time.11 Achievement gaps appear as soon as ­children start taking tests in school, and t­ hose gaps do not diminish as students advance through the grades.12 A major concern is research suggesting that ­these socioeconomic gaps in achievement have been growing over time, while gaps by race appear to be narrowing.13 By the time students reach high school, significant gaps in socioeconomic status have appeared in the curriculum to which they are exposed.14 Part of the reason for t­ hese inequities lies in the disparities in resources available across school districts, but ­there are inequities within districts as well.15 Looking at the entire education system, t­ hese inequities translate into disparities by socioeconomic status in educational attainment. In the United States such disparities abound, as they do in countries around the world.16 With re­spect to higher education, the dimension of inequity that is easiest to document is simply w ­ hether or not

230 D emand

individuals go to any college. It is similarly easy, given available statistics, to divide that enrollment between four-­year colleges and other post-­secondary institutions. Some scholars have sought to go further, by distinguishing among four-­year colleges. ­Because of the profound differences that exist among colleges in both in institutional resources and mea­sured student ability, some of ­these researchers have paid attention to socioeconomic differences across types of colleges and universities. ­Whether or not ­these refinements are made, all of the research concludes that rates of enrollment definitely differ by socioeconomic status. Considerable research attention also has gone beyond examining patterns of enrollment, finding SES differences in rates of degree completion.17 One reason why students from higher SES classes achieve higher rates of college enrollment might be that they possess superior academic qualifications. The College Board reports that, on average, SAT scores tend to rise with students’ ­family income and the education of their parents. Among students who took the SAT in 2011 / 12, the average SAT score (math plus reading) was 1156 for students whose ­family incomes w ­ ere more than $200,000, but just 894 for t­hose with incomes ­under $20,000. For ­those who had a parent with a gradu­ate degree the average was 1137, while ­those whose parents did not gradu­ate from high school averaged only 870.18 To remove the effect of a correlation such as that, researchers have put together ­tables designed to separate the in­de­pen­dent effect of aptitude or achievement from that of ­family income. Making this correction demonstrates clearly that ­there remains a strong correlation between college enrollment and socioeconomic status even for students with similar mea­sured academic aptitude. One study to show this sturdy SES effect was Proj­ect Talent, a nationally representative study of students in the high school graduating class of 1961. Among men who graduated from high school in 1961 and ­were in the top quartile of achievement, for example, the study showed that 38 ­percent of ­those also in the top socioeconomic quartile enrolled in college within a year while only 10 ­percent of t­ hose in the bottom socioeconomic quartile did. Similar patterns w ­ ere found for w ­ omen.19

Ec o n o m ic S t r a t ific a t i o n

231

­ eedless to say, this finding, that college enrollment is strongly reN lated to economic status, in­de­pen­dent of academic preparation or aptitude, is a potent challenge to the meritocratic ideal that colleges ­ought to enroll on the basis of academic merit, not economic status. This finding is a mainstay in studies of college enrollment.20 Dividing colleges between two-­and four-­year ones reveals a significant socioeconomic status distinction. As between t­hese two sectors, students from more affluent families gravitate ­toward the four-­year colleges to a greater extent than ­those from modest means. One result of this broad difference is that the statistical association between economic status and enrollment works differently between the two sectors. For enrollment in two-­year colleges, SES has its strongest positive correlation for students with the weakest academic qualifications. For enrollment in four-­year colleges, however, the patterns are reversed, with large socioeconomic differences in rates of enrollment observed at the highest level of academic preparation.21 ­Needless to say, t­ hese findings are at odds with the meritocratic ideal. Beyond the choices between enrolling or not and between two-­and four-­year institutions, the dimension perhaps most on the minds of ­those who debate the equity of the current system is the one that is central to this book—­a student’s se­lection from among the vast array of available colleges. Where, along the broad spectrum of four-­year colleges—­from Illinois State to Tennessee Wesleyan to Boston College and Stanford—do students of dif­fer­ent socioeconomic levels enroll? Just as the previous research showed that ­t hose first-­level choices are correlated with SES, it would appear that a socioeconomic correlation also strongly influences the type of college where a student ­w ill enroll. A particularly compelling illustration of this tendency is a study that tracked the college choices of more than seven hundred valedictorians from public high schools in five states. Compared with valedictorians from m ­ iddle and low socioeconomic backgrounds, ­those from high SES families applied to more colleges and w ­ ere more likely to apply to a college in the most-selective category. The share of ­these valedictorians who applied to at least one private college in the most-­selective category

232 D emand

differed markedly by the students’ SES. Whereas half of ­t hose in the lowest SES category and 58 ­percent of t­ hose in the m ­ iddle SES category applied to at least one private college in the most-­selective category, the share of t­ hose in the top SES category who did so was 80 ­percent. Not surprisingly, a higher percentage of ­these valedictorians from the high SES category ended up attending one of the most selective colleges, both public and private, than their less affluent counter­parts.22 An earlier study to examine the connection between SES and college quality, by Jerome Karabel and Alexander Astin in 1975, classified colleges by selectivity and revenues per student. Arrayed in this manner, colleges differed systematically in their tendency to enroll high SES students. Part of this association could be attributed to the positive correlation between SES and mea­sured academic ability, which they calculated to be + 0.29. Even when academic aptitude was controlled for, however, they found that enrollment in colleges at the top of e­ ither mea­sure remained correlated to ­family income. High SES students gravitated ­toward the highest-­ranked colleges. They conclude that “few students of modest social origins attend t­hose elite colleges.” The authors noted, however, that academic qualifications ­were a more potent predictor of enrollment in ­these top colleges.23 Have ­these economic disparities widened? Have the longstanding, documented advantage that high SES students have enjoyed in getting into four-­year colleges, especially the rich and selective ones, actually grown over time? This is a central question in assessing the equity of American higher education t­ oday. To answer this question, several studies have utilized data from a succession of national surveys of college-­age individuals, covering several dif­fer­ent cohorts.24 For the most part ­these studies make no distinctions among two-­ and four-­year colleges. One study that compared rates of enrollment only in four-­year colleges, using two cohorts separated by nearly two de­cades, found that the gap in enrollment rates between the lowest and highest income quartiles remained roughly constant. Over ­those two de­cades, while the percentage of high school gradu­ates from the lowest income quartile who enrolled in college increased from

Ec o n o m ic S t r a t ific a t i o n

233

42 ­percent to 49 ­percent, the corresponding rate for the highest SES group had a parallel increase, from 85 ­percent to 91 ­percent, thus preserving the gap.25 Another, examining two l­ater cohorts, found that the gap in college enrollment rates had increased over the nearly two de­cades, from 39 percentage points for the 1979–1982 cohort to 51 points for the 1997–2000 one.26 Another study performed a statistical test comparing the influence of students’ income on the probability of attending college. It found a “dramatic” increase in the effect of income on college enrollment. For the earlier cohort, ­family income was statistically related to college attendance only for ­those in the top half of test scores. For the ­later cohort, income was significantly related to college attendance for all but the top ability quartile, where the effect of ­family income actually declined over the two de­cades. Another study examining overall college enrollment (where SES was mea­sured by parents’ education) similarly concluded that the gap in enrollment rates by SES increased from the 1970s to 1990s. However, t­ here was no divergence in rates of enrollment in elite colleges.27 Echoing findings for other developed countries, ­these studies show that the general expansion of college-­ going did not have the effect of reducing economic gaps in educational attainment.28 Beyond rates of enrolling in any college, or any four-­year college, the question that has attracted special attention is the trend in access to the most selective colleges—­the “elite” colleges. Astin and Oseguera (2004) used data from the Freshman Survey to see if t­ here had been an increase in the tendency for high SES students to gravitate ­toward ­these colleges. In this study elite colleges ­were defined strictly on the basis of SAT scores (in 1999), with ­those in the top 10 ­percent being so designated. 29 They found that this tendency had in fact become more pronounced, estimating that the share of students in the most affluent income quartile who attended ­these most selective colleges had been rising since the mid-1980s. Notably, the share from the lowest income quartile who attended ­those colleges had remained steady, leaving the share of elite college places for ­those in the ­m iddle 50 ­percent of income to shrink. They con-

234 D emand

cluded, therefore, that American colleges had become more stratified over time.30 Another study came to much the same conclusion using data from four successive nationally representative samples of high school gradu­ates from 1972 to 2004. 31 The findings of that study can be illustrated by tracing the share of students attending a college ranked as highly competitive or most competitive. Over the period studied, this share increased in each of the four SES quartiles, but it increased the most in the two highest quartiles. The percentage point increases by quartile w ­ ere: lowest quartile, 0.4 percentage points; second, 2.3 percentage points; third, 6.2 points; and highest, 5.7 points.32 By 2004, the authors calculated, some 17 ­percent of top SES students attended one of ­these highly competitive colleges, compared to just 5 ­percent for the next-highest SES students and only 2 ­percent for the lowest SES students. Did In­equality Increase?

As discussed in Chapter  8, a fundamental part of the pro­cess by which the market for college became increasingly a national one was a re-­sorting of customers across colleges over time to produce new patterns of attendance. Having lost their hold over top students living nearby, colleges w ­ ere forced to compete for students across wider expanses of territory. The most selective increasingly enrolled good students who lived farther away, leaving lower-­ranked colleges to fight for the remaining students, t­ hose with more modest qualifications. The evidence presented in Chapter 8 is consistent with this academic stratification story. The question at hand is ­whether this pro­cess of scholastic segregation brought with it more segregation along economic lines. To address this question, we turn to three mea­sures of socioeconomic status available in the Freshman Survey: f­ amily income, parents’ education, and private school attendance. Using each of t­ hese mea­sures, we can compare changes over time within each of the seventeen categories of colleges, where calculations are based on surveys at dif­fer­ent points in time of students who attended precisely

Ec o n o m ic S t r a t ific a t i o n

235

the same colleges. First-­year students ­were asked to estimate their parents’ income (for the previous year) by choosing one of several income categories. ­These survey responses can be used to estimate a dollar value of f­ amily income for each student and to examine distributions by income across college categories already described. The resulting dollar values are expressed in constant 2008 dollars. Although such estimates are surely inexact, ­there is no reason to believe that they are systematically biased so as to distort comparisons between large groups of students.33 Perhaps the most obvious question is w ­ hether average incomes have been diverging between students attending the vari­ous groups of colleges. Using the Freshman Survey data, Figure 9.1 answers this question. It shows how average income for students in each group of colleges differed from that year’s mean income for all students in four-­ year colleges.34 Over the entire period from the first to the third survey waves (1972 to 2008 / 09), the inflation-­adjusted average ­family income of college students increased smartly, reflecting the rise in real incomes enjoyed over this period by ­those at middle-­and upper-­ income levels. In 2008 dollars, the enrollment-­weighted average f­ amily income of students starting four-­ year colleges increased from roughly $100,000 in 1972 to $123,200 in 1989 / 90 and to $145,600 in 2008 / 09. Showing deviations in income from the national average for four-­year college students makes it easier to identify trends indicating ­either growing or shrinking stratification by income. 35 Average ­family income across the categories of colleges did diverge. At the most selective, wealthiest private colleges, the average income of students surged ahead of the national mean, while average ­family income among the lowest-­ranked public and private colleges fell ­behind. The students attending private colleges whose average SAT scores ranked at or above the 90th percentile in 1970 had the highest average income in each of the three survey waves, and the increases for ­those colleges ­were also the largest. For the private colleges in the 98–99th percentiles, a group that includes Middlebury, Prince­ton, and Wesleyan, average f­ amily income of first-­year students exceeded the average for all first-­year students by $64,100 in 1972, $82,700

236 D emand

$80,000 $60,000 $40,000 $20,000 $0 −$20,000 −$40,000 −$60,000

Pub

Priv

90+

−97,242 −62,503 −60,401

1,114

−7,241 −11,057

−64,812 −42,838 −33,845

80–90 −1,646

−4,311 −1,473

−360

11,440

HBCU

23,648

−21,926 −20,157 −11,540

56,281

74,514 83,065

64,168

82,792 99,803

42,982

68,717 92,950

59,856

75,290 91,799

32,274

59,202 40,177

32,860

21,320 36,501

26,431

49,715 35,144

6,790

10,477 3,717

−1,431

7,319 10,942

22,032

−5,428 −2,407

2008/09

6,051

1989/90

Public

−18,714

1972

−7,172

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−$100,000

25–50

−$80,000

0–25

Average for college category minus U.S. average (2008 dollars)

$100,000

Figure 9.1. ​­Family Income by College Category and Survey Wave, Relative to National Average Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. Note: To calculate average income by category and year, income for each respondent was estimated to be the midpoint of bounded categories. For the top income category, the estimate was average Adjusted Gross Income (AGI) for taxpayers above that level, calculated by year from the IRS Statistics of Income reports. In current dollars, t­ hese values w ­ ere $88,393 for the $50,000 or more category in 1971; $91,103 for $150,000 or more category in 1988; $303,469 for the $150,000 or more category in 1989; $677,804 for the $250,000 or more category in 2007; and $604,930 for the $250,000 or more category in 2008). The enrollment-­weighted means for all first-­year students in four-­year colleges w ­ ere $99,774 for 1972, $123,225 for 1989 / 90, and $145,570 for 2008 / 09.

Ec o n o m ic S t r a t ific a t i o n

237

in 1989 / 99, and $99,800 in 2008 / 09 (again, in constant 2008 dollars). By contrast, the incomes of students who enrolled in public institutions in the bottom half of the selectivity ranking (for example, Oakland University and the University of South Carolina), private colleges in the bottom quarter (Iowa Wesleyan, Berry College) and the HBCUs all fell further b ­ ehind during the latter half of the four-­decade period. For students attending HBCUs, this divergence by income was stunning. By 2008 / 09 the first-­year students in public HBCUs had ­family incomes averaging nearly $200,000 less than students entering the private colleges in the 98th to 99th percentile. Over the course of nearly four de­cades, the gap in students’ average ­family income between ­those two groups of colleges grew from about $125,000 in 1972 to $197,000 in 2008 / 09. Between students attending public institutions in the bottom half of the SAT distribution (0 to 50th percentiles) and ­those attending the top-­ranked private colleges (99+ ­percent) the gap in average ­family income increased from about $67,000 in 1972 to $104,000 in 2008 / 09. The gaps in average ­family income across t­hese broad categories of colleges, in short, had ballooned. ­Going beyond averages, Figure 9.2 shows estimates of the distribution of f­ amily incomes by college group. Using, as before, entering students’ self-­reported f­ amily incomes, the percentage of students whose f­ amily income fell into each of four income categories, all expressed in constant 2008 dollars, can be estimated at less than $50,000, $50,000 to $100,000, $100,000 to $250,000, and $250,000 and over.36 Like Figure 9.1, Figure 9.2 reveals yawning gaps between students in the vari­ous college groups. Whereas a quarter of the students attending one of the 98th to 99th percentile private colleges in 2008 / 09 had ­family incomes of more than $250,000 (shown by the striped bar ­running from 0.75 to 1.00), virtually no student in a public HBCU did (see the top of the last bar at the far right). Among the students attending one of the vast number of 0 to 50th percentile public universities, only 7  ­percent came from families in this rarified income range. If we make comparisons based on the highest two income groups, denoting incomes of $100,000 or more, the

238 D emand

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2

99+

0–50

90+

100–250

1972

2008/09

1972

Private Public

Public

50–100

2008/09

1972

2008/09

1972

2008/09

1972

50–80 80–90

Private

LT50

2008/09

1972

2008/09

1972

0–25 25–50 50–60 60–70 70–80 80–85 85–90 90–95 95–98 98–99

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

0

2008/09

0.1

HBCU

GE250

Figure 9.2. ​Percentage Distribution of F ­ amily Income by College Category and Survey Wave Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show the proportion of first-­year students by college category whose estimated ­family incomes fell into each of four categories, defined in terms of 2008 dollars: less than $50,000; $50,000 less than $100,000; $100,000 less than $250,000; and $250,000 or more. Responses concerning ­family income, by category, in surveys for 1972, 1989, 1990, 2008, and 2009 refer to income in 1971, 1988, 1989, 2007, and 2008, respectively. Using numbers by income category in each survey, the numbers by categories corresponding to the four income groups above expressed in 2008 dollars w ­ ere estimated by interpolation.

gaps seem even bigger. Whereas 60 ­percent of the students who attended a college in the 98th to 99th percentile private group had f­ amily incomes of $100,000 or more in 2008 / 09, less than 10  ­percent of ­those attending public HBCUs did. It is helpful, if not strictly accurate, to illustrate differences such as ­these across the college categories by using as examples colleges whose students ­were part of our sample. For example, in the group of colleges containing the University of South Carolina (public colleges

Ec o n o m ic S t r a t ific a t i o n

239

in the 0 to 50th percentile of SATs in 1970), the percentage of students with ­family incomes of $100,000 or more in 2008 / 09 was 39 ­percent. The corresponding share differed by a wide margin across college groups. Among the public HBCUs in the sample, such as Winston-­Salem State, only 9 ­percent of students came from such families. Among private HBCUs, such as Howard, the share was 21 p ­ ercent; in private colleges in the bottom quartile of SAT scores, such as Iowa Wesleyan, it was 36 ­percent; in private colleges between the 60th and 70th percentiles, such as Michigan’s Calvin College, it was 44 ­percent; in public institutions between the 80th and 90th percentile, such as the University of Mas­sa­chu­setts at Amherst, it was 46  ­percent; in private institutions in the 95th to 98th percentile range, such as Northwestern, it was 62 ­percent. That range, from 9 ­percent at colleges like Winston-­Salem State to 62 ­percent at colleges like Northwestern, speaks as loudly as any statistic of the socioeconomic stratification of American four-­year colleges. Turning from ­these marked differences across types of colleges in a single year, the figure allows us to address as well the question of what happened over time. What do t­ hese distributions tell us about changes in the incomes of students by college group? One change was the across-­the-­board increase in the share of students with ­family incomes of $250,000 or more. ­Here is a vivid reflection of the country’s changing income distribution, with t­ hose at the top seeing disproportionately large increases in their incomes, pushing more families over the constant-­ dollar $250,000 income threshold. Shifting our gaze to include the top two income groups yields a consistent pattern in all four groups of public institutions (except the HBCUs). In t­ hese public institutions the share of students from families making $100,000 or more increased steadily, from 1972 to 1989 / 90 and then again to 2008 / 09. By 2008 / 09 public colleges in the top half of the SAT distribution had a bigger share of affluent students by this mea­sure than any category of private colleges in the bottom half of SATs. What socioeconomic segregation that occurred was not so much between public and private as between high and low selectivity.

240 D emand

Despite ­these across-­the-­board increases in the shares of high income students, the shares of students from the lowest income category, $50,000 or less, mostly held their own, and actually increased in private colleges above the 60th percentile of SAT scores. Evidently low-­income students ­were not being pushed out of four-­year colleges. In fact, they ­were gaining as a share in selective private colleges, a tribute to Pell Grants and colleges’ own need-­based aid policies. But with growth at the top and constant or increasing shares at the bottom, students from the $50,000 to $100,000 category w ­ ere squeezed out, leaving what one inside observer called a “bifurcated student body,” with low-­income students dependent on generous financial aid on the bottom and wealthy students paying full freight at the top. 37 This squeezing was especially evident at the top-­ranked public institutions, where the share of students from the $50,000–­$100,000 income category declined from 46 ­percent in 1972 to 30 ­percent in 2008 / 09. In short, this summary of income distributions, like Figure 9.1 showing average incomes, provides evidence of an increasing concentration of affluent students in elite colleges, where “affluent” is defined in terms of a constant dollar threshold. Keep in mind that ­these results cannot be the result of a change in the composition of colleges being compared over time, since exactly the same colleges are represented in each of the bars. The trends shown by both of ­these figures could have been the result, however, simply of rising incomes among the affluent, not any increase in their tendency to cluster at elite colleges. T ­ here can be ­little doubt that the high-­SAT colleges, both private and public, enrolled a disproportionately high-­ income set of students, but w ­ hether this disproportionality has grown more severe over time remains unanswered. To get at that question—of ­whether ­there has been a rise in the concentration of affluent students in elite colleges—we can trace the destinations of a fixed percentage of all surveyed students, that group representing the most affluent ones. The Freshman Survey responses for students in each of the three survey waves enrolling in the sample of 188 colleges identifies the 10  ­percent of ­those roughly 100,000

Ec o n o m ic S t r a t ific a t i o n

241

students with the highest self-­reported f­ amily incomes.38 To review, the full sample included all first-­year students at the same 188 colleges in each of three survey waves. Did the already disproportional presence of affluent students at selective colleges such as Amherst and Prince­ton increase further over ­these four de­cades? To determine this, we can examine enrollments at two sets of selective colleges. The first set consists of eigh­teen of the most-­selective private colleges in the country. T ­ hese ­were the private colleges in the 98th to 99th and the 99th and above percentile categories in the sample of 188 colleges with data from the Freshman Survey. 39 The second, more expansive set of selective colleges contains all the colleges among the 188 whose students scored in the top 10 ­percent of four-­year colleges in 1970. ­These fifty-­six colleges ­were mostly private, but among their number ­were also some prominent public universities, including Georgia Tech and Stony Brook.40 Although ­these selective colleges ­were certainly enrolling more than their share of rich students throughout of the period of study, that share increased only a ­little over the course of ­these four de­ cades. As shown in ­Table 9.1, the high-­income students ­were roughly twice as likely as the average college student in the sample to end up in one of the eigh­teen elite private colleges, with their share being slightly less than 20  ­percent in 1972 and slightly more two and then four de­cades ­later. The percentage of places at ­these top colleges occupied by top-­income students did increase over this long period, but the increase was modest, almost imperceptible. ­There is little evidence h ­ ere—­based on the enrollment of top-­income students in ­these eigh­teen elite colleges—to suggest a growing concentration of ­children of the rich. The story is much the same if one looks instead at the more expansive list of selective colleges. Among the fifty­six highly selective (90th+ percentile) colleges, the share of freshmen class spots claimed by ­these richest-10 ­percent students did increase, but only by a l­ ittle.41 Taken together, ­these calculations suggest that ­there was virtually no increase in the concentration of the highest-­ income students in selective or super-­selective colleges over the nearly four-­decade period. At the same time, as we have seen, ­there

242 D emand

was an increase in very high income students in t­ hose colleges, owing to the real increases in the incomes of ­those at the top of the nation’s income distribution. In 2017 a study based on an extraordinary data set offered, for individual colleges, estimates of students’ ­family incomes. Using millions of tax returns, Stanford economist Raj Chetty and a team of coauthors calculated and made public statistics describing parents’ incomes by college. The most stunning, and newsworthy, findings to emerge from their work ­were the extraordinarily high f­ amily incomes for the private elite colleges. For example, the f­amily income of the average student at Prince­ton was $519,400. It was $472,200 at Stanford, and $505,000 at Harvard. Equally impressive ­were the large shares of students at such colleges whose parents ­were in the very highest echelons of the income distribution. At Prince­ton, for example, 65  ­percent of the students w ­ ere from families in the top 10 ­percent. The corresponding shares for Stanford and Harvard w ­ ere lower, but high nonetheless: 56 ­percent and 57 ­percent, respectively.42 To see how students’ economic backgrounds differed across types of colleges, Figure  9.3 shows the percentage of students by college

­Table 9.1  Share of Students from Richest 10 ­Percent of Families Who Enrolled in Selective Colleges Survey wave 1972

1989/90

2008/09

18 highly selective private colleges

19.8

20.4

20.6

56 selective private and public colleges

15.2

15.3

15.6

Of t­ hose enrolled, percentage from richest 10% of families

Note: Figures in the first row of the t­ able show the percentage of students in one of eigh­teen private highly selective colleges who ­were among the 10 ­percent of students with the highest ­family incomes in the specified survey wave. Figures in the second row of the ­table show the percentage of students in one of fifty-­six selective public and private colleges who ­were among the 10 ­percent of students with the highest f­ amily incomes in the specified survey wave. For details on the list of colleges and the methodology, see online appendix ­Table 9.1. Source: Freshman Survey results for 188 colleges; author’s calculations.

Ec o n o m ic S t r a t ific a t i o n

243

0.50 0.40 0.30 0.20 0.10

Pu b

Priv

90+

0.034

0.340

HBCU 0.095

80–90 0.277

0.227

0.161

0.537

0.508

0.508

0.427

Public 0.374

0.329

0.323

0.216

0.288

0.197

From top 10%

0.151

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

25–50

0.00

0–25

Fraction of students from the most affluent 10 percent of families

0.60

Figure 9.3. ​Share of Students from Most Affluent 10 ­Percent of Families Source: Chetty et al. (2017) and Equality of Opportunity Proj­ect, accessed January 22, 2017, http://­w ww​.­equality​-­of​-­opportunity​.­org​/­college​/­. Based on tax return data covering parents of college students born in the period from 1980 through 1982. Parents’ income was averaged for the five years child turned fifteen to nineteen, expressed in 2015 dollars. Note: Bars show author’s calculation of weighted averages for 979 colleges, classified by 17 categories.

category whose parents ­were among the top 10 ­percent in income. Colleges with the highest shares of affluent students w ­ ere the private ones in the top echelons of selectivity, ­those in the highest 5 ­percent of 1970-­era SAT scores, with more than half of all students having parents in the top tenth. As was evident in the comparisons of income based on student reports in Figures  9.1 and 9.2, f­amily incomes tended to rise with selectivity in both the private and public sectors. At any level of selectivity, income levels at private colleges ­were higher than t­ hose in public ones. At the bottom of this comparison ­were the HBCUs, with only 3  ­percent of students at the public HBCUs coming from top-10 ­percent income families. ­These data confirm, with greater precision, the income in­equality that was evident from the Freshman Survey, although it tells us no more about w ­ hether this in­equality has been growing over time.

244 D emand

Besides ­family income, a second mea­sure that is available for documenting socioeconomic disparities across colleges is parental education, a tried and true mea­sure of socioeconomic status. Over the nearly four de­c ades covered by this study, the average educational attainment of students’ parents increased substantially, as it did in the general population, making the college degree an increasingly common credential. To illustrate, the share of first-­year students whose ­fathers w ­ ere college gradu­ates increased from 37 ­percent in 1972 to 58 ­percent in 2008 / 09. A more demanding marker of educational attainment looks to advanced degrees. For example, Figure  9.4 identifies students who had at

0.30 0.20 0.10 0.00

Pub

Priv

90+ 0.06

0.01

0.06 0.06

−0.06 −0.02

−0.20 −0.11 −0.11

80–90 0.02 0.03 0.02

0.00 0.03

HBCU

0.05

−0.07 −0.08 −0.06

0.25 0.34 0.32

0.28 0.36 0.31

0.16 0.25 0.28

0.22 0.26 0.24

0.07 0.13 0.12

0.09 0.07 0.09

0.08 0.13 0.08

0.01 0.02 0.00

0.03

0.00 0.01

2008/09

Public

0.00

1989/90

−0.04 −0.03

1972

−0.05 −0.03 −0.03

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−0.20

25–50

−0.10

0–25

Average for college category minus U.S. average

0.40

Figure 9.4. ​Parent Had a Gradu­ate Degree Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The enrollment-­weighted means for all first-­year students in four-­year colleges ­were 0.160 in 1972, 0.274 in 1989 / 90, and 0.334 in 2008 / 09.

Ec o n o m ic S t r a t ific a t i o n

245

least one parent with a gradu­ate degree. In 1972 only 16 ­percent of college entrants had a parent with such a degree. By 1989 / 90 this share had increased to 27  ­percent and by 2008 / 09 it was 33 ­percent. To identify more easily trends beyond this general advance in educational attainment, the figure shows deviations from the national mean. What­ever e­ lse can be concluded from this figure, it tells an unambiguous story about socioeconomic differences between college groups at each of the three points in time. One generalization it suggests is that the socioeconomic status of private college students tends to be higher than that of students at public colleges. A second is that, within the private or public sectors, socioeconomic status tends to be highest in colleges with the highest average SATs. Third, public HBCUs are in a world apart, serving students with markedly lower socioeconomic status than any other college group. All three of ­these patterns have clear parallels in the differences of average ­family income shown in Figure 9.1. ­Whether ­these patterns have changed over time is much less clear. Viewing t­ hese deviations from the mean, we can see that some gaps among the college categories did grow larger over the period, in two re­spects. First, this mea­sure of socioeconomic status surged above the national average between 1972 and 1989 / 90 in private colleges with SATs above the 85th percentile. Second, public HBCUs fell ­behind the national average over the next two de­cades. By 2008 / 09 some 13 ­percent of first-­year students in the public HBCUs had a parent with an advanced degree, a percentage far below the national average (33 ­percent) and even farther below that for the top-­ranked private colleges (65 ­percent). ­There is one more mea­sure of socioeconomic status in addition to ­family income and parents’ education that can be gleaned from the Freshman Survey: attendance at a private high school. As an indicator of high socioeconomic status, this is a rougher, less reliable mea­sure than the first two, for a c­ ouple of reasons. First, many Catholic parochial schools have traditionally served families across the income spectrum, and ­those schools accounted for some three-­quarters

246 D emand

of private high school enrollment in the early 1970s.43 (By 2009 they accounted for a l­ ittle less than a half.)44 Second, many affluent students did not attend private schools, but instead went to public schools in wealthy school districts. Notwithstanding t­ hese considerations, it remains a fact that students attending private high schools tend on average to be more affluent than ­those who attend public high schools.45 The Freshman Survey contained questions in two of the survey waves—­the first and the third—­about what kind of high school they attended. Figure  9.5 shows the averages for the college groups for ­these two waves. The figures exclude foreign students ­because the significance of private school attendance for them is apt to be dif­fer­ent in unknown ways from what it is for American

0.15 0.10 0.05 0.00 −0.05

0.17

0.08

0.15

0.09

0.09

0.17

0.14

0.20

0.15

Priv

90+

80–90

Pub −0.14 −0.12

0.11

0.19

0.01

0.19

0.09

−0.02

0.06

0.10

HBCU −0.02 −0.03

0.20

0.03

−0.07 −0.07

0.03

0.06

0.04 −0.01

0.08

2008/09

−0.06 −0.05

1972

0.05

Public

0.02

Private

50–80

0–50

99+

98–99

90–95

95–98

85–90

80–85

70–80

60–70

50–60

−0.15

25–50

−0.10

0–25

Share who attended private HS minus U.S. average

0.20

Figure 9.5. ​Attended Private High School Source: Freshman Survey, students from 188 colleges, two waves (1972 and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The enrollment-­weighted means for all American first-­year students in four-­year colleges born in the United States w ­ ere 0.168 in 1972 and 0.182 in 2008 / 09.

Ec o n o m ic S t r a t ific a t i o n

247

students and therefore an even less dependable marker of high socioeconomic status. The comparisons reveal sharp increases in the three highest-­ ranked groups of private colleges, ­those above the 95th percentile of 1970 scores. For ­these three groups of private colleges, the share of entering students coming from private high schools increased by 5  percentage points or more. In contrast to t­hese increases, ­there ­were declines in the share of students from private high schools in the private colleges below the 70th percentile of 1970-­era SATs and in the g­ iant category of public colleges below the median. The decline in the share of students who had attended private schools was especially steep for private colleges in the 60th to 70th percentile of SATs. Of the twelve colleges in this group, four are Roman Catholic, suggesting that a good portion of the 1972 private school gradu­ ates could have attended Catholic schools, t­ hose whose enrollments fell the most. Taken together, the evidence collected from the successive waves of data from the Freshman Survey and presented h ­ ere provides very strong evidence of SES disparities across the spectrum of colleges but weak evidence, at best, that t­ hose disparities have grown appreciably over time. ­There is absolutely no doubt that socioeconomic gaps between types of colleges have been a feature of the college landscape for a long time, and that some of ­these gaps are very substantial. Students at private colleges tend to have higher f­ amily incomes, have better-educated parents, and are more likely to have attended a private high school. Similarly, the socioeconomic status of students attending more selective institutions tends to be higher than that for lower-­ranked institutions. Over and above ­those three generalizations, the students at HBCUs, both private and public, ­were significantly ­behind other college students on t­hese scales of socioeconomic status. In short, the evidence of systematic socioeconomic disparities is indisputable. However, evidence that ­these disparities have been increasing over ­these de­cades is much less clear. It is true that gaps in students’ average ­family incomes have widened, but that is ­because incomes have diverged, not ­because a

248 D emand

more concentrated group of rich families is sending their c­ hildren to elite colleges. The Diversity Effect

Our empirical review has revealed two seemingly contradictory trends. On the one hand, the racial and ethnic diversity of American four-­year colleges advanced dramatically over the four de­cades covered by this study. Based on our Freshman Survey sample, the share of first-­year students at predominantly white colleges who ­were nonwhite or foreign increased from 8.1 ­percent to 21.3 ­percent between 1972 and 2008 / 09.46 One might have thought that this opening up of previously white colleges and universities would have boosted socioeconomic equality. Yet, as shown above, ­there is no evidence at all of movement in an egalitarian direction. The deep socioeconomic in­equality that characterized the vast array of four-­year colleges in 1972 did not diminish. If anything, it inched up a notch or two. How can the putative egalitarian effects of rising diversity be reconciled with the stubborn inequities by socioeconomic status? To explain this apparent paradox, we must look at the composition of the nonwhite students attending previously white colleges. In ­doing so we ­will see that HBCUs had a role to play. The key to this apparent paradox, perhaps obvious in hindsight, is that, in their efforts to diversify their student bodies, the elite colleges did not have to rely on low-­income students. T ­ here w ­ ere nonwhite students in middle-­and upper-­income brackets, and their numbers ­were growing. Indeed, the average incomes of students who ­were not white Americans increased at t­ hese most selective colleges. Between 1972 and 2008 / 09, for example, the average ­family income of nonwhite and foreign students at the most selective private colleges increased from $26,000 above the national average to $56,000 above the national average (in constant 2008 dollars). The other categories of highly selective private colleges witnessed similar increases for their nonwhite and foreign students, as shown in Figure  9.6. The largest increases in ­family income for ­these students occurred in the

Ec o n o m ic S t r a t ific a t i o n

249

$60,000 $40,000 $20,000 $0 −$20,000

Pub

Priv

90+

−8,466

−2,552

14,901

13,461

4,780

14,468

15,697

29,538

26,044

−6,193

763

11,773

8,808

−2,048

−7,037

5,680

36,035

16,238

24,104

35,480

56,498

67,336

48,923

3,322

−5,769

−8,930

5,283

26,951

31,106

32,766

47,905

76,557

52,397

56,795

−13,921 −22,843

21,443

10,127

705

8,147

2,189

−21,694

−54,192 −28,577 −29,148

80–90

11,132

HBCU

−6,251

2008/09

Public

20,543

1989/90

−12,659

1,972

13,694 −10,543

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−$60,000

25–50

−$40,000

0–25

Average for college category minus U.S. average (2008 dollars)

$80,000

Figure 9.6. ​­Family Income, Nonwhite and Foreign Students Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The enrollment-­weighted means for all first-­year students in four-­year colleges ­were $67,830 in 1972, $89,123 in 1989 / 90, and $102,524 in 2008 / 09, in constant 2008 dollars.

95th to 98th percentile private colleges (which includes, for example, Bowdoin, Northwestern, and Rochester). At ­these colleges the average income of nonwhite and foreign students, which had exceeded the national average by $15,600 in 1972, increased by 2008 / 09 to a level that was an astonishing $76,500 above the national average. Though this gap was still less than the $92,000 above-­average advantage for all students at ­these colleges, it is clear that increasing diversity did not have to come at the cost of admitting large numbers of needy students.47

250 D emand

Other mea­sures of socioeconomic status reveal the extent to which students at public HBCUs ­were being left ­behind the vast population of students at other four-­year colleges. The share with a parent who had a gradu­ate degree, 11  percentage points b ­ ehind the national average in 1972 and 1989 / 90, fell nearly twice as far b ­ ehind by 2008 / 09. A negligible share of t­hese students attended private school e­ ither in 1972 or 2008 / 09. Like their private counter­parts, the public HBCUs remained stubbornly segregated. In North Carolina, black public school students who attended one of the state’s public HBCUs w ­ ere on average far more racially isolated in college than they had been in eighth grade. Among all the African American eighth graders in public school in 2004 who would eventually enrolled in one of the fifteen campuses of the University of North Carolina system, they w ­ ere in an eighth grade that was on average 34.4  ­percent black, but they ­were in a college that was on average 65.0 ­percent black.48 Another clue in the diversity-­SES interaction can be seen in the proportion of black first-­year students who had attended private high schools. Figure 9.7 shows this proportion (not as deviations from the average) for the vari­ous college groups. It is evident that the percentage jumped markedly in several of the categories over the four-­decade period. In fact the percentage of black students entering college from private high schools increased in five of the seven categories of private colleges above the 70th percentile in SAT scores. The largest increases occurred in the 85th to 90th and 95th to 98th percentile categories. This influx of affluent black students into elite private colleges has a mirror image—­a corresponding exodus from historically black institutions. T ­ here is no more dramatic change in the fortunes of any subsector of American four-­year colleges than the steep decline in the economic status of students attending HBCUs. Figure  9.1 pulls no punches. It shows that, in terms of average ­family income, students at HBCUs fell far b ­ ehind the national average for college students. In 1972 the average student’s ­family income at private HBCUs was $33,000 below the national average, compared to a gap of $60,000

Ec o n o m ic S t r a t ific a t i o n

251

0.40

Share attended private high school

0.35 0.30 0.25 0.20 0.15 0.10 0.05

Private 1972

Public

Pub

Priv

90+

80-90

50-80

0-50

99+

98-99

95-98

90-95

85-90

80-85

70-80

60-70

50-60

25-50

0-25

0.00

HBCU

0.06 0.23 0.22 0.24 0.15 0.19 0.15 0.24 0.24 0.38 0.30 0.08 0.09 0.11 0.10 0.18 0.04

2008/09 0.13 0.23 0.16 0.21 0.23 0.28 0.29 0.23 0.38 0.33 0.36 0.05 0.10 0.10 0.17 0.16 0.04

Figure 9.7. ​Attended Private High School, Black Students Source: Freshman Survey, students from 188 colleges, two waves (1972 and 2008 or 2009). Note: The enrollment-­weighted means for all first-­year students in four-­year colleges ­were 0.151 in 1972 and 0.150 in 2008 / 09.

in public HBCUs (in 2008 dollars). ­These gaps increased a small degree in the next twenty years. A development of major importance was the growing tendency for the academically strongest and most affluent black college-­bound students to choose predominantly white colleges over HBCUs. A study using Freshman Survey data traced the share of blacks in several categories who chose black institutions versus predominantly white ones. The study compared cohorts entering college in 1970 and 1978. Among ­those blacks who ranked in the top quarter of their high school class, the percentage who chose to attend HBCUs declined from 42  ­percent in 1970 to 31  ­percent in 1978. Among ­those whose ­fathers ­were ­lawyers, the

252 D emand

share choosing HBCUs declined from 48 ­percent to 36 ­percent. For ­children of physicians, the decline was an astounding 22 percentage points, declining from 59  ­percent to 37  ­percent, all in less than a de­cade.49 The exodus of bright and affluent black students from the colleges that many of their parents had attended continued apace, judging from the figures in this chapter. In the two de­cades ­after 1990, the already large gaps in average ­family income between the HBCUs and the national average ballooned, increasing to $64,000 for private HBCUs and a nearly unbelievable $97,000 for the public ones. Students in HBCUs fell b ­ ehind as well in the percentage who had a parent with a gradu­ate degree. In the two de­cades between our second and third wave of survey data, ­those in private HBCUs fell below the national average by 6  percentage points. The biggest change occurred in the students attending the public HBCUs. Not only did they fall further ­behind the advancing national average, their percentage was the only one of all seventeen college groups to decline in absolute terms. While this indicator increased nationwide from 27  ­percent to 33  ­percent, the percentage for students enrolling in public HBCUs declined, from 16 ­percent to 14 ­percent, ending the period nearly 20  percentage points below the national average for entering college students. Among the black students who ­were choosing predominantly white colleges over HBCUs w ­ ere many ­children of affluent and highly educated parents. Quite a few had attended private high schools, and a good share of t­ hose students headed for the prestigious private selective colleges. By 2008 / 09, of the black first-­year students entering private colleges in the three most selective categories, more than 30 ­percent had attended a private high school. As ­these affluent students of color demonstrate, the pro­cess of racial integration did not have to bring with it socioeconomic integration, a truth vividly documented in Shamus Khan’s account of minority students at the prestigious St. Paul’s school.50

Ec o n o m ic S t r a t ific a t i o n

253

Visions of Meritocracy

The road to meritocracy has been a bumpy one for Amer­i­ca’s colleges. ­After World War II the market for college gradually became less local and regional in scope, and more national. This transformation was pushed along by advances in transportation and communications, by the spread of standardized tests, and by an avalanche of comparative information about colleges. As this transition was occurring, the racial discrimination of the past, once enforced by state governments in the South and border states and, to a lesser extent by tradition elsewhere, that kept many colleges predominantly or entirely white, crumbled in the 1960s. With the widening geo­g raph­ i­cal expanse of the market came increased stratification by academic aptitude. Highly selective colleges, besieged by ­eager applicants from across the country, became ever more selective. ­These developments would appear to be strong evidence that the market for college has become more meritocratic, with enrollment patterns reflecting objectively mea­sured aptitude or other yardsticks of deservedness rather than characteristics unrelated to intellectual or personal merit. What made the road bumpy, and this story of growing meritocracy more complicated, is that the rise in scholastic segregation did not bring with it a decline in the economic stratification that had been a hallmark of the college scene. If anything, as we have seen, patterns of privilege became more firmly established as students sorted themselves into dif­fer­ent kinds of colleges. Based on several dif­fer­ent mea­sures of socioeconomic status, data on successive waves of first-­year students suggest that economic distinctions between colleges held steady or grew slightly over this period. The colleges at the very top of the SAT / selectivity pecking order at the beginning of the period—­particularly the private institutions—­continued to attract students whose parents and schooling ­were, in comparison to other students in their age cohorts, increasingly affluent. It is worth stressing that the economic gaps identified ­here are wholly within the four-­year college market. That is, t­ hese patterns occurred within a larger educational landscape featuring even bigger economic

254 D emand

gaps between ­those who enrolled in any postsecondary training and ­those who did not, and between t­hose who chose two-­year versus four-­year institutions. As enrollment rates of high school gradu­ates from the lowest income groups advanced, so too did rates for the most affluent, leaving the gap between the two virtually unchanged.51 To answer why an increasing emphasis on objective criteria might result in a rise in economic stratification, one is naturally drawn to certain advantages possessed by affluent families. As hinted at above, a host of advantages ­were available to ­those with resources. ­These included SAT-­prep courses (and tutoring), opportunities to travel abroad, volunteer, or take unpaid internships, and the resources to learn and play a sport, like soccer or lacrosse, that is valued at elite private colleges. Affluence could also bring with it the chance to attend a private school; it almost certainly afforded the chance to attend a well-­financed public school full of students whose parents ­were professionals. 52 And, as Chapter 10 shows, affluence lessens the need for high school students to work for pay, thus freeing time for other activities, including school work. Jerome Karbel and Alexander Astin note, “Strict adherence to meritocratic criteria tends to ­favor the affluent.”53 Still one more advantage enjoyed by white and affluent applicants was the widespread admissions preference accorded to ­children of alumni, a widely-­accepted practice that undoubtedly served, in the words of Christopher Jencks and David Riesman, to “provide upper-­and upper-­m iddle class c­ hildren with enormous advantages over the other p ­ eople’s ­children.”54 Armed with a potent mix of resourcefulness, financial wherewithal, and time-­honored institutional advantages, therefore, affluent families ­were able to hold their own in an increasingly meritocratic admissions environment.55

Ec o n o m ic S t r a t ific a t i o n

255

10 Sorting by Seriousness

“The American undergraduate does not revel in the discovery of truth.” So stated Michigan sociologist Robert Cooley Angell in his 1928 study of undergraduate life.1 Of course this was a sweeping generalization, and of course it was enunciated many years ago. But, to the dismay of many a professor in con­temporary American colleges and universities, ­there remains a discernible portion of the undergraduate population for whom Angell’s pronouncement seems all too accurate. Any sober assessment of baccalaureate education in this country must acknowledge this fact. So, too, the likelihood that colleges differ from each other in the portion of their undergraduates who “revel in the discovery of truth.” This chapter explores fault lines across the vast array of four-­year colleges that are, in turn, expected and surprising. It begins with the frank admission that freshmen surely enter college with more than ­human capital on their minds. And it notes recent, unsettling re-

search on academic effort in intellectual engagement. To see how colleges might differ in the academic orientation of their students, and how, if at all, t­ hese patterns might have changed over time, we return to the Freshman Survey for information relevant to entering students’ inclinations to engage intellectually. We find that the diversity we have observed with re­spect to resources, test scores, and socioeconomic status is matched by equally big differences in the bookishness and aspirations of entering students. What Do Freshmen Want?

As recounted in Chapter 8, most economists who study higher education stick closely to the script of college as a builder of ­human capital. College can also provide students with information about their own abilities, and it can serve as a signal to employers of a person’s qualifications. When l­abor markets work well, the additional ­human capital a student gains from college enable him or her to earn a higher salary.2 But this economic view is surely too narrow, as observation suggests, popu­lar culture illustrates, and ethnographic studies make plain. Students and their families are surely interested in personal and social aspects as well. College is seen as a time for maturing, and for making new friends and connections. Parents send their ­children off to a safe place to try ­things out, to grow up. Freed from parental oversight, many college students experiment with alcohol, ­later bedtimes, and new sets of acquaintances, in addition to an academic program that is radically less structured than what they experienced in high school. Real­ity dictates a more holistic view of the college experience, acknowledging that students have a variety of aims. Popu­lar movies have presented an exaggerated picture of college life. In the 1925 film, The Freshman, comic star Harold Lloyd poked fun at the subservience of studiousness to the frenzy over football. John Belushi provided an updated portrait of collegiate hedonism in his 1978 film Animal House. Such parodies exaggerate the recreational aspects of college, no doubt, although the 2014 documentary The Ivory Tower

S o r t i n g b y S e r i o u s n e s s

257

includes video footage from a college bacchanal at Arizona State that is decidedly realistic. Although such portraits are surely not representative of all colleges, virtually all students have a variety of interests other than academic ones, students differ in the weight they place on ­these interests, and colleges differ in the types of students they enroll. Acknowledging such differences among college students, some scholars have devised schemes to classify college students according to their academic orientation. One widely cited typology, published in 1966, differentiated students along two dimensions: how much they identified with their college and how involved they ­were with the world of ideas. This two-­by-­t wo breakdown produced four “types.” One of t­ hese, no doubt many professors’ ideal student, is the academic type, the student who combines an attachment to college with an interest in ideas. The academic student’s nonintellectual counterpart—­ attached to college but not ideas—is the collegiate type. H ­ ere is the school-­color-­wearing fan or the fraternity or sorority member, actively engaged in the ready-­made social scene so prominent at many colleges and studying only enough to get by. The third type is the vocational student, unengaged in college or ideas, whose aim is just to attend classes and get the degree demanded by the ­labor market. The fourth is the nonconformist, a favorite of some professors, who is interested in ideas but not conventional college activities.3 In their ethnographic study Educated for Romance: W ­ omen, Achievement, and College Culture, Dorothy Holland and Margaret Eisenhart studied w ­ omen attending two dif­fer­ent colleges in the 1980s. As their title suggests, they discovered that friends, parties, and romantic relationships occupied a large share of students’ attention. For ­these students, “schoolwork was relatively unimportant.”4 Some of the students they observed did devote serious effort to coursework, but the decision to work hard was viewed merely as an individual prerogative, not a requirement. Among ­these students, academic success did not confer status. Rather, studiousness was more likely to result in being singled out as a grind, a “brain.” In general, students showed more interest in the details of peer interactions and dating than in

258 D emand

the content of their courses. And few of them had serious ­career plans. In short, schoolwork was significant for most of ­these students ­because it competed for time with more highly valued peer activities.5 The authors observe, “Schoolwork seemed to interfere with peer activities by demanding a lot of time. Peer-­related activities, on the other hand, ­were viewed as more fun and potentially more rewarding. Thus, they threatened to overwhelm schoolwork.”6 Another study of ­women in college, undertaken a generation ­later, is Elizabeth Armstrong and Laura Hamilton’s Paying for the Party: How College Maintains In­equality. As noted in Chapter  7, their study draws on close observation of a handful of students assigned to the same dormitory at one large state university in the Midwest, one that featured a social scene with tailgating on football weekends, elaborate themed parties, a panoply of Greek letter fraternities and sororities, and copious amounts of alcohol.7 Among the ­women they followed closely, the authors identified big differences in socioeconomic backgrounds and priorities as college students. Some, who grew up in affluent families, arrived at college with advantages that would serve them well in the university’s social scene—­from tinted hair to tanned skin. (Sororities sought out ­women who ­were “cute” and “bubbly,” and blondes had an acknowledged edge.)8 For ­these affluent students, college was expected to be a time for making friends and developing a rich social life, and the financial resources at their disposal made this pos­si­ble, allowing them, for example, to dress well and pay sorority dues. In contrast, ­women from less privileged backgrounds, many of whom worked for pay at college, could aspire, often unsuccessfully, to achieve similar social status, or instead focus on academic success. Evidence of students’ motivations for attending college can also be derived from statistical studies of enrollment choices. Using large samples of data on where students applied or matriculated, some economists have used information on college choices to infer what college characteristics students appear to desire, as part of larger studies of what f­actors influence demand. Some of this research is entirely consistent with economists’ instinct to focus on academic quality

S o r t i n g b y S e r i o u s n e s s

259

and the potential for skill acquisition, as we saw in Chapter 8. But ­there is also statistical evidence suggesting that at least some applicants are seeking a place where they can have fun. One study revealed that colleges whose football or basketball teams had won championships or finished with high rankings attracted more applications, including more applications from students with high SAT scores. ­These effects ­were strongest for males, African-­A mericans, out-­of-­state students, and ­those who had played sports in high school.9 Another study showed that colleges whose ranking in the Prince­ton Review improved regarding quality of life likewise attracted more applicants.10 Still another statistical study, provocatively entitled “College as Country Club,” provided evidence consistent with the idea that students are swayed in their choices by consumption amenities. Using data for high school gradu­ates in 1990 and 2004, the authors found that students’ choices of a university ­were positively associated with its expenditures on student ser­vices and auxiliaries, ­whether it had a Division I football or basketball team, and the share of its students who w ­ ere members of fraternities or sororities.11 So much for establishing empirically what we have known anecdotally—­that students have nonacademic reasons for attending college. The pressing questions from a national perspective are ­whether this nonacademic aspect differs systematically across colleges, ­whether it has been growing over time, and what it all means for colleges and national well-­being. Indeed, t­ here is evidence that American college students are not working as hard t­ oday as in the past. A 2011 study found that the average college student in the 1960s spent more than forty hours per week attending class and studying, but by 2003 the average time spent d ­ oing t­ hose t­ hings had declined to twenty-­seven 12 hours a week. A study of students enrolled in the University of California system in 2010 found that 19  ­percent spent less than eigh­teen hours a week ­going to class and studying, and 22 ­percent reported ­ doing less than half of the assigned reading in their 13 classes. When confronted with t­ hese findings, some of my own undergraduate students at Duke suggested that better tools for information retrieval permitted them to spend less time studying. But

260 D emand

one obvious possibility was that college faculty may simply have adjusted to lower student interest in study by reducing the amount of work required. More than one observer has suggested that faculty and students have struck an implicit bargain: we w ­ on’t burden you with school work if you leave us alone to do our research.14 Meanwhile, evidence from several sources suggested that many American college students w ­ ere not learning much in college, casting doubt on the success of colleges to teach useful skills. In 2003 the National Assessment of Adult Literacy found that only about one in three college gradu­ates could draw accurate inferences from two editorials with contrasting content or could accurately read a three-­variable graph relating age, exercise, and blood pressure.15 And Richard Arum and Josipa Roksa’s higher education bestseller of 2011, Academically Adrift, found that only about half of the students they studied made significant gains on a well-­validated test of critical thinking between the beginning of freshman and the ­m iddle of sophomore year. A year ­later, with se­n ior data in hand, they concluded that more than a third of college students failed to make significant gains on critical thinking between freshman and se­n ior years.16 ­Those students who failed to make significant gains on critical thinking tests w ­ ere more likely to have read and written l­ittle in their classes. ­Here we confront issues of paramount importance for both higher education and national prosperity. Although the college party scene is not something most professors want to think about very much, it is a topic with broad implications. ­Because learning depends so much on students’ own efforts, low levels of academic engagement have serious implications concerning the effectiveness of our baccalaureate education as a w ­ hole. In the example of the young Sonia Sottomayor, we see the opposite of low student engagement. In her we see a student who actively took part in her college education—in the editing she did to respond to an instructor’s corrections, in the interviews she conducted to write a term paper, in the hours of library research she did for her se­nior thesis, and in her passionate advocacy outside of class.

S o r t i n g b y S e r i o u s n e s s

261

I have no new information on the activities or academic effort of college students to add to the studies already cited. But I do have a rich trove of information on highly relevant personal characteristics of students at the time they enrolled in college. If one believes, as I do, that work habits and attitudes tend to persist, this information provides an unparalleled portrait of differences across colleges. In the remainder of this chapter, we turn to data collected in the Freshman Survey about characteristics of students that relate directly to ­these issues. We examine survey responses that reveal habits of work, seriousness of purpose, and academic aspirations. Although the information covers students at the outset of their college ­careers rather than in the midst of them, the analy­sis h ­ ere contributes to our understanding in two main ways. First, it differentiates among vari­ous categories of colleges. We suspect that students in some colleges are more studious or industrious than ­those in other colleges, but rarely do we have evidence beyond SAT scores to make such comparisons. Are ­there identifiable differences across colleges in academic orientation? We have certainly seen stark differences in high school grades, as shown in Chapter 8. ­Here we have information across college categories, categories that go beyond public versus private or that focus entirely on elite colleges. The second way this analy­sis contributes to our knowledge about the college market is the ability to look for trends over time. We can do so by comparing students who attended the exact same sets of colleges, but separated in time by two de­cades, and for some mea­sures, four. Habits and Inclinations

­ here may be no better indication of a serious approach to the acaT demic side of college than how much entering students studied the year before college, during their last year in high school. Among the questions posed to the first-­year students in the Freshman Survey was a series of items asking about how students had spent their time during their se­nior year in high school. ­These time-­use questions appeared in only the last two waves, 1989 / 90 and 2008 / 09, making

262 D emand

Private 1989/90 2008/09

Public

Pub

Priv

90+

50–80

80–90

0–50

99+

98–99

90–95

95–98

85–90

80–85

70–80

50–60

60–70

25–50

0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00

0–25

Share of first-year students

it pos­si­ble to examine trends, by college category, over a two-­decade period.17 Among the questions included in the survey was one very impor­tant indicator of academic seriousness: how much time t­ hese students had spent studying and d ­ oing homework. Across all first-­year students nationwide, the survey reveals a decline over the two-­decade period in time spent studying in high school, a trend that parallels the decline uncovered by the study of college students cited above. And this decline in time spent studying in high school was true nearly across the board. Figure 10.1 shows, by college category, the percentage of first-­year students who said they had studied an average of ten hours or more a week as se­n iors in high school. Averages by category are shown for the

HBCU

0.2 0.2 0.2 0.2 0.28 0.32 0.29 0.38 0.42 0.47 0.45 0.13 0.18 0.24 0.24 0.25 0.15 0.16 0.16 0.21 0.21 0.24 0.24 0.24 0.38 0.43 0.5 0.49 0.1 0.15 0.19 0.21 0.17 0.07

Figure 10.1. ​Share of Students Who Studied More than Ten Hours a Week in High School Source: Freshman Survey, students from 188 colleges, two waves (1989 or 1990 and 2008 or 2009). Note: The figure summarizes responses regarding time spent studying in high school. The question posed to first-­year students was, “During your last year in high school, how much time did you spend during a typical week ­doing the following activities?” The figure shows the percentage who said their average number of hours was ten or more per week. The enrollment-­weighted averages for all first-­year students in four-­year colleges ­were 19 ­percent for 1989 / 90 and 16 ­percent for 2008 / 09. See text for method used for time-­use questions.

S o r t i n g b y S e r i o u s n e s s

263

1989 / 90 and 2008 / 09 waves. For the general population of students at four-­year colleges, the percentage declined from 19 ­percent in 1981 to 16 ­percent in 2008 / 09, surely an unsettling trend for the nation. The share reporting ten or more hours declined in eleven of the seventeen categories, and another four categories showed small, statistically insignificant increases.18 Thus in only two categories ­were first-­year students in 2008 / 09 more likely than comparable students who attended the very same set of colleges in 1989 / 90 to report having studied ten hours a week or more in high school. ­These ­were the highly selective, high-­SAT private colleges, the same ones that ­were welcoming a bigger share of the nation’s A and A+ students. ­Here was a stark consequence of the frenzy to gain admission to elite colleges—­a growing gap in academic effort. The categorical survey responses from the Freshman Survey can be used to calculate the average number of hours the students reported spending in their se­nior year. Based on the enrollment-­weighted average for the 188 colleges in the sample, the average number of hours ­these entering students spent studying declined from 6.1 hours a week in 1989 / 90 to 5.6 hours a week in 2008 / 09. In the two de­cades between the two surveys, a host of new ways to spend time had come into being, including pagers, computers, cell phones, smart phones, instant messaging, and Facebook. W ­ hether or not the decline in time spent on homework can be laid to t­ hese innovations, the decline was significant (8  ­percent overall) and it touched first-­year students at most, but not all, colleges.19 To accentuate the differences across our categories of colleges, Figure 10.2 displays for the 1989 / 90 and 2008 / 09 waves the survey responses as deviations from each year’s national mean.20 The figure makes it easy to see the extraordinary gap in studying between students enrolling in the most selective colleges and all the rest. In 1989 / 90 the students starting at a college in one of the top four private college categories had studied at least three hours more per week, on average, than the average first-­year student in all four-­ year colleges. And for the students at colleges in the top two groups, the difference was four hours a week. Put another way, the students

264 D emand

5.0 4.0 3.0 2.0 1.0 0.0

4.06

5.35

5.35

Priv

90+

Pub −1.82 −0.93

4.32

4.27

0.99

3.54

3.60

0.08

2.99

1.39

0.63

1.77

1.52

0.93

2.19

1.46

0.93

1.47

0.97

0.58

0.14

1.03

−0.1

0.37

0.19

HBCU

−0.22

0.25

2008/09

−1.09 −1.11

1989/90

0.45

Public

0.08

Private

80–90

0–50

50–80

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−2.0

25–50

−1.0

0–25

Average hours/week minus U.S. average

6.0

Figure 10.2. ​Hours per Week Spent Studying in High School Source: Freshman Survey, students from 188 colleges, two waves (1989 or 1990 and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The question posed to first-­year students was, “During your last year in high school, how much time did you spend during a typical week ­doing the following activities?” The figure summarizes responses for “studying / homework.” Means by category w ­ ere calculated using midpoints of survey categories and the lower bound for the top category (twenty hours). The enrollment-­weighted means for all first-­year students in four-­year colleges ­were 6.1 hours for 1989 / 90 and 5.6 hours for 2008 / 09. See text for method used for time-­use questions.

entering one of the private colleges in the top two categories had put in, on average, about 70 ­percent more hours studying than the average student entering all four-­year colleges.21 In contrast, students in the numerically largest category, ­those enrolling in public institutions in the bottom half of the 1970 SAT lineup, reported in 1989 / 90 spending about one hour a week less than average studying, or about five hours a week. In comparison to this average, the students at colleges in the top two categories of private colleges had put in twice

S o r t i n g b y S e r i o u s n e s s

265

the number of hours studying in their se­nior year of high school.22 Only the students in public HBCUs in the latter period reported less—­just 3.8 hours a week. As unequal as ­these amounts of studying ­were in 1989 / 90, the gaps in high school study hours only grew larger over the next two de­cades. By 2008 / 09 the first-­year students entering the most selective private colleges (98th percentile and above) reported studying about five and a third hours more per week than the average first-­ year student in the country. Compared to the students at the below-­ median public institutions, they put in more than twice the amount of study time in high school. This growing gap in time spent studying in high school is a signal development, one that reflects both a worrisome downward trend for the ­g reat mass of entering college students and the contrasting over-­heated competition to garner one of the prized places in the country’ top colleges. In sum, the question about studying in high school reveals a significant—­a nd surprising—­gulf that distinguished the most selective private colleges from all the rest. Among the host of features that distinguished students at ­t hese selective colleges, another must be added: studying. It is noteworthy that the pattern of differences in studying across college categories revealed in Figure 10.2 is remarkably similar to the pattern we saw in Chapter 8 for high grades. Indeed, the studying was undoubtedly the principal pathway to ­those high grades.23 In the frenzy of intense efforts to gain admission to the most selective colleges, nothing was more directly ­under the control of students themselves than their own effort. Affluence alone would not suffice, though it might well ease the way, as we have seen. The Freshman Survey is a rich source of other details about how students used their time as se­niors in high school, and t­ hese details reinforce this impression of growing gaps between types of colleges mea­sured in terms of academic focus of their students. The next two figures summarize differences and trends for two activities that compete with studying in a high school se­nior’s decisions about how to spend the available hours of the week. Figure 10.3 displays the average number of hours per week devoted to socializing, which is

266 D emand

2.0 1.5 1.0 0.5 0.0 −0.5 −1.0 −1.5 −2.0

Priv

90+

80–90

Pub −0.17 −2.84

−0.17 −2.38

HBCU −1.12 −0.86

−0.59 −0.14

0.93 1.03

0.01 −0.28

−2.25 −1.71

−1.74 −0.66

0.62 −0.25

−1.12 −0.36

1.63

Public

−0.07

0.25 −0.07

0.88 −0.30

0.45 0.62

−1.00 −0.14

0.16

2008/09

−0.82 −0.10

1989/90

−0.43

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−3.0

25–50

−2.5

0–25

Average hours/week minus U.S. average

the sum of two activities named in the survey: “socializing with friends” and “partying.” The weighted average for all first-­year students was 16.5 hours a week in 1989 / 90, more than double the average amount of time spent on studying. Students attending the top-­ranked private colleges ­were decidedly below-­average on this score, but they w ­ ere outdone, in 1989 / 90, by ­those attending HBCUs, who reported socializing even less. That HBCU distinction evaporated, however, in the 2008 / 09 wave. For that most recent wave, it was the students enrolling in the most selective private colleges who,

Figure 10.3. ​Hours per Week Spent Socializing in High School Source: Freshman Survey, students from 188 colleges, two waves (1989 or 1990 and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The question posed to first-­year students was, “During your last year in high school, how much time did you spend during a typical week ­doing the following activities?” The figure summarizes responses for the sum of responses for “socializing with friends” and “partying.” The enrollment-­weighted means for all first-­year students in four-­year colleges ­were 16.5 hours for 1989 / 90 and 13.9 hours for 2008 / 09. See text for method used for time-­use questions.

S o r t i n g b y S e r i o u s n e s s

267

2.0 1.5 1.0 0.5 0.0

Pub

Priv

90+

80–90

0.19

1.42

1.78

1.47

1.67

HBCU

0.04

−0.17 −0.05

−0.08 −0.18

−0.13 −0.07

−0.57 −0.80

−0.76 −0.91

−0.16 −0.39

Public −0.11 −0.34

0.17 −0.05

0.29

0.08 0.05

0.09

0.42 0.30

0.20 −0.11

0.44

2008/09

−0.21 −0.23

1989/90

0.34

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−1.0

25–50

−0.5

0–25

Average hours/week minus U.S. average

again, stood out for how low their reported hours ­were. This increased distinctiveness among students at ­these selective colleges would be, like the patterns for studying, consistent with a growing intensity among ­those competing for spots in the top colleges. Another activity that competes for time is the infamous time sink, tele­vi­sion. As shown in Figure  10.4, the number of hours watching tele­vi­sion in the two waves of the survey seem rather modest (4.8 and 4.3 hours a week, respectively), as do the differences across most of the college categories, and t­ hese patterns held steady over the two

Figure 10.4. ​Hours per Week Spent Watching Tele­v i­sion in High School Source: Freshman Survey, students from 188 colleges, two waves (1989 or 1990 and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The question posed to first-­year students was, “During your last year in high school, how much time did you spend during a typical week ­doing the following activities?” The figure summarizes responses for “watching TV.” The enrollment-­ weighted means for all first-­year students in four-­year colleges ­were 4.8 hours for 1989 / 90 and 4.3 hours for 2008 / 09. See text for method used for time-­use questions.

268 D emand

de­cades being examined. The students of two groups of colleges stand out—in opposite directions: ­those entering the most selective private colleges (98th percentile and above) and t­ hose attending the HBCUs. By 2008 / 09 students entering the private colleges in the highest-­SAT category watched tele­vi­sion a half an hour less per week than average, while ­those entering HBCUs averaged an hour and a half more, a difference of two hours a week. To put in perspective the patterns we observed for studying, it is necessary to turn to the activity that, except for sleep, most directly competes with studying for the time of most high school se­n iors: working for pay. As shown in Figure  10.5 the amount of time students spent working for pay during their se­nior year varied markedly across the students entering colleges in the dif­fer­ent categories. In fact, the time spent working is nearly the mirror image of the pattern for studying. The four groups of first-­year students who studied the most also worked the least in paid jobs, and by roughly the same number of hours. This is no mere coincidence. The negative correlation between working and studying is clearest for the numerically largest college category, the public institutions that ­were below the  median of SATs. In both survey waves ­these students reported studying one hour a week less than average and working one hour a week more than average. In the most selective private colleges, the two averages are also nearly perfect mirror images. ­These opposing patterns serve as a reminder that students’ academic efforts are subject to practical constraints, some of which are strongly correlated with economic status. Without suggesting that economic distinctions determine the observed gaps in studying, this figure is an emphatic reminder that high school students with afterschool jobs necessarily have fewer hours left over for all other activities, including studying. T ­ hose whose ­family circumstances afforded them the freedom to study more had a palpable advantage over ­those who did not have that luxury. Still, time spent working does not control time spent studying any more than the reverse. The week contains many hours, and t­ here are many ways to spend them, and even the students who have a job can usually find time to study. Perhaps the

S o r t i n g b y S e r i o u s n e s s

269

0.0 −1.0 −2.0 −3.0

Pub 0.15 −0.33

−0.47 −0.55

0.24 −0.38

HBCU −2.69 −1.97

0.30 0.47

Priv

90+

80–90

1.00 0.98

−4.99 −4.65

−4.83 −4.94

−3.78 −3.12

Public −3.23 −3.32

−1.61 −1.24

−1.85 −1.74

−1.45 −1.34

−0.22 −0.13

−0.72 −0.01

2008/09

0.46 −0.50

1989/90

0.13 −0.56

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−5.0

25–50

−4.0

0–25

Average hours/week minus U.S. average

1.0

Figure 10.5. ​Hours per Week Spent Working for Pay in High School Source: Freshman Survey, students from 188 colleges, two waves (1989 or 1990 and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The question posed to first-­year students was, “During your last year in high school, how much time did you spend during a typical week ­doing the following activities?” The figure summarizes responses for “working (for pay).” The enrollment-­ weighted means for all first-­year students in four-­year colleges ­were 9.5 hours for 1989 / 90 and 7.8 hours for 2008 / 09. See text for method used for time-­use questions.

most telling fact that emerges from the trends shown in Figures 10.3 and 10.4 is this: the increase in time studying among students entering the top private colleges occurred over a period when their hours working did not decrease. Something more than mere affluence was spurring them to step up their academic game in their se­nior year of high school. The last category of time use shown h ­ ere is quantitatively less impor­tant than studying, socializing, or working, but it is suggestive of a subtle change over time. The Freshman Survey asked about

270 D emand

1.5

1.0

0.5

Pub

Priv

90+ 0.01

0.15

0.11

1.76

HBCU 0.74 −0.01

80–90 −0.18 −0.50

−0.08 −0.05

0.34 −0.24

0.35 −0.44

0.24 −0.35

0.15 −0.48

Public 0.33 −0.39

0.36 −0.40

0.31 −0.44

0.44 −0.31

0.09 0.17

0.26 −0.34

2008/09

0.20 −0.25

1989/90

0.32 −0.43

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−0.5

25–50

0.0

0–25

Average hours/week minus U.S. average

2.0

Figure 10.6. ​Hours per Week Spent Volunteering in High School Source: Freshman Survey, students from 188 colleges, two waves (1989 or 1990 and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The question posed to first-­year students was, “During your last year in high school, how much time did you spend during a typical week ­doing the following activities?” The figure summarizes the sum of responses for “volunteer work.” The enrollment-­weighted means for all first-­year students in four-­year colleges w ­ ere 2.2 hours for 1989 / 90 and 2.2 hours for 2008 / 09. See text for method used for time-­use questions.

two activities that may well signal special efforts that students can make to enhance their chances of getting into the college of their dreams: volunteer work and talking with teachers outside of class. Although ­these activities certainly might be driven by motives as pure as the wish to help ­others or the love of learning, they could also be part of a strategy to polish a college application, e­ ither directly, by way a rec­ord of community ser­v ice, or indirectly, by way of better grades or teachers’ recommendations. Of t­ hese two uses of time, only results for volunteering are shown, largely ­because the patterns ­were so similar for time spent talking to teachers.24 Figure  10.6

S o r t i n g b y S e r i o u s n e s s

271

shows the patterns for volunteering. In 1989 / 90 students attending private colleges ­were generally below the national average in hours of volunteer work, but by 2008 / 09 they ­were above the average, a swing of more than half an hour in most categories. ­These swings are consistent with a rise in what Demerath has called “hypercredentialing,” wherein students trying to get into selective colleges ­were increasingly building up their resumes with meritorious activities such as volunteering.25 The biggest changes in reported volunteering occurred among students attending private HBCUs, who reported volunteering an average of almost four hours a week. ­Those at public HBCUs remained above-­average in both waves. ­These reports on students’ use of time in high school represent a highly revealing indicator of impor­tant habits and propensities. The trends are especially revealing, in that each pair of bars in ­every graph represents students who attended the identical colleges, only separated in time by two de­cades. As leopards seldom change their spots, confirmed goof-­offs in high school rarely become conscientious college students. On the other side of the coin are t­ hose who ­were diligent students in high school, and they, not surprisingly, tend to enroll at selective colleges. Not coincidentally, the latter group spent fewer hours per week working for pay. But the headline finding from ­t hese comparisons is the increased time spent studying in high school by ­those who ended up attending the most selective colleges. This escalation in effort looks like a direct response to the increased difficulty of gaining admission to a top selective college. Aspirations

Although past be­hav­ior might be a better predictor of students’ inclination t­ oward academic study in college than expressed attitudes, it is illuminating to see what students say about their aspirations for college and beyond. The Freshman Survey asked students a series of questions about their ultimate objectives a­ fter college. Fortunately for the sake of tracing trends over time, many of the ques-

272 D emand

tions have been left unchanged over the years. Particularly promising for assessing seriousness of purpose is a series of potential accomplishments. Based on ­these survey questions, an indicator of ambition can be created using three, admittedly rare, achievements: “becoming an authority in my field,” “making a theoretical contribution to science,” and “writing original works (poems, novels, ­etc.).” (Adults know ­these are singular accomplishments; first-­year college students clearly do not.) Students ­were asked to rate the importance of each to them, from “not impor­tant” to “essential.” Students who indicated it was “essential” for them to achieve at least one of ­those accomplishments ­were identified. Since t­ hese questions w ­ ere asked in e­ very one of the three survey waves, we can trace the responses of students who attended exactly the same set of colleges over nearly four de­cades and compare the results. For all first-­year students over this period the percentage who answered “essential” for at least one of t­ hese accomplishments was remarkably unchanged, at roughly 26 to 28 ­percent. One group of students stood out from the rest, however. As Figure  10.7 makes clear, ­those attending HBCUs ­were real outliers, revealing dramatically more ambitious expectations. This tendency remained steadfast over four de­cades. This is especially the case for t­ hose g­ oing to the private HBCUs, whose share exceeded the national average by an astonishing 15 percentage points or more. T ­ hese high stated aspirations are stunning. Among the remaining college groups it is pos­ si­ble to discern two other in­ter­est­ing regularities. First, students at public colleges simply had more modest academic aspirations than ­those in private colleges. Second, we see a steady increase in academic orientation among students attending private colleges in the top four SAT classes, a finding that harmonizes with t­ hose seen in time usage. The Freshman Survey asked students about their likely majors in college. Among the most rigorous college majors are ­those in science, technology, engineering, and mathe­matics, the so-­called STEM fields. Although not necessarily synonymous with academic seriousness, the resolution to major in one of t­ hese fields would surely tend

S o r t i n g b y S e r i o u s n e s s

273

0.25 0.20 0.15 0.10 0.05

Pub

Priv

0.10 0.19

0.06

0.16 0.28

0.15

80–90

90+ 0.01

0.06

HBCU

0.03

0.00 −0.02 −0.01

0.00 −0.03 −0.01

−0.01 −0.03 −0.04

0.04 0.07 0.07

0.03 0.03 0.09

0.02 0.02 0.06

0.00 0.00

0.02

0.02

0.01

Public

0.04

0.00 −0.01 0.00

0.00 0.00 0.02

0.01 0.00

0.01 −0.01 −0.02

0.00

2008/09

0.00

0.00

0.01

0.00

1989/90

0.02

1972

0.00

Private

50–80

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−0.05

25–50

0.00

0–25

Average for college category minus U.S. average

0.30

Figure 10.7. ​Share of First-­Year Students for Whom Scholarly or Artistic Success Is “Essential,” Relative to All First-­Year Students, 1972, 1989 / 90, and 2008 / 09 Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The figure summarizes the percentage of first-year students for whom at least one of the following was “essential”: “becoming an authority in my field,” “making a theoretical contribution to science,” or “writing original works (poems, novels, etc.).” The enrollment-weighted means for all first-year students in four-year colleges were: 26.8 percent for 1972, 27.6 percent for 1989/90, and 25.5 percent for 2008/09.

to be correlated with seriousness. Figure 10.8 shows the percentage of students in each of the college categories who indicated a probable STEM major. The differences across the college categories are huge, led by the unusually high shares for the public institutions in the 90+ percentile group, a result surely of the unusual makeup of the colleges that ended up in this category in our sample of 188.26 ­A fter ­these top public colleges, the categories with the highest shares of probable STEM majors ­were the private ones in the three most selective categories. In sharp contrast, students in the private colleges

274 D emand

0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00

Pub

Priv

90+

0–50

80–90 0.05 0.04

0.38

0.03 −0.07

0.00 −0.09

0.09

0.29

0.06

−0.02

0.28

0.00

HBCU

0.03 −0.01

−0.06 −0.03 −0.02

0.17 0.19

0.16

0.11 0.11

0.06

0.14 0.16

0.09

0.02 0.06

0.01

0.06 0.06

0.09

0.04

0.00

Public

0.00

−0.03 −0.03 −0.03

−0.08 −0.09 −0.06

2008/09

0.00 −0.05 −0.03

1989/90

−0.08 −0.06 −0.07

1972

−0.10 −0.08 −0.09

Private

50–80

99+

98–99

95–98

90–95

80–85

85–90

70–80

60–70

50–60

−0.10

25–50

−0.05

0–25

Average for college category minus U.S. average

0.40

Figure 10.8. ​Share of First-­Year Students Planning to Major in STEM Field, Relative to All First-­Year Students, 1972, 1989 / 90, and 2008 / 09 Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The figure summarizes the percentage, among first-­year students who indicated their “probable field of study,” chose one of the following: biology, engineering, other technical, math / statistics, or physical science. The enrollment-­weighted means for all first-­year students in four-­year colleges w ­ ere 20.9 ­percent for 1972, 21.9 ­percent for 1989 / 90, and 26.0 ­percent for 2008 / 09.

below the 80th percentile of 1970 SATs and the public institutions in the bottom half w ­ ere consistently less likely than average to be considering a STEM field. Evidently, colleges at lower SAT ranges simply held ­little appeal for STEM-­minded students. Among students enrolling at HBCUs, the percentages w ­ ere especially low only in 1972. Turning to trends over the four de­cades, we see a small but steady decline in STEM interest in the vast below-50th percentile public category. But for students attending top-­ranked private colleges, the four-­decade span was one of steadily increasing shares of students

S o r t i n g b y S e r i o u s n e s s

275

aiming for STEM majors, yet one more indication of a growing concentration of academically serious students in ­these colleges. Another major that accounts for a sizable share of undergraduates is business, and ­here we can see another mirror image. Unlike STEM majors, not all colleges offer it, or have not always offered it, so differences across colleges tend to reflect availability as well as student interest. Figure  10.9 shows the resulting patterns. It indicates a division among private colleges, with students attending colleges below the 70th percentile in 1970 SATs showing above-­average

0.10 0.05 0.00 −0.05 −0.10 −0.15

Pub

Priv

0.11 −0.03

0.10

0.02 −0.02 −0.02

80–90

90+

HBCU −0.09 −0.17 −0.07

0.02 −0.01 −0.01

−0.01 −0.03 −0.06

0.00 0.01 0.02

−0.13 −0.21 −0.08

−0.12 −0.19 −0.09

−0.07 −0.16 −0.06

0.02 −0.03 −0.02

Public −0.03 −0.10 −0.04

0.01 0.05 −0.02

0.09 0.01

0.00 −0.01 −0.01

0.07 0.00

0.07

0.00

2008/09

0.06

0.05

0.03

0.03

1989/90

0.00

1972

0.06

Private

50–80

0–50

99+

98–99

90–95

95–98

85–90

80–85

70–80

60–70

25–50

−0.25

50–60

−0.20

0–25

Average for college category minus U.S. average

0.15

Figure 10.9. ​Share of First-­Year Students Planning to Major in Business, Relative to All First-­Year Students, 1972, 1989 / 90, and 2008 / 09 Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The figure summarizes the percentage, among first-­year students who indicated a “probable field of study,” chose business. The enrollment-­weighted means for all first-­year students in four-­year colleges w ­ ere 10.3 ­percent for 1972, 24.5 ­percent for 1989 / 90, and 17.9 ­percent for 2008 / 09.

276 D emand

rates of projected business majors, and ­those above the 80th percentile showing below-­average rates. In short, STEM and business majors tend not to coexist u ­ nder the same roof. In colleges where one thrives, the other languishes, and vice versa. ­A fter all, STEM majors all belong to the traditional liberal arts curriculum, a set of disciplines embraced by the colleges boasting high SAT scores as well as high prestige. Preparing for the “Party Pathway”

As the brief references to previous studies cited in this chapter’s first section indicate, academic pursuits are not necessarily the top priority for students g­ oing to college. And in a number of impor­tant re­spects, colleges accommodate the nonacademic motives of students, a subject discussed in Chapter 7. To gain some sense of how student expectations might fit into the acknowledged opportunities that college affords for socializing, we turn to tabulations based on two questions, in addition to the information discussed above regarding time spent on socializing in high school. Alcohol is widely available to college students, no m ­ atter what the ­legal drinking age might be. Are t­ here any differences across college groups in the share of students who use alcohol? One question in the Freshman Survey that remained the same over the four de­cades covered ­here is ­whether students drank beer in the previous year. For first-­year students in the country, the percentage who reported they had done so was 56 ­percent in 1972, 62 ­percent in 1989 / 90, and just 41 ­percent in 2008 / 09. Figure 10.10 reports the average by college category relative to the national average. By a wide margin, students ­going to HBCUs w ­ ere more likely to abstain, with rates of beerdrinking far below the national average. The 22 to 34 percentage point gaps are extraordinary by any mea­sure. In the remaining college groups, the historically white ones, relative rates of exposure to beer increased in private colleges above the 80th percentile in SAT and increased from 1989 / 90 to 2008 / 09 in public colleges in the top half of that distribution.

S o r t i n g b y S e r i o u s n e s s

277

0.05 0.00 −0.05 −0.10 −0.15 −0.20 −0.25

Pub

Priv

−0.27 −0.28 −0.28

−0.28 −0.32 −0.22

80–90

90+ 0.00

HBCU

−0.04 −0.06

0.08 0.05

0.02 −0.01

0.11

0.03

0.02 −0.01 0.07

0.00

0.03 −0.03 0.05

−0.05 −0.01

0.05 0.09

0.03 −0.02 −0.04

0.03 0.09 0.11

0.03

0.00

Public

0.09 −0.03

0.04 −0.02 0.03

0.00 −0.03

−0.06 −0.01 −0.05

2008/09

0.05

1989/90

−0.05 −0.09 −0.08

1972

−0.02 −0.04 −0.06

Private

50–80

0–50

99+

98–99

90–95

95–98

85–90

80–85

70–80

60–70

25–50

−0.35

50–60

−0.30

0–25

Average for college category minus U.S. average

0.15 0.10

Figure 10.10. ​Share of First-­Year Students Who Drank Beer in High School, Relative to All First-­Year Students, 1972, 1989 / 90, and 2008 / 09 Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. When asked w ­ hether, during the previous year, they drank beer, the figure shows the percentage who answered “occasionally” or “frequently.” The enrollment-­weighted means for all first-­year students in four-­year colleges ­were 56.4 ­percent for 1972, 62.0 ­percent for 1989 / 90, and 41.1 ­percent for 2008 / 09.

The last indicator of academic focus, or lack thereof, comes from questions in the 1972 and 1989 / 90 waves asking ­whether students expected to join a fraternity or sorority.27 See Figure  10.11. Least likely to say they had such plans in 1972 w ­ ere students in the top-­ ranked public and private colleges. Most likely ­were ­those starting out at HBCUs, followed closely by t­ hose g­ oing to private colleges in the 85th to 90th percentile group, a group including Claremont McKenna, Gettysburg, and Tulane. Between the first and second waves of the survey, the rates at private colleges below the 70th per-

278 D emand

0.15

0.05 0.00 −0.05 −0.10

Pub

Priv

90+

0.10

0.11 0.07

−0.12 −0.05

HBCU

0.13

0–50

80–90 −0.02 −0.09

0.00 0.04

0.05

0.10

0.05

0.09

−0.02 −0.01

0.05 0.06

−0.18 −0.14

0.05 −0.01

−0.08 −0.11

0.07 0.02

0.00

0.06 −0.03

0.04

0.05

1989/90

Public 0.01 −0.02

1972

−0.01

Private

50–80

99+

98–99

95–98

90–95

80–85

85–90

70–80

60–70

50–60

−0.20

25–50

−0.15

0–25

Average for college category minus U.S. average

0.10

Figure 10.11. ​Share of First-­Year Students Considering Joining a Fraternity or Sorority, Relative to All First-­Year Students, 1972 and 1989 / 90 Source: Freshman Survey, students from 188 colleges, two waves (1972 and 1989 or 1990). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. Students w ­ ere asked, “What is your best guess as to the chances that you ­w ill join a social fraternity or sorority?” The figure shows the percentage who answered that t­ here was “some chance” or “a very good chance.” The other pos­si­ble responses w ­ ere “no chance” and “very ­little chance.” The enrollment-­weighted means for all first-­year students in four-­year colleges w ­ ere 57.2 ­percent for 1972 and 61.1 ­percent for 1989 / 90.

centile reverted ­toward the national mean (which increased form 57 ­percent to 61 ­percent), and the percentage at the highest-ranked private colleges fell further below the national average. An unexpected channel by which colleges have become more unequal parallels the increased differentiation in terms of academic readiness. Students at the most selective institutions w ­ ere studying harder in high school compared to most of their peers across the country. This intensification of effort in high school among top students is no coincidence. It occurred over the same two de­cades when

S o r t i n g b y S e r i o u s n e s s

279

their grades ­rose relative to the average and their SAT scores r­ ose as well. Taken together, t­hese advances by students who went to the most selective colleges reflect a remarkable intensification of effort at the top. Over a two-­decade period when high school students ­were spending less time studying (but earning higher grades), the students aiming for colleges like Bowdoin, Davidson, Brown, and Stanford ­were buckling down. They ­were taking Advanced Placement courses, volunteering for local charities, and taking SAT prep courses. Studying more was just part of the larger campaign to gain admission to a top college. Affluence itself did not guarantee admission to one of t­hese desired colleges, though it helped. Coming from an affluent ­family made it pos­si­ble to attend a private school or a public school in a good school district, where many AP courses ­were available. It made an expensive SAT prep course, perhaps a tutor as well, affordable. And it made part-­time work unnecessary, clearing the student’s calendar for sports, volunteering, and studying. But even the ­children of affluent parents had to be active partners in a f­ amily’s efforts to gain admission to one of the most selective colleges. For the vast majority of high school se­niors, efforts of that magnitude ­were not necessary. Getting admitted to Indiana State or Tennessee Wesleyan was ­orders of magnitude less daunting than getting into Pomona or Prince­ton. For the vast majority of applicants, ­there seemed to be l­ ittle penalty for working less in high school. In this contrast—­between the efforts required to get into a top college and ­those needed to get into a less selective one—we see the roots of an increasingly bifurcated college market. The colleges at the top are the envy of the world, in part ­because the undergraduates who go ­there have striven mightily to get in.

280 D emand

11 Sorting by Belief?

Just as restaurants serve dif­fer­ent kinds of food, colleges offer versions of education that can differ a good deal from one college to the next, a fact that Chapter  2 amply illustrates. Colleges differ in size, curricular requirements, social rules, and athletics, to name just a few re­spects. ­These are differences in supply, not demand. To be sure, such differences in supply might influence demand, but the college market is a long way from being ruled by some version of Say’s Law, with supply creating its own demand. Having described variations in colleges, in par­tic­u­lar differences in religious orientation, it is fitting now to consider demand, exploring to what extent birds of a feather do flock together. One virtue of the Freshman Survey is its inclusion of a host of student characteristics by which demand could be described. Indeed, previous chapters in this part of the book have already described patterns of enrollment by race and language (Chapter 4), academic preparation of students (Chapter 8),

socioeconomic status (Chapter  9), and habits and aspirations (Chapter  10). ­Here we turn to two dimensions of demand rarely dealt with in the economics of education—­politics and religion. No aspect of college seems more likely to influence patterns of demand—as restaurant menus shape clienteles—­than religion. Historians of higher education tell us that a principal reason why religious bodies established colleges was to provide places where their young ­people could safely pursue their studies in an environment not threatening to their religious upbringing. ­Today, Catholic families wishing to find colleges with a “commitment to a faithful Catholic education” can consult the Newman Guide to Choosing a Catholic College. This specialized college guide reports, for example, on the share of faculty and students who are Catholic, the number of days per week confession is scheduled, and w ­ hether dormitories allow opposite sex visitors into students’ bedrooms.1 Similar intelligence is available on evangelical Protestant colleges.2 That students might sort themselves by religion in choosing which college to attend is suggested by the existence of colleges with formal religious affiliations, as described in Chapter 2. As to po­liti­cal sorting, ­there is less reason to believe that it occurs, but the question certainly invites asking, particularly since the emergence of the so-­called Christian Right in the 1980s and the close association that developed between evangelical Protestantism and conservative politics. Most notably, several private colleges founded by widely recognized evangelical leaders became known not only for their religious orientation but also for the po­liti­cal leanings of their leadership. Among the most prominent examples ­were Liberty University, whose founder, Jerry Falwell, openly supported conservative ­causes and candidates, and Regent University, founded by televangelist and 1988 Republican presidential candidate Pat Robertson.3 Parallel examples of left-­leaning colleges are harder to identify, perhaps b ­ ecause, as some critics would argue, liberal leaning among universities and university faculty is sufficiently pervasive as to be unremarkable.4 Like the data source and format used in previous chapters, the data presented h ­ ere highlights differences across colleges regarding

282 D emand

the religious affiliation, religious practice, and po­liti­cal orientation of their entering students. Like much of what has come before, the source of data is the Freshman Survey, which allows us to consider not only how students differ by category of college but how dif­fer­ent generations of students attending t­ hese same colleges compared. As in the preceding chapters, comparisons across the seventeen categories of colleges for three survey waves begin with 1972 and end with 2008 / 09. To the extent that ­these categories, as narrow as some of them are, nevertheless combine dissimilar colleges, the comparisons using ­these categories tends to understate the true amount of variation that may exist between individual colleges. We can address this aggregation prob­lem somewhat by further dividing one of the seventeen categories into three pieces. As noted elsewhere, the calculations used throughout the book using the Freshman Survey for colleges in the 25th to 50th percentile category are based on data from twenty-­four colleges. The colleges in this category are further subdivided into three smaller groups: one consisting of the six Catholic colleges in the category, one made up of five colleges identified as Protestant evangelical, and one consisting of the thirteen remaining colleges in the original category. 5 Religion

For Amer­i­ca’s first colleges, religious belief was central to their reason for being, as it was central to the lives of so many of the Eu­ro­pean settlers who inhabited this continent. The country’s first college, Harvard, was founded for reasons that combined the ecclesiastical with the practical. In his history of Harvard, Samuel Eliot Morison writes, “Enthusiasm for education was one aspect of that desire to know and do the ­will of God that bound the puritans together, and led them to brave the sea, the wilderness, and the New ­England climate. Their Congregational churches must have a learned clergy, cost what it might. A puritan minister must be able to expound the Sacred Scriptures from the original Hebrew and Greek, and be cognizant of what the Church F ­ athers, the Scholastic Phi­los­o­phers, and

S o r t i n g b y B e l i e f ?

283

the Reformers had written, in Greek and Latin.”6 Through the nineteenth-­century period of college establishment, or­ga­nized religion continued to have a role in the founding of most colleges. Owing to the longevity of so many of ­these colleges, and to the religiosity of the American public, the college landscape has retained to this day a remarkably large share of institutions with religious roots, although many of ­these colleges have over time distanced themselves from ­those beginnings. How does this landscape of supply correspond to patterns of demand? For many con­temporary urban parents who bring their sons and d ­ aughters to ­these colleges t­oday, the religious allusions embedded in the mottos and images of ­these colleges must surely seem anachronistic, if ultimately inconsequential in shaping the college’s con­temporary program. For many other parents, such as ­those who consult the Newman Guide or other­wise care about the religious content of their ­children’s education, ­these religious ele­ments can be of the first importance. As we saw in Chapter 7, for many colleges such religious antecedents continue to define their mission and shape their program of education. No description of demand in the American college market can be complete without at least noting the religious backgrounds of the students they enroll. Nor does the importance of religion apply exclusively to private colleges; a complete description must include patterns of enrollment in the public as well as private sectors. The most straightforward way to describe the religious component of demand is to ask how students identify themselves. Figure  11.1 summarizes the stated religious preferences of first-­year students across the seventeen college categories in each of the three survey waves. Asking about “your current religious preference,” the survey typically offered students more than a dozen choices, depending on the survey year. For the graph, ­these ­were combined into five: Jewish, Protestant, Roman Catholic, Other, and No religion. The survey in 1972 was unusual, being one of a few over the history of the Freshman Survey to offer only t­ hose five categories. This unusual feature appears to be the reason for the large share in the “Other” category for

284 D emand

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

0–25 25–50 50–60 60–70 70–80 80–85 85–90 90–95 95–98 98–99

Catholic

0–50 50–80 80–90

Protestant

Other

90+

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

99+

Public

Private

Jewish

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

1972

2008/09

0

Private Public HBCU

No religion or response

Figure 11.1. ​Religious Preference of First-­Year Students, by Category and Survey Wave Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show the percentage distribution of students’ indicated religious preference by college category and survey wave. The survey asked students to indicate “Your current religious preference.”

HBCUs in 1972, and for that reason we do not discuss the estimates for HBCUs for that survey year.7 The bars in Figure 11.1 show the resulting religious distribution for each college category, with the percentage giving no response added to the “No religion” response at the top of each bar. The religious group most unevenly distributed across the categories is Jewish, shown at the bottom of each bar. The colleges in some categories had virtually no Jewish students. Aside from their absence in HBCUs, which is not surprising given their paucity of white students, the virtual absence of Jewish students at both the private and public colleges in the lowest SAT percentiles is very striking. In sharp

S o r t i n g b y B e l i e f ?

285

contrast, the categories containing the most selective private colleges revealed a sizable presence of Jewish students—10  ­percent or more in the three most selective categories. Percentages of Roman Catholics and Protestants also differed across the categories, but the variation was quite a bit less. The share of Catholics was less than 10 ­percent in the sample’s public HBCUs in all three waves and less than 20 ­percent in the private HBCUs (which included one Catholic college, Xavier) and the two most selective categories of private colleges. By contrast, Catholics consistently made up more than a quarter of the students in the categories containing most of the public colleges as well as the private colleges below the 90th percentile of SATs. Protestants, reflecting their greater numbers in the population, outnumbered Catholics in most categories. In the HBCUs, Protestants made up by far the predominant religious group. A noteworthy difference across the college categories was the share of entering students who indicated no religious preference. The categories with the largest share of students to give this response w ­ ere the highest-­ ranked private colleges, ­those with 1970 era SATs at the 90th percentile or higher, where at least a quarter of entering students gave this answer. The lower level of religiosity suggested by ­these negative responses in the top private colleges is also reflected in religious practice. Comparing the bars over the four-­decade span reveals above all how stable t­ hese differences ­were between the college categories. But changes did occur over ­these de­cades in how entering students classified themselves. For most categories, the percentage of students who indicated a religious preference went up in most of the categories between 1972 and 1989 / 90. But this trend reversed itself over the succeeding two de­cades, with the share indicating a religious preference declining in all but one of the seventeen college categories, the exception being the private HBCUs. ­These declines in the share of college students who indicated any religious affiliation paralleled a steady weakening of the hold of or­ ga­nized religion in the country. Church attendance in the adult population, for example, steadily declined over time. In 1972 some

286 D emand

37  ­percent of Americans said they attended church ­every week or almost ­every week. By 2012 that share had fallen to 32 ­percent. The decline was much sharper for ­ those attending less often—­ from 44 ­percent in 1972 to 25 ­percent in 2012. Meanwhile the percentage saying they never attend church more than doubled over that same period, rising from 18 ­percent to 43 ­percent. 8 It should not be surprising, therefore, that a decline in religious observance in the population would show up among college students. What is worth noting, however, is the wide and growing differences across the categories of colleges, with students at the lower-­SAT private colleges and the HBCUs more likely to have traditional religious affiliations while students at the top colleges increasingly ­were turning away from or­ga­nized religion altogether. Diversity is a byword of higher education ­today, an aspect honored by colleges, but religion is not an index by which diversity is usually mea­sured. Yet ­there is ­every reason to think that religious diversity offers some of the same educational benefits that are associated with racial and economic diversity. One ready mea­sure that can be used to reflect diversity, or its absence, is an index of homogeneity applied to religious affiliation, calculated as the sum of the squared proportions for dif­fer­ent groups.9 The higher the value, the less the diversity. If ­every student in a college professed the same religion, the index would be 1.0. Lacking information on religious distributions for individual colleges, we can calculate this index of religious homogeneity for the students attending all the colleges in each of the seventeen college categories used in this study. B ­ ecause colleges even within a category differ in the religious makeup of their students, combining colleges in this way tends to increase mea­sured diversity, thus decreasing the value of this index. Figure 11.2 shows the calculated index of religious homogeneity for each of the seventeen categories in all three survey waves, all expressed as a deviation from the national mean. For the predominantly white colleges, the figure reveals an unmistakable correlation between average SAT and mea­ sured religious homogeneity: the higher the average SAT of students, the less homogenous (more diverse)

S o r t i n g b y B e l i e f ?

287

0.25 0.15

Pub

Priv

0.26 0.37

0.08

0.12 −0.02 0.28

50–80

90+

HBCU −0.06 −0.05 −0.02

−0.05 −0.03 −0.02

−0.04 −0.03 −0.02

0.02 0.02 0.03

−0.07 −0.11 −0.06

−0.07 −0.10 −0.06

−0.08 −0.10 −0.08

Public −0.08 −0.07 −0.05

−0.07 −0.07 −0.02

0.00 0.00 −0.03

0.02 −0.01

2008/09

−0.05 −0.01 −0.01

0.00 0.01 −0.02

1989/90

0.03

0.03

0.01

0.02

1972

0.02

−0.01 −0.01

Private

80–90

0–50

99+

98–99

95–98

90–95

85–90

80–85

70–80

60–70

50–60

−0.15

0–25

0.05 0.00 −0.05

25–50

Average for college category minus U.S. average

0.35

Figure 11.2. ​Religious Homogeneity, by College Category and Survey Wave Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. Religious homogeneity is mea­sured by the Herfindahl index, the sum of the squared proportions for each indicated religious group. Diversity is indicated by more negative values. The figure is based on responses about the “current religious preference” of the student. By college category and year, proportions w ­ ere calculated for each of the following five categories: Jewish, Catholic, Protestant, Other, and the remaining category, which includes students who answered “none” and t­ hose who did not give an answer. For each category, the Herfindahl index was calculated by summing the squares of ­these five proportions. Higher values of the index indicate a heavier concentration of students into one or more groups. Lower values indicate more even distributions of students in each of the religious groups. The enrollment-­weighted means for the index w ­ ere 0.299 for 1972, 0.346 for 1989 / 90, and 0.334 for 2008 / 09.

the collections of students are in each category. Notably, this conclusion applies to public and private colleges alike. The historically black colleges are the outliers, with positive indices that indicate spectacularly more homogeneity in religious background compared to the predominantly white colleges. As shown in Figure 11.1 the stu-

288 D emand

dents attending HBCUs tended overwhelmingly to be Protestant. The chances that a student at an HBCU would run into another student who was not a Protestant was simply not very high. The reader is reminded, however, that calculations based on categories that combine survey responses for students attending dif­ fer­ent colleges tend to overstate the a­ ctual religious diversity of individual colleges if t­ here exist differences across t­ hose colleges. If colleges with similar religious traditions could be grouped together and examined separately, it would be pos­si­ble to obtain a more accurate gauge of a­ ctual religious diversity as experienced by students. This can be illustrated by further subdividing one of the seventeen college categories, that for private colleges with 1970-­era SATs in the 25th to 50th percentile range. Owing to the large numbers of colleges in this category, we can obtain a more detailed division while maintaining the required minimum group size of five by separating out the six Roman Catholic colleges and five denoted as evangelical Protestant colleges from the rest. Not surprisingly, breaking up this category in this way produces religiously more homogeneous collections of colleges. Whereas the twenty-­four colleges in the combined category displayed calculated homogeneity indices quite close to the national average, two of the three subgroups—­for Catholic colleges and evangelical Protestant ones—­show up as much less diverse than average. In 1972 the subgroup of six Catholic colleges and that of five evangelical Protestant colleges showed high degrees of religious homogeneity. For the six Catholic colleges, however, mea­sured religious homogeneity declined over the four de­cades, owing to increasing numbers of non-­Catholics who enrolled in them. Their Catholic share decreased, from 78 ­percent in 1972 to 56 ­percent in 2008 / 09.10 Close ­behind the Catholic colleges in 1972 in mea­sured homogeneity ­were the five evangelical Protestant ones. For ­these colleges, mea­ sured homogeneity went down in 1989 / 90, only to return in 2008 / 09 to virtually the same index value observed for 1972.11 We are left with two conclusions about religious homogeneity across the vast array of college types, subject to the caveat that such conclusions

S o r t i n g b y B e l i e f ?

289

are based on a relatively small sample of colleges. First, ­these half dozen Catholic colleges became over time more religiously diverse. Second, the HBCUs ­were some of the least religiously diverse colleges as a group, owing to their large share of African Americans, and the strong tendency of African Americans to be Protestants. Another, equally telling religious indicator is observance, as distinct from religious background. Students w ­ ere asked if they had attended any religious ser­vice in the previous year. To be sure, this question may appear to set a rather low bar for religious observance, but it does serve as a clear demarcation. Another advantage of using this criterion is that, like other mea­sures used in this chapter, information on it is available in all three survey waves. For all college students in the nation, this percentage fluctuated over the four de­ cades, rising from 76 ­percent in 1972 to 84 ­percent in 1989 / 90, before returning to near its 1972 level in 2008 / 09, to 77  ­percent. Figure 11.3 shows how the students at colleges in each category in each survey wave differed from that year’s national average, highlighting the considerable differences in this reported be­hav­ior by category. In 1972 first-­year students in private colleges at the 60th to 70th percentile w ­ ere most likely to have attended a religious ser­ vice; some 82 ­percent reported that they had done so. Least likely to have done so ­were students enrolled at the top-­ranked private colleges, with only 64  ­percent of ­these students having done so. By 2008 / 09 the colleges whose students w ­ ere most likely to have attended a ser­vice ­were the public HBCUs, and the gap between their propensity and that of students at private colleges in the 98th to 99th percentiles was a stunning 27  percentage points: 92  ­percent versus 65 ­percent. Although a few trends are evident in this mea­ sure, therefore, it is the differences across the categories that are most noteworthy. Students at the most selective private colleges reported rates consistently 10  percentage points below each wave’s national average. In fact, the figure suggests a generally negative relationship between the average SAT score of a college and the percentage of its students who reported attending a religious ser­vice in their last year of high school.12

290 D emand

0.15 0.10 0.05 0.00 −0.05

Pub

0.00

0.02

0.09

0.08

0.15

0.11

HBCU −0.08 −0.05 −0.06

−0.05 −0.01 −0.06

Priv

90+

80–90

0–50 0.03 0.01

0.03 0.01

0.02

0.00

0.02 0.01

−0.02 −0.02 −0.02

0.02

0.05 0.07

−0.11 −0.10 −0.14

0.06

0.03 0.03

−0.13 −0.11 −0.13

0.01

0.01 0.01

−0.10 −0.07 −0.08

0.00

0.02

2008/09

Public −0.04 −0.04 −0.05

0.04

1989/90

−0.04 −0.02 −0.01

1972

0.03

Private

50–80

99+

98–99

95–98

90–95

80–85

85–90

70–80

60–70

50–60

−0.15

25–50

−0.10

0–25

Average for college category minus U.S. average

0.20

Figure 11.3. ​Attended a Religious Ser­v ice in the Past Year, by Category and Survey Wave Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The figure summarizes the percentage, among first-­year students who reported that, in the previous year, they had attended a religious ser­v ice. (The question was, “Indicate which activities you did during the past year.”) The enrollment-­weighted means for all first-­year students in four-­year colleges w ­ ere 75.8 ­percent for 1972, 83.9 ­percent for 1989 / 90, and 77.4 ­percent for 2008 / 09.

Turning to the three detailed categories, shown in Figure 11.4, reveals that in terms of this mea­sure students at the evangelical Protestant colleges in the 25th to 50th percentiles w ­ ere unsurpassed in their rates of religious observance. In 1972 they led all other groups, with an 86  ­percent rate of attendance, and in 2008 / 09 they ­were virtually tied with students at the public HBCUs. Students at the corresponding Catholic colleges also reported above-­average attendance, but less than t­ hose at the evangelical colleges. Taken together, Figures 11.3 and 11.4 suggest that religiosity of entering college students, mea­sured using this modest indicator, did differ markedly by

S o r t i n g b y B e l i e f ?

291

Average for detailed college category m inus U.S. average

0.15 0.10 0.05 0.00 −0.05 −0.10 1972 1989/90 2008/09

Private 25–50, Catholic 0.07 0.04 0.06

Private 25–50, Evangelical 0.10 0.06 0.14

Private 25–50, Other −0.06 −0.03 −0.05

Figure 11.4. ​Attended a Religious Ser­v ice in the Past Year, Private Colleges in 25th to 50th SAT Percentile, Religious Affiliation Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year ­students in four-­year colleges. The figure summarizes the percentage, among first-­year students who reported that, in the previous year, they had attended a religious ser­v ice. (The question was, “Indicate which activities you did during the past year.”) The enrollment-­weighted means for all first-­year students in four-­year colleges ­were 75.8 ­p ercent for 1972, 83.9 ­p ercent for 1989 / 90, and 77.4 ­p ercent for 2008 / 09.

college category. ­There ­were no discernible trends over the four de­ cades in t­ hese differences. As a dimension of demand in higher education, religion is rarely an aspect to which scholars devote any attention. Perhaps that omission reflects the extreme secularism of the acad­emy. What­ever the reason, religion is included h ­ ere as a noteworthy aspect of demand in this market. Thanks to the nearly four-­decade span covered by the Freshman Survey data, it is easy to observe wide differences between categories of colleges. Especially noteworthy is the religious heterogeneity and the presence of Jewish students in the most selective private colleges. Students at ­these colleges w ­ ere also least likely to be religiously observant. In contrast is the increase in church atten-

292 D emand

dance between 1989 / 90 and 2008 / 09 (relative to the national mean) for the HBCUs and evangelical Protestant colleges. Po­liti­cal Views

From its beginnings in 1967 the Freshman Survey has faithfully included one generic question asking students about their po­liti­cal leanings: “How would you characterize your po­liti­cal views?” The survey offered students in e­ very survey wave the same five pos­si­ble answers: far right, conservative, middle-­of-­t he-­road, liberal, and far left. Although ­these terms surely could mean dif­fer­ent t­ hings to dif­fer­ent generations of students, the categories themselves never changed. As one all-­purpose index of po­liti­cal leaning, we h ­ ere use the share of students who chose e­ ither far right or conservative. Summarizing the survey responses using the usual treatment provides an unparalleled look at how incoming college students have differed po­liti­cally across consistently defined categories of colleges and over nearly four de­cades. Figure 11.5 summarizes the findings in terms of the standard graph, with percentages shown as deviations from the weighted average for all college students. That national average of the share who said they w ­ ere far right or conservative, by the way, increased from 16 ­percent in 1972 to 24 ­percent in 1989 / 90 and then remained unchanged in 2008 / 09. Two main conclusions emerge from Figure  11.5. First, students who described themselves as conservative using e­ither label attended colleges with lower average SATs than t­ hose who did not so identify. Except for the highest SAT group in the public colleges and the HBCUs, this generalization stands. The explanation for the first exception, for the 90th and above percentile public colleges, no doubt lies in the fact that three military academies make up half the six colleges included in the Freshman Survey sample, and t­ hese students would be expected to hold conservative views. As for the students in HBCUs, their liberal leaning is in sync with ­those of African Americans in general.

S o r t i n g b y B e l i e f ?

293

0.05 0.00 −0.05

90+

Priv

Pub

−0.12 −0.09 −0.06

0.01

80–90

0.05

−0.03 −0.06

0.10 0.03

0.00 0.01 −0.02

−0.05 −0.04 −0.04

0.03 −0.02 −0.01

−0.11 −0.05 −0.04

0.00 −0.03 −0.10

−0.13 −0.04 −0.04

0.03

0.00

0.04 0.05 −0.01

0.04

0.02 0.04 −0.02

0.03

0.03 0.03 0.02

−0.03

0.03 0.05 0.02

Public

−0.04

0.02

0.02

0.03

2008/09

0.01

1989/90

0.02

1972

0.00

Private

50–80

0–50

99+

98–99

90–95

95–98

85–90

80–85

70–80

60–70

25–50

−0.15

50–60

−0.10

0–25

Average for college category minus U.S. average

0.10

Figure 11.5. ​Po­l iti­cal View Is Conservative or Far Right, by Category and Survey Wave Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The figure summarizes responses to the question, “How would you characterize your po­l iti­cal views?” The pos­si­ble responses w ­ ere “Far right,” “Conservative,” “Middle-­of-­the-­road,” “Liberal,” and “Far left.” The figure shows the percentage who gave one of the first two answers. The enrollment-­weighted means for all first-­year students in four-­year colleges ­were 15.5 ­percent for 1972, 23.7 ­percent for 1989 / 90, and 24.2 ­percent for 2008 / 09.

The second conclusion arising from Figure 11.5 is that, between 1989 / 90 and 2008 / 09, the highest-SAT private colleges, t­ hose above the 95th percentile, took a sharp leftward turn, with their conservative share diverging markedly from the national average. The share of self-­described conservatives g­ oing to t­ hese colleges, already below the national average, fell much further b ­ ehind in the latter two de­ cades. In the 1989 / 90 survey wave, the share of students at colleges in the 98th percentile and above who reported conservative or far-­

294 D emand

right views was 4 percentage points below the national average, and ­those attending the 95th to 98th percentile colleges ­were at the national average. Twenty years ­later, the share of students at ­these same colleges (95th and above) describing themselves this way had fallen far below the national average, to 10  percentage points or more below the national average. Another category of colleges whose students had become much less likely to be self-­reported conservatives w ­ ere ­those at the private HBCUs. Meanwhile the views of students attending less selective public colleges and universities like Illinois State (­those that had been in the bottom half of the SAT distribution in 1970) w ­ ere more likely than average, by 3 percentage points, to report being conservative or far right. ­These leftward shifts are all the more noteworthy when compared to the increases in conservative sentiment that occurred in our five evangelical Protestant colleges in the 25th to 50th percentile, shown in Figure 11.6. This figure shows the percentage of conservative students for the three subsets of the private 25th to 50th percentile category. Already more conservative than college students in general in 1989 / 90, students at ­these colleges became markedly more conservative, on average, in the succeeding two de­c ades, with such self-­ described conservatives growing from 5 percentage points above the national average to 18  percentage points above it by 2008 / 09. In just two de­cades the gap in the conservative share between students attending ­those evangelical colleges and the students at the top-­ ranked private colleges increased spectacularly, from 9 percentage points in 1989 / 90 to 30 points in 2008 / 09. Perhaps this increase in segregation by po­liti­cal leaning ­ought not be terribly surprising, given the existence of colleges publicly aligned with conservative po­ liti­cal organ­izations, such as Liberty, Oral Roberts, and Regent. ­Here is one suggestion that the echo chamber effect decried in the media might also be playing out in the form of increased sorting by po­ liti­cal belief across colleges. Notably, the prominence of po­liti­cally conservative colleges stands in contrast with the absence of corresponding left-­leaning colleges. This asymmetry reflects not an absence of liberal sentiment, most

S o r t i n g b y B e l i e f ?

295

Average for detailed college category minus U.S. average

0.20 0.15 0.10 0.05 0.00 −0.05 −0.10 1972 1989/90 2008/09

Private 25–50, Catholic 0.00 0.03 0.02

Private 25–50, Evangelical 0.05 0.05 0.18

Private 25–50, Other 0.01 −0.03 −0.05

Figure 11.6. ​Po­l iti­cal View Is Conservative or Far Right, Private Colleges in 25th to 50th SAT Percentile, Religious Affiliation Source: Freshman Survey, students from 188 colleges, three waves (1972, 1989 or 1990, and 2008 or 2009). Note: Bars show deviations from enrollment-­weighted mean for all first-­year students in four-­year colleges. The figure summarizes responses to the question, “How would you characterize your po­l iti­cal views?” The pos­si­ble responses w ­ ere “Far right,” “Conservative,” “Middle-­of-­the-­road,” “Liberal,” and “Far left.” The figure shows the percentage who gave one of the first two answers. The enrollment-­weighted means for all first-­year students in four-­year colleges ­were 15.5 ­percent for 1972, 23.7 ­percent for 1989 / 90, and 24.2 ­percent for 2008 / 09.

likely, but rather its pervasiveness in academia, especially among the faculty who teach in the selective colleges and research universities.13 Despite the liberal leanings of most faculty, t­ here remains diversity among students, though that diversity may have weakened in the last two de­cades. On net, colleges prob­a bly contribute modestly to the larger tendency of Americans to insulate themselves from contrary po­liti­cal viewpoints. Colleges and Democracy

Without meaning to, the market for baccalaureate education plays a role in shaping and sustaining American democracy. T ­ here are three ways in which this market may exert po­liti­cal influence. 296 D emand

The Sorting Effect

Self-­segregation by po­l iti­cal leaning has prob­ably been a feature of modern democracies for de­ cades if not centuries. Most of the readers of this book know from their own experience how common it is for Demo­crats to socialize with other Demo­crats, and Republicans with other Republicans. In his book The Big Sort, Bill Bishop pres­ents evidence that this tendency has been intensifying in the United States. ­There is evidence as well that Americans have been sorting themselves geo­graph­i­cally, increasingly choosing to live in communities of like-­minded p ­ eople.14 Meanwhile, counties and other local areas have become increasingly identified with one party or the other.15 Among the forces contributing to this geographic segregation has been cable tele­vi­sion and the echo-­chamber effect that has been associated with it.16 Accompanying this sorting are unmistakable signs that the gulf in po­liti­cal dialog between right and left has widened appreciably, resulting in legislative stalemate and a presidential campaign in 2016 more bitterly contested than any election since 1860. Quite apart from their descriptive value, the answers to t­hese questions are impor­tant to the extent one believes that diversity of background and viewpoint in a student body might be a beneficial component to a college education. Not only might religious diversity be a valuable attribute, so might po­liti­cal diversity. In this age of cable tele­vi­sion and the multiplicity of sources of news and opinion, many observers have worried that po­liti­cal debate has given way to dozens of separate echo chambers, where no one need confront a contrary viewpoint. One suspects that developments such as ­these have also affected colleges and college students. Certainly t­ here is evidence of it in the figures showing po­liti­cal attitudes by college category, Figures 11.5 and 11.6. Students at selective private colleges have been less conservative than average, and they tended to move left over time. ­Going in the other direction w ­ ere students at evangelical Christian colleges. To be sure, t­ hese conclusions are based on aggregated data on students at a limited number of colleges, grouped into categories or S o r t i n g b y B e l i e f ?

297

subcategories of colleges. To document changes in po­liti­cal sorting fully would require more granular information than examined in this book, such as data that details the friendship networks of individual students. The pres­ent findings are suggestive, however. Colleges appear to have become more po­liti­cally differentiated over time. While diversity of other kinds has advanced over time, this one has not. Pluralism

The second po­liti­cal consequence vis­i­ble in the college market is its contribution to the honored attribute of pluralism. Long celebrated by social commentators, from Alexis de Tocqueville to con­temporary pundits, pluralism is often associated with the country’s nonprofit sector. De Tocqueville famously remarked on the number and variety of nongovernmental organ­izations he observed during his famous visit to the United States: “Americans of all ages, all conditions, all minds constantly unite. Not only do they have commercial and industrial associations in which all take part, but they also have a thousand other kinds: religious, moral, grave, futile, very general and very par­tic­u ­lar, im­mense and very small.”17 Protected by the First Amendment and subsidized by favorable tax provisions, nonprofit organ­izations—­service providers, advocacy groups, religious congregations, and iconic cultural institutions—­embody the Found­ers’ intention to encourage variety in the manner that citizens or­ga­nize to deal with social needs. The market for college is a surprisingly robust exemplar of pluralism, although it is seldom recognized as such. In the vast nonprofit sector, only religion and advocacy offer a wider variety of orga­ nizational aims or foundational doctrines. This variety is richly illustrated by the reviews of college aims published in course cata­logs and con­temporary websites, such as ­those described in Chapters 2 and 7. Despite the remarkable similarity in some aspects of their organ­ ization and modes of operating, such as the form of their curricula, the organ­ization of their instructional staff into departments by

298 D emand

discipline, and the requirements they have established for graduation, colleges pres­ent striking contrasts in other aspects, especially with re­spect to religious doctrine. Among the colleges described ­here are ones that have held fast to their original religious princi­ples, as out of step as ­those might be with con­temporary urban secular culture. The gap is especially evident for the evangelical Protestant colleges, who proclaim a learning environment founded on a collective expression of Christian teaching and faith. Not surprisingly, students appear to sort themselves in ways that reflect such official doctrines, but the correspondence is far from perfect. What is germane for pluralism is that, through its diversity, the college industry itself maintains and protects vastly divergent outlooks on the world of ideas. ­Because ­these contrasting world views so often correlate with po­l iti­cal affiliation, colleges, by way of the durability of their identities, offer a safe haven for families and students who do not care to enter the mainstream environment, e­ ither intellectually or po­liti­cally. Democracy’s Classroom

A third way in which colleges as a ­whole have consequences for politics and democracy arises out of the learning itself. From the classroom to the late-­night bull session, the college experience shapes students’ ­future roles as citizens and leaders, just as surely as they shape their roles as workers and parents. Some observers suspect that college has a radicalizing effect on some students, as it appeared to in the days of Kent State, while ­others have noted the extent to which colleges are deeply embedded in the status quo. The liberal bent of many professors has been both noted and documented, but how much of that rubs off on students is uncertain. In any case, the research that has sought to determine the effect of college on po­ liti­cal attitudes has failed to produce any definite conclusions.18 More impor­tant than partisan tilt may be the skills and expectations that students carry with them a­ fter graduation. As to the impact of coursework on po­liti­cal attitudes, t­ here is considerable

S o r t i n g b y B e l i e f ?

299

research but few definitive conclusions. A less obvious effect is on the skills that college students develop as a result of the scores of meetings many of them attend or convene as members of student organ­ izations. In her book Diminished Democracy Theda Skocpol argues that, with the steady decline and disappearance of civic organ­izations like the Masons, Elks, and Odd Fellows, Americans have lost what amounted to a vast classroom where citizens, in their roles as members, could hone their skills in demo­cratic participation and leadership.19 For college students, the hiking club, the major speakers committee, the sororities and fraternities, and student government, not to mention the Young Republicans and Young Demo­crats, all provide members with kinds of learning opportunities similar that Skocpol extols in the civic organ­i zations of the past. Not only are the skills that students acquire as members of student organ­ izations useful in their f­uture employment in workplaces built around teamwork, therefore, ­those skills ­will also be valuable when they, as ­future college gradu­ates, enter the po­liti­cal realm as citizens and leaders. ­Because of the growing economic stratification of colleges, this training for demo­cratic involvement is necessarily occurring in settings that are more and more separate and unequal. For the most privileged, this advanced civics education is taking place in the com­ pany of other, similarly well-­off young ­people, and in increasingly opulent surroundings. To be sure, ­there are ser­vice proj­ects at soup kitchens, low-­income schools, and occasionally developing countries, all of which may serve as antidotes to some of the effects of affluence. At the other end of the prestige range, students of humbler origins learn similar lessons in civic engagement, though the unstated assumptions for them about the efficacy of civic engagement ­will prob­ably be dif­fer­ent from ­those made by their counter­parts at the elite colleges. In historically black colleges the civics education must surely be distinctive from e­ ither of t­ hese. Through the 1960s t­ hese colleges educated a large share of African American leaders, but HBCUs seem markedly less central in this regard than they once ­were.

300 D emand

Aside from their in­equality in resources and student aptitude, colleges also differ from each other ­because their clienteles differ in ways not mea­sured by test scores or ­family socioeconomic status. Religion is one of t­ hose. Po­liti­cal orientation is another. Successive waves of data from the Freshman Survey reveal wide variation across college categories in both of ­these, as well as several trends. The differences across colleges serve pluralism, although t­ hose differences also work against diversity. It seems likely that colleges as a w ­ hole exert an influence on politics and governance that deserves greater scrutiny.

S o r t i n g b y B e l i e f ?

301

P A R T IV

Consequences

12 Outcomes

If colleges are the firms in this industry, students and their parents the consumers, and tuition net of scholarships the price of enrolling, what is the commodity itself—­the ­thing being produced, sold, and purchased? And, since some colleges seem better at producing it than ­others, how does its quality vary? Up to now, we have devoted ­little attention to this rather basic aspect of the market for baccalaureate education, not ­because the topic is unimportant, but ­because it is so elusive. Although scholars can describe aspects of the output of the educational production pro­cess, mea­sur­ing it convincingly is nearly inconceivable. One could as easily mea­sure the output of an art museum or a f­amily. This difficulty has not prevented some empirically minded researchers from trying, however. Chief among the mea­sures they have turned to are graduation rates and the earnings of alumni, but such mea­sures are deeply flawed, as t­ hose who use them ­will usually admit.

Thus far we have gone about the business of describing the market for baccalaureate education without paying much attention to outcomes. The story of in­equality in the college market is indeed largely one of inputs and of the matching of students to par­tic­u­lar kinds of colleges within the wider industry. But this chapter turns to the daunting topic of outcomes, first by reviewing the kinds of outcomes that scholars have associated with college attendance in general. The explanation, perhaps a defensive one, for the paucity of empirical research that can be used to mea­sure the outcomes of college, especially in a study that is so concerned as this is, is with differences across colleges. We then look at several unorthodox mea­sures, created to be consistent with the approach taken throughout the book. That is, they both reflect differences across colleges as well as changes over time, retaining their meaning so as to allow for apples-­to-­apples comparisons. The mea­sures used are far from perfect, but they are unlike what can be found in most previous empirical work on higher education in the United States. Next, two normative criteria for judging the college market as a ­whole are explained and applied. The first such standard is w ­ hether colleges make it pos­si­ble for students to rise above modest origins. The second is ­whether the in­equality we observe can be justified on the basis of economic efficiency. In this context we turn to research that seeks to uncover empirically the causal effects of college as well as evidence that ­these effects differ across colleges. The Effects of ­Going to College

“Teach ­these boys and girls nothing but Facts. Facts alone are wanted in life. Plant nothing ­else, and root out every­thing ­else. You can only form the mind of reasoning animals upon Facts: nothing ­else ­will ever be of any ser­vice to them.” So stated Charles Dickens’ schoolmaster and spokesman for factually based education, Thomas Gradgrind.1 Of the vari­ous outcomes of college, certainly the acquisition of factual knowledge is the most obvious, if not necessarily the most impor­tant. Scholars who have studied the effects of college on students agree that the effects are both numerous and difficult to

306 C onsequences

mea­sure. In addition to acquiring factual knowledge, students learn how to acquire new knowledge, evaluate evidence, and make judgments. Some of their learning occurs in classrooms and labs, some of it comes from studying, and some of it arises through interaction with the p ­ eople they meet at college.2 In all this, subject m ­ atter learning is easiest to mea­sure. ­There is ­little evidence, however, that such gains differ by the characteristics of colleges, including selectivity.3 As impor­tant as factual knowledge may be in earlier grades, the experts who write about college place more importance on generalizable skills, such as the ability to discern relevant facts and assess the relative importance of values as well as facts. One test designed to assess such generalizable skills is the Collegiate Learning Assessment, which aims to mea­sure critical thinking and complex reasoning as well as skill in writing. One study based on student per­for­mance on that test, Arum and Roksa’s Academically Adrift, found that scores for most students did not rise very much during college, suggesting that not much learning occurred. A major analytical challenge in assessing the effects of college on general cognitive skills is what authors of one review of the lit­er­a­ture as the “confounding effects of maturation.”4 As with gains in knowledge, ­there is scant evidence of differences in learning across colleges, with one consistent exception: students at colleges with “strong and balanced” general education curricula tend to demonstrate bigger gains in critical thinking. 5 In addition, some research suggests that selective liberal arts colleges tend to test students’ higher-­order cognitive skills more than less selective ones do. Other research indicates that institutions with fraternities and sororities “inhibit growth in critical thinking.”6 Despite ­these few pieces of evidence, solid findings on cognitive learning are hard to come by, largely b ­ ecause of the difficulties in assessing learning and the methodological challenges in identifying the separate contribution of college to a person’s intellectual growth.7 Owing to t­ hese difficulties, scholars of higher education often content themselves with easily mea­sured, rough approximations, such as degree attainment. Indeed, college graduation rates have captured the attention of scholars and policy makers alike.

Ou t c o m e s

307

One recent study has noted that, although rates of college attendance has increased impressively, the rate of completion for enrolled students has declined. Comparing the cohort of Americans who graduated from high school in 1972 to t­ hose who graduated in 1992, the percentage who ever attended college r­ose from 48  ­percent to 71 ­percent.8 But the percentage of t­ hose enrollees who finished college (within eight years) fell from 51 ­percent to 46 ­percent.9 The authors lay most of the blame for this falling rate of completion on colleges, not students. In par­tic­u­lar, they point to the two-­year and less selective four-­year colleges, where many of the students in the 1992 cohort enrolled. They estimate that three-­fourths of the decline in college completion rates is due to the types of institutions where students have disproportionally enrolled. ­These are the same groups of colleges that have been falling ­behind in the unequal competition for resources described in Chapter 5. In addition to cognitive and analytical skills, psychologists have stressed vari­ous aspects of personal growth that occur during college. Friendships and romantic relationships as well as other, less personal connections all affect students’ self-­esteem, attitudes, and values. As impor­tant as t­hese aspects are, t­here exist few studies that demonstrate that college, apart from the maturation pro­cess itself, has any consistent effects on what is called psychosocial development, including identity, self-­esteem, social relationships, attitudes, and values. In par­tic­u­lar, researchers have uncovered no consistent differences between colleges, or types of colleges, in ­these impor­tant aspects of personal development.10 ­There is also evidence suggesting that college may have long-­lasting effects on a wide variety of nonpecuniary but mea­sur­able aspects of the quality of life. It turns out that t­ here is a rather long list of quality-­ of-­life indicators that are statistically associated with years of education.11 For example, having “highbrow” tastes in ­music is strongly associated with high levels of education.12 To be sure, the evidence is based on correlation, which means that college itself may not be the cause of the statistical association. Rather, it may be that personal characteristics other than years of education which are correlated

308 C onsequences

with education might instead explain the differences. One characteristic that might have this effect is income. The findings described below eliminate the potentially confounding effects of income by statistically controlling for it. Thus the comparisons can be viewed as between individuals with the same income but dif­fer­ent amounts of education. Perhaps the most comprehensive mea­sure of quality of life is happiness, as expressed on surveys asking p ­ eople how happy they are. T ­ hose who had four years of college had the highest scores, on average, even when income was controlled for, which itself is associated with greater happiness. Aside from the higher income that might result from college, gradu­ates also have more prestigious jobs and report higher job satisfaction than ­those with similar incomes who completed less schooling. In addition, college gradu­ates are less likely to be unemployed. Gradu­ates report better health and lower rates of smoking. They have lower rates of divorce and separation than do individuals with less education, again holding income constant. A college education is also associated with distinctive attitudes. ­Those with four years of college are less likely than ­others to f­ avor spanking as a means of discipline. They are more likely to believe that p ­ eople can be trusted. They are more likely than o ­ thers to be able to delay gratification. And they are less likely than o ­ thers to have been arrested.13 Some outcomes of college work through peer effects, and recent research does provide some evidence of causal effects. Students’ membership in fraternities and consumption of beer, for example, have been shown to be influenced by the be­hav­ior of roommates or other students in a student’s dormitory.14 ­There is evidence as well of the effect of interracial contact on race-­related attitudes. Students who, through random room assignment, had close exposure to a student of another race w ­ ere more likely in the following year to choose to room with such a student.15 Although t­ here is no evidence at this point that such effects on preferences are long-­lasting, ­these studies establish a link that is not merely correlational. It is one ­thing to list the pos­si­ble outcomes of college. It is quite another t­hing to mea­sure them. The difficulty of mea­sure­ment

Ou t c o m e s

309

arises, most of all, from the multiplicity of ­things colleges do or, rather, the multiplicity of ways the experience of g­ oing to college can affect students. To an extent not matched at the K-12 level, education at the college level does many t­ hings. Even putting aside research and ser­vice, the kinds of topics covered in college vary widely, from the practical and job-­related to the theoretical and otherworldly, as Chapter  2 illustrates. This diversity in subject ­matter does not capture the full range of ­things that are taught and learned in college, of course. College students are expected to grow in their ability to discern relevant questions to pursue, analyze prob­lems, and make cogent arguments, and, as we have seen, how much growth occurs ­will depend on a student’s own choices and effort. The complexity of t­hese vari­ous objectives pres­ents a daunting challenge to anyone wishing to mea­sure the outcomes of college. In devising proxies for learning, scholars have used three kinds of outcome mea­sures: graduation or progression rates, direct tests of learning, and ­labor market outcomes. Of ­these, graduation rates are prob­ably the easiest to implement and defend, but this is not without its prob­lems. The biggest drawback is that degree attainment says not the first ­thing about content or quality. Using it as an outcome mea­sure in the policy context also invites unintended effects in the form of watering down requirements and quality in order to push more students out the door with a degree in hand. ­There are other practical prob­lems with this mea­sure as well, such as penalizing institutions with many part-­t ime students or ­those whose students transfer to other colleges to complete their degrees. L ­ abor market outcomes have the virtue of letting employers assess the quality of instruction with their paychecks, but ­these mea­sures have prob­lems of their own. They are available only for t­ hose who enter the l­abor force, they have ­little meaning in the first few years ­after graduation, and they completely miss many impor­tant dimensions of education that have no economic payoff. Direct mea­sures of learning are at once most attractive and most problematic, requiring ­great expense and differentiation if done correctly.

310 C onsequences

The outcome mea­sure favored by economists, not surprisingly, is earnings. Standard microeconomic theory posits that in well-­ functioning ­labor markets workers ­will be paid in proportion to their productivity. If education imparts skills, then workers with more education should be paid more. Indeed, t­ here are few statistical regularities more solid than the substantial earnings premium enjoyed by college gradu­ates, compared to ­those with less education. In 2012, for example, among men with full-­time year-­round employment, t­ hose with at least four years of college earned an average of $34,969 more than ­those who had just a high school diploma. For ­women, the comparable earnings advantage was $23,280.16 In the economic model, college-­educated workers earn more than high school gradu­ates ­because employers’ demand for college gradu­ates is strong and ­because such workers are relatively scarce. Over the last three de­cades, this earnings advantage for college gradu­ates has grown, as discussed in Chapter 4. The reasons for this growth are the subject of ongoing debate, but many economists would place considerable emphasis on profound changes that have occurred in the American economy. Knowledge-­intensive industries have grown at the expense of manufacturing, and all industries have seen increases in the need for educated workers. Despite an increasing supply of college gradu­ates, the demand for them has grown even faster. This growth in demand has been fueled by major industrial shifts, as manufacturing has given way to business ser­vices and education, and by the growing demand within each industry for skilled workers able to use computers and deal with increasingly complex job demands.17 The consequence of t­ hese changes has been a doubling of the college earnings advantage for men, which increased from $17,411 (expressed in 2012 dollars) in 1979 to the $34,969 figure in 2012.18 The existence of this earnings advantage is not the end of the story, however. Two questions in par­tic­u­lar arise. First, what portion of this difference can be attributed to attending college? Second, what explains that portion? To answer the first question, researchers must somehow remove the influence of unmea­sured personal characteristics that might cause college-­goers to differ systematically

Ou t c o m e s

311

from ­those who do not go, which is exactly the same prob­lem that arises in explaining the differences in the quality-­of-­life mea­sures discussed above. Such se­lection bias would imply that differences in average earnings overstate the true effect of g­ oing to college, and overcoming this bias has proven to be a formidable challenge for researchers. We return to se­lection bias below. Granted that at least some of the observed differences in earnings associated with postsecondary training is due to the amount and quality of that training and assuming that wages reflect differences in workers’ productive value, what is it about college that is responsible for the statistical association between productive value and postsecondary training? Economists offer two answers. The first, and the one regarded by the majority of scholars to be the principal function, is the transformation of individuals through learning, or ­human capital investment. Postsecondary education, according to this view, like its K-12 cousin, enhances students’ productive worth. Some of this increased ­human capital has payoffs in the ­labor market. A second, alternative view is that higher education works mainly by identifying able individuals rather than transforming them. This view focuses on the information contained in simply possessing a college degree. Like the h ­ uman capital view, this explanation accepts the essential accuracy of the neoclassical economics model of competitive ­labor markets and its implication that more productive workers ­w ill be more valuable to employers, holding constant their supply. Where the signaling explanation diverges from the h ­ uman capital view comes down to what, exactly, the contribution of college is. In the signaling view of the world, most of the skills or attributes that ­will be valuable to employers have already been baked in by the time students have finished high school. All that colleges have to do is to identify the young ­people most richly endowed with ­those abilities. In this signaling explanation, that is the primary function of college—to identify and certify talent. Students who gain admission and manage to persevere to graduation day can then march boldly into the ­labor market armed with a certificate of talent. In this explanation, any ­actual learning that occurs is incidental to, not the

312 C onsequences

result of, college. The role of college in this world then is to certify skills, not instill them. From society’s perspective, it m ­ atters a ­great deal which one of t­ hese two views is the more accurate, for if the signaling view is right, then the w ­ hole college enterprise is l­ittle more than a terribly wasteful way of certifying talent. Few observers think this view is an accurate portrayal of real­ity, however. Changing Patterns of Some College Outcomes

Not only does the “output” of college education resist mea­sure­ment, most mea­sures that can be thought of are not available for individual colleges. Nor are they available far enough into the past to cover the four-­decade period of the pres­ent study. Nevertheless, in keeping with the unapologetic approach taken in this book of looking ­under the lamppost for mea­sures that can be compared across colleges and over time, adopting a set of mea­sures that lend themselves to the study at hand requires linking all mea­sures used to par­tic­u­lar colleges. The outcome mea­sures used ­here are based on three types of achievements of new college gradu­ates: earning a PhD, being admitted to a highly selective law school, and receiving a prestigious postgraduate scholarship. Although mea­sures such as ­these have rarely been used to assess college outcomes, their use is not entirely without pre­ce­dent. A study published in 1969 sought to determine which American colleges and universities produced leaders through American history, and it focused almost entirely on a small number of elite institutions, almost invariably beginning with Harvard. Gradu­ates of that college, for example, accounted for more signers of the Declaration of In­de­pen­dence than any other (eight out of fifty-­six) and more Rhodes Scholars than any other.19 Similarly, in numbers of gradu­ates being listed in Who’s Who, Harvard led all colleges in each of four sample years from 1928 to 1962. Summarizing all the mea­ sures it used, the study concluded that nine colleges and universities led all o ­ thers in having produced distinguished alumni. In order, they are Harvard, Yale, Prince­ton, Columbia, Michigan, Pennsylvania, California, Chicago, and Cornell. None of t­ hese names is a

Ou t c o m e s

313

surprise, especially considering the large number of gradu­ates from the two public universities. Nor is it surprising that most of the attention of this study was devoted to elite, selective institutions. Another study that examined the postgraduation achievements of college gradu­ates did so to determine the most frequently followed paths to leadership positions in business. The most common paths ­were by way of a bachelor’s degree at a top-­ranked undergraduate college, MBA program, or law school, with an upper-­class f­ amily background being a plus no ­matter what college or gradu­ate school a person attended. Law turned out to be an impor­tant alternative to business school as a path to business leadership.20 By tracing the undergraduate colleges of individuals a­ fter they gradu­ate, the approach ­here is similar to the ones taken in ­these previous studies. The calculations differ from ­these studies, however, in that they keep track of all individuals, not just ­those who graduated from elite colleges. The first mea­sure used is based on surveys of ­those who receive PhDs from American universities. The National Science Foundation’s annual Survey of Earned Doctorates contains information on each person who earned a PhD at an American university, including the recipient’s undergraduate college. The survey makes it pos­si­ble, therefore, to count the number of students from each college who went on to earn PhDs, and to do this by field. ­Because the survey has been conducted for many years, it is pos­si­ble to gather this information for periods at the beginning, m ­ iddle, and end of our four-­decade period. In an effort to make the findings more representative, ­these are calculated as three-­year averages. A second outcome mea­sure is admission to one of three prestigious and highly selective law schools in the country.21 H ­ ere again, the aim in selecting this indicator is to be able to link students from par­tic­ u­ lar colleges to desirable postgraduation outcomes. Unpublished data from three law schools for periods near the beginning and end of the four de­cades are used, as well as for at least three years’ entering classes in each of t­ hose periods. The three law schools w ­ ere Harvard, New York University, and Duke, all of them highly ranked schools.22 In the early 1970s, one published ranking put Harvard on top and

314 C onsequences

NYU in the eighth spot.23 Another ranking about the same time had Harvard tied for seventh, Duke tied for fifteenth, and NYU out of the top twenty.24 In the 2015 U.S. News ranking of law schools, Harvard was ranked second, NYU sixth, and Duke eighth.25 The third outcome used is considerably less common than obtaining a PhD or gaining admission to one of three prominent law schools. It is receipt of one of three very prestigious national scholarships—the Rhodes, Marshall, and Truman Scholarships. Of ­these, the Rhodes is the oldest and most famous, having been established in 1903 and counting among its recipients many p ­ eople who would l­ater rise to national prominence. About twenty Americans win this scholarship each year. The Marshall and Truman Scholarships, established by Congress in 1953 and 1977, respectively, are also quite prestigious and nearly as rare. ­Because ­these scholarships provide information on the undergraduate colleges of past winners, it is pos­si­ble to use ­these awards in the same way as for the previous two markers of postgraduation achievement. Owing to the small numbers of individuals involved, it is feasible to track this mea­sure annually. In order to smooth the series, they are presented h ­ ere in three-­year moving averages. It is worth asking how well t­ hese three mea­sures meet the apples-­ to-­apples requirement—­that is, ­whether the meaning of the indicator remains the same over time. All three mea­sures entail a se­lection pro­cess, and it does seem quite pos­si­ble that the criteria for se­lection might have changed over time. The PhD involves as well the students’ own per­for­mance and perseverance in a wide variety of university degree programs. ­There seems to be ­little reason to believe that the difficulty or significance of ­these programs have changed systematically over the time span being examined. In the pre­sen­ta­tions that follow, two aspects of ­these achievements are especially noteworthy. The first is the extreme concentration in a relatively small number of colleges, a concentration most evident, not surprisingly, with the highly coveted Rhodes and Marshall Scholarships. The second, and equally in­ter­est­ing, aspect is the change in the shares of college groups over time. The first of ­these two aspects is abundantly clear in Figure 12.1. The four bars show

Ou t c o m e s

315

100% 90% 80%

Proportion of students

70% 60% 50% 40% 30% 20% 10% 0%

Percent national enrollment, 2010

Average percent PhD recipients, 2009–11

Percent Harvard Law, classes of 2014–16

Percent Rhodes Scholars, 2010–13

Private colleges SAT: 98–100 percent Private colleges SAT: 95–98 percent Private colleges SAT: 90–95 percent Private colleges SAT: 50–90 percent Private colleges SAT: 0–50 percent and HBCUs Public colleges SAT: 90–100 percent Public colleges SAT: 50–90 percent

Public colleges SAT: 0–50 percent and HBCUs

Figure 12.1. ​The Disproportionate Success of Elite College Gradu­ates, Three Mea­sures Source: IPEDS; Survey of Earned Doctorates; Harvard Law School; Rhodes Scholarship. Note: Each bar shows the percentage distribution of students in specified group of colleges who graduated from each of the eight indicated clusters of colleges.

316 C onsequences

the shares of each quantity made up by each of eight categories of colleges. The clear message is the disproportional success of gradu­ates from the most elite private colleges and universities (private colleges whose 1970-­era SATs ­were in the 98th percentile and higher). Enrolling just 1.9 ­percent of all undergraduates in 2010, ­these colleges accounted for 11 ­percent of all PhDs awarded by American universities, 39  ­percent of the students at Harvard Law School, and a whopping 52  ­percent of the Rhodes Scholars awarded. No other group of colleges comes close to matching the success of this group. The next-highest group of private colleges, t­hose occupying the 95th to 98th percentiles, enrolling about the same share of students, is also overrepresented among each of the achievements, but the success for gradu­ates of this category of colleges is more muted, as they garnered 5 ­percent of the PhDs, 11 ­percent of the places in Harvard Law School, and 2  ­percent of the Rhodes Scholarships. The other college category that showed greater than proportionate success is public universities whose 1970-­era SATs ranked at the 90th percentile or higher. With total enrollment of 4 ­percent of all students, gradu­ates of ­these institutions accounted for 7 ­percent of all PhDs, 6 ­percent of Harvard Law students, and 12.5 ­percent of Rhodes Scholars. The two groups with the weakest rec­ords in producing recipients of t­hese achievements ­were the lowest-­ranking private and public institutions. Relative to their enrollments, they had roughly comparable rates of success, except for getting into Harvard Law, where the lowly private colleges did better than the comparable public ones. Doctorates

Of the outcome mea­sures examined h ­ ere, earning a PhD is the most commonly achieved one. In 2011 / 12 American universities awarded 170,000 doctorates, of which about 68,000 ­were PhDs. This total amounted to about 4 ­percent of the number of bachelor’s degrees awarded in that year.26 Since the mid-1970s this share has declined, as the academic job market in most fields suffered from an oversupply

Ou t c o m e s

317

of job-­seekers. What PhDs that ­were awarded went disproportionately to gradu­ates of the top-­ranked private colleges. It is easy to see the comparative success of students by college group over time by looking at each group’s share of all PhD recipients at the beginning, m ­ iddle, and end of the period. T ­ hese shares are shown in 27 Figure 12.2. The large enrollments of the public universities are reflected in their categories’ big proportions. Two noteworthy trends show up in t­ hese changing shares of all PhDs awarded. The first is the drop in the share of the private colleges in the 50th to 90th percentiles. Whereas this group accounted for 16 ­percent of all PhDs in the period from 1974 to 1976, their share dropped to 13 ­percent in the period from 1992 to 1994 and remained ­there in the most recent

0.30

Proportion of students

0.25 0.20 0.15 0.10 0.05 0.00

0–50

50–90

90–95

95–98

98–100

Private colleges 1974–1976

0–50

50–90

90–100

Public colleges 1992–1994

2009–2011

Figure 12.2. ​Share of All PhDs by College Group, Three-­Year Averages, 1974–1976, 1992–1994, and 2009–2011 Source: Survey of Earned Doctorates. https://­ncsesdata ​.­nsf​.­gov​/­webcaspar​/­a nd http://­w ww​.­norc​.­org ​/­R esearch​/­Projects​/­Pages​/­survey​-­of​-­earned​-­doctorates​-­%28sed%29​ .­a spx and http://­w ww​.­nsf​.g­ ov​/­statistics​/­s4094​/­. FICE and unitid codes ­were used to match recipients with their undergraduate college. Note: HBCUs combined with corresponding 0–50th percentile groups.

318 C onsequences

period. The share of the lowest-­ranked group of private colleges also fell slightly. The second trend is a rise over time in the share of PhDs accounted for by the highest-­ranked private and public colleges, from 10 ­percent to 11 ­percent for the top private colleges and from 5 ­percent to 7 ­percent for the top publics. In light of their disproportionate contribution to the ­future PhD workforce, it is illuminating to focus on shares for top-­ranked private and public schools. ­Table 12.1 compares the shares for the 90th and above percentile private and public colleges in 1972–1974 and 2007–2009 for fourteen fields. Although the institutions together enrolled only a tenth of all undergraduates, they accounted for much bigger percentages of PhDs, ranging in 1972–1974 from 26.6 ­percent in chemistry to 42.6 ­percent in physics. Over the entire period, the shares for top colleges fell in only four fields—­English, physics, engineering, and business and management. Among the increases, the biggest ­were in economics and biological sciences. In economics this increase was entirely due to the private elite colleges, whose increase more than compensated for a decline in the numbers from the public sector. As shown in Figure 12.3 the very top private colleges had an extraordinary increase, from 16 ­percent to 22 ­percent, in the share of gradu­ates ­going into economics. In sum, top colleges generally strengthened their position as suppliers of successful PhD candidates over ­these four de­cades. Top Law Schools

The dominance of top private colleges is considerably more pronounced in admission to three highly ranked law schools, especially Harvard. As shown in Figure  12.4, although private colleges in the 98th percentile and above enrolled only about 2 ­percent of all undergraduates, their gradu­ates accounted for more than half of Harvard Law School students in the classes of 1974 to 1977. By comparison, the reliance on ­these top private colleges was comparatively muted at NYU and Duke, whose shares of t­hese colleges’ gradu­ates was just 41  ­percent and 23  ­percent, respectively (see Figures  12.5 and 12.6).

Ou t c o m e s

319

27.2% 6.5%

Business and Management

Note: Share of total-­year college enrollment

21.1%

Chemistry 27.8%

36.0%

Physics

Engineering

33.1%

30.7%

Arts and M ­ usic

Mathe­matics and Statistics

29.6%

En­glish and Lit­er­a­t ure 20.7%

29.3%

Foreign Languages

Biological Sciences

30.1%

22.9%

Sociology

History

22.6%

Psy­chol­ogy

3.8%

10.9%

10.5%

5.5%

6.6%

5.9%

5.4%

5.0%

5.2%

5.7%

4.3%

5.4%

6.3%

5.3%

7.1%

6.6%

5.4%

Pub90+

5.9%

22.7%

23.6%

19.6%

30.3%

31.8%

24.7%

32.2%

25.9%

32.0%

30.0%

22.9%

21.9%

26.8%

37.3%

25.2%

22.4%

Priv90+

4.1%

12.4%

12.0%

7.1%

9.7%

7.1%

7.7%

4.5%

6.6%

5.7%

7.3%

6.4%

7.0%

7.9%

6.4%

8.6%

7.3%

Pub90+

2007–2009

0.7%

−0.7%

−2.3% 1.0%

1.4% −0.5%

−3.7%

1.5% 1.5% 1.5%

−4.5%

3.1%

−5.7% −1.4%

1.2%

−1.2%

−4.2%

2.3%

4.1%

−3.0%

−2.7%

0.1%

−2.6%

0.0%

6.4%

2.6%

−0.1% 2.7% 1.5%

3.0%

1.1%

0.0% 3.0%

0.0%

1.0%

2.6%

0.1%

9.2%

−0.7%

9.9% −2.0%

0.6%

1.0%

2.0%

−1.0%

3.3%

All 90+

1.9%

Pub90+

1.3%

Priv90+

Change

Source: Survey of Earned Doctorates. https://­ncsesdata​.n ­ sf​.­gov​/w ­ ebcaspar​/­and http://­w ww​.­norc​.o ­ rg​/R ­ esearch​/­Projects​/P ­ ages​/­survey​-­of​-­earned​-­doctorates​ -­%28sed%29​.­aspx; and http://­w ww​.­nsf​.­gov​/­statistics​/­s4094​/­. FICE and unitid codes w ­ ere used to match recipients with their undergraduate college. Note: HBCUs are combined with corresponding 0–50th percentile groups. Engineering is the sum of Aerospace Engineering, Chemical Engineering, Civil Engineering, Electrical Engineering, Industrial Engineering, Materials Engineering, Mechanical Engineering, and Other Engineering.

Other Disciplines

Natu­ral Sciences

Humanities

27.4% 28.8%

Social Sciences

Po­l iti­cal Science and Public Administration

26.2%

TOTAL (of selected disciplines)

Economics

21.1%

TOTAL (of all disciplines)

Priv90+

1972–1974

­Table 12.1  Share of PhDs with Bachelor Degrees from 90th + Percentile Public and Private Colleges

0.30

Proportion of students

0.25 0.20 0.15 0.10 0.05 0.00

0–50

50–90

90–95

95–98

98–100

1974–1976

0–50

50–90

90–100

Public colleges

Private colleges 1992–1994

2009–2011

Figure 12.3. ​Share of Economics PhDs, by College Group, Three-­Year Averages, 1974–1976, 1992–1994, and 2009–2011 Source: Survey of Earned Doctorates, https://­ncsesdata​.­nsf​.­gov​/­webcaspar​/­a nd http://­w ww​ .­norc​.­org ​/­R esearch​/­Projects​/­Pages​/­survey​-­of​-­earned​-­doctorates​-­%28sed%29​.­a spx and http://­w ww​.­nsf​.­gov​/s­ tatistics​/­s4094​/­. FICE and unitid codes ­were used to match recipients with their undergraduate college. Note: HBCUs are combined with corresponding 0–50th percentile groups.

Over the next thirty years, however, the dominance of the highest-­ ranked private colleges declined at all three law schools, precipitously so at Harvard, from 52 ­percent to 39 ­percent. One ­factor in this decline in the share of the 98th and above private colleges may have been the growing allure of Wall Street, combined with the near-­ calamitous effects of the ­Great Recession on the market for ­lawyers. Offsetting this declining share ­were consistent increases in the share of students from public universities above the average but excluding the top. Although it might be tempting to see ­these changes over time as an indicator that top private colleges have slipped, the trends might equally well be a reflection of relative shifts in the highly lucrative ­labor markets in business, finance, and consulting as they related to the turbulent market in law.

Ou t c o m e s

321

0.60

Proportion of students

0.50 0.40 0.30 0.20 0.10 0.00

0–50

50–90

90–95

95–98

98–100

0–50

Private colleges 1974–77

50–90

90–100

Public colleges 2014–16

Figure 12.4. ​Share of Harvard Law Students by College Group, 1974–1977 and 2014–2016 Classes Source: Unpublished information provided by Harvard Law School.

National Scholarships

Our third marker of postcollege success, the prestigious Rhodes, Marshall, and Truman Scholarships, are so rare that annual data are plagued by vicissitudes and variability. To make it easier to discern trends, three-­year moving averages of the distribution of t­ hese prizes by year have been calculated. ­These are shown in Figures 12.7 to 12.9. Lines show the cumulative percentage for college groups beginning with the dominant one, for private colleges in the 98th and above category, with intervals above that line showing the shares of other college groups. Figure 12.7 shows that, for the years from 1971 to about 2000, the share of Rhodes Scholarships accounted for by gradu­ates of colleges in the top private group generally declined, starting at over 50 ­percent in the early 1970s and declining to a low of about 40 ­percent, ­a fter which it recovered. This degree of dominance is remarkable, considering the very small number of gradu­ates

322 C onsequences

Proportion of students

0.40

0.30

0.20

0.10

0.00

0–50

50–90

90–95

95–98

98–100

0–50

Private colleges

50–90

90–100

Public colleges

1976–78

2014–16

Figure 12.5. ​Share of NYU Law Students by College Group, 1976–1978 and 2014–2016 Classes Source: Unpublished data provided by NYU Law School.

0.25

Proportion of students

0.20

0.15

0.10

0.05

0.00

0–50

50–90

90–95

95–98

98–100

0–50

Private colleges 1977–79

50–90

90–100

Public colleges 2014–16

Figure 12.6. ​Share of Duke Law Students by College Group, 1977–1979 and 2014–2016 Classes Source: Unpublished data provided by Duke Law School.

Ou t c o m e s

323

90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

19 7 19 1 73 19 7 19 5 77 19 7 19 9 8 19 1 83 19 8 19 5 87 19 8 19 9 9 19 1 93 19 9 19 5 97 19 9 20 9 0 20 1 03 20 0 20 5 07 20 0 20 9 11

Three-year rolling average of proportion of total number of scholarships

100%

Award year Private, SAT: 98–100 pct

Private, SAT: 95–98 pct

Private, SAT: 90–95 pct

Private, SAT: 50–90 pct

Private, SAT: 0–50 pct and HBCUs

Public, SAT: 90–100 pct

Public, SAT: 50–90 pct

Public, SAT: 0–50 pct and HBCUs

Figure 12.7. ​Distribution of Rhodes Scholarship Winners, by College Group Source: Rhodes Scholarship.

available (about 2 ­percent of the total). No other college group comes close to it. The top-­ranked public schools (90th percentile and above) reached a high mark of only 17 ­percent share in 2005. Moreover, ­there is no evidence that the share of the top private colleges declined over time. For the Marshall Scholarship, however, t­here is such a trend. In this case the top private college share reached 70 ­percent, in the late 1970s, and then began a long decline, reaching a low of 33 ­percent in 2007. Picking up the slack, in par­tic­u­lar, ­were above-­average public institutions, which together accounted for 36 ­percent in that year. To be sure, ­those public institutions had roughly 30 ­percent of enrollment, compared to the 2 ­percent for top private colleges. Although somewhat less concentrated than the Rhodes, the Marshall Scholarship was still heavi­ly weighted ­toward the elite private institutions.

324 C onsequences

90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

19 7 19 1 73 19 7 19 5 77 19 7 19 9 8 19 1 83 19 8 19 5 8 19 7 8 19 9 9 19 1 93 19 9 19 5 97 19 9 20 9 0 20 1 03 20 0 20 5 07 20 0 20 9 11

Three-year rolling average of proportion of total number of scholarships

100%

Award year Private, SAT: 98–100 pct

Private, SAT: 95–98 pct

Private, SAT: 90–95 pct

Private, SAT: 50–90 pct

Private, SAT: 0–50 pct and HBCUs

Public, SAT: 90–100 pct

Public, SAT: 50–90 pct

Public, SAT: 0–50 pct and HBCUs

Figure 12.8. ​Distribution of Marshall Scholarship Winners, by College Group Source: Marshall Scholarship.

Rounding out ­these scholarships is the Truman Scholarship. Reflecting perhaps the ­humble roots of the president whose name it carries, it displays a strikingly dif­fer­ent, more egalitarian distribution. The top private colleges are again overrepresented, but less extremely so. For this scholarship, the top private colleges account for only about 20 ­percent of the winners. The second biggest contributor for most of the period was private colleges in the 50th to 90th percentile. As with the Rhodes Scholarship, t­ here is no indication of a change in the distribution across colleges in successful outcomes. Taken together the indicators examined in this section paint a portrait of spectacular in­equality in outcomes. The indicators are imperfect; they share the same virtues and weaknesses. Their strength lies in their very existence over most or all of the entire period, thus meeting the apples-­to-­apples test. To be sure, circumstances do

Ou t c o m e s

325

90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

19

78 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02 20 04 20 06 20 08 20 10 20 12

Three-year rolling average of proportion of total number of scholarships

100%

Award year Private, SAT: 98–100 pct

Private, SAT: 95–98 pct

Private, SAT: 90–95 pct

Private, SAT: 50–90 pct

Private, SAT: 0–50 pct and HBCUs

Public, SAT: 90–100 pct

Public, SAT: 50–90 pct

Public, SAT: 0–50 pct and HBCUs

Figure 12.9. ​Distribution of Truman Scholarship Winners, by College Group Source: Truman Scholarship.

change, and the significance of t­ hese achievements might have been altered in the pro­cess. Both the PhD and law degrees ­were tarnished by soft ­labor markets, beginning in the 1970s for the PhD and a­ fter 2008 for the law degree. Nevertheless, both sets of indicators remained solid mea­sures of early gradu­ate success. The prestigious scholarships may have retained their original celebrity the best, even if se­lection policies might have shifted slightly. But ­these mea­sures also have two major weaknesses as reflections of outcomes. One is their relative rarity, most obvious for the Rhodes, which reduces their value as indicators of any but the very top students. The second is the deeper concern that postgraduate achievements of any variety surely cannot be attributed entirely to college. Some economists have taken on this knotty challenge of identifying a college’s value-­added, as noted below.

326 C onsequences

Passing Judgment

When we examine the market for baccalaureate education in the United States, we behold a scene of spectacular disparities. They reveal themselves as differences across colleges in tangible resources, academic qualifications of entering students, and outcomes for gradu­ates. ­There can be l­ ittle doubt that the adjective “unequal” is appropriate. How should we judge this market? Is this in­equality necessarily a bad ­thing? As we have seen, the existing unequal setup is not simply some historical accident. Rather, it is the product of forces inside and outside the college market itself that have combined to make ­things more unequal. The resulting disparities would appear to be at odds with cherished notions of equality of opportunity. Less obviously, the disparities might also reflect, from the standpoint of the nation’s economic well-­being, a wasteful allocation of scarce resources. It is fitting to conclude this consideration of outcomes, as limited as it may be by the lack of comparable mea­sures, by borrowing two conventional normative yardsticks for passing judgment on the college market. One of ­these standards is intergenerational equity. College has often been viewed as an instrument for upward mobility. Some recent research bears directly on this norm. A second, related question is w ­ hether the system that we have, with the brightest students attending the wealthiest colleges, is a sensible way to allocate the nation’s scarce resources. A Path for Upward Mobility?

A widely accepted guideline for judging the fairness of economic systems is the likelihood that talented but lowly born individuals can rise above their origins to achieve higher economic status as adults. To what degree do individuals’ economic positions depend on their own merit, as opposed to the economic position of their parents? Economists often mea­sure the ease of economic mobility by comparing the incomes of f­athers and sons; the less they diverge, the less economic mobility t­ here is. The United States, in spite of its

Ou t c o m e s

327

aspirations to be a land of opportunity, actually does poorly on this mea­sure when compared to other developed countries.28 In addition, research indicates that intergenerational mobility in the United States declined ­a fter 1980.29 It is natu­ral to won­der what role colleges play in this intergenerational transmission of economic status. In their recent empirical tour de force, Raj Chetty and coauthors generated new information that is directly relevant to this question. Two aspects seem especially worth mea­sur­ing. The first aspect, sometimes termed “access” in higher education discussions, asks simply how many low-­income students the college enrolled. If a college makes no room for the less advantaged, it can hardly aspire to contribute to upward mobility, except in some indirect way, such as through its research. One ser­ viceable mea­sure of such openness is the percentage of students whose parents ­were in the bottom one-­fifth of the income distribution. Chetty and his team calculated this percentage for e­ very four-­year college in the land, using data from some 30 million tax returns for Americans born between 1980 and 1982 who subsequently went to college, as well as their parents’ tax returns. Using t­hese researchers’ calculations for nearly a thousand colleges, Figure 12.10 shows average percentages for each of the seventeen college categories (see the black bars). The figure reveals wide differences in colleges’ enrollment of low-­income students. Except for public HBCUs, students from the bottom quintile are underrepresented in e­ very category of college, relative to their share in the population. The proportions are lowest in the most selective private colleges, ­those at the 90th percentile and above. No more than 4 ­percent of the students at ­these colleges are from the bottom income quintile. Aside from the HBCUs, the highest shares of low-­income students are in the least selective private and public colleges. If colleges have the potential to contribute to upward mobility by providing students of modest means the chance to gradu­ate and enter the workforce with a college degree, t­hese percentages would suggest that most four-­year colleges in the United States do not avail themselves of this chance to any appreciable degree.

328 C onsequences

0.20 0.15 0.10 0.05

Private

Public

Pub

Priv

90+

50–80

80–90

0–50

99+

98–99

90–95

95–98

85–90

80–85

60–70

70–80

50–60

25–50

0.00

0–25

Proportion of students from bottom 20%; proportion also ascending to top 20%

0.25

HBCU

Pct lowest 20% 0.09 0.07 0.05 0.06 0.05 0.05 0.05 0.04 0.04 0.04 0.04 0.09 0.07 0.08 0.06 0.17 0.24 Mobility rate 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.03 0.03 0.03 0.03

Figure 12.10. ​Colleges and Intergenerational Mobility: Two Mea­sures Source: Chetty et al. (2017) and Equality of Opportunity Proj­ect, accessed January 22, 2017, http://­w ww​.­equality​- ­of​- ­opportunity​.o ­ rg​/c­ ollege​/.­ Note: The black bars to the left in each pair give the proportion of students whose parents’ income was in the bottom 20 ­percent at the time they went to college. The gray bars on the right give the mobility index, which is the proportion of students meeting two requirements: their parents’ income was in the bottom 20 ­percent at the time they went to college and they themselves went on to have incomes in 2014 that ­were in the top 20 ­percent for their birth cohort in that year. Based on tax return data covering parents of college students born in the period from 1980 to 1982. Parents’ income was averaged for the five years child turned fifteen to nineteen, expressed in 2015 dollars. Bars show author’s calculation of weighted averages for 984 colleges, classified by 17 categories.

A second, more exacting benchmark of intergenerational mobility, proposed by Chetty and his collaborators, asks how many of ­those low-­income students ended up, as young adults, rising to the top one-­fifth of the income distribution. For the 1980–1982 cohorts of millennials, Chetty and his team recorded the income of each student in 2014, a year when they ­were in their early thirties, to determine where each one ranked, income-­wise, in their age cohort nationwide.30 Armed with this information, the researchers devised a mea­sure of income mobility, again, for each and ­every college. This mea­sure combined information on students’ origins with

Ou t c o m e s

329

i­nformation on their postcollege economic success. The mea­sure is the percentage of students at a college who both came from a ­family in the bottom fifth of the income distribution (the first mea­sure described above) and in their early thirties had achieved a level of income that placed them in the top one-­fifth in their age cohort. 31 For example, 2 ­percent of Prince­ton’s students born between 1980 and 1982 had parents in the lowest one-­fi fth of incomes. Of ­those Prince­ton students, almost two-­thirds, 65.9 ­percent, ascended the income ladder to enjoy incomes in the top fifth by 2014. Prince­ton’s mobility index of 1.3 ­percent is simply the product of t­ hose two percentages. Another university with the same mobility index, Northern Michigan University, featured a bigger share of students who grew up in the bottom fifth (7.1  ­percent) but a lower rate of success in ascending into the top fifth ­after college (18.7 ­percent). To see how this index of mobility differs across our seventeen college categories, look at the gray bars in Figure 12.10. It is clear from the height of ­these bars that a much smaller share of the students at ­these colleges managed to ascend from the bottom to the top income quintiles. Judged by this exacting standard, the private selective institutions did no worse, on average, than the less selective colleges in ­either sector. It is worth stressing that their calculated mea­sures of mobility, as impressive as they are in their detail, are not determined by, but rather only associated with, the colleges themselves. But, as the authors point out, the mea­sures are highly suggestive. Colleges that enroll students of modest means, let alone see them ascend to the top quintile, are colleges making an effort to engender intergenerational mobility. The Best of All Pos­si­ble Worlds?

The second normative perspective to judge the college market is economic efficiency. This perspective questions w ­ hether the elaborate tracking system that has evolved represents the best use of this nation’s scarce resources. Does it make economic sense for selective, prestigious colleges to control such a disproportionate share of all

330 C onsequences

the resources devoted to the sector? Are they more productive, in the sense of producing more education, than colleges lower down in the hierarchy of colleges? A number of economists have addressed the narrow question of economic payoff, using wages, the ­labor market’s verdict on how valuable colleges’ gradu­ates are, to mea­sure output, and gradu­ates of prestigious colleges do indeed tend to earn more than gradu­ates of less selective colleges. Using a massive data set that allowed her to match ­people’s earnings from tax returns with where they had attended college, Caroline Hoxby estimated that gradu­ates of the most selective colleges w ­ ill earn on average nearly twice as much over their lifetimes as ­will gradu­ates of colleges with average SATs of 1000.32 But ­these differences in average earnings almost surely overstate the contribution of college: t­here is ­every reason to think that gradu­ates of less selective colleges prob­ ably lack some desirable traits possessed by gradu­ates of the more selective colleges. It would be a m ­ istake, therefore, to attribute all the difference in earnings to the effect of the college experience. Researchers have devised a variety of approaches to deal with this sort of statistical bias, caused by the failure to account for unobserved traits. 33 One creative attempt to get around such se­lection bias is a study by Mark Hoekstra, who compared the earnings in ­later life of white men in Texas who scored just above or below the score cutoff required to attend their state’s flagship university. All of ­these applicants who clustered around this cutoff arguably w ­ ere very similar, except for which side of the line they fell. But the side they landed on turned out to be highly impor­tant. ­Those who w ­ ere just above it and ­were admitted went on to earn some 20 ­percent more than ­those who had to ­settle for a lesser university.34 For ­these students very close to the cutoff, the chance to attend the more selective university had an arguably causal effect on their ­later earnings. Although this finding offers compelling evidence of the effect of attending a more selective institution, it may not be valid to generalize that result for students elsewhere in the achievement distribution. Another strategy devised for overcoming se­lection bias was to compare the outcomes of other­w ise similar college applicants,

Ou t c o m e s

331

some of whom went to college A and ­others of whom went to college B. In so ­doing, the reasoning went, all relevant but unmea­sured characteristics could be assumed to be equal, making the students’ attendance at one or the other college more or less random, thus outflanking the pesky prob­lem of se­lection bias. One attempt to do this, by Stacy Dale and Alan Krueger, is based on comparisons of groups of students who had been admitted to and rejected by the same set of colleges but who ultimately chose to attend dif­fer­ent schools. Following this method, the authors found, in contrast to most studies, that students who attended more selective colleges did not earn more. Paradoxically, they also found that students who chose to attend colleges with higher tuitions did earn more. Although innovative, the approach they took can be faulted if students who chose less selective over more selective colleges w ­ ere somehow systematically dif­ fer­ent from the typical college applicant, who almost always chooses the most selective colleges he or she gets into.35 Still another effort to base comparisons of dif­fer­ent colleges on earnings for “virtually identical” students, by Caroline Hoxby, is part of a larger study with the ambitious aim to calculate the productivity of ­every single college in the country. Hoxby started by asking the same question that previous studies had taken on—­what is the contribution of attending a par­tic­u­lar college to a person’s earnings? She compared the postgraduation earnings for pairs of groups of similar students who, based on their SAT scores, w ­ ere judged to be “on the ­bubble” of admission to Vanderbilt, for example, but between which only one group was ultimately admitted and went to Vanderbilt. Of the other group some may have gone to Yale while o ­ thers went to less selective colleges. Hoxby treated the assignment of students in ­these two groups to colleges—to Yale or Vanderbilt, for example—as random, making it pos­si­ble to attribute any differences in earnings between t­ hose students entirely to where they went to college. Using a massive data set including all tax returns in the nation, she compared the ­later earnings (to age thirty-­two) of such paired groups of students, employing the same method used to establish a ranking of all tennis players based on the outcomes of

332 C onsequences

tennis matches.36 Each instance in which two colleges enrolled two presumptively identical students is like a tennis match, and it is pos­si­ble to see which of the two colleges in each pair “wins” in the task of preparing its student by observing which student in that pair earns more. It is an ingenious approach; it depends for its execution on a wealth of information on thousands of students. 37 By examining t­ hese presumptively other­w ise identical students, ­these sets of comparisons are intended to yield the additional productive value a student gets from attending one college rather than another— the value-­added. ­A fter applying this method to her massive data set, Hoxby concludes that this value-­added is higher for gradu­ates of more selective colleges than for ­those of less selective ones, although average value-­added does not rise as steeply with college selectivity as average earnings itself. This is a power­ful finding, for it implies that the most able, academically ready students are best able to take advantage of the rich resources available at the most selective colleges. Expressed differently, the social return from devoting educational resources is higher for t­hese more able students than it would be for students less well-­prepared for college.38 On top of this value-­added mea­sure, Hoxby adds two more potentially quantifiable aspects of outcomes that can be attributed to colleges. One is the social value of work done by gradu­ates who choose to work in government or the nonprofit sector, where salaries are below what ­those gradu­ates could have earned in the for-­profit sector. Another potential outcome is innovation that spills out from gradu­ates to raise productivity in other firms and industries.39 Few would doubt the social value in ­either of ­these aspects, but no one before Hoxby had tried to quantify them. Adding ­these three aspects of output together, Hoxby offers a bold and unpre­ce­dented estimate of how colleges differ in their economic output. To address the ultimate question of productivity requires comparing ­these outcomes to the social cost of operating t­ hese colleges. A fact made clear in Chapter 5 is that the cost of educating college students rises as one moves up the selectivity hierarchy. The question Hoxby asks is, does the output per student—as mea­sured by the

Ou t c o m e s

333

three aspects described—­rise as fast as the costs do? Her answer is yes, on average. Some colleges do a better job than o ­ thers, but on average the output per dollar spent for the most selective colleges is no more or less than it is for less selective ones. This conclusion has an enormous implication. It offers an economic rationale for the in­equality we see across American colleges. Although resources per student rises, and in places, sharply, with selectivity, so too does productivity. The vastly richer resources at the top colleges are not wasted, ­these findings imply. Brighter, better prepared students are able to put to good use t­ hose additional resources at the colleges where they enroll. This conclusion rests, however, on the accuracy of the method of computing value-­added. If that method has not succeeded in squeezing all the se­lection bias out, the value-­added of selective colleges ­will tend to be overstated. If, for example, admissions officers are able to spot characteristics in applicants that suggest ­future success, over and above the observed difference in test scores, then the students admitted to the more selective colleges in t­hese paired comparisons w ­ ill tend to be more able on average than the students who w ­ ere not. ­Here the analogy with the ranking of tennis players would break down, by requiring that two colleges get the chance to influence pairs of more or less identical students, to see which college “wins” ­these match-­ups more often. ­Whether or not such bias exists, it is clear that, more than anyone e­ lse, Hoxby has looked at the w ­ hole market for college and has tried to make sense of its structure, in par­tic­u­lar the matching of bright students with rich schools. Implications beyond the College Market

As stated at the outset, the primary purpose of this book is descriptive, not predictive or normative. And most of the description contained h ­ ere applies in one way or the other to inputs—­such as majors offered, financial resources, number of faculty, aspects of competition, or characteristics of incoming students. This heavy focus on inputs has resulted in part from the need to look ­under the lamp-

334 C onsequences

post in the search for consistent mea­sures that can be traced over the four de­cades covered by the book. The pres­ent chapter diverges from this focus on inputs and considers outcomes of college both as a general topic and through several mea­sures devised to fit snugly into the divisions drawn to categorize colleges. As a general topic, two t­ hings are clear about them. First, outcomes are multifaceted. College has many effects, touching on personality, habits, health, attitudes, skills, and knowledge. As such, ­these effects defy easy mea­sure­ment. Only one study has gone beyond the simplest mea­sures and compared outcome for par­tic­u­lar colleges. The major analytical prob­lem that arises with any outcome mea­sure is determining causation, which is difficult b ­ ecause of se­lection bias or bias due to omitted variables. The second general point is that the production pro­cess involves the effort, skills, and inclinations of the student, as the example of Sonia Sotomayor illustrates. This chapter uses unorthodox output mea­sures defined by college category, which show that some outputs can be even more unequal than the most unequal of input mea­sures. Taken as a w ­ hole, the economywide effects of the patterns and trends we see in the college market fall into two categories. One set might be termed growth and efficiency. Our system features colleges that are sharply stratified in terms of financial resources and student quality. If the brightest students are the ones who can make best use of the rich resources at the most selective colleges, then the existing system ­will look efficient. It w ­ ill be one best designed to achieve a high rate of economic growth. Hoxby’s calculations suggest this is the case. The second set of effects is distributional. Not only do affluent students tend to end up at rich colleges, this tendency is amplified by way of marriage patterns. Evidence suggests that educational homogamy, the tendency for ­people to marry partners with the same educational attainment, has risen in the United States since the 1960s.40 And this trend in marriages has increased ­house­hold income in­equality.41 If the increasing educational stratification across colleges increases educational homogamy, then ­there is reason to believe a result w ­ ill be further income in­equality.42

Ou t c o m e s

335

13 Why It M ­ atters

The period covered in this book is, not coincidentally, a time when I had the chance to witness, firsthand, many of the developments it chronicles, albeit from the myopic vantage point of one person. I have been, by turns, student, professor, administrator, and parent of college students. I had the good fortune to teach undergraduates at three excellent universities, one public and two private. At Duke University, where I have spent most of t­ hese years, I held several administrative positions, in which I had the chance to work with the admissions and financial aid offices, to help plan a new residential hall and a recreation center, to take part in discussions about tuition increases, to consider new modes of instruction, and to encourage good teaching practices. Given the necessarily limited perspective that ­those roles offered me, I could not have known, purely from what I observed, what kind of macro-­level changes w ­ ere taking place all around me.

Soon a­ fter I arrived at Duke as a faculty member, the university’s chancellor, following a half de­cade of crippling inflation, issued a meticulously reasoned report calling for full-­scale retrenchment in hopes of avoiding what he saw as other­wise inevitable bud­get deficits.1 But the ensuing de­cades, marked by intensifying income in­equality in the country, would bring growth and riches to my university, not the retrenchment predicted by the chancellor. And that same good fortune befell other selective colleges as well. As the evidence presented in this book makes clear, American colleges ­were becoming more unequal during t­ hese years, and Duke was one of the unwitting beneficiaries of the ironic but potent tailwinds spawned by that growing in­equality of income. The extent of the in­equality among colleges, copiously documented in this book, is stunning in its extent and breathtaking in the variety of its manifestations. Yet no one who worked at colleges like Duke set out to create a more unequal higher education industry. Quite the contrary. Leaders in higher education, when they speak at all on the subject, abhor in­equality and praise policies to extend the benefits of a college education to disadvantaged groups, especially policies to help needy students attend the college of their choice. Moreover, public policies enacted or decreed at the federal level, with few exceptions, pushed back against discriminatory practices of the past, and colleges embraced ­these policies. By the turn of the ­century, thanks to demography and the end of racial discrimination in college admissions, college enrollments would show a degree of racial and ethnic diversity unimagined in 1970. Yet we are left with a system of baccalaureate education that is in many ways even more unequal than it was in 1970. Equality ­isn’t every­thing, of course. It is pos­si­ble that this system, with all its disparities, represents the best use of the nation’s limited resources. Certainly, the sterling reputation that Amer­i­ca’s top colleges have achieved around the world would suggest that they are ­doing their job effectively. In deciding how to respond to the findings in this book, I could view them through the lens provided by the dif­fer­ent perspectives offered by my own exposure to the college market. The first perspective



W h y I t ­M a t t e r s

337

is that of consumer—as student and ­later as parent. The feature of the college market that looms largest from this vantage point is surely the market’s highly segmented nature. In deciding where to apply to college, the prospective freshman and his or her parents quickly narrow the range of choices from a thousand to no more than a dozen or two, constrained less by distance than by admissions standards. Thanks to published rankings, college guides, and the internet, information is plentiful. But some of this information is also of uncertain value. For example, many observers would judge the features valued by U.S. News to be weaker indicators of quality than the kinds of outcomes incorporated in the Washington Monthly rankings. And no one has to tell prospective matriculants that the stakes are high or that evidence of academic merit—­grades and test scores— is indispensable. The second perspective through which I observed the college market was from positions deeply embedded inside the industry, especially my administrative stints at Duke. As one of the colleges fortunate enough to enjoy the fruits of the Matthew effect, Duke had the luxury of choice—in admissions policies and in programmatic direction. The third perspective I have been able to take is the lens of public policy, a perspective that is, a­ fter all, the day job of a professor in a school of public policy. The pres­ent book has something to contribute to each of ­these three audiences—­ student / customer, college / firm, and policy analyst. Before enunciating the choices the book lays out for each audience, it is useful to step back and recall how the nation’s college market got ­here. Four De­cades of Change

By almost any measure—­resources, quality of students, or prestige—­the market for baccalaureate education in 1970 was highly unequal. The country was emerging from a turbulent period of protest, reaction, and reform. The formerly segregated colleges of the South ­were now enrolling black students, but the historically black colleges and universities created by Jim Crow segregation lived on. The nation’s population was moving south and west, but its colleges hardly ever

338 C onsequences

budged from the place they ­were founded. In markets for most goods and ser­vices, retail firms move so they can be close to their customers, but in this market the firms remained fixed in place. Many of the nation’s most prestigious colleges, anchored in the Northeast, drew students from an ever-­expanding national, and international, geo­graph­i­cal area. As admissions officers for ­these schools more and more looked to test scores and other objective markers of academic ability, applicants from the Northeast found the road to Yale and Amherst considerably more challenging than their parents had. The social force that exerted the most potent influence on the market for bachelor’s degrees was the growing income in­equality in the American economy. Beginning about 1980 the incomes of ­house­holds in the top rungs of the income distribution commenced to swell, while incomes for almost every­one e­ lse ­were stagnating. For the wealthiest of the colleges, this intensifying in­equality returned a dividend that was as handsome as it was unanticipated. Part of the dividend resulted from the surge in the demand for places in t­ hese selective colleges. Admission to one of them became a fiercely sought-­ after prize, one not subject to the usual bidding pro­cess encountered in most markets. Affluent families came to realize that their rising incomes alone could not guarantee one of t­ hese prized admissions slots, a realization that gave birth to feverish efforts to gain an advantage in the fiercely competitive tournament for admission. The second piece of the in­equality dividend was the surge in donations spawned by the new affluence at the top. In a perfect illustration of the enigmatic Matthew effect, t­ hese new donations, b ­ ecause donors tended to give to their own alma maters, gravitated to the very institutions that ostensibly needed them the least. For as long as anyone could remember, colleges have vied with each other to enroll students, owing to the proliferation of private colleges. But this competition took on dif­fer­ent forms, one for the selective colleges, another for the nonselective ones. The college market was, in truth, profoundly segmented. Into this mix came a wild card, in the form of a spectacularly successful proj­ect launched by the venerable news weekly U.S. News and



W h y I t ­M a t t e r s

339

World Report. The magazine’s idea to devote a few pages in one issue to a numerical ranking of colleges, in 1983, quickly captured the nation’s attention and, before long, became a fixture in the college marketplace. As it came to eclipse the very magazine that spawned it, the ranking lubricated the already active forces of competition. For more than a c­ entury, the market, deeply segmented by prestige and exclusivity, had been marked by keen competition—­among colleges for students and among students and their families for admission. Among the most selective colleges, thanks to the declines in transportation costs and the existence of a nationally recognized test in the SAT, the U.S. News rankings acted at once as a source for comparison shoppers and a focal point for colleges’ strategic investments to increase their attractiveness. ­Because they had the resources to do so, ­these selective colleges put money into the types of expenditures that would both improve their quality and boost their U.S. News rank. ­These investments included hiring more faculty and upping their pay. ­These selective colleges could afford to remain true to their educational philosophies, usually embracing the liberal arts, rather than veering very far in the direction of vocational training. For less-­selective colleges—­the modestly positioned private colleges and the public institutions in the lower SAT ranges—­rankings and the pro­cess of competing for students took on a faintly desperate aspect. The private colleges in t­ hese ranges ­were often entirely dependent on tuition to pay their cost of operations. Likewise, the comparable public institutions relied on state appropriations, which in turn depended on the number of students they enrolled. Unfortunately, the country’s demographic trends had decreed a decade-­long decline in the number of eighteen-­year-­olds. In the grim competition for students that accompanied that decline, private colleges scurried to award no-­ need scholarships to promising applicants, hoping by ­doing so to improve the statistics they could report to U.S. News and to convince lesser applicants to enroll as well. But this practice of discounting tuition was inherently risky, as it opened colleges up to the destructive competition of a bidding war. For less-­selective colleges ­these tactics could be no more than a holding pattern, not a sus-

340 C onsequences

tainable strategy that would put them on a firm financial footing. To achieve longer-­term stability, ­these colleges would have to revise course offerings ­toward occupationally oriented majors such as nursing, business, or recreation. The urge for survival, then, had the effect of further separating the haves from the have-­nots. Such supply-­side developments, and the growing in­equality they accompanied, can be illustrated by revisiting the three colleges featured in Chapter 2. By 2016 Illinois State was still catering almost exclusively to Illinois residents with middling academic rec­ords, its ac­cep­tance rate hovering around 80 ­percent and the average SAT of its enrolled students about 1160, both indicators remaining roughly where they stood in 1970. Its undergraduate enrollment had grown by more than a fifth over the period, to 18,400. Reflecting its shift away from its formerly primary role of teacher training, the university introduced dozens of new majors. Some of ­these ­were in traditional arts and sciences disciplines, such as biology, En­glish, physics, and philosophy. But many more of the new majors w ­ ere pragmatic ones typical of colleges lower down the prestige ranking, majors that offered skills that could lead directly to middle-­level, white-­collar jobs. Among the college’s new majors in this group w ­ ere athletic training, construction management, criminal justice, fashion design and merchandising, insurance, interior design, and recreation and park management. In recent years the university had added several new buildings, including a center devoted to student fitness and another for performing arts. In its competition for faculty, Illinois State strug­gled to stay competitive. In the forty-­five years from 1970 / 71 to 2015 / 16, the average compensation for its full professors barely kept up with inflation, rising a scant 2 ­percent in real value. The university managed to increase the pay of assistant professors, however, by 22 ­percent over the same period.2 Boston College, by contrast, seemed to blossom in the de­cades ­after 1970, becoming more selective, more academic, and wealthier. In 2016 its ac­cep­tance rate was 29 ­percent, and the average SAT score of its enrolled students had risen more than 100 points, to 1430. Undergraduate enrollment was more than a third larger than it was



W h y I t ­M a t t e r s

341

in 1970. Like Illinois State, Boston College instituted dozens of new majors. Corresponding to its position as a selective private college, most of t­ hese ­were in traditional arts and sciences disciplines: art history, astronomy, biochemistry, cinema and media studies, comparative literature, linguistics, mathe­matics / computer science, and neurosciences, to name only some. Owing to its undergraduate business program, the college also added a number of majors in business as well, including accounting, marketing, and finance. To accommodate more of its growing student body on campus, the college acquired more than a hundred acres of real estate, making room for new dormitories and its law school, among other uses. In 2016 it was building a new recreation center. In sharp contrast to the stagnation of pay for full professors at Illinois State, full professors at Boston College enjoyed a stellar period of rising real pay, with average compensation increasing by 74 ­percent. The inflation-­adjusted pay for assistant professors r­ose faster as well, increasing by 55 percent. Such was the lot of the favored colleges in the top echelon of wealth and prestige.3 Our third illustrative college, Tennessee Wesleyan, appeared to hold its own, but strug­gled mightily in terms of faculty compensation. Its tiny enrollment grew by half over the four and a half de­cades, to 1,000 in 2016. It remained, like Illinois State, relatively unselective, admitting two out of ­every three applicants. The average SAT for its students was 1010. Over the period the college added new majors, some dozen and a half. About half of t­ hese ­were traditionally academic ones, like creative writing, biology, psy­chol­ogy, and theater. The remainder ­were practical ones with close occupational linkages; t­ hese included early childhood education, exercise sciences, sport management, and criminal justice. In 2016 the college was constructing a new campus center and a fitness complex. In the forty-­five years ­after 1970 the college’s full professors fell far b ­ ehind in terms of pay, however, with their average inflation-­adjusted compensation falling by 23 ­percent. For assistant professors, the college managed to increase pay just enough to keep up with inflation. While Tennessee Wesleyan was able to sustain its operation and pro-

342 C onsequences

vide both liberal arts and practical learning, therefore, it did so by allowing their se­nior faculty to languish financially.4 On the demand side of the market, in the mid-1980s high school se­niors began to consult the new source of information provided by college rankings, which became steadily more exact, as regional lists ­were added to the famous rankings of national universities and liberal arts colleges. The growing attention to rankings did ­little to upset the preexisting hierarchy of colleges; rather it served to solidify it. In terms of the average SATs of entering students, the rank order of colleges changed remarkably ­little over the four de­cades, and the concentration of high-­scoring students, already pronounced in 1970, became more concentrated over the period, with the top private colleges attracting a growing share of the top high school students. Thus the academic stratification that plainly denoted in­equality in 1970 became even more pronounced in the ensuing de­cades. This increasing scholastic concentration could be seen even in the study habits of students, as the most hardworking students increasingly chose the most selective colleges. ­These trends fit in well with the parallel narrative of ever more objective and merit-­based admission standards at selective colleges. What could not have been a predicted result of heavier reliance on meritocratic admissions criteria was the stubbornness of the socioeconomic stratification. Such stratification had been a sturdy feature of American higher education for de­cades, of course, and the patterns for 1972, based on ­family income, parental education, and private school attendance, reveal it. But it was not clear w ­ hether this stratification would intensify over the period. In fact it did, but only modestly, as shown in Chapter 9. Thus the demand for college remained deeply riven by economic gaps. ­These gaps are dramatically revealed in the work of Raj Chetty and his coauthors, which provides an estimate of the share of students whose parents ­were among the nation’s most affluent 10 ­percent. Consider our three colleges. High-­income students constituted just 5 ­percent of the students at Tennessee Wesleyan. By comparison, their share was 22  ­percent at Illinois State and 55 ­percent at Boston College.5 Despite a revolution in civil rights,



W h y I t ­M a t t e r s

343

billions spent on financial aid, and the professed commitment of colleges to greater equality, stratification by economic class, inherited from the past, remained firmly in place. The Importance of the College Market

What significance do ­these findings hold for us? If the market for baccalaureate education w ­ ere inconsequential, the patterns and trends documented in this book could be dismissed as mere curiosities. But in fact this market has far-­reaching social and economic significance. For well over a ­century colleges have played a central role in determining which individuals in each succeeding cohort of young adults assume which positions as working adults. As a result of the complicated coproduction that occurs during the college years—­a pro­cess that requires as much from students as it does from professors and college facilities—­these young p ­ eople come out the other side, while by no means fully formed, ready to work or to gain more specialized training. Some scholars ­will quibble about how much of what colleges do ­really amounts to building new h ­ uman capital, as opposed to signaling inherent talent, but my view, and I am in the vast majority of economists on this point, is that colleges and the experience they provide to students do indeed build ­human capital. ­There is no better illustration than Sonia Sotomayor’s years as a Prince­ton student of the essential interplay among courses, professors, laboratory facilities, written assignments, myriad contacts with other students, and extracurricular activities ranging from protests to parties that creates the new knowledge, skills, and perspective that we ascribe to the college gradu­ate. Its technology may be complicated and its effects idiosyncratic, but college is no mere holding cell where fully formed minds merely bide their time, waiting for their chance to enter the working world. It is ­because of their outsize influence on the lives of their students that colleges, as a w ­ hole, have the power­ful effect that they do. Seen from the celestial heights occupied by social scientists who examine aggregate data, the market for baccalaureate education in-

344 C onsequences

fluences the broader society and economy in two main ways. First, this market is a vast mechanism for determining how the limited resources in the economy—­workers, students, buildings, books, lab equipment, and the like—­will be put to use. Which professors w ­ ill teach at which colleges? Which colleges ­will have the most up-­to-­ date laboratories, classrooms, and dining halls? How large w ­ ill classes be? How ­will students use their time? How much time ­will professors spend meeting with students and reading their written work? Perhaps most importantly, which students ­will go to which colleges? When critics voice concerns about large lectures or the dearth of writing assignments in college courses or declines in time spent studying, they are expressing worries that resources are not being put to their most productive uses. And the unhappy consequence of such misallocation is that the nation is falling short of what it might ideally achieve, that its national income w ­ ill be too small and its economic growth too slow. Numerous figures in this book provide evidence relevant to resource allocation. A central stylized fact that has emerged is this one: the students who finished high school with the strongest academic rec­ords ended up attending the colleges that possessed the richest set of resources. This pattern of matching students with colleges surprises no one who is familiar with American higher education. As I have noted, this matching pattern is akin to academic tracking in high schools. A significant finding that emerges from the evidence presented h ­ ere is that this matching, and the academic stratification it creates, has become more exact, more extreme, in the de­cades since 1970. What this stratification means for resource allocation is that the academically best students have been exposed to the country’s best educational resources. And, as a byproduct of this assignment, ­those students have had the added advantage of having as their college classmates mostly students just as bright and well prepared as they are. Students in ­these selective colleges encountered curricula rich in theoretical and classical arts and sciences subjects, while students at the less selective colleges received instruction that tended to the prosaic, practical, and vocational.



W h y I t ­M a t t e r s

345

This pattern of matching students with colleges has potent implications for the allocation of resources. No doubt it is responsible in part for the extraordinary achievements of the gradu­ates of our top colleges. It explains why spots in Amer­i­ca’s top colleges are sought ­after by students around the world, and why affluent American parents are so keen to have their ­children gain admission to one of them. But this two-­sided concentration of resources and academic talent also means that students with less promise ­will gather in more humbly appointed dormitories and classrooms and be taught by less accomplished instructors with heavier teaching loads. Is this a sensible pattern of allocating scarce resources? It would be, from an economist’s point of view, if the bright student gets more out of attending the rich college with the most prominent professors and with similarly good students alongside. One economist who has addressed this issue head-on is Caroline Hoxby, whose efforts to mea­ sure the value-­added from college are described in Chapter 12. Her calculations suggest that the returns that the nation reaps from the education of bright students attending rich colleges exceeds the social return from educating less promising students at modestly endowed colleges. She finds, in fact, that, from less selective to the most selective colleges, t­ hese social returns (more precisely, the social marginal benefit) rise more or less in step with costs. The implication of this finding is a happy one for the nation’s allocation of resources: despite the disproportionate amount of resources the country devotes to the likes of Amherst and Stanford, the students who enroll ­there make better use of ­those resources than less able students would. This finding rests in large part on her estimates of the additional earnings a student can expect by attending a more selective college. If ­these estimates overstate the ­actual causal effect, however, then the conclusion that the current system represents an efficient allocation of resources is harder to assert with confidence. At a more impressionistic level, the college market looks reasonably efficient. Owing to its decentralized setup and its large number of firms, and considering the copious amount of information that potential consumers have about colleges, the market has most of the

346 C onsequences

essential ingredients to be well-­functioning. Information is imperfect, to be sure. And the fixed locations of suppliers further makes conditions less than ideal for efficient operation. Another oddity that is worth noting in assessing the market’s efficiency is commercial sports, the strange but per­sis­tent appendage of many American universities. Although spectator sports have become a fundamental, if unacknowledged, core function for the universities that run them, ­these commercial enterprises surely represent a diversion, both financial and cognitive, from the educational mission of ­these universities. One feature that suggests that t­ hese enterprises might represent a misallocation of resources is the fact that the tax law largely treats them as tax-­exempt educational functions rather than the largely unrelated commercial businesses that they are. We cannot know for sure how much t­ hese enterprises—­a nd the attention devoted to them by students, alumni, and the public—­diminish the educational quality of American higher education as a ­whole. But my guess is that it is worth worrying about and that it affects the quality of education in the m ­ iddle ranges of the SAT distribution of colleges. In addition to its role in resource allocation, the college market has a second far-­reaching influence—on the distribution of income, privilege, and power. Not every­one goes to college, and that disparity is one impor­tant reason for in­equality. As late as the nineteenth c­ entury only a fraction of young men in the United States attended college, and ­these students ­were by no means chosen randomly from the population. What distinguished the American experience with higher education from its Eu­ro­pean counter­parts was the rapid increase in the rate of college-­going, spurred by the emergence and growth of land-­grant and other public colleges in the late nineteenth and first quarter of the twentieth centuries. As significant as the Morrill Acts of the nineteenth ­century ­were, however, the G.I. Bill and the National Defense Education Act following World War II w ­ ere at least their equal in raising the share of Americans with a college degree. Despite the demo­cratizing thrust of this rise in college attainment, in­equality persisted in three significant re­spects. One of ­these, racial



W h y I t ­M a t t e r s

347

segregation and discrimination, despite its historical stranglehold on colleges throughout the South, had been curtailed if not eliminated in the college market by the 1970s. The other two aspects of in­equality in the college market ­were much more resistant to change. One was the socioeconomic bias in who goes to college and who gradu­ates with a bachelor’s degree. Study ­after study over the years revealed a marked socioeconomic class correlation with both college attendance and graduation, correlations that remained strong even when comparisons ­were restricted to students with similar academic aptitude.6 The third per­sis­tent aspect of in­equality in the baccalaureate market showed up in comparisons across colleges—­where, not w ­ hether, a student attended college. As the bachelor’s degree became more commonplace, distinctions among colleges took on added significance. ­These are the distinctions given par­tic­u­lar attention in this book. Chapter 9 documents wide gaps in economic and socioeconomic status across the seventeen categories of colleges, and the per­sis­tence of ­these gaps over time. The average ­family income of students attending the most selective private colleges is breathtaking, compared to the incomes of students attending the less selective public institutions or the HBCUs, and ­these gaps grew larger during the four de­cades covered by the study. To be sure, not every­one who enrolled at the private selective colleges came from an affluent ­family. Certainly Sonia Sotomayor did not. But, in terms of average income or the share with high incomes, the trends ­were unmistakable. That this socioeconomic stratification not only survived but intensified is remarkable, given the concerted efforts by government and colleges themselves to eliminate discrimination and advance equity. Not only had racial discrimination been outlawed, colleges all over the country, both private and public, ­adopted affirmative action policies in admissions. And, by all appearances, admissions criteria became more objective, coming to rely more on standardized tests, grades, and documented achievements, and less on connections and recommendations. But the equalizing effect of ­these

348 C onsequences

developments was countered by other forces. In par­tic­u­lar, three aspects of college admissions policies worked to retard more egalitarian admissions: legacy preferences, cuts in need-­based aid, and preferences for nonrevenue athletes. Most prominent among ­these was the favorable consideration routinely given to c­ hildren of alumni. The same colleges that professed their desire to enroll more eco­ nom­ically and racially diverse entering classes continued to ­favor ­these legacy applicants, a preference that surely works against diversity. A second development that tended to discourage economic diversity was departures from need-­blind admissions or full-­coverage financial aid for ­those admitted. Exhibit A was the rise of no-­need, “merit” scholarships used to entice high-­scoring students. Some colleges deci­ded to limit the number of applicants they accepted ­under a need-­blind rule. Other colleges simply dropped any pretense whatsoever of following the need-­blind princi­ple. Significantly, public colleges joined the movement away from need-­based aid, as states enacted their own merit aid scholarship programs. A third ­factor that leaned against egalitarian admissions was colleges’ desire to have winning athletic teams. Preferences for promising athletes had a mixed effect on economic diversity, since football and basketball players often increased the economic diversity of freshman classes. But preferences for other athletes tended not to.7 On top of ­these admissions policies, the other force that worked to maintain economic stratification was the diffuse but power­ful efforts by parents and students to prepare themselves for the college application pro­cess. Starting often before high school, ­children of highly educated parents ­were playing soccer, field hockey, tennis, and lacrosse. They ­were volunteering, traveling abroad, and working as unpaid summer interns. Heeding the advice of school and private counselors, they signed up for test-­prep courses, often taking the SAT or ACT multiple times. They took Advanced Placement courses and enrolled in private schools offering challenging curricula. Through such preparation t­ hese parents and students armed themselves to compete in the progressively credential-­driven environment of college admissions. For their part, the most prestigious colleges



W h y I t ­M a t t e r s

349

­ ere able to fill their entering classes with ever more accomplished w students who w ­ ere also racially diverse, thanks to the growing number of minority applicants from private schools and affluent families. Choices

No one in higher education had as their objective the increase in in­ equality that has occurred since 1970. Yet it occurred nonetheless. What implications follow from this growing in­equality depends, as so often is the case, on one’s point of view. Consider the three perspectives identified at the beginning of this chapter—­consumer, supplier, and policymaker. If you are the parent of a high school se­nior who aspires to enroll in one of the nation’s selective colleges, you do not need a book to tell you that this is a stressful, highly competitive pro­cess. What might be surprising, however, is the extent of the gaps in resources between the dif­fer­ent categories of colleges, or the ways in which students differ across colleges. As a rule, high academic standards are usually found at the same colleges that boast manicured lawns and well-­paid professors. Such colleges tend to offer majors in the traditional liberal arts disciplines rather than in applied fields designed to prepare students for immediate employment. To be sure, colleges differ as well in ways that have ­little to do with resources, most importantly in their religious affiliation or lack thereof. But ­these regularities should not obscure a crucial truth about college education: the student’s choices—of courses, activities, and effort—­are indispensable inputs to the production pro­cess. Both opportunities and distractions abound. Second, colleges themselves face choices. Most pointedly, the “selective” colleges, ­those who have the luxury of picking which applicants to accept, have the power to increase or decrease the share of their students who come from modest backgrounds. If colleges, singly or as a w ­ hole, want to reduce the degree of economic stratification that exists across the spectrum of colleges, they must act accordingly. ­There is no shortage of pos­si­ble remedies. As economists

350 C onsequences

Caroline Hoxby and Sarah Turner have demonstrated, ­there are hundreds of low-­income, highly able high school se­niors across the country, many of them outside metropolitan areas, who are t­ here to be contacted and informed that colleges want them and that financial aid is available. 8 To identify some of ­these students, colleges’ admissions offices could redirect a few of their visits each year from affluent suburbs to nontraditional recruitment areas. In evaluating applications, they could do more to neutralize the advantages of affluence, such as by giving less weight to experiences, like unpaid internships, that are more accessible to affluent applicants. And, in making financial aid offers, they could take steps to lessen or eliminate the debt burden on the neediest students. Lastly, they could reduce the preference they give to legacies. That most of the selective colleges continue to f­ avor legacies reveals that “excellence” must not be the sole institutional objective. Put another way, if colleges are to be compared to churches, it appears that caring for the congregation logically extends to the next generation as well. If any college wishes to take steps to increase their share of low-­income students, it must be willing to sacrifice other objectives. That might mean scrimping on renovations, professors’ salaries, or additions to the endowment, or it might mean turning in less impressive statistics to U.S. News. It is unreasonable, however, to expect colleges to do much re­distribution beyond what is in their own private best interest. We turn, therefore, to the third set of choices, t­ hose in the realm of public policy. U.S. public policy t­oward higher education does suggest a consensus that too much in­equality is a bad ­thing. States have long played the major role in supporting higher education, in the form of institutional appropriations and across-­the-­board tuition discounts, and t­ hese have been augmented in recent years with expensive merit aid programs, like Georgia’s HOPE scholarship. If states wanted to address in­equality within their systems, they could certainly restructure their programs to take financial need into account. At the federal level, the issues are well known and options are regularly debated. Several policy levers are available. Congress could certainly put more money into Pell Grants, making it pos­si­ble for



W h y I t ­M a t t e r s

351

low-­income students to attend more than just the least expensive colleges. Or the government could pursue the idea of rewarding affordability, high graduation rates, and good rec­ords of enrolling low-­income students when it doles out financial aid funds. Devising metrics that do not penalize colleges for admitting low-­income students is essential, however.9 Such remedies are easier to list than implement, of course. For the state or federal government, expanding redistributive policies would require cutting other programs or raising taxes, two decidedly unpromising paths in the current po­ liti­cal environment. Merit and In­equality

­ very college was founded for a reason. The found­ers, among them E philanthropists, religious organ­izations, and state legislatures, invariably saw colleges as benign instruments for advancing the greater purposes of sustaining that which was good in society and creating the conditions for making ­things better. Historian Samuel Eliot Morison tells us that Harvard College was just one component in a larger set of initiatives, including common schools and printing presses, which w ­ ere “made by the ruling class of New ­England and supported by the ­people at large.”10 The fact that access to college has been largely confined to c­hildren of the wealthy and influential should surprise no one, since that was their original purpose. As a practical ­matter, colleges have performed much the same role in Amer­i­ca that they have everywhere in modern socie­ties—to educate the ­children of the ruling upper-­middle class and, in effect, to reproduce t­ hose elites. One need not be a Marxist to acknowledge this fact. At the same time, however, it would have been unwise to allow colleges simply to replicate elites from one generation to the next, without allowing entry by the clever and energetic young persons of modest means to ascend into positions of influence. Writing almost a half a ­century ago, sociologists Christopher Jencks and David Riesman reflected on the role of colleges in balancing the imperatives of reproduction and mobility. Higher education, they argued,

352 C onsequences

serves this country’s need “to prevent its elite from decaying into a hereditary aristocracy.”11 They pointed to two ave­nues in this American college system for upward mobility: the limited spots in elite colleges set aside for meritorious applicants regardless of economic status and places at nonelite colleges.12 As demonstrated by the new evidence on intergenerational mobility reviewed in Chapter 12, colleges in e­ very category enroll a handful of students who do indeed vault from the bottom to the top quintiles, though the numbers are small. To their credit, American colleges are more “meritocratic” than they ­were fifty years ago. Racial discrimination, once explicit and accepted, has virtually vanished. Objective mea­sures of academic merit have assumed a paramount role. Yet, as we have seen, the affluent and influential have a­ dopted strategies to give their c­ hildren ­every advantage in this new, merit-­based environment. The term “meritocracy” was coined by sociologist Michael Young in his 1958 satirical novel about a dystopian regime in which status was strictly determined by intelligence tests. Rather than being a vehicle for social mobility, this system worked instead to perpetuate the position of ­those in a position to do well on the tests.13 Ironically, more than a few of the developments documented in ­these pages bear a resemblance to that fictional outcome. The market for baccalaureate education has long exerted a quiet but pervasive influence on American society. ­A fter all, the country’s first college predated the Constitutional Convention by a c­entury and a half. As colleges proliferated and matured, they became an industry of im­mense importance. In fact, it is difficult to think of another American industry whose total impact on the country has been larger. This industry trained inventors, entrepreneurs, technocrats, jurists, legislators, and presidents. Also teachers, accountants, nurses, and real estate agents. The industry both solidified and unsettled social hierarchies. It s­ haped millions of lives, including ­those of notable gradu­ates like Sonia Sotomayor and Pat Conroy. Over the de­cades the industry felt the force of sweeping changes. The modes of production practiced inside the colleges slowly absorbed the fruits of



W h y I t ­M a t t e r s

353

successive technological revolutions—­from electric lights to personal computers. Colleges witnessed firsthand a revolution in civil rights as well as widespread changes in gender roles. Immigration profoundly changed the racial and ethnic makeup of the population, and college enrollments. Yet all the while the structure of this industry remained remarkably unchanged. In par­tic­u­lar, a durable hierarchy has undergirded this structure, a hierarchy mea­sured in prestige and wealth. This hierarchy stretches from the scores of unselective colleges, both public and private, at one end, to a handful of elite colleges at the other. Colleges ­were and are unequal—in the money they take in, the money they are able to put aside, the money they spend, and the quality of their facilities, faculty, and students. This book offers clear evidence that ­these disparities have grown deeper. In assessing ­these changes, we would do well to keep in mind the potential for good that colleges as a ­whole have to serve the country, as Thomas Jefferson wrote, to train ­future leaders who would be “called to that charge without regard to wealth, birth, or other accidental condition ­ ill be a continuing challenge for Amer­i­ca’s or circumstance.”14 This w market for colleges.

354 C onsequences

Appendix Table A.1 Shares by College Category of Total Undergraduate Enrollment in 1,157 Four-­Year Institutions Table A.2 The Dwindling Share of Places at Elite Colleges

Notes References Acknowl­edgments Index

Appendix

Table A.1  Shares by College Category of Total Undergraduate Enrollment in 1,157 Four-­Year Institutions Type and percentile of 1970 SAT Private

Public

HBCUs

1972

1989

2008

2013

0–25

0.038

0.042

0.051

0.053

25–50

0.044

0.043

0.050

0.051

50–60

0.036

0.033

0.031

0.030

60–70

0.032

0.031

0.032

0.031

70–80

0.030

0.030

0.028

0.027

80–85

0.013

0.013

0.012

0.011

85–90

0.018

0.017

0.017

0.017

90–95

0.025

0.024

0.024

0.024

95–98

0.019

0.019

0.016

0.015

98–99

0.010

0.010

0.010

0.010

99+

0.010

0.010

0.008

0.008

0–50

0.372

0.378

0.372

0.379

50–80

0.205

0.205

0.200

0.200

80–90

0.074

0.072

0.071

0.071

90+

0.038

0.039

0.041

0.040

Private

0.010

0.011

0.010

0.009

Public

0.024

0.025

0.025

0.023

1.000

1.000

1.000

1.000

Source: IPEDS, Delta Cost Proj­ect; http://­w ww​.­deltacostproject​.­org​/­delta​-­cost​-­project​ -­d atabase. Note: Covers the 1,157 institutions that awarded bachelor’s degrees for which data on standardized tests were available circa 1970. See the online appendix for a description of this sample. Numbers in the table show the undergraduate enrollment in each college category as a share of all full-time, degree-seeking undergraduates in those 1,157 four-year institutions in the fall of the indicated year. These proportions are used to calculate enrollment-weighted national averages in all figures showing deviations from the national mean.

Table A.2  The Dwindling Share of Places at Elite Colleges 1972

1989

2010

Undergraduate enrollment in 2-­and 4-­year institutions (a)

7,964,380

11,742,531

18,078,672

Undergraduate enrollment in 4-­year institutions (a)

5,208,700

6,591,642

10,397,797

65.4%

56.1%

57.5%

Undergraduate enrollment in 4-­year colleges, full-­time (a)

4,178,724

4,984,995

8,086,375

Undergrad enrollment in 1, 157 4-­year colleges, full-­time (b)

3,784,008

4,415,155

5,872,606

Enrollment in top-5% private colleges (using 1970 SAT percentiles) (b)

150,345

171,920

199,108

As ­percent of 4-­year college enrollment

3.60%

3.45%

2.46%

Enrollment in top-1% private colleges (using 1970 SAT percentiles) (b)

39,557

43,502

49,034

As ­percent of 4-­year college enrollment

0.95%

0.87%

0.61%

As ­percent of 2-­and 4-­year institutions

Note: Percentages show, as a percentage of full-­time four-­year college enrollment, the enrollment in fifty-­eight colleges ranking in the enrollment-­weighted top 5 ­percent of colleges in 1970 by average SAT and the comparable percentage based on the fifteen colleges in the top 1 ­percent. Source for enrollments of top fifty-­eight colleges: sample of colleges that existed as four-­year institutions from 1970 to 2010 for which SAT data w ­ ere available circa 1970. Sources: (a) U.S. Department of Education, Digest of Education Statistics 2012, ­Table 240, “Total undergraduate fall enrollment in degree-­g ranting institutions.” Figures for 1972 estimated using the ratio of undergraduates to total enrollment for totals in ­Table 223. Digest of Education Statistics 2013, ­Table 303.70, “Total undergraduate fall enrollment in degree-­g ranting postsecondary institutions, by attendance status, sex of student, and control and level of institution: Selected years, 1970 through 2023.” (b) Colleges for which SAT data ­were available circa 1970 and operated throughout the period. See text.

Notes

1: Unequal Colleges

1. Sue Murphy, “1976: Smaller, More Diverse,” Daily Prince­tonian, September 6, 1972, 1; Kerry North, “Bowen Addresses Freshman Assembly,” Daily Prince­tonian, September 6, 1972, 1. 2. Sotomayor (2013), 135. 3. Sotomayor (2013), chaps. 17 and 18. 4. U.S. Department of Education, Digest of Education Statistics 2013 (May 2015), T ­ able 305.10, 424. 5. Integrated Postsecondary Education Data System (IPEDS), National Center for Education Statistics, accessed December 12, 1915, http://­nces​.­ed​.­gov​ /­ipeds​/­datacenter​/­Default​.­a spx. 6. David Leonhardt, “As Wealthy Fill Top Colleges, New Efforts to Level the Playing Field,” New York Times, April 22, 2004; Jeffrey Selingo and Jeffrey Brainard, “The Rich-­Poor Gap Widens for Colleges and Students, Chronicle of Higher Education, April 7, 2006, accessed May 25, 2017, www​.­chronicle​ .­com​/­a rticle​/­T he​-­R ich​-­Poor​- ­Gap ​-­Widens​-­for​/­2052; Jordan Weissman, “Amer­i­ca’s Top Colleges Have a Rich-­K id Prob­lem,” Atlantic Monthly, May 24, 2013, reported that 70 ­percent of the students at the nation’s most com­ petitive colleges come from the top quarter of the income distribution; Richard Pérez-­Peña, “Generation ­L ater, Poor Are Still Rare at Elite Colleges,” New York Times, August 25, 2014; Amanda Cox, Kevin Quealy, and Gregor Aisch, “A Bigger Affluence Gap Than Expected at Elite Colleges,” New York Times, January 19, 2017, 3. See also Dan Golden, “Extra Credit: At Many Colleges, The Rich Kids Get Affirmative Action,” Wall Street Journal, February 20, 2003.

7. Based on examination of tax return data for parents of college students born in 1991. The top 1 ­percent included parents with incomes of $630,000 or more. T ­ hose in the lowest 60 ­percent had incomes of $65,000 or less. 8. Intergenerational mobility is mea­sured by the “intergenerational earnings elasticity,” the coefficient in a regression of the logarithm of a son’s adult earnings on the logarithm of his ­father’s earnings. See, for example, Solon (2002), 62, or Corak (2013), 3. 2: System, Industry, or Crazy Quilt?

1. Robert Stevens quoting an unnamed “commentator”; Bowen, Kurzweil, and Tobin (2005), 66n87. 2. U.S. Department of Education, Digest of Education Statistics 2013, ­Table 317.40, accessed May 11, 2017, https://­nces​.­ed​.­gov​/­pubsearch​/p ­ ubsinfo​.­a sp​?p ­ ubid​ =­2015011. 3. Writing in 1980, sociologist David Riesman noted the reluctance of universities to differ too much from the norm, worried that students, also fearful of signing up for a collegiate program too far from the ordinary, simply would not come. 4. DiMaggio and Powell (1983). 5. See Brint et al. (2009), esp. 630–633. The authors identify this line of thought as the “new institutionalism.” 6. Among orga­n izational sociologists, this idea is associated with adherents to the “old institutionalism.” Brint et al. (2009) explains the per­sis­tence of such distinguishing features in the context of college curricula by noting the highly decentralized nature of colleges and academic disciplines. They note that their findings regarding the diffusion of curricular changes across colleges are largely but not wholly consistent with the new institutionalism: “Unlike the new institutionalism, we have shown that models embraced by influential actors in the field do not necessarily replace, or even partially replace, existing models. Instead, in highly decentralized and loosely controlled orga­n izational fields, where costs for non-­compliance are low, organ­i zations can select among competing models without penalty”; Brint et al. (2009), 631. 7. The descriptions are based on information gleaned from college cata­logs for 1970 and a well-­k nown college guide of the day, Cass and Birnbaum’s Comparative Guide to American Colleges, which featured entries for each of some 1,288 colleges, including information on size, location, founding date, admissions, academic programs, and student life. 8. Burke (1982), 222. 9. Cass and Birnbaum (1972), 278–279.

362

NOTES TO P AGES 1 0 – 1 9

10. Standardized ACT and SAT aptitude test scores for 1970 w ­ ere combined and recentered to be comparable to con­temporary SAT scores. Based on their average 1970 scores, colleges w ­ ere ranked and then assigned to percentiles based on their total number of undergraduates in that year. 11. Illinois State University Bulletin, 1970–1971, 11. 12. Majors ­were deemed vocational, or professional, if the courses offered in them, on graduation, might reasonably meet an employer’s expectations for job-­related skills. Also included in this category was home economics. 13. Binstock (1970), quotation on 310; see also 311–313. 14. Cass and Birnbaum (1972) do not specify the year for which their tuition and fees figures are based. Since scores and faculty compensation figures are based on the 1970–1971 year, it seems reasonable to assume the tuition and fees numbers apply to that year or a nearby year. 15. The cata­log included this policy statement regarding fraternities and sororities: “Acting upon the Student Senate recommendation, the University Council concurred that the University adopt a policy of neutrality ­toward social fraternities and sororities. Consequently, hereafter the University ­w ill neither oppose nor encourage its students to form or join such social fraternities or sororities as may be established in the town of Normal. However, the University w ­ ill not invite social fraternities or sororities to its campus, nor ­w ill it register them as student organ­i zations or accord them a status that would involve or suggest approval, regulation, or support.” (Illinois State University Cata­log, 1970–1971, 31) 16. Cass and Birnbaum (1972), 278–279. 17. Boston College Bulletin, 1969–1970, iv. 18. The requirement was more stringent for students in nursing and education, who had to take more than two courses in theology. 19. Course titles included “History of Religions,” “Modern Judaism,” “Sociology of Religion,” “Challenges to Con­temporary Man,” and “Law and Morality.” 20. Cass and Birnbaum (1972), 60–61. 21. Ibid. 22. Ibid. 23. Steiner, Eymonerie, and Woolf (1971). 24. The college offered seven majors in the business school: accounting, economics, finance, management, management and computer science, marketing, and general business. For the purpose of comparing majors across three colleges, all of the business majors are grouped together. 25. Cass and Birnbaum (1972), 60–61. 26. To be sure, ­there ­were many more Methodist-­a ffiliated colleges than ­those carry­ing the name “Wesleyan,” including Boston University, Duke, and Southern Methodist University.

NOTES TO P AGES 1 9 – 2 3

363

27. Tennessee Wesleyan College Bulletin, 1970–1972, 3. 28. Cass and Birnbaum (1972), 642. 29. Ibid. 30. Mills (1963), 313. 31. Brint, Riddle, and Hanneman (2006), ­Table 1. 32. Hoxby (2009). 33. For some purposes, where such fine gradations are e­ ither less impor­tant or unattainable, some categories are collapsed, but in all such cases, e­ very category corresponds exactly to one of the seventeen or is the sum of two or more of the seventeen. Details about classifying colleges appear in this chapter’s last section. 34. See Figures 5.7 and 5.5 for expenditures and endowments, respectively, in Chapter 5. 35. ­Because the categories used in the entire study have finer gradations on the private side of the market, and b ­ ecause I found enough college cata­logs to do so, I deci­ded that noteworthy differences by cluster would be easier to spot, if they exist at all, by comparing thinner slices on the private side. 36. As explained in the last section of this chapter, the construction of the seventeen categories used in most of the book was determined in part by the availability of Freshman Survey data. The five illustrative clusters employed ­here and in Chapter 7 w ­ ere constructed from t­ hose seventeen. 37. Cata­logs of Shepherd College (West ­Virginia), Armstrong State College (Georgia), and Kansas State University. 38. By 2010 Ohio State’s average student SAT score had increased by more than a hundred points, well above the median, but for the purpose of comparisons that include Ohio State in this book, that university remains grouped where it was in 1970, which follows the method applied to all of the groupings used in comparisons over time. 39. University of Kentucky Bulletin, 1970–1971, 8, accessed February 18, 2017, http://­eris​.­u ky​.­edu​/­catalog​/­xt70p26q0b4m ​_­1​?.­ 40. University of Kentucky Bulletin, 1970–1971, accessed May 5, 2017, http://­eris​ .­u ky​.­edu​/­catalog​/­xt70p26q0b4m ​_­1​?­, and The Ohio State University Bulletin, 1970–­1971, accessed May 5, 2017, http://­registrar​.­osu​.­edu​/­scheduling​/­old​ _­book3​_ i­ nfo​/­course​_­catalog​_­1970​_­1971​.­pdf. 41. Bulletin of the Citadel, 1970–1971, https://­a rchive​.­org​/­details​/­catalogofcitadel​ 1970cita​.­ 42. Conroy (2002), 101. According to Conroy, just 83 cadets graduated out of a starting class of 240. 43. See Chapter 10. 44. Percentages are weighted averages of figures in Cass and Birnbaum, provided by gender, using full-­time enrollments of men and ­women for weights.

364

NOTES TO P AGES 2 3 – 3 0

45. Cass and Birnbaum (1972), 482, 615–616. 46. James Howell, historical scores for Ohio State, accessed June 9, 2015, www​ .­jhowell​.­net​/­cf​/­scores​/­OhioState​.­htm. 47. Also included among high-­SAT public institutions are the country’s most famous ser­vice academies, but their 1970 cata­logs ­were not included in the repository used h ­ ere. 48. Founded as the Georgia School of Technology in 1885, it began with an enrollment of 129; see Marla Edwards and John D. Toon, “Georgia Institute of Technology (Georgia Tech),” New Georgia Encyclopedia, accessed May 7, 2016, www​.­georgiaencyclopedia​.­org​/­a rticles​/­education​/­georgia​-­institute​ -­technology​-­georgia​-­tech. 49. University of California, Davis, General Cata­log, 1970–1971, 10. 50. The College of William and Mary General Cata­log, 1970–1971, 3; https://­a rchive​ .­org ​/­details​/ ­bulletinofcolleg657coll. 51. Its name in 1970 was Howard Payne College. For the most part, institutions are referred to by their current name. 52. Bulletin of Howard Payne College 1970–1971, accessed May 25, 2017, https://­ texashistory​.­u nt​.­edu​/­a rk:​/­67531​/­metapth46512​/­​?­q​=­howard%20payne%20 college%201970. 53. The 2014 website included almost identical wording: “A personal commitment to Jesus Christ as Savior is the controlling perspective from which the educational enterprise is carried out.” 54. Binstock (1970), 320–321 writes: “The Catholic colleges run the gamut of types of institutions and maintain constituencies from all class levels and diverse academic backgrounds and abilities, representing a kind of parallel but smaller mainstream within the broader secular one of American higher education. But the denominational colleges tend to cluster together, limited in funds, offering limited and concrete occupational academic programs to a narrow range of constituencies.” 55. Binstock (1970), 365–366. 56. Oglethorpe College Bulletin 1970–1971, p. 9, accessed May 3, 2017, https://­ archive​.­org​/­details​/­oglethorpecolleg7071ogle. 57. University of New Haven, Cata­log 1970–1971, Day Division, p. 3, accessed May 3, ­ etails​/­u niversityofne197172univ. 2017, https://­a rchive​.­org​/d 58. York College of Pennsylvania General Cata­log 1970–1971, 5, accessed May 3, 2017, https://­a rchive​.o ­ rg​/­stream​/­yorkcollege19701971york#page​/­n11​/­mode​/­2up. 59. Bowdoin College Cata­logue, accessed May 5, 2017, http://­d igitalcommons​ .­bowdoin​.­edu​/­course​-­catalogues​/­252​/.­ 60. Bulletin of the California Institute of Technology: Information for Students 1970– 1971, p. 117, accessed May 5, 2017, http://­caltechcampuspubs​.­l ibrary​.­caltech​ .­edu​/9 ­ 8​/1 ­ ​/­1970​-1 ­ 971​.­pdf.

NOTES TO P AGES 3 0 – 3 4

365

61. Davidson College Cata­log, 1970/71, accessed May 5, 2017, https://­a rchive​.­org​ /­details​/d ­ avidsoncollegec19701971. 62. Brint et al. (2009), 607. 63. Brandeis University Bulletin 1970/1971, 5, accessed May 5, 2017, https://­a rchive​ .­org​/­details​/­generalcatalog7071bran. For a thorough history of quotas at Harvard, Yale, and Prince­ton, see Karabel (2005). 64. Davidson College Cata­log, 1970/71, accessed May 5, 2017, https://­a rchive​.­org​ /­details​/d ­ avidsoncollegec19701971. 65. Allen and Jewell (2002), 246–248. 66. See Harris (1971). 67. Clotfelter (2004), 154. 68. Florida Memorial College Annual Cata­logue 1970–1972, 8, accessed May 3, 2017, https://­a rchive​.­org​/­stream​/­floridamemorialc28flor#page​/­n1​/­mode​ /­2up. 69. Oakwood College Bulletin, 1970–1971, 26–27, accessed May 5, 2017, https://­ archive​.­org​/d ­ etails​/­oakwoodbulletin197071oakw. 70. Geiger (2004), 14. 71. For example, about half of the gradu­ates of Harvard College in the 1600s went on to become ministers; see Handlin and Handlin (1970), 9. 72. To get an idea of the material colleges in the nineteenth c­ entury expected applicants to have covered, see sample questions Trinity College (North Carolina) sent to prospective applicants as preparation for that college’s entrance exam, in Amy McDonald, “Getting Past the Gates,” Dev­il’s Tale (blog), January 5, 2010. http://­blogs​.­l ibrary​.­duke​.­edu​/­r ubenstein​/­2010​/­01​/­05​ /­getting​-­past​-­the​-­gates​/.­ 73. See Veysey (1973), 31; Handlin and Handlin (1970), 30, 36, 41, and 61; Bledstein (1976), 223–247. C ­ olleges after the Civil War w ­ ere reformed in part by students themselves, by way of associations that encouraged intellectual debate and upstanding character. ­These organ­i zations included many of the Greek letter fraternities still found on many colleges campuses t­ oday; see Bledstein (1976), 252–254. 74. Handlin and Handlin (1970), 25. 75. Geiger (2000), 146–147. 76. Ibid. 77. Bledstein (1976), 296. 78. Chapter 7 discusses the operation of commercial sports enterprises. 79. See Goldin and Katz (1999), 39–48, and Clotfelter (2004), 237n10. 80. Goldin and Katz (1999), 40. 81. Veysey (1973), 44. 82. Goldin and Katz (1999), 44. 83. Goldin and Katz (1999), 39–48.

366

NOTES TO P AGES 3 4 – 4 0

84. The Public Health Ser­v ice Act (1944), which launched a period of tremendous growth in spending on public health ­a fter World War II, was a significant step t­ oward the creation of the National Institutes of Health; see National Institutes of Health, accessed January 25, 2009, http://­history​.­n ih​ .­gov​/­exhibits​/­history​/­docs​/­page​_ ­06​.­html). See also Morison (1965), 419 and 464; National Science Foundation, accessed January 25, 2009, www​.­nsf​.­gov​ /­about​/h ­ istory​/.­ 85. For discussions of the multiplicity of functions in American research universities, see Goldin and Katz (1999). 86. Clotfelter et al. (1991), 31. 87. Ibid. 88. Karabel (2005), 45, 87. 89. Karabel (2005), 89. 90. For a discussion of desegregation in higher education, see Clotfelter (2004), chap. 6. 91. Glazer (1973), 90–91. 92. Karabel (2005), 520–521. 93. Andy Thompson, “20% of White Members of Harvard’s Class of 2014 Followed a Parent ­There,” Chronicle of Higher Education, May 27, 2014, accessed May 5, 2017, http://­w ww​.­chronicle​.­com​/ ­blogs​/­t icker​/­20​- ­of​-­white​ -­members​-­of​-­harvards​-­class​-­of​-­2014 ​-­followed​-­parents​-­t here​-­survey​-­shows​ /­78607. 94. Michael Olivas, quoted in Adam Liptak, “A Hereditary Perk the Founding ­Fathers Failed to Anticipate,” New York Times, January 15, 2008, http://­w ww​ .­nytimes​.­com​/­2008​/­01​/­15​/­us​/1 ­ 5bar​.­html, accessed April 28, 2017. Chapter 6 takes up this issue in more detail. 95. Higher education: OECD, Education at a Glance 2014 (October 2014), ­Table B2.1, accessed March 2, 2015, www​.­oecd​.­org ​/­edu​/­Education​-­at​-­a​- ­Glance​ -­2014​.p ­ df; Health care: World Bank, 2013, accessed June 15, 2015, http://­data​ .­worldbank​.o ­ rg​/­indicator​/S ­ H​. ­X PD​.­TOTL​. Z ­ S​/.­ 96. Research on higher education has shown that peer effects are impor­tant in a variety of ways, both good and bad. See, for example, Sacerdote (2001); Kremer and Levy (2008); Carrell, Malmstrom, and West (2008); or Carrell, Hoekstra, and West (2015). 97. See Figure 4.3 in Chapter 4. 98. Economists refer to this practice, like that engaged in by airlines, as “price discrimination.” 99. ­There are other unusual aspects of colleges and universities, especially about their internal organ­i zation, but t­ hese are not directly pertinent to the study at hand. Tenure and the unusual freedom of faculty are two examples. See Coleman (1973). Positional competition is discussed below.

NOTES TO P AGES 4 1 – 4 6

367

100. As noted below, for some analyses one of t­ hese groups was further divided into three subgroups. 101. See Appendix ­Table A.1 to see the share of all four-­year college students enrolled in colleges in each of the seventeen categories by sample year. 102. The generally smaller enrollments of private colleges are reflected in the average number of first-­year students who w ­ ere surveyed. Assuming first-­year students constitute roughly a fourth of all undergraduates, t­ hese sampled colleges would have had undergraduate enrollments ranging from about 3,500 to 11,200 for private institutions and 17,600 to 47,400 for public ones in the sample. 3: Snapshot, Circa 1970

1. Report of the President’s Commission (1970), 17, 233–266. 2. John Kifner, “4 Kent State Students Killed by Troops,” New York Times, May 5, 1970, 1, 17; Jerry M. Lewis and Thomas R. Hensley, “The May 4 Shootings At Kent State University: The Search For Historical Accuracy,” Ohio Council for the Social Studies Review 34 (Summer): 9–21, accessed May 19, 2017, https://­web​.­a rchive​.­org ​/­web​/­20080509131525​/­http://­dept​.­kent​.­edu​ /­sociology​/ ­lewis​/­LEWIHEN​.h ­ tm. 3. Chronicle of Higher Education, August 31, 1970, 1. 4. Ibid. 5. Jean E. Engelmayer and Melissa I. Weissberg, “Reflecting on the 1969 Student Strike,” Harvard Crimson, April 9, 1984, accessed May 5, 2017, http://­w ww​.­thecrimson​.­com​/­a rticle​/­1984​/­4/​ ­9​/r­ eflecting​- ­on​-t­ he​-­1969​ -­student​-­strike​/.­ 6. Columbia History, accessed May 5, 2017, http://­w ww​.c­ olumbia1968​.­com​ /­history​/.­ 7. For the Duke Allen Building takeover, February 13, 1969, see Duke University Libraries, “Student Activism at Duke University,” accessed ­ 932740. May 5, 2017, http://­g uides​.­l ibrary​.­duke​.­edu​/­c​.­php​?­g ​=­289450&p​=1 For an account of the University Hall takeover at Harvard, see Jean E. Engelmayer and Melissa I. Weissberg, “Reflecting on the 1969 Student Strike,” Harvard Crimson, April 9, 1984, accessed May 5, 2017, http://­w ww​ .­thecrimson​.c­ om​/a­ rticle​/1 ­ 984​/­4/​ ­9​/­reflecting​- ­on​-­the​-­1969​-­student​-­strike​/.­ 8. Stadtman (1973), 9. 9. On April 20, 1969, black students who had taken over Cornell’s Willard Straight Hall left, some with r­ ifles and ammunition. Tyler Alicea, Cornell Daily Sun, http://­cornellsun​.c­ om​/ ­blog​/­2014​/­04​/­18​/­students​-­took​- ­over​ -­w illard​-­straight​-­hall​- ­45​-­years​-­ago​/­. For Columbia, April 23, 1968, see www​ .­columbia1968​.­com​/­history​/­. Accessed June 13, 2014.

368

NOTES TO P AGES 4 8 – 5 5

10. For Wisconsin protests, including firebombing of Sterling Hall in 1970, see https://­w ww​.­l ibrary​.­w isc​.­edu​/­a rchives​/­exhibits​/­sterling​-­hall​-­bombing​-­of​ -­1970​/­. Accessed May 5, 2017. 11. Gerald Tebben, “Columbus Mileposts: May 6, 1970: Riots Force Closure of Ohio State’s Campus,” Columbus Dispatch, May 6, 2012, accessed June 17, 2015, www​.­d ispatch​.­com​/­content​/­stories​/­local​/­2012​/­05​/­06​/­r iots​-­force​ -­closure​-­of​-­ohio​-­states​-­campus​.­html. 12. Report of the President’s Commission (1970), 18. 13. “Presidents Prepare for Disorders, Seek Ways to Keep Campuses Open,” Chronicle of Higher Education, August 31, 1970, 1. Statement was by Richard A. Harvill, president of the University of Arizona and of the land-­grant college association. 14. “Threat Seen to Academic Freedom,” New York Times, July 13, 1970, 25. Brewster added, “But the one that worries me most, I think, is the conventional, old-­fashioned issue of academic freedom and the use of established power to try to make the university an instrument for conformity rather than an instrument for exploration.” Accessed May 5, 2017, http://­w ww​ .­nytimes​.­com​/­1970​/­07​/­13​/­a rchives​/­threat​-­seen​-­to​-­academic​-­freedom​.­html​?­​ _­r ​=1 ­. 15. Report of the President’s Commission (1970), 2. 16. “Grading Softer? Students or Teachers Better? What­ever the Cause, Grades Are G ­ oing Up,” Chronicle of Higher Education, March 1, 1971, 1. 17. The most comprehensive documentation of grade inflation known to the author is produced by Stuart Rojstaczer on his website, accessed July 13, 2016, www​.­gradeinflation​.­com​/­. He attributes the rapid rise in average grades a­ fter 1963 in large part to the pressure exerted by the Vietnam draft. 18. In 1973, David Riesman wrote, “The abdication of the adults from any monitoring of campus residences occurred with extraordinary speed in the last five years.” Riesman, “Commentary and Epilogue,” in Riesman and Stadtman (1973), 438. 19. Cheit (1971), 17. 20. Baumol and Eymonerie (1970), 185. 21. Ibid. 22. Jellema (1971), 25. 23. Corbin Gwaltney, “Many Private Colleges Facing Financial Crisis,” Chronicle of Higher Education, May 3, 1967, 1. 24. Robert Jacobson, “Small Colleges Face Shortage of Students,” Chronicle of Higher Education, August 31, 1970, 1. 25. Steiner, Eymonerie, and Woolf (1971), 226. 26. John Crowl, “100 Institutions Reported Facing Fiscal Disaster,” Chronicle of Higher Education, September 27, 1971, 1, 4. See also Jellema (1971).

NOTES TO P AGES 5 5 – 5 9

369

27. Cheit (1971), viii, 137. For a careful study that examines liberal arts colleges in the 1980s, see Breneman (1994). 28. Cass and Birnbaum (1972), xix and entries for indicated colleges. All of the colleges listed ­were in the sample of 188 except for Boston College, which was one of the three illustrative colleges in Chapter 2. In our notation, ­those in the 188 ­were in the following groups: Texas Christian, private 60th to 70th percentile; College of Charleston, public 50th to 80th; Catholic University, private 70th to 80th; Iowa State, public 80th to 90th; Georgia Tech, public 90th-­plus; Grinnell, private 90th to 95th; UC Santa Cruz, public 90th-­plus; Prince­ton, private 98th to 99th. 29. Schudson (1972), 45. 30. The test also was a reasonably good predictor of first-­year college grades (R 2 = .15) (Lemann 1999). 31. In 1960 the SAT administered 400,000 tests, compared to 133,000 for the ACT. By 1970 the two w ­ ere about equal, 1,000,000 for the SAT and 993,000 for the ACT; see Schudson (1972), 58. 32. See Schudson (1972), 35. 33. Ibid., 64. 34. College Board president Frank Bowles in 1960, quoted in Schudson (1972), 62. 35. See, for example, Reynolds (1927), 23–24, Handlin and Handlin (1970), 36, and Burke (1982), 214. 36. Out of the 470 whose parents w ­ ere not deceased, some 63 had ­fathers with blue collar occupations. That number was exceeded by ­those whose f­ athers ­were doctors, l­ awyers, or manufacturers; see Angell (1928), 232, T ­ able G. 37. The parents of four-­fifths of ­these undergraduates ­were homeowners. By comparison the rate of homeownership in the state in 1920 and 1930 was 59 ­percent. Angell (1928), 234, ­Table J, and U.S. Census, Historical Census of Housing ­Tables, accessed June 10, 2014, www​.­census​.­gov​/­hhes​/­w ww​/­housing​ /­census​/h ­ istoric​/­owner​.h ­ tml. 38. Jencks and Riesman (1968), 96. 39. Reynolds (1927), 24. 40. For a list of the colleges included in ­these samples, see Reynolds (1927), 3–7. 41. The report stated, “For the g­ reat majority of our boys and girls, the kind and amount of education they may hope to attain depends, not on their own abilities, but on the f­ amily or community into which they happened to be born.” Quoted in Bowen, Kurzweil, and Tobin (2005), 34. 42. Harris (1972), 61. 43. Ibid., 61. 44. Hansen and Weisbrod (1969), ­Table 5, 183. 45. Bourdieu (1973); Bowles and Gintis (1976).

370

NOTES TO P AGES 5 9 – 6 6

46. Jencks and Riesman (1968), 97. 47. Included in the 1909 survey w ­ ere ­these five colleges destined to be in the 99th and above percentile of 1970-­era SATs: Columbia, Harvard, MIT, Stanford, and Yale. Their average per student income in 1909 was $453. Next highest ­were Chicago, Cornell, Johns Hopkins, and Prince­ton, whose average was $344. 48. Some recent studies that incorporate athletics into a broader consideration of colleges and universities include Shulman and Bowen (2002); Clotfelter (2011); Lindo, Swensen, and Waddell (2012); and Lindo, Siminski, and Swensen (2016). 49. Conroy (2002), 138. 50. Clotfelter (2011). 51. ­A fter 1981, the Ivy League colleges dropped out of Division I-­A . 52. See Lifschitz, Sauder, and Stevens (2014). 4: Outside Forces

1. For evidence on the college earnings premium over time, see OECD (2012), 105. For evidence more generally on income in­equality over time in the United States, see Aaronson and Mazumder (2008), 164. 2. Freeman (1976). 3. Levy and Temin (2007). 4. Expressed as ratios (multiplied by 100) to their incomes in 1967, the inflation-­adjusted incomes in 1981, 2006, and 2012, respectively, w ­ ere, for the ­m iddle 20 ­percent, 116, 140, and 132; for the top 5 ­percent, 101, 212, and 198; and for the top 1 ­percent, 96, 302, and 270 (reaching a peak of 323 in 2007). Growth rates ­here and elsewhere are exponential, based on the equation X1 = X0ert, where X0 is the initial value, X1 is the final value, and t is the number of years between them. See also OECD (2011), 23. Between the mid-1980s and late 2000s incomes in the bottom decile in the United States ­rose at 0.5 ­percent a year, compared to 1.9 ­percent a year for the top decile. 5. Average inflation-­corrected ­family incomes in 1981 and 2012, respectively, ­were, for the bottom quintile, $16,480 and $15,534; for the second, $35,837 and $38,184; and for the ­m iddle, $54,812 and $62,464. Source: U.S. Census Bureau, accessed January 29, 2014, www​.c­ ensus​.­gov​/h ­ hes​/­w ww​/­income​/­data​ /­historical​/­inequality​/­index​.­html. 6. Average income for the top 1 ­percent, as calculated by Piketty and Saez, in 2015 dollars was $359,462 in 1967, $356,917 in 1981, and $1,053,398 in 2015.Spreadsheet “TabFig2015prel.xls, sheet ‘­Table A4,’ ” accessed on May 6, 2017, attached to entry for “Income In­equality in the United States, 1913– 1998” with Thomas Piketty, Quarterly Journal of Economics, 118(1), 2003, 1–39.

NOTES TO P AGES 6 6 – 7 8

371

­ ables and Figures updated to 2015 in Excel format, (June 2016), accessed T May 6, 2017, http://­eml​.­berkeley​.­edu​/­∼saez​/.­ 7. The mea­sure that economists use is the elasticity of men’s earnings with re­spect to their ­father’s earnings. Aaronson and Mazumder (2008) pres­ent evidence for the United States over time. 8. Saez and Zucman (2014). About 160,000 families made up the top 0.1 ­percent of wealth holders in 2011, each having at least $20 million. 9. Wolff (2012), T ­ able 2; Wolff (2013); U.S. Census Bureau, Current Population Survey, Annual Social and Economic Supplements, accessed February 11, 2014, www​.­census​.­gov​/­hhes​/­w ww​/­income​/­data​/­historical​/­inequality​/­index​ .­html. 10. Beginning in 2000 list price tuition rates have risen much faster even than incomes of the top 5 ­percent families. 11. In 2009, applicants applied to a median of four colleges. Pryor et al. (2009). 12. See also Baum (2014), fig. 4, p. 6, and Clotfelter (1996). 13. In the post-2000 period, write Beaudry et al. (2013), 3, “having a BA is less about obtaining access to high paying managerial and technology jobs and more about beating out less educated workers for the Barista or clerical job.” See also Valletta (2016). 14. Duncan and Murnane (2011), 11. Their calculations are based on data from the Consumer Expenditure Survey. 15. See, for example, Clotfelter (1985), chap. 2. 16. Hoxby (2009), 16, notes this source of support. 17. Bound and Turner (2006). 18. ­Those institutions and their states are Dartmouth, NH; Brown, RI; Harvard, MA; Yale, CT; Columbia and Cornell, NY; Prince­ton, NJ; and Penn, PA. 19. Reflecting the current census methodology, Hispanic is considered an ethnicity, and nonwhites of Hispanic origin are counted in Hispanic as well as their racial category. Thus the classifications are not mutually exclusive, and some individuals may be counted in two categories. 20. See Clotfelter (2004), chap. 6. 21. The gap-­based segregation index between white and nonwhite students in public four-­year colleges and universities declined in the South from 0.99 in 1961 (indicating almost complete segregation) to 0.78 in 1976 to 0.35 in 1998. In the border states, the comparable index values w ­ ere 0.67, 0.33, and 0.22. This segregation index is the percentage gap between the maximum and a­ ctual exposure rate of whites to nonwhites, or (n-­E) / n, where n is the percentage nonwhite among all college students (the maximum pos­si­ble exposure rate) and E is the a­ ctual exposure rate, the percentage nonwhite in the average white student’s college; see Clotfelter (2004), T ­ able A6.1, 175. 22. Reflecting how terms have changed over the four de­cades, the 1972 survey offered students t­ hese racial categories: White / Caucasian; Black / Negro / Afro-​

372

NOTES TO P AGES 7 8 – 8 9

American; American Indian; Oriental; Mexican-­A merican ​/ Chicano; Puerto Rican-­A merican; and Other. 23. Enrollment-­weighted averages based on Freshman Survey responses for first-­year students entering the 188 colleges in the sample. 24. U.S. Department of Education, Digest of Education Statistics 2012, T ­ able 239, accessed April 29, 2017, https://­nces​.­ed​.­gov​/­programs​/­d igest​/­d12​/­tables​/­dt12​ _­239​.­a sp. 25. Ibid. 26. The enrollment-­weighted averages referred to ­here and in figures showing deviations from the national average use as weights the full-­time, degree-­ seeking undergraduates in all colleges in each of the 17 defined categories, based on calculations for all 1,157 institutions which existed as four-­year institutions in 1970 and for which 1970-­era achievement scores ­were available. See Appendix ­Table A.1 for t­ hese enrollment weights calculated for selected years. 27. Ibid., T ­ able 310, 448. 28. Ibid., T ­ able 239, 338. 29. See Buchmann and DiPrete (2006), 516. T ­ here remain gender differences in undergraduate majors, however, as noted by E ­ ngland (2010). Between 1971 and 1996, she finds that the sex segregation of fields of study for U.S. bachelor degree recipients fell, but then slightly r­ ose to 2006 (155). 30. Digest of Education Statistics 1972, 94. 31. Information on years colleges became coed taken from College Express, “Years that Men’s Colleges Became Coed,” accessed February 14, 2014, www​ .­collegexpress​.­com​/­l ists​/­l ist​/­years​-­that​-­mens​-­colleges​-­became​-­co​-­ed​/­366​/.­ 32. See Figure 4.2 in the online appendix, which shows the share of w ­ omen by college category and survey wave. 33. Wendell Cox and Jean Love, “40 Years of the US Interstate Highway System: An Analy­sis,” Highway and Motorway Fact Book, June 1996, accessed March 14, 2014, www​.­publicpurpose​.­com​/­freeway1​.­htm. 34. Calculations based on data from the U.S. Department of Transportation, Bureau of Transportation Statistics, accessed July 8, 2014, www​.­r ita​.d ­ ot​.­gov​ /­bts​/­sites​/­r ita​.d ­ ot​.­gov​.­bts​/­fi les​/­publications​/­national​_­transportation​ _­statistics​/­html​/­table​_ ­01​_­4 0​.­html; and the Census Bureau, Current Population Reports, accessed July 9, 2014, http://­eire​.c­ ensus​.­gov​/­popest​/­data​ /­counties​/­tables​/­CO​-­EST2001​-­12​/­CO​-­EST2001​-­12​- ­0 0​.­php. 35. Smithsonian Institution, Amer­ic­ a by Air, accessed March 14, 2014, http://­ airandspace​.­si​.­edu​/e­ xhibitions​/­a merica​-­by​-­a ir​/­online​/­flyacross​/­index​.­cfm. 36. According to Hoxby (2009), 8, the cost of a ten-­m inute coast-­to-­coast phone call (in 2005 dollars) was $25.91 in 1970, $3.97 in 1990, and $2.61 in 2005. According to the Federal Communication Commission’s Universal Ser­vice Monitoring Report for 2010, the cost per minute fell from 2005 to 2010, in

NOTES TO P AGES 9 1 – 9 6

373

constant dollars, by 3 ­percent, ­Table 7.8, accessed March 14, 2014, http://­ transition​.­fcc​.g­ ov​/­wcb​/­iatd​/­monitor​.­html. 37. Clotfelter et al. (1991), 98; Zumeta et al. (2012), 62. 38. See, for example, UNESCO Institute for Statistics, Global Educational Digest ­ ocuments​ 2011, accessed July 12, 2016, www​.­u is​.­u nesco​.­org​/­Education​/D /­ged​-2 ­ 011​- ­en​.­pdf. 39. Basic Educational Opportunity Grants ­were subsequently renamed Pell Grants. See also Zumeta et al. (2012), 69. 40. The law was the M ­ iddle Income Student Assistance Act of 1978. See Clotfelter et al. (1991), 101; Zumeta et al. (2012), 70. 41. The total expenditure, in 1988 dollars, increased from $10.2 billion in 1970 / 71 to $20.6 billion in 1980 / 81. Clotfelter et al. (1991), ­Table 4.4, 99. 42. Calculated from Zumeta et al. (2012), fig. 4.1, 77. See also Clotfelter et al. (1991), 100. 43. For an example of its calculation, see Clotfelter et al. (1991), 97. 44. Clotfelter et al. (1991) 90, 95. 45. Zumeta et al. (2012), 80. 46. By 2015 the inflation-­adjusted value of the maximum Pell Grant award was just 2 ­percent less than the corresponding amount in 1977: College Board (2015), fig. 24, 32. See Mettler (2014), 10–15, for a discussion of the funding of Pell Grants. 47. The percentage distribution for federal aid in 2014 / 15 was Pell Grants, 18.7 ­percent; veterans and military grants, 9.4 ­percent; work-­study, 1.0 ­percent; loans, 59.4 ­percent; and tax benefits, 11.3 ­percent: College Board (2015), fig. 6, p. 17. 48. Matthews (2012). 49. Including tuition and fees for three quarters, the total was $958.50. University of California, Berkeley, 1970–71 General Cata­logue, May 15, 1970, 66, accessed February 11, 2016, http://­d igitalassets​.­l ib​.­berkeley​.­edu​/­general​ catalog​/t­ ext​/­1970​_­1971​_ ­intro​.­pdf. 50. For each semester tuition was $112.50 and fees ­were $86.50. Cata­log of the University of North Carolina 1970, 26–27, accessed February 11, 2016, http://­ library​.­d igitalnc​.­org​/­cdm​/r­ ef​/c­ ollection​/­yearbooks​/­id​/­13526. Section 9 of Article IX of the state constitution states: “The General Assembly s­ hall provide that the benefits of The University of North Carolina and other public institutions of higher education, as far as practicable, be extended to the ­people of the State ­free of expense,” accessed January 22, 2016, www​.­ncga​ .­state​.n ­ c​.­us​/­L egislation​/­constitution​/­a rticle9​.­html. 51. Hansen and Weisbrod (1969). See Chapter 3 of this volume. 52. Kane et al. (2005), 101, 106. 53. Long (2011); Clotfelter, Hemelt, and Ladd (2016); Jenna Johnson, “U-­Va. to Scale Back Financial Aid Program for Low-­and Middle-­Income Students,”

374

NOTES TO P AGES 9 7 – 1 0 1

Washington Post, August 6, 2013, www​.­washingtonpost​.­com​/ ­local​/­education​ /­u​-­va​-­to​-­scale​-­back​-­fi nancial​-­a id​-­program​-­for​-­low​-­a nd​-­m iddle​-­income​ -­students​/­2013​/­08​/0 ­ 6​/3 ­ 3c424f8​-f­ ec6​-­11e2​-­9711​-­3708310f6f4d​_ ­story​.­html. 54. See, for example, Dynarski (2000), and Cornwell and Mustard (2008). 55. For evidence on the regressivity of lottery finance, see Clotfelter and Cook (1989), chap. 4. 56. Quoted in Doyle and Hartle (1985), 8, 9. 57. U.S. News and World Report, “About,” accessed July 12, 2016, www​.­usnews​.­com​ /­info​/­features​/­about​-­usnews. 58. Cass and Birnbaum (1972). 59. For national universities, ­there ­were fifteen dif­fer­ent mea­sures used in that year. For regional institutions, just fourteen. U.S. News and World Report, “1997 Amer­i­ca’s Best Colleges” (1997), 54. 60. In addition to checking for internal consistency, the magazine checked submitted figures with Moody’s Investors Ser­vice, a com­pany in the business of advising lenders, as well as the NCAA and the College Board. U.S. News and World Report, 1997 Amer­i­ca’s Best Colleges, 55. 61. Ibid., 55. 62. The Car­ne­g ie Foundation for the Advancement of Teaching devises the most widely used classification of colleges and universities, and U.S. News bases its own classification on that. Car­ne­g ie had switched Berea from Comprehensive II to Baccalaureate I. Machung (1998), 13. 63. Ibid., 15–16. 64. Griffith and Rask (2007). 65. Bowman and Bastedo (2009). 66. Monks and Ehrenberg (1999). 67. Some scholars of American higher education do believe that colleges’ sponsorship of football teams has been seen as a form of marketing for the college and that conference membership has been viewed by some colleges as a mark of status, a definition of a college’s reference group. See Lifschitz, Sauder, and Stevens (2014). 68. Belasco, Rosinger, and Hearn (2015) examine the decision by some colleges to make the SAT an optional rather than a mandatory part of applicants’ admissions dossiers. 69. According to the college’s then-­president, U.S. News announced in 2006 that, for a college that deci­ded not to submit information asked for in the survey and which they could not locate using published sources, it would assign a default value for the mea­sure equal to one standard deviation below the mean. Myers (2007). 70. Myers (2007). 71. Private interview, administrators, Sarah Lawrence College, June 22, 2014. Another case was Reed College, which chose not to participate in the 1995

NOTES TO P AGES 1 0 1 – 1 0 8

375

survey. In response, U.S. News initially moved Reed down to the bottom tier of colleges, but l­ ater, using data it collected itself, treated the college as before (Machung 1998), 16. 72. U.S. News, “National University Rankings,” accessed February 12, 2017, http://­colleges​.­usnews​.­rankingsandreviews​.­com​/­best​-­colleges​/­best​-­colleges​ /­rankings​/­national​-­u niversities; and Kevin Carey, “Introduction: A Dif­fer­ent Kind of College Ranking,” Washington Monthly, September–­October 2016, 19–22, 80, http://­wmf​.­washingtonmonthly​.­com​/­college​_ g­ uide​/­2016​/ ­W M​ _­2016​_ ­Embargoed​_ ­R ankings​.­pdf. 5: The In­equality Dividend

1. Matthew 25:29 (King James Version). 2. Merton (1968). Although Merton’s may have been the first application of the parable to social science, it was not, by his own report, its first adaptation as an empirical regularity. A scientist in 1875 applied Matthew 25:29 to the rate of dehydration of alcohols in organic chemistry. Merton notes, “Evidently the Matthew effect transcends the world of h ­ uman be­hav­ior and social pro­cess” (n50). 3. OECD (2011). 4. Alvaredo et al. (2013), fig. 1. Figures are for income including capital gains. Excluding capital gains, the share of the top 1 ­percent ­rose from 8 ­percent to 17 ­percent. 5. Saez and Zucman (2014). 6. See, for example, Atkinson, Piketty, and Saez (2011), Reardon (2011), and Wolff (2012). 7. Winston (2004), 339. See also Winston (1999), 17. 8. Another piece of evidence supporting the charitable interpretation one might add to Winston’s case for colleges as charities is the widespread practice of admissions preferences for minority students. 9. In 1917, four years ­a fter the income tax was enacted. 10. Wicksteed (1910), bk. 1, chap. 4, Library of Economics and Liberty, accessed January 8, 2017, www​.­econlib​.­org​/­l ibrary​/­Wicksteed​/­wkCS0​.­html. 11. Bowen (1980), 19. 12. This idea is noted by Geiger (2004), 4. 13. Quoted in Brint (2005), 25. 14. See also Clotfelter (1996), 253–255. 15. Bowen (1970), 81. 16. Bowen (1980), 20. 17. Geiger (2004), 17. 18. U.S. Department of Education, Digest of Education Statistics 2012, T ­ ables 232 and 234.

376

NOTES TO P AGES 1 0 8 – 1 2 1

19. Richard Pérez-­Peña, “Best, Brightest and Rejected: Elite Colleges Turn Away Up to 95%,” New York Times, April 8, 2014, accessed April 29, 2017, https://­w ww​.­nytimes​.­com​/­2014​/­04​/­09​/­us​/­led​-­by​-­stanfords​-­5​-­top​-­colleges​ -­acceptance​-­rates​-­hit​-­new​-­lows​.­html​?­​_­r ​= 0 ­. 20. ­These are the fifteen institutions in the 99th percentile and above SAT category in 1970, that is, the top 1 ­percent in the enrollment-­weighted top 1 ­percent of all U.S. colleges and universities. All are private. They are Amherst, Brandeis, Cal Tech, Columbia, Dartmouth, Harvard, Harvey Mudd, Haverford, MIT, Rice, Sarah Lawrence, St. John’s, Stanford, Swarthmore, and Yale. 21. Calculations based on sample of colleges that existed as four-­year institutions from 1970 to 2010 for which SAT data w ­ ere available circa 1970. For 1972 the estimated 39,557 undergraduates in the top 1 ­percent private colleges w ­ ere 0.95 ­percent of the estimated 4,178,724 full-­time undergraduates in four-­year institutions. In 2010 t­ here ­were an estimated 49,034 in top 1 ­percent colleges, 0.61 ­percent of the estimated 8,086,375 full-­time undergraduates. Top 1 ­percent based on the enrollment in the fifteen colleges ranking in the enrollment-­weighted top 1 ­percent of colleges in 1970 by average SAT. Source for total full-­time undergraduate enrollment in degree-­granting four-­year institutions: U.S. Department of Education, Digest of Education Statistics 2012, ­Tables 223, 240; Digest of Education Statistics 2013, ­Table 303.70. The growth in total enrollment reflects the addition of new four-­year colleges and growth among existing ones. Among the 1,157 colleges that operated as four-­year institutions in 1970 and for which SAT scores ­were available, the increase in total enrollment was 55 ­percent. 22. For 1972 the estimated 150,345 undergraduates in top 5 ­percent private colleges w ­ ere 3.60 ­percent of the estimated 4,178,724 full-­time undergraduates in four-­year institutions. In 2010 t­ here ­were an estimated 199,108 in top 5 ­percent colleges, 2.46 ­percent of the estimated 8,086,375 full-­time undergraduates. Top 5 ­percent enrollment is based on the enrollment in fifty-­eight colleges ranking in the enrollment-­weighted top 5 ­percent of colleges in 1970 by average SAT. The source for enrollments of top fifty-­eight colleges: sample of colleges that existed as four-­year institutions from 1970 to 2010 for which SAT data ­were available circa 1970. Source for total full-­time undergraduate enrollment in degree-­granting four-­year institutions: U.S. Department of Education, Digest of Education Statistics 2012, T ­ ables 223, 240; Digest of Education Statistics 2013, ­Table 303.70. See Appendix ­Table A.2. Riesman (1980, 67) notes that, in the de­cade following Sputnik, selective colleges, while they w ­ ere attracting more able students, did not increase their enrollments. Similarly, Bound, Hershbein, and Long (2009), 120, note that the supply for college places overall did not keep pace with demand over the period 1972–2004.

NOTES TO P AGES 1 2 3 – 1 2 4

377

23. Pérez-­Peña, “Best, Brightest and Rejected.” 24. So noted by Bound, Hershbein, and Long (2009). 25. Rates are average annual exponential growth rates. Sources of tuition, all accessed May 6, 2017, are: Harvard: Kharblog, http://­kwharbaugh​.­blogspot​ .­com​/­2005​/­02​/­educational​-­costs​.­html (1980) and College Calc, http://­w ww​ .­collegecalc​.­org​/­colleges​/­massachusetts​/­harvard​-­university​/(­ 2000); Wellesley: Bulletin of Wellesley College, http://­repository​.­wellesley​.­edu​/­cgi​/­v iewcontent​.­cgi​ ?­article​=1 ­ 078&context​=­catalogs (1980) and CollegeCalc, http://­www​.­collegecalc​ .­org​/c­ olleges​/­massachusetts​/­wellesley​-­college (2000); Swarthmore: Swarthmore College, http://­w ww​.­swarthmore​.­edu​/­Documents​/­administration​/­ir​/­Tuition​ .­pdf (1980) and CollegeCalc, http://­w ww​.­collegecalc​.o ­ rg​/c­ olleges​/p ­ ennsylvania​ /­s warthmore​-­c ollege​/­(2000); and Duke: Duke Archives, https://­l ibrary​ .­duke​.­edu​/r­ ubenstein​/­uarchives​/­history​/­articles​/­statistics (1980) and CollegeCalc, http://­w ww​.­collegecalc​.­org​/­colleges​/­north​- ­carolina​/­duke​ -­university​/­(2000)​.­ 26. In a report to the university’s trustees, Pye (1980), 11, wrote “While public institutions, ironically, may prosper in times of inflation, private institutions strug­gle to remain solvent. Costs increase with inflation in both sectors, but the private university must pay increases . . . ​in large part from tuition revenues, not from legislative appropriations. . . . ​The gap between tuition at private and public schools increases at the same time that many parents are facing serious prob­lems maintaining their standards of living and educating their c­ hildren.” The report also called for increased oversight over the university’s forestry school and its marine laboratory on the North Carolina coast. 27. Hoxby (2009), 16; Bound, Lovenheim, and Turner (2010), 141–142. 28. One alternative explanation is offered by Hoxby (2009), 16, who suggests that it resulted in large part from the caliber of students that colleges w ­ ere attracting. 29. Taxpayers who make gifts in the form of appreciated assets not only enjoy a tax deduction for the market value of the asset, they also avoid the tax that would have been paid on the capital gains had the asset been sold instead. 30. Annie Lowrey, “The 2010 Slate 60,” Slate, accessed June 19, 2014, www​.­slate​ .­com​/­a rticles​/ ­l ife​/­the​_ s­ late60​.­html. 31. For a brief review of the lit­er­a­t ure, see, for example, Auten, Clotfelter, and Schmalbeck (2000). One recent study showed that donations to colleges and universities ­were positively correlated to a state’s average income and h ­ ouse values (Brown, Dimmock, and Weisbenner 2012). 32. Auten, Clotfelter, and Schmalbeck (2000), ­Table 2. 33. Ry Rivard, “Deeper Pockets,” Inside Higher Ed, February 12, 2014, accessed April 29, 2017, www​.­insidehighered​.­com​/­news​/­2014​/­02​/­12​/­donors​-­gave​-­34​ -­billion​- ­colleges​-2 ­ 013.

378

NOTES TO P AGES 1 2 4 – 1 2 8

34. Emma Roller, “Stanford Campaign Brings in $6.2-­Billion, a Rec­ord for Higher Education,” Chronicle of Higher Education, February 8, 2012, accessed April 29, 2017, www​.­chronicle​.­com​/­a rticle​/­Stanford​-­R aises​- ­62​-­Billion​ /­130698​/­; Tamar Lewin, “Report Says Stanford Is First University to Raise $1 Billion in a Single Year,” New York Times, February 20, 2013, accessed April 29, 2017, http://­w ww​.­nytimes​.­com​/­2013​/­02​/­21​/­education​/­stanfords​ -­f und​-­raising​-­topped​-­1​-­billion​-­in​-­2012​.­html; David Abel, “Harvard Looks to Raise $6.5 B by 2018,” Boston Globe, September 21, 2013, accessed April 29, 2017, https://­w ww​.­bostonglobe​.­com​/­metro​/­2013​/­09​/­21​/­harvard​-­launches​ -­record​-­billion​-­capital​-­campaign​/­IRIICJYhmhyfhQSzCOvIEO​/­story​.­html. 35. Kevin Carey, “How Taxpayers Are Helping to Finance Harvard’s Capital Campaign,” Chronicle of Higher Education, September 24, 2013, accessed April 29, 2017, http://­w ww​.­chronicle​.­com​/­blogs​/­conversation​/­2013​/­09​/­24​ /­how​-­taxpayers​-­a re​-­helping​-­to​-­fi nance​-­harvards​-­capital​-­campaign​/.­ 36. ­T here was a modest degree of volatility in rank over time among the 121 institutions in the sample over the decadal years from 1970 to 2010. The university with the largest gap between highest and lowest rank was the Illinois Institute of Technology. In contrast, Harvard was ranked first or second in each of the five decadal years. See online appendix ­Table 5.1. 37. Institutions with gaps w ­ ere included if they ­were missing no more than five years total out of the forty-­three years covered by the data and w ­ ere not missing both of the beginning two years (1969 and 1970) or both of the ending two years (2010 and 2011). Complete data on donations for all forty-­three years ­were available for sixty-­t wo universities. The number of universities included in the sample, by number of missing years, was fifteen with one year missing, sixteen with two, thirteen with three, seven with four, and eight with five. 38. Support in year t (Xt) was estimated as Xt* = (1/2)(Xt-1 + Xt+1). Support in years t and t+1 ­were estimated as Xt* = Xt-1 + (1 / 3)(Xt+2-­Xt-1) and Xt+1* = Xt-1 + (2/3)(Xt+2-­Xt-1). 39. Although they w ­ ere grouped using the same criteria as the sample of 188 colleges used for analyzing the Freshman Survey data, t­ hese 167 are not the same or a subset of the larger group of colleges. 40. The standard errors of the estimates shown in Figure 5.4 w ­ ere small due to the large numbers of colleges in most categories. The largest for 2009 was for the private colleges in the 99th and above percentiles, which was $565. This implies a 95 ­percent confidence interval of $28,384 to $30,598, around the estimate of $29,491. 41. Lau and Rosen (2015) take a contrary view, based on analy­sis of a limited number of recent years. Based on data for 2003 to 2010, they conclude that in­equality in financial position across universities did not grow.

NOTES TO P AGES 1 2 8 – 1 3 3

379

42. Samuel H. Williamson, “Daily Value of the Dow Jones Average in the United States, May 2, 1885 to Pres­ent,” accessed January 9, 2017, www​ .­measuringworth​.­com​/D ­ JA​/­. Value as of the first trading day in July. 43. As noted above, the tax code makes it especially advantageous to gifts in the form of appreciated assets. 44. Piketty (2014). 45. In 2005 the twenty largest university endowments recorded an average net real return of 12.3 ­percent, compared to an average of 4.4 ­percent for the Standard and Poor’s 500 index. Lerner et al. (2008), 208, 213. 46. National Association of College and University Business Officers, “2013 NACUBO-­Commonfund Study of Endowments,” accessed July 7, 2014, www​.­nacubo​.­org​/­Research​/­NACUBO ​- ­Commonfund ​_ ­Study​_­of​ _­Endowments​.­html. 47. For more detail, see footnote in t­ able on asset allocations in National Association of College and University Business Officers, “2013 NACUBO-­ Commonfund Study of Endowments,” accessed July 7, 2014, www​.­nacubo​ .­org​/­Research​/­NACUBO​- ­Commonfund​_ ­Study​_­of​_ ­Endowments​.­html. 48. NACUBO, “2013 NACUBO-­Commonfund Study of Endowments. The proportion of endowments invested in alternative strategies r­ ose monotonically with endowment size in FY 2013. 49. Lerner et al. (2013), 213. 50. Ehrenberg and Smith (2001). 51. NACUBO, “Average Annual Effective Spending Rates for U.S. College and University Endowments and Affiliated Foundations, FY 2013 to 2004, ­ 013 accessed May 10, 2017, www​.­nacubo​.­org​/­Documents​/­EndowmentFiles​/2 NCSEPublicTablesSpendingRates​.­pdf; Another study found that universities cut spending out of endowment when markets went down, essentially “hoarding,” rather than attempting to smooth spending. See Brown, Dimmonck, and Weisbenner (2012). 52. Per-­student figures ­were calculated by dividing endowment by the total number of students, both undergraduate and gradu­ate. 53. Several mea­sures of in­equality, like the Gini coefficient, are based on proportional differences rather than algebraic differences, and they similarly fail to show increases in in­equality. Based on the sample of 705 colleges with endowment data in 1970, 1990, and 2013, t­ here is no evidence of an increase in in­equality, using conventional mea­sures that focus on proportional mea­sures of in­equality, like the Gini coefficient. For this sample, the Gini was .774 for 1970, .776 for 1990, and .759 for 2013. In other words, almost no change. If one looks at the 925 colleges and universities with data for 1990 and 2013, the big difference is much more in­equality for 1990, and therefore a decline in in­equality. The Gini based on ­those two

380

NOTES TO P AGES 1 3 3 – 1 3 6

years was .817 and .767 for 1990 and 2013, respectively. A pos­si­ble explanation for the difference in the 1990 values is that the larger sample includes Emory but the smaller one does not, and adding that university, due to the large size of its endowment, increases the calculated degree of in­equality. 54. Winston and Lewis (1997), 181. 55. This was the top decile in Winston and Lewis’s 2,773 institutions. 56. Winston and Lewis (1997). 57. Owing to the small number of universities in it, the public institutions in the 90th percentile and above uses all available universities with data available for each of the three years. 58. In the 50th to 90th percentile private sample used in Figure 5.6, for example, half of the twenty sampled institutions had the word “college” in their names, whereas only one of the twenty public institutions in this SAT group did. As a consequence, the institutions in the private-­sector categories tend to be smaller and often lack the research apparatus associated with universities. This difference between the private and public institutions applies not only to ­these random samples used for Figure 5.6 but to all the comparisons in the book using variants of the seventeen categories, b ­ ecause t­ hose categories w ­ ere not based on institution size or other characteristics that typically differentiate colleges from universities. 59. Research I universities are the largest and best-­f unded research universities. See Courant and Turner (2016). T ­ able 5.2 shows average salaries for full professors, which excludes fringe benefits. 60. See also Kane et al. (2005), fig. 7, 118. 61. Ibid., 119. 62. More precisely, the average subsidy, as defined in federal data sets, is equal to average education and related expenditures minus average net tuition, where the latter is defined as the sticker price of tuition minus average grant aid awarded to students. Education and related expenses includes spending on instruction, student ser­v ices, and a portion of general support and maintenance costs associated with ­these functions. It is worth noting that education and related expenditures may be an imperfect mea­sure of resources devoted to educating students since it includes spending on research done by faculty that is not sponsored by an outside funding agency. 6: Zero-­Sum Competition

1. Jack Maggarrell, “The Enrollment Roller Coaster: Colleges Fear the Big Dip,” Chronicle of Higher Education 17 (September 5, 1978): 1. 2. Duffy and Goldberg (1998), 88–89. 3. Ibid., 41.

NOTES TO P AGES 1 3 6 – 1 4 9

381



4. Breland et al. (2002), ­Table 2.5. 5. Primarily state governments, but also federal and local. 6. Fiske (1979). 7. Adding to the appeal of ­these campus photos ­were ­couples talking or holding hands, suggesting, according to Fiske (1979), “latent sexuality.” 8. Fiske (1979). 9. Duffy and Goldberg (1998), 55–56. 10. Reflecting their strong scholastic credentials, the ac­cep­tance rate for t­ hese applicants was 77 ­percent, compared to 28 ­percent for the entire pool of applicants; Duffy and Goldberg (1998), 56. 11. Harr (1984), 55. 12. Ibid., 50. 13. Ames (1984), 2, quoted in Duffy and Goldberg (1998), 62. 14. Duffy and Goldberg (1998), 179. 15. Don Hossler, “The Role of Financial Aid in Enrollment Management,” New Directions for Student Ser­vices 89 (Spring 2000), accessed January 12, 2017, www​.­uccs​.­edu​/­Documents​/­retention​/­2000%20The%20Role%20of%20 Financial%20Aid%20in%20Enrollment%20Management​.­pdf. 16. Arriving for the 1987 Fiesta Bowl, its football team disembarked from its airplane wearing ­battle fatigues; Paul Orehovec, “Enrollment Management and Higher Education: The History of Enrollment Management at the University of Miami,” Enrollment Management & Higher Education, April 8, 2011, ­ rg​/­id68​.­html. 4, accessed January 12, 2017, http://­enrollmentmanagement​.o 17. The College Board sells lists of names to more than a thousand colleges and universities. See Ry Rivard, “Micro-­Targeting Students,” Inside Higher Ed, October 24, 2013, accessed April 29, 2017, https://­w ww​.­i nsidehighered​.­com​ /­news​/­2013​/­10​/­24​/­political​-­campaign​-­style​-­targeting​-­comes​-­student​ -­search. 18. Paul M. Orehovec, “Enrollment Management and Higher Education: History of Enrollment Management at the University of Miami,” April 8, 2011, accessed April 29, 2017, http://­enrollmentmanagement​.­org​/­id68​.­html. 19. The claim made in one consulting firm’s description of enrollment management, that the approach taken addresses “all aspects of the institution’s mission,” was typical of the broad view taken by its proponents. R. B. Wilkinson et al., A Practical Guide to Strategic Enrollment Management Planning in Higher Education, Educational Policy Institute, November 2007, p. 6, ­ df​/S ­ EM%20Guide​ accessed January 12, 2017, www​.e­ ducationalpolicy​.­org​/p .­pdf. 20. Among the case studies published by Huron Consulting Group was “Revamping Enrollment Management and Marketing at the University of Alabama in Huntsville,” accessed April 29, 2017, https://­w ww​

382

NOTES TO P AGES 1 4 9 – 1 5 5

.­huronconsultinggroup​.­com​/­resources​/h ­ igher​- ­education​/­enrollment​ -­management​-­marketing​-­university​-­a labama​-­huntsville. 21. Purdue University, Enrollment Management Annual Report 2011–2012, accessed January 12, 2017, www​.­purdue​.­edu​/­enrollmentmanagement​/­documents​ /­emannualreport2011​.­pdf. 22. Quirk (2005). 23. Ibid. 24. Maguire Associates, Indiana State University, Summary Modeling Report, May 2012, 8. 25. Ibid., 16. 26. Ry Rivard, “Using FAFSA against Students,” Inside Higher Ed, October 28, 2013. 27. Kirp (2003), 113. 28. This trend was also revealed by surveys of financial officers. See, for example, Kevin Kiley, “Price of a Bad Economy,” Inside Higher Ed, May 7, 2013. 29. Breland et al. (2002), ­Table 6.13, 128. 30. Woo and Choy (2011), 4, 11, 12. 31. Santiago Merea, “Tuition Discount Shake-­Up,” Business Officer Magazine, June 2010, accessed April 31, 2014, www​.­nacubo​.­org​/­Business​_­Officer​ _ ­Magazine​/­Magazine​_ ­A rchives​/­June​_ ­2010​/­Tuition ​_ ­Discount ​_ ­Shake​-­Up​ .­html. 32. Burd (2013), 4. 33. California State Auditor, The University of California: Its Admissions and Financial Decisions Have Disadvantaged California Resident Students, Report 2015-107, March 2016, accessed April 29, 2017, www​.­auditor​.­ca​.­gov​/­pdfs​ /­reports​/­2015​-­107​.­pdf. 34. Stephanie Saul, “Public Colleges Chase Out-­of-­State Students, and Tuition,” New York Times, July 7, 2016, accessed April 29, 2017, https://­w ww​.­nytimes​ .­com​/­2016​/­07​/­08​/­us​/­public​-­colleges​-­chase​-­out​-­of​-­state​-­students​-­a nd​-­t uition​ .­html. 35. Based on data for the period from 1986 to 2012 for public research universities, Bound et al. (2016) estimated equations suggesting that foreign enrollment increased some 12 ­percent for ­every 10 ­percent fall in appropriations. 36. Purdue, Enrollment Management Annual Report 2011–2012, 12. 37. In an explicit attempt to recruit affluent students, officials at the University of Alabama explained how they employed census data to direct their promotional material at students attending Texas high schools located in high-­income areas; see Quirk (2005). 38. Griffith (2011). Income level of students was proxied by the share of students receiving Pell Grants. 39. Jaquette, Curs, and Posselt (2015).

NOTES TO P AGES 1 5 6 – 1 6 3

383

40. Mission of Amherst College, accessed April 29, 2017, https://­w ww​.a­ mherst​ .­edu​/­a mherst​-s­ tory​/­facts​/­m ission​.­ 41. Harr (1984), 54. 42. U.S. News and World Report, Best Colleges, 2014 Edition, 76–81. 43. Some colleges that offer merit aid are listed on the U.S. News website, accessed July 12, 2016, http://­colleges​.­usnews​.­rankingsandreviews​.­com​/­best​ -­colleges​/r­ ankings​/m ­ ost​-­merit​-­a id​/­page+3. 44. Karabel (2005), 517. 45. See Figure 5.2. 46. Clotfelter (1996). 47. Ibid. 48. “Mercedes syndrome” attributed to Williams director of admissions Phil Smith; see Harr (1984), 58. See also Duffy and Goldberg (1998), 195. 49. Machung (1998) argued that early decision grew rapidly between 1993 and 1998. See also Fallows 2001. 50. Fallows (2001). Part of the difference is due to differences in characteristics, but Avery, Fairbanks, and Zeckhauser (2003) show that this advantage remains when other student characteristics are accounted for. 51. This one move had the effect of reducing the ac­cep­tance rate from 53 to 51 ­percent and increasing the yield rate from 25 to 27 ­percent. Emory University pursued a similar policy in the 1990s by favoring the academically strongest candidates in their pool according to how many contacts they had made with the college during the application pro­cess; see Golden (2001). 52. Ibid. 53. Golden (2001) reported one especially effective version of the waiting list, employed by Car­ne­g ie Mellon, which asked students to pay a $400 deposit to be on an exclusive waiting list. Of the sixty who w ­ ere accepted off this list, fifty-­seven enrolled, an impressive 95 ­percent yield rate. 54. Kutner (2014). 55. They included Baylor, Bucknell, Claremont McKenna, Clemson, and George Washington. See also Beckie Supiano, “ ‘U.S. News’ Removes 2 More Colleges from Its Rankings,” Chronicle of Higher Education, May 14, 2013, accessed April 29, 2017, www​.­chronicle​.­com​/ ­blogs​/­headcount​/­u​-­s​-­news​-­removes​-­2​ -­more​-­colleges​-­from​-­its​-­rankings​/­35005; Scott Jaschik, “More Fudging of SAT Averages,” Inside Higher Ed, April 16, 2013, accessed April 29, 2016, https://­w ww​.­insidehighered​.­com​/­news​/­2013​/­04​/­16​/a­ nother​- ­college​-­admits​ -­calculating​-­incorrect​-­sat​-­averages. 56. Henry Adams noted Harvard’s preference for members of Boston’s elite before the Civil War; see Howell and Turner (2004), 327. 57. Karen (1991), 368. 58. Howell and Turner (2004), 329, figures for 1986 / 87.

384

NOTES TO P AGES 1 6 4 – 1 7 1

59. Karabel (2005), 520–521. 60. Andy Thompson, “20% of White Members of Harvard’s Class of 2014 Followed a Parent T ­ here,” Chronicle of Higher Education, May 27, 2014. Michael Olivas noted that legacy status at most universities “is a near-­perfect proxy for being white.” Quoted in Adam Liptak, “A Hereditary Perk the Founding ­Fathers Failed to Anticipate,” New York Times, January 15, 2008, accessed April 29, 2017, http://­w ww​.­nytimes​.­com​/­2008​/­01​/­15​/­us​/­15bar​.­html. 61. Shulman and Bowen (2001), fig. 2.3, 41. 62. Espenshade, Chung, and Walling (2004), fig. 1, 1443. 63. For legacies and the two comparison groups of nonlegacies, the percentage who said they wanted to enter into one of t­ hese professions was: medical doctor, 10 ­percent (versus 22 ­percent and 12 ­percent, respectively); professor / scientist, 8 ­percent (versus 9 ­percent and 11 ­percent); engineer, 2 ­percent (versus 3 ­percent and 8 ­percent); see Martin and Spenner (2009), 642, 643. 64. Clotfelter (2011), T ­ able 8A.1, 248. 65. Miami (Ohio), Penn State, University of Michigan, and University of North Carolina at Chapel Hill. 66. They w ­ ere Duke, Georgetown, Northwestern, Rice, Stanford, Tulane, Notre Dame, and Vanderbilt; see Shulman and Bowen (2001), fig. 2.4. 67. Between 1976 and 1989 the gap for high-­profile sports increased from 199 to 237 points in the public Division I-­A universities and from 180 to 284 in the private ones. Similarly, Espenshade, Chung, and Walling (2004), fig. 1, 1443, in their study of ten selective institutions, found that the percentage point advantage in admissions for recruited athletes in their sample increased from 25 ­percent in the 1980s to 32 ­percent in 1993 to 35 ­percent in 1997. 68. Breland et al. (2002), T ­ able 4.7. 69. Bowen, Kurzweil, and Tobin (2005), ­Table 5.1, 105. 70. Espenshade, Chung, and Walling (2004), fig. 1, 1443. 71. One reason given for this preference is that it offered a way of limiting the number of Jewish students; Howell and Turner (2004), 327. 7: Evolution in the Core Business



1. Quoted in Geiger (2004), 99. 2. Gardner (2005), 97. 3. Kirp (2005), 116; Fallows (2005), 40. 4. Kirp (2005), 116. 5. For example, Kirp (2005) and Schneider (2005) write that liberal arts majors are declining. 6. College of Mount Saint Vincent, “Areas of Study,” accessed April 17, 2016, https://­mountsaintvincent​.e­ du​/­academics​/­u ndergraduate​- ­college​/­a reas​- ­of​ -­study​/.­

NOTES TO P AGES 1 7 1 – 1 7 8

385

7. See, for example, Riesman (1980), 76; Geiger (2004), 93; Sperber (2005), 133; and Stuart Rojstaczer, accessed May 5, 2016, http://­stuartr​.­com​/­grade​ -­inflation​/­a nd www​.­gradeinflation​.­com​/.­ 8. See, for example, Riesman (1980), xi; Kirp (2003), 15–16; Geiger (2004), 5, 116–119; Lewis (2006), 14. 9. Noting the growth in non-­instructional categories, Webber and Ehrenberg (2009) have found favorable effects on per­sis­tence. 10. Botstein (2005), 220. 11. Kidd (2005), 196. 12. See Ehrenberg and Zhang (2004) and Vedder (2004), 47, for discussions of the rise in nonregular faculty. 13. Riesman (1980), 162. 14. Recall the definitions of t­ hese groups from Chapter 2: (1) low-­SAT (0–50th percentile) public colleges; (2) high-­SAT (80th + percentiles) public colleges; (3) lowest-­SAT (0–25th percentiles) private colleges; and (4) very-high-­SAT (90th + percentiles) private colleges. 15. University of Southern Mississippi, accessed April 29, 2017, http://­catalog​ .­usm​.e­ du​/c­ ontent​.p ­ hp​?c­ atoid​=1 ­ 0&navoid​= 6 ­ 32. According to the Gallup website, its Reputational Model “is one of the most sophisticated reputation models in the world.” Gallup, “Reputation Index Model,” accessed April 29, 2017 www​.­wingia​.­com​/­en​/­services​/ ­branded​ _­solutions​/­corporate​_ ­branding​_ ­reputation​_ ­index​_ ­model​_ ­e​/­3/​ ­13​/.­ 16. William and Mary, “Our Curriculum,” accessed April 29, 2017, www​.­wm​.­edu​ /­a s​/­u ndergraduate​/­curriculum​/­index​.­php. 17. University of California, “Admissions: UC San Diego,” accessed April 17, 2016, http://­admission​.­u niversityofcalifornia​.­edu​/­campuses​/­san​-­d iego​/.­ 18. Georgia Tech, Ivan Allen College of Liberal Arts, “Dean’s Welcome,” accessed May 1, 2017, www​.­iac​.­gatech​.­edu​/­about​/­overview​/­deans​-­welcome. 19. University of Florida, Undergraduate Cata­log 2017–17, Mission Statement, accessed April 29, 2017, https://­catalog​.­u fl​.­edu​/­ugrad​/­current​/­uf​-­m ission​ /­Pages​/h ­ ome​.a­ spx. 20. Riesman (1980), 162. 21. Lee University, the Pentecostal college, stated in its 2014 website, “A personal commitment to Jesus Christ as Savior is the controlling perspective from which the educational enterprise is carried out.” And Trevecca Nazaerene University used the identical sentence in 2014 that it had used in 1970: “The college has maintained a strong Christian evangelical emphasis and motivation for ser­v ice.” 22. William Carey, “Mission and Vision,” accessed April 26, 2015, www​.­wmcarey​ .­edu​/­m ission​-­vision. 23. “Elon University Mission Statement,” accessed April 26, 2015, www​.­elon​.­edu​ /­e​-­web​/­administration​/­m ission​_­statement​.­x html.

386

NOTES TO P AGES 1 7 8 – 1 8 2

24. Saint Mary’s College, “Prepare for a Bold ­Future,” accessed April 26, 2015, www​.­stmarys​-­ca​.­edu​/­about​-­smc. 25. Saint Leo, accessed April 26, 2015, www​.s­ aintleo​.­edu​/.­ 26. Oglethorpe College, “About,” accessed April 29, 2015, www​.­oglethorpe​.e­ du​ /­about​_­us​/.­ 27. University of New Haven 2015 mission statement, accessed April 29, 2015, www​.­newhaven​.­edu​/a­ bout​/m ­ ission​-­statement​/­. For comparison, see University of New Haven Cata­log, 1971–1972, https://­a rchive​.­org​/­details​ /­universityofne197172univ. 28. www​.m ­ ocavo​.­com​/ Y ­ ork​- ­College​- ­of​-­Pennsylvania​-­1970​-­1971​-­Volume​-­29​ /­294864​/7 ­ ​?­browse​= ­true; York College, accessed April 29, 2015, www​.­ycp​.­edu​ /­about​-­us​/.­ 29. About Elon, accessed April 29, 2017, https://­w ww​.­elon​.­edu​/­e​/­about​/.­ 30. Howard Payne University website, accessed April 29, 2017,​/­w ww​.­hputx​.­edu​ /­academics​/s­ chools​/.­ 31. In 2005 this requirement was replaced by a requirement that 80 ­percent of the board be Christian and twenty-­four of the forty-­four members be Presbyterians; Inside Higher Ed, May 9, 2012. 32. North Carolina History Proj­ect, Shane Williams, “Wake Forest University,” accessed July 6, 2015, www​.­northcarolinahistory​.­org​/­encyclopedia​/­736​ /­entry. 33. Wake Forest University, “Pro Humanitate,” accessed April 30, 2015, http://­ admissions​.­w fu​.­edu​/­our​-­philosophy​/­pro​-­humanitate​/.­ 34. Swarthmore College, “Why the Liberal Arts M ­ atter,” accessed April 17, 2016, www​.­swarthmore​.­edu​/­aydelotte​-­foundation​/­why​-­l iberal​-­a rts​-­matter. 35. Brandeis University Bulletin, 2017–18, accessed April 17, 2016, www​.b ­ randeis​ .­edu​/­registrar​/ ­bulletin​/­provisional​/­college​/.­ 36. Mount Holyoke, “The Liberal Arts,” accessed April 17, 2016, www​.­mtholyoke​ .­edu​/­about​/­l iberalarts. 37. Lehigh University, “Colleges & Departments,” accessed April 17, 2016, www1​.­lehigh​.­edu​/­academics​/­colleges. 38. Bucknell, “Value of a Liberal Arts Education,” accessed April 17, 2016, www​ .­bucknell​.­edu​/­info​-­about​-­attending​-­bucknell​/­how​-­your​-­l ife​-­w ill​- ­change​ /­value​-­of​-­a​-­l iberal​-­a rts​-­education​.­html. 39. Bates, “Liberal Arts at Work,” accessed April 17, 2016, www​.­bates​.­edu​/­career​ /­students​/­why​-­l iberal​-­a rts​/.­ 40. Kalamazoo College, “Programs by Division,” accessed April 17, 2016, www​ .­k zoo​.­edu​/­academics​/.­ 41. Goucher College, “Goucher’s New Curriculum,” accessed April 17, 2016, www​.­goucher​.­edu​/­academics​/­l iberal​-­education​-­curriculum. 42. Agnes Scott College, “Why a Liberal Arts College?,” accessed April 17, 2016, www​.­agnesscott​.­edu​/­about​/­at​-­a​-­glance​/­why​-­a​-­l iberal​-­a rts​-­college​.­html.

NOTES TO P AGES 1 8 3 – 1 8 5

387

43. Savannah State University, accessed May 1, 2017, http://­w ww​.­savannahstate​ .­edu​/­president​/­index​.­shtml. 44. Its 2015 website states, “Florida Memorial University serves a culturally diverse student population through programs in liberal and professional education.” 45. Oakwood also included a curiously detailed recitation of ethical responsibilities of employees, including admonitions against theft, embezzlement, bribery, and the use of work time to pursue personal activities. 46. For the purposes of making comparisons, I have designated majors that I deem primarily vocational, in contrast to primarily liberal arts. T ­ hese are gross generalizations, of course, and many of ­those choices could be subject to second-­g uessing. A list of the majors deemed primarily vocational are given in the online appendix. 47. Hearn and Belasco (2015), 400. 48. Brint et al. (2012b), 594. 49. Brint et al. (2012a), 29. 50. The mean values ­were very similar. 51. Each of the named majors, including ­those in parentheses, was ­adopted by at least seven of the twenty-­three sampled public universities in the bottom half of the SAT distribution. 52. ­These majors ­were added by at least seven of the sixteen private colleges ranked below the 25th percentile. 53. ­These majors ­were added by at least three of the seven sampled public universities at the 90th percentile and above. 54. Each of ­these was ­adopted by two of the eight very-­high-­SAT private colleges. 55. Brint et al. (2012b), 599–600, 603. 56. This finding is consistent with the work of Hearn and Balasco (2015), 402–406, who found that colleges ­were most likely to hold onto humanities majors if they w ­ ere selective, old, small, prestigious, and financially strong. 57. ­These academic majors could become “practical,” of course, to the extent that they could prepare students for doctoral training, a connection that actually grew stronger in economics, as Chapter 12 demonstrates. 58. Darolia and Koedel (2016) provide evidence of such a hierarchy by calculating an index of academic rigor (based on test scores and per­sis­tence) for ­every major offered in all the public colleges and universities in Missouri. In private correspondence, the authors calculated indices for eight majors that ­were offered by one or both of two institutions, one selective and one not selective. Within each institution the ranking of the majors was the same. From most to least rigorous, they ranked as follows: mathe­matics; physics; chemistry; computer science; social work; physical education; criminal

388

NOTES TO P AGES 1 8 6 – 1 9 0

justice / corrections; and parks, recreation, and leisure. Where a major was offered by both colleges, the major at the more selective college had a higher index value. 59. In 1997 the U.S. News formula for determining college rank gave 20 ­percent weight to four aspects of faculty, three of which w ­ ere percentage of faculty with a PhD or highest available degree in the field, student / faculty ratio, and percentage faculty who w ­ ere full-­time (Monks and Ehrenberg 1999), 4. 60. Clotfelter (1996), 192–199. 61. Such findings might be consistent with a decline in teaching load, of course, depending on how much time faculty spend on each course taught. See Clotfelter (1996), 187. 62. Ibid., 186. 63. Ehrenberg (2012), 199. 64. Ibid., 194, citing U.S. Department of Education, Digest of Education Statistics 2009, ­Tables 249 and 253. 65. Ehrenberg (2012), 194. 66. ­W hether greater reliance on nontenure-­track faculty weakens instruction remains an open question. Ehrenberg and Zhang (2004) analyzed panel data for colleges and universities from 1987 to 2001, which implied that increases in part-­time or nontenure-­track faculty ­were associated with decreases in graduation rates. But a recent study at Northwestern found that students taught by nontenure-­track instructors learned more than ­those taught by faculty on the tenure track; see Figlio, Schapiro, and Soter (2013). 67. Public ser­vice spending grew at 1.7 ­percent a year, operations and maintenance at 0.5 ­percent, and auxiliaries at 0.4 ­percent; see Ehrenberg (2012), 204. 68. All clusters are based on 1970-­era average SATs. 69. Tenure-­track faculty w ­ ere deemed to be t­ hose with the titles “assistant professor,” “associate professor,” or “professor,” including t­ hose with distinguished or named chairs. Terms indicating nontenure-­track status ­were the positions of “lecturer” or “instructor,” or positions modified with one or more of t­ hese terms “adjunct,” “visiting,” “acting,” “research,” “interim,” “emeritus,” or “instructional.” 70. See Handlin and Handlin (1970), 36. 71. Keller (2004), 48. 72. Scott Carlson, “Spending Shifts as Colleges Compete on Students’ Comfort,” Chronicle of Higher Education, July 28, 2014. 73. Andrew Luo, “Students Flock to Au Bon Pain,” Duke Chronicle, August 24, 2012. 74. Allie Grasgreen, “Luring Students with a Stadium,” Inside Higher Ed, October 30, 2013.

NOTES TO P AGES 1 9 0 – 1 9 7

389

75. Courtney Rubin, “Making a Splash on Campus,” New York Times, September 19, 2014, accessed May 1, 2017, www​.­nytimes​.­com​/­2014​/­09​/­21​ /­fashion​/­college​-­recreation​-­now​-­includes​-­pool​-­parties​-­a nd​-­r iver​-­r ides​.­html​?­​ _­r ​= 0 ­. 76. The college’s website noted with pride that High Point ranked number one in a survey of student satisfaction with their college dining options, the survey run by Aramark, a com­pany that operates dining facilities on college campuses. Viewbook accessed May 5, 2016, https://­issuu​.­com​/­highpoint​ university​/­docs​/­hpuseniorviewbook2015. 77. The description of the Wanek Center and University Center II includes descriptions of living areas and dormitory accommodations, including this: “Each bathroom has a granite vanity with two u ­ nder mounted sinks, large vanity mirror, maple cabinetry with generous storage space for personal items, private rest­room, lighted shower, and marble tile floors.” Highpoint, “Residence Life,” accessed July 18, 2016, www​.­highpoint​.­edu​/­residencelife​ /­dormitory​/­u niversity​-­center​/­. See also www​.­dailymail​.­co​.­u k​/­news​/­a rticle​ -­2537212​/­Is​-b ­ est​-­u niversity​-­youve​-n ­ ever​-­heard​-­The​-­700​-­m illion​-­Disney​-­style​ -­campus​-­complete​-­ice​-­cream​-­truck​-­movie​-­theatre​-­Skee​-­ball​.­html; www​ .­huffingtonpost​.c­ om​/­2014​/­01​/­09​/­college​-o ­ f​-­your​-­wildest​-­dreams​_­n _​ ­4508077​ .­html; www​.­bloomberg​.­com​/ ­bw​/­a rticles​/­2012​- ­04​-­19​/ ­bubble​-­u​- ­dot​-­high​ -­point​-­university. 78. Armstrong and Hamilton (2013), 51. 79. Ibid., 53. When parties got out of hand, the system called for warnings before police would arrive, allowing underage revelers to make themselves scarce. 80. Ibid., 53–54. 81. Ibid., 54. 82. Wright State University, www​.­w right​.­edu​/­campus​-­l ife; Bloomsburg Univer­ irginia Wesleyan College, www​.­w vwc​.­edu​/­l ife​ sity, www​.­bloomu​.­edu​/­; West V -­at​-­w vwc; William Carey University, wmcarey​.­edu​/­departments​/­student​ -­services; Tennessee Wesleyan University, www​.­t wcnet​.­edu​/­student​-­l ife​/;­ Saint Leo University, www​.­saintleo​.­edu​/­campus​-­l ife​.­a spx; and Southern Utah University, www​.­suu​.­edu​/­college​-­l ife​.­html, all accessed September 24, 2015. 83. Jacob, McCall, and Stange (2013). 84. James C. Hearn and Jarrett B. Warshaw, Mission-­Driven Innovation, Washington, DC: Council of In­de­pen­dent Colleges (July 2015), 15, accessed May 11, 2015, https://­w ww​.­cic​.­edu​/­r​/­r​/­Documents​/­CIC​-­Hearn​-­Report​-­2015​.­pdf. 85. Babcock and Marks (2011). 86. See Stuart Rojstaczer, “Grade Inflation,” accessed May 5, 2016, http://­stuartr​ .­com​/­grade​-­inflation​/­a nd www​.­gradeinflation​.­com​/.­

390

NOTES TO P AGES 1 9 7 – 2 0 1

87. Also known as the Football Bowl Subdivision. 88. Clotfelter (2011), 106. 89. They include the University of North Carolina at Charlotte, the University of Alabama at Birmingham, and Georgia State University. 90. See Pope and Pope (2009), and Humphreys and Mondello (2007). 8: Scholastic Segregation

1. This point of view is succinctly summarized by Oreopoulos and Salvanes (2011), 159: “Economists usually view schooling as a financial investment: that is, individuals spend money and time to acquire (or perhaps signal) ­human capital, in hopes of greater lifetime wealth and consumption in return.” 2. The more widely accepted view holds that education enhances a person’s ­human capital, which in turn increases productivity, which leads to a better wage. This alternative, “signaling,” view sees college as a g­ iant sorting mechanism, whereby the most talented students attend college. The fact that college gradu­ates earn more than nongraduates is a reflection of the ability t­ hose students had when they enrolled. 3. Long (2004), 283, 286, found that t­ hese mea­sures did indeed correlate with students’ choices, the first more strongly than the second. 4. See Monks and Ehrenberg (1999) or Griffith and Rask (2007). 5. Bowman and Bastedo (2009). See also Griffith and Rask (2007), who show ­these rankings do have an impact on college choices, especially for students not expecting to receive aid. 6. The authors used spending on student ser­vices and auxiliary ser­v ices as proxies for consumption amenities, including dormitories. They also found that the presence of fraternities and Division I athletics increased the chance of choosing an institution. Jacob, McCall, and Stange (forthcoming). 7. Hoxby (1997, 2009). 8. Hoxby (1997), 11. 9. Ibid., T ­ able 1a, 46. 10. Ibid., T ­ able 3, 49. 11. The sample contained 731 private and 390 public colleges. Hoxby (1997), T ­ able 5, 52. 12. Zelizer (1985). 13. In their book, Jencks and Riesman (1968), 97, acknowledge as much: “­There is no vocal opposition to keeping American society open, to encouraging the poor-­but-­able to get ahead, or to equality of opportunity. Nevertheless, well-­to-do parents also have an understandable impulse to make sure that their c­ hildren ­w ill enjoy the same privileges that they do. While such parents

NOTES TO P AGES 2 0 2 – 2 1 3

391

hope their c­ hildren ­w ill earn their privileges in fair competition, the parental commitment to fairness seldom takes pre­ce­dence over ­family loyalty. If a middle-­class child cannot win out on merit alone, his parents ­w ill try to obtain some sort of special treatment for him.” 14. Lareau (2011). 15. See Demerath (2009), especially chap. 2. 16. Cohen (2006), 1. 17. Noting the ways in which affluent families benefit from exceptions to purely meritocratic criteria, Jencks and Riesman (1968), 97, conclude, “When all such exceptions to meritocratic princi­ples are taken together, they provide upper-­and upper-­m iddle-­class ­children with enormous advantages over the other p ­ eople’s c­ hildren.” 18. Bound, Hershbein, and Long (2009) expand on this point. 19. Ibid., ­Table 5. 20. Handwerk et al. (2008). 21. Lemann (1999), 110–114, 220–227, provides an entertaining account of the history of Kaplan and the truth-­in-­testing legislation. 22. Bound, Hershbein, and Long (2009), ­Table 6, 134. See also Belasco, Rosinger, and Hearn (2014). 23. The two highest income categories ­were $35,000–­$50,000 and $50,000 and above. 24. Buchman, Condron, and Roscigno (2010), fig. 1, 447. The survey was the National Educational Longitudinal Study, eighth graders in 1988. 25. Jenny Anderson, “A Hamptons Summer: Beach, Horses, and SAT Prep,” New York Times, August 13, 2012, accessed May 1, 2017, www​.­nytimes​.­com​/­2012​/­08​ /­14​/­education​/­a​-­hamptons​-­summer​-­surfing​-­horses​-­a nd​-­hours​-­of​-­sat​-­prep. 26. According to Buchman, Condron, and Roscigno (2010), 439, courses offered by the Prince­ton Review ranged in cost from $1,000 to $1,200. 27. For an analy­sis of this strategy, see Vigdor and Clotfelter (2003). 28. At one public high school serving an affluent student body, the percentage of students classified as “other health impaired” tripled between 1996 and 2002. This classification allowed t­ hese students special accommodation in completing tests in school; see Demerath (2009), 19–21. 29. Abrams (2005), 43–44. 30. Abigail Moor, “Accommodations Angst,” New York Times, November 4, 2010, accessed May 1, 2017, www​.­nytimes​.­com​/­2010​/­11​/­07​/­education​/­edlife​ /­07strategy​-­t​.­html. 31. Ramey and Ramey (2010), Figs. 1 and 3. 32. For an account of how rugby is being used to increase the chances for some students, see John Otis, “Need Scholarship? Join Scrum,” New York Times, May 22, 2014, B11. Also see Stevens (2009) for a discussion of the value of applicants knowing how to play certain sports favored by elite colleges.

392

NOTES TO P AGES 2 1 3 – 2 1 8

33. According to Demerath (2009), 125, the frenzy to get into desirable colleges is the “logical local response to an intensifying global academic arms race. It is a practical outcome of the community’s emphasis on individual advancement and the school’s policies of institutional advantaging, parents’ proprietary relationship ­toward the school, students’ identities for control and success, and teachers’ conceptions of their work as preparing students for higher education markets.” 34. Among the techniques used by some counselors is a college application boot camp, where students gather to work on their application materials. Frank Bruni, “Naked Confessions of the College-­Bound,” New York Times, June 15, 2014, accessed May 1, 2017, www​.­nytimes​.­com​/­2014​/­06​/­15​/­opinion​/­sunday​ /­frank​-­bruni​-­oversharing​-­in​-­admissions​-­essays​.­html​?­​_­r ​= 0 ­. 35. Foreign students are excluded from the calculations. 36. For only one group was the trend the other way—­the highest-­scoring public institutions that included three ser­vice academies. It seems pos­si­ble, even likely, that the inclusion of t­ hese three academies is responsible for this anomaly. Another f­ actor may be Georgia’s HOPE scholarship, awarded to Georgia residents with good grades in high school to attend colleges in the state. Since Georgia Tech is one of the colleges in this college group, an increase in the share of in-­state students might also have contributed to this result. 37. The average for ­those who lived in the Northeast was less than 19 miles, compared to averages of 95 miles in the South and Midwest and 149 miles in the West. Griffith and Rothstein (2009), 623. 38. For institutions reporting ACT scores, t­ hose scores ­were converted to their SAT equivalents, and college-­w ide weighted averages w ­ ere calculated according to the number of matriculants reporting scores from each test. The resulting SAT equivalents from 1970 w ­ ere converted to their re-­centered equivalent SAT scores. For 2008 / 09 institutional average scores ­were estimated as the average of the 25th and 75th percentile scores. 39. To express all achievement test scores in terms of a comparable scale, the ACT for students reporting just that score was converted to the corresponding composite SAT using a crosswalk provided by the College Board. The 1989 / 90 SAT scores w ­ ere also re-­centered to make them comparable to the 2008 / 09 scores. 40. Weighted averages for each year ­were calculated using averages for each group, where each group’s weight is its share of total four-­year enrollment in the given year. Owing to the large number of student surveys covered in each category, the standard errors applying to t­ hese and other estimates using the Freshman Survey tend to be small. For Figure 8.4, the largest standard errors w ­ ere for the private colleges in the 99th and above category. For 2007 / 08, it was 0.0097, implying a 95 ­percent confidence interval for the

NOTES TO P AGES 2 1 8 – 2 2 4

393

percentage of students who reported studying more than ten hours a week of 0.602 to 0.640 around the mean of 0.621. The smallest standard error applied to the 0 to 50th percentile public colleges was 0.0041, which implied a confidence interval of 0.185 to 0.201 around the mean of 0.193. 41. Fryer and Greenstone (2010) compare the economic return from attending HBCUs in the 1970s and 1990s. For a reflection on the changing role of HBCUs, see Ta-­Nehisi Coates, “Homecoming at Howard,” New York Times, October 29, 2013, accessed May 11, 2017, www​.­nytimes​.­com​/­2013​/­10​/­30​ /­opinion​/­coates​-­homecoming​-­at​-­howard​.­html. 42. Long (2004), 294. 43. Avery and Hoxby (2004), 274–276. In an experiment at one college, Monks (2009) also finds weak evidence that students prefer getting a scholarship to an equivalent reduction in net price without a scholarship. 9: Economic Stratification

1. Richard Pérez-­Peña, “Generations L ­ ater, Poor are Still Rare at Elite Colleges,” New York Times, August 25, 2014, accessed May 1, 2017, www​.­nytimes​ .­com​/­2014​/­08​/­26​/­education​/­despite​-­promises​-­l ittle​-­progress​-­in​-­drawing​ -­poor​-­to​-­elite​-­colleges​.­html; Douthat (2005). 2. Burd (2013). 3. Pérez-­Peña, “Generations ­L ater.” 4. Useem and Miller (1975), 122. 5. Kirp (2003), 31, states, “It used to be said that higher education replicated the nation’s in­equalities. These days it is making things worse.” 6. For two of the many studies motivated by concern over equity, see Stevens (2009), chap. 8, or Bailey and Dynarski (2011). 7. See, for example, Winston (1999), Sacerdote (2001), and Kremer and Levy (2008). 8. Among ­others, Murray (2012). Also see Becker, Hubbard, and Murphy (2010), who argue that success in the marriage market is affected by education. 9. See Bastedo and Jaquette (2011), 318. 10. Laughlin (2010), ­Table 9. 11. Duncan and Murnane (2011), 11. 12. See, for example, Clotfelter, Ladd, and Vigdor (2009). 13. Reardon (2011). 14. Ingels and Dalton (2013). 15. For comparisons of teacher characteristics across schools with differing proportions of low-­income students, see Clotfelter et al. (2007). 16. For a comparative study of socioeconomic stratification in higher education, see Shavit, Arum, and Gamoran (2007).

394

NOTES TO P AGES 2 2 4 – 2 3 0

17. As noted in Chapter 3, a study in 1928 revealed that undergraduates at the University of Michigan came disproportionately from middle-­and upper-­ class families. 18. U.S. Department of Education, Digest of Education Statistics 2012, ­Table 173. 19. Source: Proj­ect Talent: T ­ able 5-1 in “One Year Follow-up Studies,” Proj­ect Talent, American Institutes of Research, Pittsburgh, 1966, 93, in U.S. Office of Education, Students and Buildings, An Analy­sis of Selected Federal Programs for Higher Education, 1968, ­Table A-8, reproduced in Harris (1972), 61. An earlier study beginning in 1957, based on high school se­n iors in Wisconsin, yielded very similar patterns. Among se­n iors in the highest intelligence group, the share who attended college ranged from 52 ­percent for t­ hose in the lowest SES group to 91 ­percent for t­ hose in the highest. For females the corresponding percentages w ­ ere 28 ­percent and 76 ­percent, respectively (Sewell and Shah 1967), T ­ able 3. Still other national surveys, covering cohorts of youth in the 1970s, 1980s, and 1990s, have confirmed t­ hese early findings. 20. See, for example, figs. 2a and 2b in Belley and Lochner (2007), which pres­ents a similar two-­way comparison of college enrollment rates by ­family income and Armed Forces Qualification Test (AFQT) quartiles. 21. See, for example, findings from an analy­sis of the High School and Beyond survey, cited in Clotfelter et al. (1991), ­Table 2.12. For tabulations on enrollment in four-­year colleges, see also Kane (2004), T ­ able 8.2. 22. Radford (2013), 99, 111, 146. 23. Karabel and Astin (1975), 394. 24. ­These ­were the National Longitudinal Study of 1972 (NLS), covering the high school class of 1972; High School and Beyond, class of 1982; National Education Longitudinal Study (NELS), class of 1992; Educational Longitudinal Study of 2002 (ELS), class of 2004. 25. Roksa et al. (2007), 169. 26. National Survey of Youth, which focused on students in the high school classes of 1979–1982 and 1997–2000; Bailey and Dynarski (2011). 27. Roksa et al. (2007), employing data from the NLS-72, High School and Beyond, and NELS, estimated logistic regressions explaining enrollment in a four-­year college and in an elite four-­year college. T ­ here ­were no controls for student ability in t­ hese regressions. 28. See Shavit et al. (2007). 29. Astin and Oseguera (2004), 325. 30. Ibid., 330–334. 31. Bastedo and Jaquette (2011). 32. Bastedo and Jaquette (2011), ­Table 3, 327.

NOTES TO P AGES 2 3 1 – 2 3 5

395

33. Categorical responses ­were converted to dollar values by assigning the midpoint of all closed categories. To estimate the average income for the top category for each survey, IRS tabulations in the Statistics of Income ­were used to calculate the mean adjusted gross income for each of the five years covered by the surveys. 34. The average ­family income for all students in four-­year colleges was estimated by the enrollment-­weighted average for all students attending one of the 188 colleges in the sample. 35. Note that ­there are more bars showing above-­average than below-­average ­family incomes. This apparent lack of balance arises b ­ ecause the seventeen college groups do not account for equal numbers of students. If averages for the college groups had been weighted by the number of students enrolled in the college in each category, they would have produced an average deviation of exactly zero. Since they are based on samples, t­ hese calculations of average income levels are subject to error. But the errors are small, owing to the large numbers of students who w ­ ere surveyed. To give a sense of the standard errors attached to the estimates for f­ amily income in Figure 9.1, the largest standard error was for private colleges in the 99th percentile and above. For 2008 / 09 it was $4,967, which implied a 95 ­percent confidence interval around the mean of $83,065 of $73,330 to $92,800. 36. Responses concerning ­family income, by category, in surveys for 1972, 1989, 1990, 2008, and 2009 referred to income in 1971, 1988, 1989, 2007, and 2008, respectively. Using numbers by income category in each survey, the numbers by categories corresponding to the four income groups above expressed in 2008 dollars w ­ ere estimated by interpolation. 37. The words ­were ­those of Barry O’Connell, professor of En­glish and American Studies at Amherst (Harr 1984), 58. 38. To identify exactly 10 ­percent for the highest income group, all the students ­were ranked in each of the three waves by the category they indicated for their f­ amily’s income in the previous year, using as many of the highest income categories as necessary to yield at least 10 ­percent of all students with income data for that survey wave. To trim the resulting number to exactly 10 ­percent, other available variables ­were used to eliminate enough students in the lowest of the selected categories to achieve the necessary number. For two of the survey waves, the median income of the student’s home ZIP code was used, followed by f­ ather’s education, to order students in that lowest category. For the 1972 wave, f­ ather’s education was used. To break ties beyond that, students ­were chosen at random ­until the resulting group constituted exactly 10 ­percent of all students who had answered the ­family income question. 39. As noted in the appendix, the colleges in the sample falling into the 98th–99th percentile category of private colleges w ­ ere Carleton, Hamilton,

396

NOTES TO P AGES 2 3 6 – 2 4 2

Johns Hopkins, Middlebury, Mount Holyoke, Prince­ton, Reed, Wesleyan, and Williams; the private colleges in the 99th and above category w ­ ere Amherst, Brandeis, Caltech, Dartmouth, Harvey Mudd, Haverford, Rice, Sarah Lawrence, and Swarthmore. 40. ­These consisted of the private colleges in the 90th–95th, 95th–98th, 98th–99th, and 99th + SAT percentile ranges, as well as the public institutions in the 90th percentile and up range. See the online appendix for a list. 41. An alternative method of identifying students in the richest 10 ­percent, using median ZIP code income, available only for the second and third waves, produced estimates that actually suggest no growth at all. Note that all the calculations h ­ ere are based on unweighted counts of first-­year students enrolling at the same 188 colleges. In light of the fact that overall college enrollment in the period grew more rapidly than enrollment for the 188 colleges, and selective institutions being overrepresented in the sample, it is reasonable to view ­these students as an increasingly unrepresentative group. This could appear to work against any set of colleges becoming more academically elite, and it is unclear how it cuts in terms of affluence. 42. Data and explanations of the methodology w ­ ere published by The Equality of Opportunity Proj­ect, accessed January 22, 2017, www​.­equality​-­of​ -­opportunity​.­org​/­. Parents’ incomes are averages of ­house­hold income over the five years when the college student turned fifteen to nineteen and are expressed in 2015 dollars. 43. In 1970 / 71 enrollment in Catholic schools made up 77 ­percent of all nonpublic secondary enrollment. U.S. Office of Education, Digest of Education Statistics 1973, ­Table 41, http://­fi les​.­eric​.­ed​.­gov​/­f ulltext​/ ­ED088240​.­pdf. 44. U.S. Department of Education, Digest of Education Statistics 2013, T ­ able 205.20, 113. 45. Using h ­ ouse­hold data for California, Buddin, Cordes, and Kirby (1998), ­Table 4, 119, show that the percentage of students enrolled in private schools ­rose with income, from 3.9 ­percent in the lowest income class to 17.4 ­percent in the highest. See also Long and Toma (1988) and Lankford and Wyckoff (1992) for evidence that private school enrollment is positively related to income. Evidence on this association is also available from a study of Harvard students who graduated in the mid-1960s. T ­ hose who had attended public schools had an average socioeconomic index value that was statistically lower than t­ hose who had attended private schools; Zweigenhaft (1993), T ­ able 1, 216. 46. Enrollment-­weighted average based on the Freshman Survey sample, excluding HBCUs. For 1989 / 90 the percentage was 13.4 ­percent. The comparable national averages including HBCUs w ­ ere 11.2 ­percent for 1972, 16.4 ­percent for 1989 / 90, and 24.0 ­percent for 2008 / 09.

NOTES TO P AGES 2 4 2 – 2 4 9

397

47. A similar pattern emerges when we look at the education of parents rather than income. For private colleges at the 95th–98th percentiles the percentage of nonwhite and foreign students who had a parent with a gradu­ate degree exceeded the national average by 9 percentage points in 1972 and by 28 percentage points in 2008 / 09. For the 99th and above category, that gap increased from 16 to 27 percentage points. See the online appendix for more on this. 48. ­These are own-­group exposure rates, which are weighted averages of the percentage black of eighth grades in public schools and among entering students in the fifteen branches of the University of North Carolina, where the weights are the number of black eighth graders by school and entrants by campus, respectively. T ­ hese exposure rates can be interpreted as the percentage black in the average black student’s eighth grade or branch campus. In 2004, 34.4 ­percent of all the eighth graders in the state w ­ ere black. When that cohort entered the University of North Carolina, 26.1 ­percent of entering students in the University system w ­ ere black. 49. Webster, Stockard, and Henson (1981), 285. The study excluded black students who attended institutions in the far west. Allen and Jewell (2002), 253–254, note the effects on HBCUs of the opening up of previously all-­white colleges, including a “sharp decline” in the proportion of affluent and academically strong students. 50. Khan (2011). 51. Roksa et al. (2007), 169. 52. Khan’s summary is apropos: “The difference between rich and poor p ­ eople can be understood very simply. Rich p ­ eople have more money than poor ­people. And they use that money to buy advantages for themselves and their ­children.” Khan (2011), 195. 53. Karabel and Astin (1975), 396. 54. Jencks and Riesman (1968), 97. 55. The tendency for socioeconomic stratification gaps to persist in the m ­ iddle of general expansion of college-­going has been documented in a number of developed countries, causing some sociologists to conclude that the inequities in the system of postsecondary education are firmly entrenched and not threatened by expansion. One hypothesis consistent with this stylized fact is “maximally maintained in­equality,” which implies that the ­children of parents with modest means w ­ ill not fill places u ­ ntil ­those of high-­status parents have all done so. This implies that expansion of higher education opportunities ­w ill not automatically lead to more equality. A key to the preservation of the privileged position of the affluent is differentiating across higher education institutions, with the prestigious research universities on top and the less prestigious institutions emphasizing

398

NOTES TO P AGES 2 5 0 – 2 5 5

vocational skills below. This applies to centralized and “diversified” systems like those in the United States, Japan, and Korea, that have higher percentages of students attending private institutions. One finding is that countries with larger private sectors (such as the United States, Japan, and K ­ orea) also have higher percentages of young ­people in higher education. Arum, Gamoran, and Shavit (2007), 3, 5, 23. 10: Sorting by Seriousness

1. Angell (1928), 4. 2. An example of an economic model of college enrollment is Lee, Shin, and Lee (2015). The authors pres­ent a two-­period model in which a person’s decision to attend and complete college is risky ­because the returns are uncertain. The authors “assume a h ­ uman capital accumulation aspect: If a student does not attend or complete college, he w ­ ill be compensated for only a fraction of zi in the l­ abor market,” where zi is a mea­sure of the person’s financial return to college (10). 3. Clark and Trow’s (1966) two-­way classification of students can be diagramed as follows: Involvement with ideas Identified with institution

Much

­L ittle

Much ­L ittle

“academic” “nonconformist”

“collegiate” “vocational”



In his 1928 book, Angell offered his own typology, one that anticipates Clark and Trow’s 1966 one: “A small minority are sincerely interested in all their academic work; a larger minority do not put their hearts into any of it; while the g­ reat mass are genuinely intent upon only a few of their subjects, commonly the more practical ones, and apathetic t­ oward the rest” (2). More recent ethnographic studies of college life paint a not radically dif­fer­ent portrait, suggesting that socializing and dating occupy a large share of students’ time and attention; ­A fter living in a freshman dormitory, for example, Nathan (2005), 101, reports: “what I observed was that engagement in the philosophical and po­l iti­cal issues of the day was not a significant part of college student culture.” See also Armstrong and Hamilton (2013) and Holland and Eisenhart (1990). 4. Holland and Eisenhart (1990), 163. 5. Ibid., 164–165. 6. Ibid., 167. 7. Armstrong and Hamilton (2013), 51. 8. Ibid., 81–83, 221.

NOTES TO P AGES 2 5 6 – 2 5 9

399

9. Using requests about sending SAT scores to colleges as a proxy for applications, Pope and Pope (2009) estimated models using data for 332 universities with big-­time football and basketball programs for the years 1983 to 2002, with fixed effects for college and year and with a separate linear time trend for each institution. Colleges whose football teams won a national championship enjoyed a 7–8 ­percent increase in applications. T ­ hose whose teams ranked in the top twenty nationally saw an increase of 2.5 ­percent. Basketball success also produced surges of applicants, such as advancing to the round of sixteen in the NCAA national tournament for men. All of ­these effects ­were short-­l ived, disappearing ­a fter one or two years. 10. Alter and Reback (2014). 11. Jacob, McCall, and Stange (2013). 12. Babcock and Marks (2011). 13. Brint and Cantwell (2014). The survey, administered via the internet, had a 33 ­percent response rate and, according to the authors, was prob­ably disproportionately made up of industrious students. All branches except Merced ­were covered, though the return rates differed by branch. The sample size was 8,823. The sample averages for hours spent per week in vari­ous activities ­were watching tele­v i­sion, 4.2 hours; surfing the internet “for fun,” 11.7 hours; socializing with friends, 10.3 hours; and partying, 3.2 hours. 14. Sperber (2005), 5, calls it a “non-­aggression pact.” Other commentators who have suggested the notion of a faculty-­student agreement are Reisman (1980) and Armstrong and Hamilton (2013). 15. Kutner et al. (2007). 16. Arum, Roksa, and Cho (2012). 17. The general question was, “During your last year in high school, how much time did you spend during a typical week ­doing the following activities?” The 1998 / 99 surveys asked about nine activities: studying / homework, socializing with friends, talking with teachers outside of class, exercise or sports, partying, working (for pay), volunteer work, student clubs / groups, and watching TV. The 2008 / 09 added four more: h ­ ouse­hold / childcare duties, reading for plea­sure, playing video / computer games, and online social networks (MySpace, Facebook, e­ tc.). The weighted means for the number of hours for all first-­year students are given in the footnotes of the figures. Means for activities not shown ­were: exercise or sports, 7.1 in wave 2 and 8.2 in wave 3; and student clubs / groups, 3.1 and 2.7, respectively. For wave 3, the weighted mean hours for the four additional activities w ­ ere ­house­hold / childcare duties, 2.4 hours; reading for plea­sure, 2.4; playing video / computer games, 2.0; and online social networks (MySpace, Facebook, ­etc.), 3.6. The total weighted average number of hours for the nine activities in wave 2 was 52.5 hours, and the total for the thirteen activities was 56.6 hours.

400

NOTES TO P AGES 2 6 0 – 2 6 3

For all the activities, eight categorical answers w ­ ere pos­si­ble, from “None” to “Over 20 hours.” To restate t­ hese survey results in terms of average hours of school work per week, each student was assigned a value for hours equal to the midpoint of the category answered. The top category (Over 20 hours) was assigned twenty hours. Respondents whose answers ­were unreasonable or suggested lack of seriousness in answering the question w ­ ere omitted from the calculations of time use. T ­ hese included students who gave the same answer to e­ very question with a 6 (11 to 15 hours), 7 (16 to 20 hours), or 8 (Over 20 hours), or who answered 7 or 8 to ­every question, or whose total hours exceeded 168, the total number of hours in a week. 18. As in other applications of the Freshman Survey, the standard errors applying to t­ hese estimates tended to be small. For Figure 10.1, the largest standard errors w ­ ere for the private colleges in the 99th and above category: 0.9 ­percent for 1998 / 99 and 1.0 ­percent for 2007 / 08. 19. ­These striking changes in reported studying cannot, by the way, be explained by changes in the gender mix that occurred at many colleges. Although females do report that they studied more in high school than males, and ­women’s share of college students did increase in the sampled colleges (from 50 ­percent in 1972 to 56 ­percent in the two ­later waves of data, using the enrollment-­weighted means), statistically controlling for gender has l­ ittle effect on the pattern shown in Figure 10.1. 20. Expressing the averages as deviations also has the virtue of correcting for general biases that might occur when more activities w ­ ere added to the survey, which they ­were between 1989 / 90 and 2008 / 09. 21. To calculate the average for any category, simply add the deviation to the national mean. For the 1989 / 90 wave, for example, mean minus the (negative) deviation was 6.1 − 1.1 = 5.0. 22. The average for the highest private category was 6.1 + 4.1 = 10.2. 23. The connection between study time and GPA extends into college. A study covering college students at 28 selective institutions shows that high school GPA is the best predictor of study time in college. The second-strongest predictor was enrollment at a private liberal arts college. Massey and Mooney (2007), T ­ able 1, 110. 24. Like the patterns for volunteering, reported time spent talking to teachers outside of class ­were almost all below the national average for private college students in 1989 / 90; they ­were above the national average in 2008 / 09. ­These findings are available in the online appendix. 25. Demerath (2009), 18. See Chapter 8 for discussion of this phenomenon. 26. As noted above, the six institutions in this group include Georgia Tech and three of the ser­vice academies, surely not a representative sample of all selective public universities.

NOTES TO P AGES 2 6 4 – 2 7 4

401

27. The corresponding question in the 2008 / 09 wave added clubs, making the answers incomparable. 11: Sorting by Belief?

1. The guide is published by the Cardinal Newman Society, accessed January 20, 2017, https://­cardinalnewmansociety​.­org​/­program​/­the​-­newman​-­guide​/.­ 2. The Council for Christian Colleges and Universities, whose mission is “to advance the cause of Christ-­centered higher education and to help our institutions transform lives by faithfully relating scholarship and ser­v ice to biblical truth,” offers a list of recommended colleges on its website, www​ .­cccu​.­org​/.­ 3. For a discussion of the Christian Right, see Wacker (2016). For information on the po­l iti­cal activities of Jerry Falwell, see Michael Duffy, “Jerry Falwell, Po­l iti­cal Innovator,” Time, May 15, 2007, and for statements of the founding princi­ples of Liberty and Regent Universities, see their websites: www​.­l iberty​ .­edu​/­a nd www​.­regent​.­edu​/.­ 4. For a discussion and many sources regarding the liberal bias of many faculty, see Gross and Fosse (2012). 5. Whereas the Catholic colleges ­were easy to identify, the Protestant evangelical ones ­were chosen based on a close inspection of their current websites. The websites of four of the five colleges clearly indicated an explicit emphasis on Bible-­based, fundamentalist Protestant faith, the fifth, Lenoir-­R hyne College, exhibited ­l ittle of this, other than its Lutheran roots. This college was included, however, due to the requirement that no category be smaller than five. The Catholic colleges in the 25th to 50th percentiles w ­ ere Benedictine College, Gonzaga University, Loyola University / New Orleans, Sacred Heart University, Seton Hall University, and Spring Hill College; the colleges deemed to be evangelical Protestant colleges ­were Eastern Mennonite University, Erskine College and Seminary, Huntington University, Lenoir-­ Rhyne University, and Waynesburg University; the remaining colleges in the “larger” category w ­ ere Buena Vista University, Chapman University, Dominican College of Blauvelt, Emory and Henry College, Geneva College, Marietta College, Monmouth College, Pacific University, Simpson College, Stephens College, the University of Tampa, Wesleyan College (GA), and Western New E ­ ngland University. 6. Morison (1965), 3. 7. Whereas the 1972 survey offered five religious groups from which to choose, the 1989 and 1990 surveys offered seventeen religious groups, ten of which ­were Protestant denominations, and the 2008 and 2009 surveys offered nineteen religious groups, of which eleven ­were Protestant. According to one

402

NOTES TO P AGES 2 7 8 – 2 8 5

official at the Higher Education Research Institute, the man­ag­ers of the Freshman Survey, the small number of choices in 1972 is the probable cause for the high percentage of HBCU students who picked “Other,” saying that it was likely that “students at HBCUs did not have a complete understanding of the term ‘Protestant,’ which moved many of them into ‘Other.’ They may have been looking for a more specific denomination given that Catholic and Jewish w ­ ere listed” (email correspondence from Kevin Eagan, June 16, 2016). Supporting the hypothesis that more students picked “Other” when t­ here ­were just five categories, the national percentage for all freshmen in the nation who chose Other fell by almost 10 percentage points from 1972 to 1973 (from 12.0 to 3.5 ­percent) when the number of categories was increased from five to seventeen. Consequently, in 1972 the nationwide percentage Protestant was 41.2, but in 1973 it was 50.2. National percentages are presented in Pryor et al. (2007). 8. American National Election Studies, Guide to Public Opinion and Electoral Be­hav­ior, ­Table 1B.5b, accessed June 12, 2016, www​.­electionstudies​.­org​ /­nesguide​/­toptable​/­tab1b​_ ­5b​.­htm. 9. This is the widely used Herfindahl index, defined in this application as H = ∑ (pi)2, where pi is the proportion of students who indicated religious group i. 10. The percentage of first-­year students who identified as Catholic in t­ hese six colleges fell from 78 ­percent in 1972 to 67 ­percent in 1989 / 90 and to 56 ­percent in 2008 / 09. 11. See Appendix Figure 11.1 in the online appendix. 12. The same correlation can be seen in the percentage of students who reported having no religious affiliation or who did not respond to the question. For this indicator, our four-­decade history shows a long-­term trend away from religious affiliation but not a steady fall. The enrollment-­weighted mean for the country went from 18 ­percent in 1972, to 16 ­percent in 1989 / 90, and then ­rose to 23 ­percent in 2008 / 09. Students at the most selective colleges ­were consistently some 15 percentage points more likely than average to be in this nonreligious group, with the propensity to be so classified rising steadily from the 60th-­to 70th-­percentile category all the way to the top. The same positive relationship between 1970 SAT and lack of religious affiliation can be seen among students at public colleges as well. Perhaps the most notable trend was in the HBCUs, where, by 2008 / 09, their students ­were reporting rates of religious affiliation more than 10 percentage points higher than the national average 13. See Scott Jaschik, “Moving Further to the Left,” Inside Higher Ed, October 24, 2012, accessed May 3, 2017, https://­w ww​.­insidehighered​.­com​/­news​/­2012​/­10​ /­24​/­survey​-­fi nds​-­professors​-­a lready​-­l iberal​-­have​-­moved​-­f urther​-­left; and Rothman, Lichter, and Nevitte (2005).

NOTES TO P AGES 2 8 7 – 2 9 6

403

14. As evidence, Bishop (2008), 302, shows that more Americans lived in counties with lopsided Demo­cratic or Republican majorities in presidential elections in 2000 than was the case in 1970. 15. Ibid. 16. See, for example, Nicholas DeFonzo, “The Echo-­Chamber Effect,” New York Times, April 22, 2011, accessed May 3, 2017, https://­w ww​.­nytimes​.­com​ /­roomfordebate​/­2011​/­04​/2 ­ 1​/ ­barack​- ­obama​-­a nd​-­the​-p ­ sychology​- ­of​-­the​ -­birther​-­myth​/­the​-­echo​-­chamber​-­effect. 17. De Tocqueville (2000). 18. Pascarella and Terenzini (2005), 275–276, 287, cite evidence of a tendency to move away from the po­l iti­cal center, ­either to the right or left, as well as a modest move t­ oward liberalism on social issues. 19. Skocpol (2003). 12: Outcomes

1. Charles Dickens, Hard Times, 1854, chap. 1. Proj­ect Gutenberg eBook, accessed May 11, 2017, http://­w ww​.­g utenberg​.o ­ rg​/­fi les​/­786​/­786​- ­0.​ ­t xt. 2. ­Because so much depends on interactions of this sort, few studies have uncovered substantial collegewide, or between-­college, effects. Write Pascarella and Terenzini (2005), 147, “The majority of colleges and universities in the American postsecondary system have impor­tant subenvironments with more immediate and power­f ul impacts on individual students.” 3. See Pascarella and Terenzini (1991), 107, 108, and Pascarella and Terenzini (2005), 146. 4. Pascarella and Terenzini (1991), 127. 5. Pascarella and Terenzini (1991), 157, and Pascarella and Terenzini (2005), 168–169. 6. Pascarella and Terenzini (1991), 157, and Pascarella and Terenzini (2005), 190–191. 7. For a discussion of conceptual issues involved in assessing the impact of college, see Astin (1993), chap. 1. 8. Bound, Lovenheim, and Turner (2010), 133. 9. Ibid., 135. 10. For an exhaustive review of the lit­er­a­t ure on the psychosocial effects of college, see Pascarella and Terenzini (1991), chaps. 5–8. 11. The comparisons group individuals by years of education, with t­ hose having sixteen or more years counted as having finished four years of college. 12. See, for example, Peterson and Kern (1996). 13. Oreopoulos and Salvanes (2011), 161, 163, 167, 170. The question used for trust was: “Generally speaking, would you say that most ­people can be

404

NOTES TO P AGES 2 9 7 – 3 0 9

trusted or that you ­can’t be too careful in dealing with ­people?” Ability to delay gratification was based on not agreeing with the statement, “Nowadays, a person has to live pretty much for ­today and let tomorrow take care of itself.” 14. Sacerdote (2001). 15. Carrell, Hoekstra, and West (2015). 16. Autor (2014), 844. 17. See, for example, Freeman (1976); Clotfelter, Ehrenberg, Getz, and Siegfried (1991), 64–69; and Clotfelter (1996), 62. 18. Autor (2014), 844. 19. From 1904 to 1964, out of 1,698 winners, Harvard had 115 Rhodes Scholar winners, barely edging out Prince­ton, with 113; see Pierson (1969), 17, 208–209. 20. Useem and Karabel (1986). 21. Hacker and Dreifus (2011), chap. 12, use admission to Harvard Law School, specifically the share of each college’s gradu­ates who enrolled t­ here between 2002 and 2008, as a mea­sure of gradu­ates’ success. They reasoned, “Why not let Harvard Law tell us how it ranks undergraduate colleges?” (67). 22. Facebook pages provided by the law schools identified the undergraduate college of each student. In addition to data for the classes of 2014–2016 at each school, I obtained books for Harvard’s classes of 1974–1977, NYU’s classes of 1976–1978, and Duke’s classes of 1977–1979. 23. Blau and Margulies (1974–1975), 44. 24. Paul Caron, “The First Law School Rankings (1975),” TaxProof (blog), October 14, 2006, accessed July 23, 2015, http://­taxprof​.­t ypepad​.­com​ /­taxprof​_­blog​/­2006​/­10​/­the​_­first​_­law​_­s​.­html. 25. Duke was tied for eighth with two other schools. See U.S. News Rankings and ­ snews​.­rankingsandreviews​ Reviews, accessed July 23, 2015, http://­grad​-s­ chools​.u .­com​/­best​-­graduate​-­schools​/­top​-­law​-­schools​/­law​-­rankings. 26. U​. ­S. Department of Education, Digest of Education Statistics 2013, May 2015, ­Table 318.30, 535–549. The 170,062 doctoral degrees awarded in 2011 / 12 included 46,836 law degrees, 9,990 degrees in education, and 45,119 in medicine and related clinical specialties, such as medicine, nursing, physical therapy, and chiropractic practice. The difference is 68,117, counted ­here as PhDs, or 3.8 ­percent of the total for bachelor’s degrees. 27. ­B ecause of the low numbers of recipients from colleges in some groups, colleges w ­ ere grouped into six categories, which are combinations of the seventeen basic categories: private, 0 to 50th percentile; private 50th to 90th; private 90+; public 0 to 50th; public 50th to 90th; and public 90+.

NOTES TO P AGES 3 0 9 – 3 1 8

405

28. Corak (2013). Intergenerational economic mobility is mea­sured by the elasticity between a ­father’s earnings and ­those of a son’s adult earnings, with a numerically large elasticity indicating less mobility. 29. Aaronson and Mazumder (2008). Chetty et al. (2016) provide related findings, showing that the chance that c­ hildren ­w ill earn more than their parents has declined over the last fifty years in the United States. 30. In cases in which data for public university systems ­were combined, I applied the system data to the largest, usually the flagship, campus. 31. ­These two conditions mean that the index is calculated simply as the product of t­ hose two percentages, the first a mea­sure of access for students from low-­income families and the second a mea­sure of economic success for t­ hose students. 32. The pres­ent value of lifetime earnings for gradu­ates of colleges in the highest SAT category was about $8.5 million, compared to about $4.5 million for colleges at the 1000 SAT level; see Hoxby (2016b), fig. 1. 33. See, for example, Brewer, Eide, and Ehrenberg (1999). 34. Hoekstra (2009). 35. See Hoxby (2009), 22. 36. For each selective college, students w ­ ere considered “on the b ­ ubble” if their SAT scores fell in a range implying a probability of admission between 40 ­percent and 60 ­percent for that college. For students in this range, admission is assumed to be random. Writes Hoxby (2015), “In the ­bubble range, admissions probabilities are such that staff appear to be flipping coins. Of course, staff are not actually flipping coins. Rather, they are influenced by minor considerations that could not plausibly have a major effect on long-­term outcomes except through the admissions decision.” (4). 37. This is the Paired Comparison Method used in tennis rankings. It requires overlap, or common support. See Hoxby (2016b). This approach rests on the assumption that admissions decisions for pairs of observationally identical students are more or less random, not driven by decisions of college admissions officers who know about traits not observed by the researcher. 38. Hoxby notes that this empirical finding supports a longstanding assumption made in theoretical models of education and the l­ abor market, the so-­called single-­crossing assumption. This is the assumption that the pro­cess of education has a lower marginal cost for students with higher ability than it does for students with less ability; see Hoxby (2016a), 1. For an example of the assumption of single crossing, see Daron Acemoglu, Lectures in L ­ abor Economics, 14.661, 2016, accessed November 24, 2016, http://­ economics​.­m it​.­edu​/­fi les​/­12265. 39. For details see Hoxby (2016b).

406

NOTES TO P AGES 3 2 8 – 3 3 3

40. Schwartz and Mare (2005). 41. Greenwood et al. (2014) estimated that without educational homogamy the Gini coefficient would have been 0.34 rather than 0.43, its a­ ctual value. 42. See Murray (2012), 52–68, who connects marriage and college stratification (“the college sorting machine”), predicting that increased educational homogamy w ­ ill produce increased cognitive homogamy. 13: Why It M ­ atters

1. Pye (1980). 2. Illinois State University, University FactBook: Fall 2014, http://­prpa​ .­i llinoisstate​.­edu​/­Book​-­14​.­pdf. Information on new buildings available at http://­masterplan​.­illinoisstate​.­edu​/­downloads​/­Master percent20Plan percent202010-2030.pdf. Faculty compensation based on surveys conducted and published by the American Association of University Professors in 1971 and 2016; Peter O. Steiner, Maryse Eymonerie, and William B. Woolf (1971), “At the Brink: Report on the Economic Status of the Profession, 1970–71,” AAUP Bulletin 57 (2): 223–285; Steven Shulman et al. (2016) “Higher Education at a Crossroads: The Economic Value of Tenure and the Security of the Profession,” Academe (March–­April): 9–13, with appendix. 3. Boston College, “Facts & Figures,” accessed January 20, 2017, www​.­bc​.­edu​/­bc​ -­web​/­about​/­bc​-­facts​.­html; Corinne Duffy, “A Look at BC Master Planning: Past, Pres­ent, and ­Future,” The Heights, September 25, 2015, http://­bcheights​ .­com​/­2015​/­09​/­21​/­a​-­look​-­at​-­bc​-­master​-­planning​-­past​-­present​-­a nd​-­f uture​/.­ See note 2 for source of faculty compensation. 4. Jackie Bowers, “Common Data Set: Fall 2016,” Tennessee Wesleyan University, Office of Institutional Research and Effectiveness, www​.­t nwesleyan​.­edu​ /­wordpress​/w ­ p​- ­content​/­uploads​/2 ­ 017​/0 ­ 1​/­CDS​_ ­2016​-­2017​_­web​_­v3​.­pdf. See note 2 for source of faculty compensation. ­Because the AAUP did not report compensation for Tennessee Wesleyan for 1970 / 71, figures for 1971 / 72 w ­ ere used and deflated for the national average increase in faculty compensation of 4.3 ­percent. 5. Chetty et al. (2017), appendix, T ­ able 2. 6. The findings of Roksa et al. (2007) suggest that the expansion of higher education has not reduced in­equality in educational attainment (chap. 7, 190). 7. See Stevens (2009), chap. 4. 8. For a description of their work and that of o ­ thers to address “undermatching,” see Beckie Supiano, “The $6 Solution,” Chronicle of Higher Education, June 16, 2014, accessed May 3, 2017, www​.­chronicle​.­com​/­a rticle​ /­The​- ­6 -​ ­Solution​/1 ­ 47125​/.­

NOTES TO P AGES 3 3 5 – 3 5 1

407

9. The methodology used in the Washington Monthly rankings illustrates how one might adjust for the background of students who enroll. 10. Morison (1965), 4. 11. Jencks and Riesman (1968), 97. 12. Ibid., 99–107. 13. Young (1958) and Celarent (2009). 14. A Bill for the More General Diffusion of Knowledge, June 18, 1779, in McClure and Looney (2016), chap. 79.

408

NOTES TO P AGES 3 5 2 – 3 5 4

References

Aaronson, Daniel, and Bhashkar Mazumder. 2008. “Intergenerational Economic ­ uman Resources 43 Mobility in the United States, 1940 to 2000.” Journal of H (Winter): 139–172. Abrams, Samuel J. 2005. “Unflagged SATs.” Education Next 5 (3) (Summer). ­ nflaggedsats​/.­ Accessed May 6, 2017. http://­educationnext​.­org​/u Allen, Walter R., and Joseph 0. Jewell. 2002. “A Backward Glance Forward: Past, Pres­ent, and ­Future Perspectives on Historically Black Colleges and Universities.” Review of Higher Education, 25(3): 241–261. Alter, Molly, and Randall Reback. 2014. “True for Your School? How Changing Reputations Alter Demand for Selective U.S. Colleges.” Educational Evaluation and Policy Analy­sis 36 (March): 346–370. Alveredo, Facundo, Anthony B. Atkinson, Thomas Piketty, and Emmanuel Saez. 2013. “The Top 1 ­Percent in International and Historical Perspective.” National Bureau of Economic Research Working Paper 19075 (May). Ames, Mary Ellen. 1984. “The Admissions Picture: A Sobering Look Ahead.” REALIA (June): 2. Angell, Robert Cooley. 1928. The Campus: A Study of Con­temporary Undergraduate Life in the American University. New York: Appleton. Armstrong, Elizabeth A., and Laura T. Hamilton. 2013. Paying for the Party: How College Maintains In­equality. Cambridge, MA: Harvard University Press. Arum, Richard, Adam Gamoran, and Yossi Shavit. 2007. “More Inclusion Than Diversion: Expansion, Differentiation, and Market Structure in Higher Education.” In Shavit, Arum, and Gamoran 2007, 1–35. Arum, Richard, and Josipa Roksa. 2011. Academically Adrift: Limited Learning on College Campuses. Chicago: University of Chicago Press.

Arum, Richard, Josipa Roksa, and Esther Cho. 2012. Improving Undergraduate Learning: Findings and Policy Recommendations from the SSRC-­CLA Proj­ect. New York: Social Science Research Council. ­ atters in College: Four Critical Years Revisited. San Astin, Alexander. 1993. What M Francisco: Jossey Bass. Astin, Alexander W., and Leticia Oseguera. 2004. “The Declining ‘Equity’ of American Higher Education.” Review of Higher Education 27 (3): 321–341. Atkinson, Anthony B., Thomas Piketty, and Emmanuel Saez. 2011. “Top Incomes in the Long Run of History.” Journal of Economic Lit­er­a­ture 49: 3–71. Auten, Gerald E., Charles T. Clotfelter, and Richard L. Schmalbeck. 2000. “Taxes and Philanthropy Among the Wealthy.” In Does Atlas Shrug? The Economic Consequences of Taxing the Rich, edited by Joel Slemrod, 392–424. New York: Russell Sage Foundation and Harvard University Press. Autor, David H. 2014. “Skills, Education and the Rise of Earnings In­equality among the ‘Other 99 ­Percent.’ ” Science 344 (6186): 843–851. Avery, Christopher, and Caroline M. Hoxby. 2004. “Do and Should Financial Aid Packages Affect Students’ College Choices?” In College Choices: The Economics of Where to Go, When to Go, and How to Pay for It, edited by Caroline M. Hoxby, 239–290. Chicago: University of Chicago Press. Avery, Christopher, Andrew Fairbanks, and Richard Zeckhauser. 2003. The Early Admission Game: Joining the Elite. Cambridge, MA: Harvard University Press. Babcock, Philip, and Mindy Marks. 2011. “The Falling Time Cost of College: Evidence from Half a ­Century of Time Use Data.” Review of Economics and Statistics 93 (May): 468–478. Bailey, Martha, and Susan Dynarski. 2011. “Gains and Gaps: Changing In­equality in U.S. College Entry and Completion.” In Duncan and Murnane 2011, 117–131. Bastedo, Michael N., and Ozan Jaquette. 2011. “­Running in Place: Low-­Income Students and the Dynamics of Higher Education Stratification.” Educational Evaluation and Policy Analy­sis 33 (3): 318–339. Baum, Sandy. 2014. Higher Education Earnings Premium: Value, Variation, and Trends. Washington, DC: Urban Institute (February). Baumol, William J., and Maryse Eymonerie. 1970. “Rising Costs and the Public Institutions: The Annual Report on the Economic Status of the Profession, 1969–1970.” AAUP Bulletin 56 (June). Beaudry, Paul, David A. Green, and Benjamin M. Sand. 2013 “The G ­ reat Reversal in the Demand for Skill and Cognitive Tasks.” National Bureau of Economic Research Working Paper 18901 (March). Becker, Gary S., William H. J. Hubbard, and Kevin M. Murphy. 2010. “The Market for College Gradu­ates and the Worldwide Boom in Higher Education for W ­ omen.” American Economic Review: Papers and Proceedings 100 (2): 229–233.

410

References

Belasco, Andrew S., Kelly O. Rosinger, and James C. Hearn. 2014. “The Test-­ Optional Movement at Amer­i­ca’s Selective Liberal Arts Colleges: A Boon for Equity or Something Else?” Educational Evaluation and Policy Analy­sis 37 (June): 206–223. Belley, Philippe, and Lance Lochner. 2007. “The Changing Role of F ­ amily Income and Ability in Determining Educational Achievement.” Journal of H ­ uman Capital 1 (1): 191–206. Binstock, Jeanne Schmidt. 1970. Design from Disunity: The Tasks and Methods of American Colleges. Unpublished dissertation, Brandeis University (February). Bishop, Bill. 2008. The Big Sort. New York: Houghton Mifflin. Blau, Peter M., and Rebecca Zames Margulies. 1974–1975. “A Research Replication: The Reputations of American Professional Schools.” Change 6: 42–47. Bledstein, Burton J. 1976. The Culture of Professionalism: The ­Middle Class and the Development of Higher Education in Amer­i­ca. New York: W. W. Norton. Botstein, Leon. 2005. “The Curriculum and College Life: Confronting Unfulfilled Promises.” In Hersh and Merrow, 209–228. Bound, John, Breno Braga, Gaurav Khanna, and Sarah Turner. 2016. “A Passage to Amer­i­c a: University Funding and International Students.” Unpublished paper (March 27). Bound, John, Brad Hershbein, and Bridget Terry Long. 2009. “Playing the Admissions Game: Student Reactions to Increasing College Competition.” Journal of Economic Perspectives 23 (4): 119–146. Bound, John, Michael F. Lovenheim, and Sarah Turner. 2010. “Why Have College Completion Rates Declined? An Analy­sis of Changing Student Preparation and Collegiate Resources.” American Economic Journal: Applied Economics 2 (3): 129–157. Bound, John, and Sarah Turner. 2006. “Cohort Crowding: How Resources Affect College Attainment.” National Bureau of Economic Research Working Paper 12424. Bourdieu, P. 1973. “Cultural Reproduction and Social Reproduction.” In Knowledge, Education, and Cultural Change, edited by R. Brown, 71–112. London: Tavistock. Bowen, Howard R. 1970. “Financial Needs of the Campus.” In The Corporation and the Campus: Corporate Support of Higher Education in the 1970s, 75–93. New York: Acad­emy of Po­l iti­cal Science. —­—­—. 1980. The Costs of Higher Education. San Francisco: Jossey-­Bass. Bowen, William G., Martin A. Kurzweil, and Eugene M. Tobin. 2005. Equity and ­ irginia Excellence in American Higher Education. Charlottesville: University of V Press. Bowles, Samuel, and Herbert Gintis. 1976. Schooling in Cap­i­tal­ist Amer­i­ca. New York: Basic Books.

R e f e r e n c e s

411

Bowman, N. A., and M. N. Bastedo. 2009. “Getting on the Front Page: Orga­ nizational Reputation, Status Signals, and the Impact of U.S. News and World Report on Student Decisions.” Research in Higher Education 50: 415–436. Breland, Hunter, James Maxey, Renee Gernand, Tammie Cumming, and Catherine Trapani. 2002. Trends in College Admission 2000: A Report of a Survey of Undergraduate Admissions Policies, Practices, and Procedures. National Association for College Admission Counseling (March). Breneman, David W. 1994. Liberal Arts Colleges: Thriving, Surviving, or Endangered? Washington, DC: Brookings Institution. Brewer, Dominic, Eric Eide, and Ronald Ehrenberg. 1999. “Does It Pay to Attend an Elite College? Cross Cohort Evidence on the Effects of College Type on ­ uman Resources 34 (Winter): 104–123. Earnings.” Journal of H Brint, Steven. 2005. “Creating the F ­ uture: The ‘New Directions’ in American Research Universities.” Minerva 43: 23–50. —­—­—. 2012. “The Orientations of US College Students: Revising the Clark-­Trow Typology.” Bildung und Erziehung 65: 195–206. Brint, Steven, and Allison M. Cantwell. 2014. “Conceptualizing, Mea­sur­ing, and Analyzing the Characteristics of Disengaged Student Populations: Results from UCUES 2010.” Journal of College Student Development 55: 808–823. Brint, Steven, Robert Hanneman, Kerry Mulligan, Kristopher Proctor, and Matthew Rotondi. 2012. “Declining Academic Fields in US Four-­Year Colleges and Universities, 1970–2006.” Journal of Higher Education 83: 583–613. Brint, Steven, Robert Hanneman, Scott Murphy, and Kristopher Proctor. 2012. “The Market Model and the Growth and Decline of Academic Fields in US Four-­Year Colleges and Universities, 1980–2000.” So­cio­log­i­cal Forum 27: 275–299. Brint, Steven, Kristopher Proctor, Scott Patrick Murphy, Lori Turk-­Bicakci, and Robert A. Hanneman. 2009. “General Education Models: Continuity and Change in the U.S. Undergraduate Curriculum, 1975–2000.” Journal of Higher Education 80: 605–642. Brint, Steven, Mark Riddle, and Robert A. Hanneman. 2006. “Reference Sets, Identities, and Aspirations in a Complex Orga­nizational Field: The Case of American Four-­Year Colleges and Universities.” Sociology of Education 79 (July): 229–252. Brown, Jeffrey R., Stephan G. Dimmock, and Scott Weisbenner. 2012. “The Supply and Demand for Charitable Donations to Higher Education.” Paper presented at National Bureau of Economic Research Conference, Cambridge, MA (September). Buchmann, Claudia, Dennis Condron, and Vincent Roscigno. 2010. “Shadow Education, American Style: Test Preparation, the SAT, and College Enrollment.” Social Forces 89: 435–461.

412

References

Buchmann, Claudia, and Thomas A. DiPrete. 2006. “The Growing Female Advantage in Higher Education: The Role of Parental Resources and Academic Achievement?” American So­cio­log­i­cal Review 71(4): 515–541. Buddin, R., J. J. Cordes, and S. Kirby. 1998. “School Choice in California: Who Chooses Private Schools?” Journal of Urban Economics 44: 110–134. Burd, Stephen. 2013. Undermining Pell: How Colleges Compete for Wealthy Students and Leave the Low-­Income ­Behind. New Amer­i­ca Foundation (May). Burke, Colin B. 1982. American Collegiate Populations: A Test of the Traditional View. New York: New York University Press. Carrell, Scott E., Mark Hoekstra, and James E. West. 2015. “The Impact of Intergroup Contact on Racial Attitudes and Revealed Preferences.” National Bureau of Economic Research Working Paper 20940 (May). Carrell, Scott E., Frederick V. Malmstrom, and James E. West. 2008. “Peer Effects ­ uman Resources 43 (Winter): 173–207. in Academic Cheating.” Journal of H Cass, James, and Max Birnbaum. (1964) 1972. Comparative Guide to American Colleges. New York: Harper and Row. Celarent, Barbara. 2009. Review of The Rise of the Meritocracy, 1870–2033 by Michael Young. American Journal of Sociology 115 (July): 322–326. Chatterji, Aaron J., Joowon Kim, and Ryan C. McDevitt. 2016. “School Spirit: Legislators’ School Ties and State Funding for Higher Education.” Unpublished paper (June). Cheit, Earl F. 1971. The New Depression in Higher Education: A Study of Financial Conditions at 41 Colleges and Universities. New York: McGraw-­H ill. Chetty, Raj, John N. Friedman, Emmanuel Saez, Nicholas Turner, and Danny Yagan. 2017. “Mobility Report Cards: The Role of Colleges in Intergenerational Mobility.” Unpublished paper (January). Chetty, Raj, David Grusky, Maximilian Hell, Nathaniel Hendren, Robert Manduca, and Jimmy Narang. 2016. “The Fading American Dream: Trends in Absolute Income Mobility Since 1940.” National Bureau of Economic Research Working Paper 22910 (December). Chickering, Arthur W., and Zelda Gamson. 1987. “Seven Princi­ples for Good Practice in Undergraduate Education.” AAHE Bulletin 39: 3–7. Clark, Burton, and Martin Trow. 1966. “The Orga­nizational Context.” In College Peer Groups: Prob­lems and Prospects for Research, edited by T. M. Newcomb and E. K. Wilson, 19–26. Chicago: Aldine. Clotfelter, Charles T. 1985. Federal Tax Policy and Charitable Giving. Chicago: University of Chicago Press. —­—­—. 1996. Buying the Best: Cost Escalation in Elite Higher Education. Prince­ton, NJ: Prince­ton University Press. —­—­—. 2004. ­After Brown: The Rise and Retreat of School Desegregation. Prince­ton, NJ: Prince­ton University Press.

R e f e r e n c e s

413

—­—­—. 2011. Big-­Time Sports in American Universities. New York: Cambridge University Press. —­—­—. 2015. “Die-­Hard Fans and the Ivory Tower’s Ties that Bind.” Social Science Quarterly 96 (June): 381–399. Clotfelter, Charles T., and Philip J. Cook. 1989. Selling Hope: State Lotteries in Amer­i­ca. Cambridge, MA: Harvard University Press. Clotfelter, Charles T., Ronald G. Ehrenberg, Malcolm Getz, and John J. Siegfried. 1991. Economic Challenges in Higher Education. Chicago: University of Chicago Press. Clotfelter, Charles T., Steven W. Hemelt, and Helen F. Ladd. 2016. “Multifaceted Aid for Low-­Income Students and College Outcomes: Evidence from North Carolina.” National Bureau of Economic Research Working Paper 22217 (May). Clotfelter, Charles T., Helen F. Ladd, and Jacob L. Vigdor. 2009. “The Academic Achievement Gap in Grades 3 to 8.” Review of Economics and Statistics 91 (May): 420–431. —­—­—. 2015. “Public Universities, Equal Opportunity, and the Legacy of Jim Crow: Evidence from North Carolina.” National Bureau of Economic Research Working Paper 21577 (September). Clotfelter, Charles T., Helen F. Ladd, Jacob L. Vigdor, and Justin Wheeler. 2007. “High Poverty Schools and the Distribution of Teachers and Principals.” North Carolina Law Review 85 (June): 1345–1379. Cohen, Paula Marantz. 2006. Jane Austen in Scarsdale. New York: St. Martin’s Press. Coleman, James S. 1973. “The University and Society’s New Demands Upon It.” In Content and Context, edited by Carl Kaysen, 359–399. New York: McGraw-­H ill. College Board. 2015. Trends in Student Aid 2015. Washington, DC: College Board. Accessed January 22, 2016. http://­trends​.­collegeboard​.­org​/­sites​/­default​/­fi les​ /­trends​-­student​-­a id​-­web​-­fi nal​-­508​-­2​.­pdf. Conroy, Pat. 2002. My Losing Season. New York: Nan A. Talese. Corak, Miles. 2013. “Income In­equality, Equality of Opportunity, and Intergenerational Mobility.” Journal of Economic Perspectives 27 (Summer): 79–102. Cornwell, C., and D. B. Mustard. 2008. “Race and the Effects of Georgia’s HOPE Scholarship.” In Who Should We Help? The Negative Social Consequences of Merit Scholarships, edited by D. E. Heller and P. Marin, 57–72. Cambridge, MA: Civil Rights Proj­ect at Harvard University. Courant, Paul, and Sarah Turner. 2016. “Faculty Deployment in Research Universities.” Paper presented at National Bureau of Economic Research conference on Productivity in Higher Education, Cambridge, MA (June 1). Dale, Stacy Berg, and Alan B. Krueger. 2002. “Estimating the Payoff to Attending a More Selective College: An Application of Se­lection on Observables and Unobservables.” Quarterly Journal of Economics 117 (4): 1491–1527.

414

References

Darolia, Rajeev, and Cory Koedel. 2016. “How High Schools Explain Students’ Initial Colleges and Majors.” CALDER Working Paper 165 (January). Davies, Scott, and David Zarifa. 2012. “The Stratification of Universities: Structural In­equality in Canada and the United States.” Research in Social Stratification and Mobility 30 (2): 143–158. Demerath, Peter. 2009. Producing Success: The Culture of Personal Advancement in an American High School. Chicago: University of Chicago Press. De Tocqueville, Alexis. (1835) 2000. Democracy in Amer­i­ca, edited and translated by Harvey C. Mansfield and Delba Winthrop. Chicago: University of Chicago Press. DiMaggio, Paul, and W. Powell. 1983. “The ‘Iron Cage’ Revisited: Institutional Isomorphism and Collective Rationality in Orga­nizational Fields.” American So­cio­log­i­cal Review 48: 147–160. Douthat, Ross. 2005. “Does Meritocracy Work?” Atlantic Monthly (November). Accessed May 1, 2017. https://­w ww​.­t heatlantic​.­com​/­m agazine​/­a rchive​/­2005​ /­11​/­does​-­meritocracy​-­work​/­304305​/.­ Doyle, Denis P., and Terry W. Hartle. 1985. “Facing the Fiscal Chopping Block.” Change 17 (July–­August): 8–10, 54–56. Doyle, William R. 2010. “Changes in Institutional Aid, 1992–2003: The Evolving Role of Merit Aid.” Research in Higher Education 51: 789–810. Duffy, Elizabeth A., and Idana Goldberg. 1998. Crafting a Class: College Admissions and Financial Aid, 1955–1994. Prince­ton, NJ: Prince­ton University Press. Duncan, Greg J., and Richard M. Murnane, eds. 2011. Whither Opportunity? Rising In­equality, Schools, and ­Children’s Life Chances. New York: Russell Sage. Dynarski, S. 2000. “Hope for Whom? Financial Aid for the M ­ iddle Class and Its Impact on College Attendance.” National Tax Journal 53: 629–661. Ehrenberg, Ronald G. 2012. “American Higher Education in Transition,” Journal of Economic Perspectives 26 (Winter): 193–216. Ehrenberg, Ronald G., and Christopher L. Smith. 2001. “The Sources and Uses of Annual Giving at Private Research Universities.” National Bureau of Economic Research Working Paper 8307 (May). Ehrenberg, Ronald G., and Ling Zhang. 2004. “Do Tenured and Tenure-­Track Faculty M ­ atter?” National Bureau of Economic Research Working Paper 10695 (August). Ehrenberg, Ronald G., Liang Zhang, and Jared M. Levin. 2006. “Crafting a Class: The Trade-­Off between Merit Scholarships and Enrolling Lower-­Income Students.” Review of Higher Education 29 (2) (Winter): 195–211. ­England, Paula. 2010. “The Gender Revolution: Uneven and Stalled.” Gender and Society 24 (2): 149–166. Espenshade, Thomas J., Chang Y. Chung, and Joan L. Walling. 2004. “Admission Preferences for Minority Students, Athletes, and Legacies at Elite Universities.” Social Science Quarterly 85 (December): 1422–1446.

R e f e r e n c e s

415

Fallows, James. 2001. “The Early-­Decision Racket.” Atlantic Monthly (September). Accessed May 6, 2017. https://­w ww​.­theatlantic​.­com​/­magazine​/­a rchive​/­2001​/­09​ /­the​-­early​-­decision​-­racket​/­302280​/.­ ———. 2005. “College Admissions: A Substitute for Quality?” In Hersh and Merrow, 39–46. Figlio, David N., Morton O. Schapiro, and Kevin B. Soter. 2013. “Are Tenure Track Professors Better Teachers?” National Bureau of Economic Research Working Paper 19406 (September). Fiske, Edward B. 1979. “The Marketing of the Colleges.” Atlantic Monthly (October). Accessed April 29, 2017. https://­w ww​.­theatlantic​.­com​/­magazine​/­archive​ /­1979​/­10​/­the​-m ­ arketing​- ­of​-­the​- c­ olleges​/­376296​/.­ Freeman, Richard B. 1976. The Overeducated American. New York: Academic Press. Fryer, Roland, and Michael Greenstone. 2010. “The Changing Consequences of Attending Historically Black Colleges and Universities.” American Economic Journal: Applied Economics 2: 116–148. Gardner, Howard. 2005. “Beyond Markets and Individuals: A Focus on Educational Goals.” In Hersh and Merrow, 97–112. Geiger, Roger. 2000. “The Era of Multipurpose Colleges in American Higher Education.” In The American College in the Nineteenth ­Century, edited by Roger Geiger, 127–152. Nashville, TN: Vanderbilt University Press. —­—­—. 2004. Knowledge and Money: Research Universities and the Paradox of the Marketplace. Stanford, CA: Stanford University Press. Glazer, Nathan. 1973. “City College.” In Academic Transformation: Seventeen Institutions ­under Pressure, edited by David Riesman and Verne A. Stadtman, 71–98. New York: McGraw-­H ill. Golden, Daniel. 2001. “Glass Floor: How Colleges Reject the Top Applicants.” Wall Street Journal, May 29. Goldin, Claudia, and Lawrence F. Katz. 1999. “The Shaping of Higher Education: The Formative Years in the United States, 1890 to 1940.” Journal of Economic Perspectives 13 (Winter), 37–62. Greenwood, Jeremy, Nezih Guner, Georgi Kocharkov, and Cezar Santos. 2014. “Marry Your Like: Assortative Mating and Income In­equality.” National Bureau of Economic Research Working Paper 19829 (January). Griffith, Amanda L. 2011. “Keeping Up with the Joneses: Institutional Changes Following the Adoption of a Merit Aid Policy.” Economics of Education Review 30: 1022–1033. Griffith, Amanda, and Kevin Rask. 2007. “The Influence of the US News and World Report Collegiate Rankings on the Matriculation Decision of High-­Ability Students: 1995–2004.” Economics of Education Review 26: 244–255. Griffith, Amanda L. and Donna S. Rothstein. 2009. “­Can’t Get T ­ here from H ­ ere: The Decision to Apply to a Selective College.” Economics of Education Review 28: 620–628.

416

References

Grigsby, Mary. 2009. College Life through the Eyes of Students. Albany: State University of New York Press. Gross, Neil, and Ethan Fosse. 2012. “Why Are Professors Liberal?” Theory and Society 41: 127–168. Hacker, Andrew, and Claudia Dreifus. 2011. Higher Education? How Colleges Are Wasting Our Money and Failing Our Kids—­and What We Can Do About It. New York: St. Martin’s Griffin. Handlin, Oscar, and Mary F. Handlin. 1970. The American College and American Culture: Socialization as a Function of Higher Education. New York: McGraw-­H ill. Handwerk, Philip, Namrata Tognatta, Richard Coley, and Drew Gitomer. 2008. Access to Success: Patterns of Advanced Placement Participation in U.S. High Schools. Prince­ton, NJ: Educational Testing Ser­v ice Policy Information Center. Hansen, W. Lee, and Burton Weisbrod. 1969. “The Distribution of Costs and Direct Benefits of Public Higher Education: The Case of California.” Journal of ­Human Resources 4 (Spring): 176–191. ­ ngland Monthly (April): Harr, Jonathan. 1984. “The Admissions Circus.” New E 48–58. Harris, Patricia. 1971. “The Negro College and Its Community.” Daedelas 100 (Summer): 720–731. Harris, Seymour E. 1972. A Statistical Portrait of Higher Education. New York: McGraw-­H ill. Hearn, James C., and Andrew S. Belasco. 2015. “Commitment to the Core: A Longitudinal Analy­sis of Humanities Degree Production in Four-­Year Colleges.” Journal of Higher Education 86 (May / June): 387–416. Hersh, Richard H., and John Merrow, eds. 2005. Declining by Degrees of Mediocrity: Higher Education at Risk. New York: Palgrave Macmillan. Hoekstra, Mark. 2009. “The Effect of Attending the Flagship State University on Earnings: A Discontinuity-­Based Approach.” Review of Economics and Statistics 91 (November): 717–724. Holland, Dorothy C., and Margaret A. Eisenhart. 1990. Educated in Romance: ­Women, Achievement, and College Culture. Chicago: University of Chicago Press. Howell, C., and S. E. Turner. 2004. “Legacies in Black and White: The Racial Composition of the Legacy Pool.” Research in Higher Education 45: 325–351. Hoxby, Caroline M. 1997. “How the Changing Market Structure of U.S. Higher Education Explains College Tuition.” National Bureau of Economic Research Working Paper 6323 (December). —­—­—. 2009. “The Changing Selectivity of American Colleges.” Journal of Economic Perspectives 23 (Fall): 1–25. —­—­—. 2015. “Computing the Value-­Added of American Postsecondary Institutions.” Unpublished paper (July).

R e f e r e n c e s

417

—­—­—. 2016a. “The Dramatic Economics of the U.S. Market for Higher Education.” National Bureau of Economic Research Reporter 3: 1–6. —­—­—. 2016b. “The Productivity of U.S. Postsecondary Institutions.” Paper presented at National Bureau of Economic Research conference on Productivity in Higher Education, Cambridge, MA (May). Humphreys, Brad R., and Michael Mondello. 2007. “Intercollegiate Athletic Success and Donations at NCAA Division I Institutions.” Journal of Sports Management 21: 265–280. Ingels, S. J., and B. Dalton. 2013. High School Longitudinal Study of 2009 (HSLS:09) First Follow-­up: A First Look at Fall 2009 Ninth-­Graders in 2012 (NCES 2014–360). U.S. Department of Education. Washington, DC: National Center for Education Statistics. Accessed May 28, 2014. http://­nces​.­ed​.­gov​/­pubs2014​ /­2014360​.­pdf#16. Jacob, Brian, Brian McCall, and Kevin Stange. Forthcoming. “College as Country Club: Do Colleges Cater to Students’ Preferences for Consumption?” Journal of L ­ abor Economics. Jaquette, Ozan, Bradley R. Curs, and Julie R. Posselt. 2015. “Tuition Rich, Mission Poor: Nonresident Enrollment Growth and the Socioeconomic and Racial Composition of Public Research Universities.” Journal of Higher Education (March): 635–673. Jellema, William W. 1971. “The Red and the Black”: Special Preliminary Report on the Financial Status, Pres­ent and Projected, of Private Institutions of Higher Learning. Association of American Colleges, typescript. Jencks, Christopher, and David Riesman. 1968. The Academic Revolution. New York: Doubleday. Kane, Thomas J. 2004. “College-­Going and In­equality.” In Social In­equality, edited by Kathryn Neckerman, 319–354. New York: Russell Sage. Kane, Thomas J., Peter R. Orszag, Emil Apostolov, Robert P. Inman, and Andrew Reschovsky. 2005. “Higher Education Appropriations and Public Universities: Role of Medicaid and the Business Cycle.” Brookings-­W harton Papers on Urban Affairs: 99–146. Karabel, Jerome. 2005. The Chosen: The Hidden History of Admission and Exclusion at Harvard, Yale, and Prince­ton. New York: Mari­ner Books. Karabel, Jerome, and Alexander W. Astin. 1975. “Social Class, Academic Ability, and College ‘Quality.’ ” Social Forces 53: 381–398. Karen, David. 1991. “ ‘Achievement’ and ‘Ascription’ in Admission to an Elite College: A Political-­Organizational Analy­sis.” So­cio­log­i­cal Forum 6: 349–380. Keller, George. 2004. Transforming a College. Baltimore: Johns Hopkins University Press. Khan, Shamus Rahman. 2011. Privilege: The Making of an Adolescent Elite at St. Paul’s School. Prince­ton, NJ: Prince­ton University Press.

418

References

Kidd, Julie Johnson. 2005. “It Is Only a Port of Call: Reflections on the State of Higher Education.” In Hersh and Merrow, 195–208. Kirp, David L. 2003. Shakespeare, Einstein and the Bottom Line: The Marketing of Higher Education. Cambridge, MA: Harvard University Press. —­—­—. 2005. “This ­Little Student Went to Market.” In Hersh and Merrow, 113–130. Kremer, Michael, and Dan Levy. 2008. “Peer Effects and Alcohol Use Among College Students.” Journal of Economic Perspectives 22 (3) (Summer): 189–206. Kutner, Mark, Elizabeth Greenberg, Ying Jin, Bridget Boyle, Yung-­chen Hsu, and Eric Dunleavy. 2007. Literacy in Everyday Life: Results for the 2003 National Assessment of Adult Literacy. Washington, DC: National Center for Education Statistics. Kutner, Max. 2014. “How to Game the College Rankings.” Boston Magazine (September). Lankford, H., and James Wyckoff. 1992. “Primary and Secondary School Choice Among Public and Religious Alternatives.” Economics of Education Review 11: 317–337. Lareau, Annette. (2003) 2011. Unequal Childhoods: Class, Race, and ­Family Life. Berkeley: University of California Press. Lau, Yan, and Harvey S. Rosen. 2015. “Are Universities Becoming More Unequal?” National Bureau of Economic Research Working Paper 21432 (July). Laughlin, Lynda. 2010. Who’s Minding the Kids? Child Care Arrangement: Spring 2005 / Summer 2006. U.S. Dept. of Commerce, Bureau of the Census (August). Lazear, Edward. 1977. “Education: Consumption or Production?” Journal of Po­liti­cal Economy 85: 569–597. Lee, Sang Yoon, Yongseok Shin, and Donghoon Lee. 2015. “The Option Value of ­Human Capital: Higher Education and Wage In­equality.” National Bureau of Economic Research Working Paper 21724 (November). Lemann, Nicholas. 1999. The Big Test: The Secret History of the American Meritocracy. New York: Farrar, Straus and Giroux. Lerner, Josh, Antoinette Schoar, and Jialan Wang. 2008. “Secrets of the Acad­emy: The ­Drivers of University Endowment Success.” Journal of Economic Perspectives 22 (Summer): 207–222. Levy, Frank, and Peter Temin. 2007. “Institutions and Wages in Post-­World War II Amer­i­ca.” Paper prepared for the New L ­ abor Market Institutions and the Public Policy Response: A Symposium to Honor Lloyd Ulman, University of California at Berkeley (October). Lewis, Harry R. 2006. Excellence without a Soul: How a ­Great University Forgot Education. New York: Public Affairs. Lifschitz, Arik, Michael Sauder, and Mitchell L. Stevens. 2014. “Football as a Status System in U.S. Higher Education.” Sociology of Education 87: 204–219.

R e f e r e n c e s

419

Lindo, Jason M., Peter Siminski, and Isaac D. Swensen. 2016. “College Party Culture and Sexual Assault.” Unpublished paper, Texas A&M University (January). Lindo, Jason M., Isaac D. Swensen, and Glen R. Waddell. 2012. “Are Big-­Time Sports a Threat to Student Achievement?” American Economic Journal: Applied Economics 4 (4): 254–274. Linneham, Frank, Christy H. Weer, and Paul Stonely. 2011. “High School Guidance Counselor Recommendations: The Role of Student Race, Socioeconomic Status and Academic Per­for­mance.” Journal of Applied Social Psy­chol­og y 41 (3): 536–558. Long, Bridget Terry. 2004. “How Have College Decisions Changed Over Time? An Application of the Conditional Logistic Choice Model.” Journal of Econometrics 121: 271–296. —­—­—. 2011. “The New Financial Aid Policies: Their Impact on Access and Equity for Low-­I ncome Students?” Unpublished draft, Harvard School of Education, October 1, 2010. Abbreviated version published in Diversity in American Higher Education: T ­ oward a More Comprehensive Approach, edited by Lisa M. Stulberg and Sharon L. Weinberg, 232–246. New York: Routledge. Long, James E., and Eugenia Toma. 1988. “The Determinants of Private School Attendance, 1970–1980.” Review of Economics and Statistics 70 (May): 351–357. Machung, Anne. 1998. “Playing the Ranking Game.” Change 30 (July–­August): 12–16. Martin, Nathan D., and Kenneth I. Spenner. 2009. “Capital Conversion and Accumulation: A Social Portrait of Legacies at an Elite University.” Research in Higher Education 50: 623–648. Massey, Douglas S., and Margarita Mooney. 2007. “The Effects of Amer­i­ca’s Three Affirmative Action Programs on Academic Per­for­m ance.” Social Prob­lems 54: 99–117. Matthews, Christine M. 2012. Federal Support for Academic Research. Washington, DC: Congressional Research Ser­v ice (October 18). Accessed July 12, 2016. http://­fas​.­org​/­sgp​/c­ rs​/­m isc​/­R41895​.­pdf. McClure, James P., and J. Jefferson Looney, eds. 2008–2016. The Papers of Thomas Jefferson, Digital Edition. Charlottesville: University of ­Virginia Press, Rotunda. Accessed October 16, 2016. http://­rotunda ​.­upress​.­v irginia ​.­edu​/­founders​ /­TSJN010202013200040079. Merton, Robert K. 1968. “The Matthew Effect in Science.” Science 159 (January 5): 56–63. Mettler, Suzanne. 2014. Degrees of In­equality: How the Politics of Higher Education Sabotaged the American Dream. New York: Basic Books. Mills, C. Wright. 1963. “The Sociology of Stratification.” In Power, Politics, and ­People: The Collected Essays of C. Wright Mills, edited by I. L. Horo­w itz, 305–324. New York: Oxford University Press.

420

References

Monks, James. 2009. “The Impact of Merit-­Based Financial Aid on College Enrollment: A Field Experiment.” Economics of Education Review 28: 99–106. Monks, James, and Ronald G. Ehrenberg. 1999. “The Impact of US News and World Report College Rankings on Admission Outcomes and Pricing Decisions at Selective Private Institutions.” National Bureau of Economic Research Working Paper 7227 (July). Morison, Samuel Eliot. 1965. Three Centuries of Harvard: 1636–1936. Cambridge, MA: Harvard University Press. Murray, Charles. 2012. Coming Apart: The State of White Amer­i­ca 1960–2010. New York: Crown Forum. Myers, Michele Tolela. 2007. “The Cost of Bucking College Rankings.” Washington Post, March 11. Nathan, Rebekah. 2005. My Freshman Year: What a Professor Learned by Becoming a Student. Ithaca, NY: Cornell University Press. National Science Foundation, Survey of Earned Doctorates, vari­ous years. Accessed May 3, 2017. https://­w ww​.­nsf​.­gov​/­statistics​/­srvydoctorates​/.­ Neal, Dan J., and Kim Fromme. 2007. “Hook ’Em Horns and Heavy Drinking: Alcohol Use and Collegiate Sports.” Addictive Be­hav­iors 32 (11): 2681–2693. Newfeld, Christopher. 2008. Unmaking the Public University: The Forty-­Year Assault on the ­Middle Class. Cambridge, MA: Harvard University Press. Oreopoulos, Philip, and Kjell G. Salvanes. 2011. “Priceless: The Nonpecuniary Benefits of Schooling.” Journal of Economic Perspectives 25 (Winter): 159–184. Organ­i zation for Economic Co-­operation and Development. 2011. Divided We Stand: Why In­equality Keeps Rising. Paris: OECD. —­—­—. 2012. OECD Economic Surveys: United States. Paris: OECD. Pascarella, Ernest T., and Patrick T. Terenzini. 1991. How College Affects Students: Findings and Insights from Twenty Years of Research. San Francisco: Jossey-­Bass. —­—­—. 2005. How College Affects Students, vol. 2, A Third De­cade of Research. San Francisco: Jossey-­Bass. Peterson, R. A., and R. M. Kern. 1996. “Changing Highbrow Taste: From Snob to Omnivore.” American So­cio­log­i­cal Review 61: 900–907. Pierson, George W. 1969. The Education of American Leaders. New York: Praeger. Piketty, Thomas. 2014. “r>g: Lessons from the History of Wealth,” TED Talk, Berlin, June 23. Accessed July 7, 2014. http://­piketty​.­pse​.­ens​.­f r​/­fi les​ /­Piketty2014Capital21cTED​.­pdf. Pope, Devin G., and Jaren C. Pope. 2009. “The Impact of College Sports Success on the Quantity and Quality of Student Applications.” Southern Economic Journal 75 (January): 750–780. Pryor, John H. et al. 2009. The American Freshman: National Norms Fall 2009. Los Angeles: Higher Education Research Institute. Accessed April 29, 2017.

R e f e r e n c e s

421

https://­w ww​.h ­ eri​.­ucla​.­edu​/­PDFs​/­pubs​/­T FS​/­Norms​/­Monographs​/­The​ AmericanFreshman2009​.­pdf. Pryor, John H., Sylvia Hurtado, Victor Saenz, José Santos, and William Korn. 2007. The American Freshman: Forty Year Trends, 1966–2006. Los Angeles: Higher Education Research Institute (April). Pye, A. Kenneth. 1980. Directions for Pro­g ress: A Report to the Duke University Board of Trustees. (September 26). Quirk, Matthew. 2005. “The Best Class Money Can Buy.” Atlantic Monthly (November). Accessed May 6, 2017. https://­w ww​.t­ heatlantic​.­com​/­magazine​ /­a rchive​/­2005​/­11​/­the​-­best​-­class​-­money​-­can​-­buy​/­304307​/.­ Radford, Alexandria. 2013. Top Student, Top School? How Social Class Shapes Where Top Students Go to College. Chicago: University of Chicago Press. Ramey, Garey, and Valerie A. Ramey. 2010. “The Rug Rat Race.” Brookings Papers on Economic Activity 41 (Spring): 129–199. Reardon, Sean F. 2011. “The Widening Academic Achievement Gap Between the Rich and the Poor: New Evidence and Pos­si­ble Explanations.” In Duncan and Murnane, 91–116. Report of the President’s Commission on Campus Unrest. 1970. Washington, DC: Government Printing Office. Reynolds, O. Edgar. 1927. The Social and Economic Status of College Students. New York: Teachers College. Riesman, David. 1980. On Higher Education: The Academic Enterprise in an Era of Rising Student Consumerism. San Francisco: Jossey-­Bass. Riesman, David, and Verne A. Stadtman, eds. 1973. Academic Transformation: Seventeen Institutions u­ nder Pressure. New York: McGraw-­H ill. Roksa, Josipa, Eric Grodsky, Richard Arum, and Adam Gamoran. 2007. “United States: Changes in Higher Education and Social Stratification.” In Shavit, Arum, and Gamoran 2007, 165–191. Rothman, Stanley, S. Robert Lichter, and Neil Nevitte. 2005. “Politics and Professional Advancement among College Faculty.” Forum 3 (1): 1–16. Accessed May 6, 2017. http://­w ww​.­conservativecriminology​.­com​/­uploads​/­5​/­6​/­1​/­7​ /­56173731​/­rothman​_ ­et​_ ­a l​.­pdf. Sacerdote, Bruce. 2001. “Peer Effects with Random Assignment: Results for Dartmouth Roommates.” Quarterly Journal of Economics 116 (2): 681–704. Saez, Emmanuel, and Gabriel Zucman. 2014. “Wealth In­equality in the United States since 1918: Evidence for Capitalized Income Tax Data.” National Bureau of Economic Research Working Paper 20625 (October). Sauder, Michael, Freda Lynn, and Joel Podolny. 2012. “Status: Insights from Orga­n izational Sociology.” Annual Review of Sociology 37: 267–283. Schneider, Carol G. 2005. “Liberal Education: Slip-­Sliding Away?” In Hersh and Merrow, 61–76.

422

References

Schudson, M. 1972. “Organ­izing the ‘Meritocracy’: A History of the College Entrance Examination Board.” Harvard Educational Review 42: 44–69. Schwartz, Christine R., and Robert D. Mare. 2005. “Trends in Educational Assortative Mating from 1940 to 2003.” Demography 42 (4): 621–646. Sewell, William H., and Vinnal P. Shah. 1967. “Socioeconomic Status, Intelligence and the Attainment of Higher Education.” Sociology of Education 40 (Winter): 1–23. Shavit, Yossi, Richard Arum, and Adam Gamoran, eds. 2007. Stratification in Higher Education: A Comparative Study. Stanford, CA: Stanford University Press. Shulman, James, and William G. Bowen. 2001. The Game of Life: College Sports and Educational Values. Prince­ton. NJ: Prince­ton University Press. Skocpol, Theda. 2003. Diminished Democracy: From Membership to Management in American Civic Life. Norman: University of Oklahoma Press. Solon, Gary. 2002. “Cross-­Country Differences in Intergenerational Earnings Mobility.” Journal of Economic Perspectives 16 (Summer): 59–66. Sotomayor, Sonia. 2013. My Beloved World. New York: Knopf. Sperber, Murray. 2005. “How Undergraduate Education Became College Lite—­a nd a Personal Apology.” In Hersh and Merrow, 131–144. Stadtman, Verne A. 1973. “Constellations in a Nebulous Galaxy.” In Academic Transformation: Seventeen Institutions ­Under Pressure, edited by David Riesman and Verne A. Stadtman, 1–12. New York: McGraw-­H ill, 1973. Steiner, Peter O., Maryse Eymonerie, and William B. Woolf. 1971. “At the Brink: Report on the Economic Status of the Profession, 1970–71.” AAUP Bulletin 57 (June). Stevens, Mitchell L. 2009. Creating a Class. Cambridge, MA: Harvard University Press. Useem, Michael, and J. Karabel. 1986. “Pathways to Top Corporate Management.” American So­cio­log­i­cal Review 51: 184–200. Useem, Michael, and S. M. Miller. 1975. “Privilege and Domination: The Role of the Upper Class in American Higher Education.” Social Science Information 14: 115–145. Valletta, Robert G. 2016. “Recent Flattening in the Higher Education Wage Premium: Polarization, Skill Downgrading, or Both?” National Bureau of Economic Research Working Paper 22935 (December). Vedder, Richard. 2004. ­Going Broke by Degree: Why College Costs Too Much. Washington, DC: American Enterprise Institute, 2004. Veysey, Laurence. 1973. “Stability and Experiment in the American Undergraduate Curriculum.” In Content and Context: Essays on College Education, edited by Carl Kaysen, 1–64. New York: McGraw-­H ill. Vigdor, Jacob L., and Charles T. Clotfelter. 2003. “Retaking the SAT.” Journal of ­Human Resources 38 (Winter): 1–33.

R e f e r e n c e s

423

Wacker, Grant. 2016. “The Christian Right.” National Humanities Center. Accessed June 9, 2016. http://­n ationalhumanitiescenter​.­org ​/­t serve​/­t wenty​ /­t keyinfo​/­chr​_­rght​.­htm. Webber, Douglas A., and Ronald G. Ehrenberg. 2009. “Do Expenditures Other Than Instructional Expenditures Affect Graduation and Per­sis­tence Rates in American Higher Education.” National Bureau of Economic Research Working Paper 15216 (August). Webster, David S., Russell L. Stockard, and James W. Henson. 1981. “Black Student Elite: Enrollment Shifts of High Achieving, High Socio-­Economic Status Black Students.” College and University 56 (Spring): 283–291. Wicksteed, Philip H. 1910. The Common Sense of Po­liti­cal Economy. London: Macmillan. Accessed January 8, 2017. www​.e­ conlib​.­org​/ ­l ibrary​/ ­Wicksteed​/­wkCS0​ .­html. Winston, Gordon. 1999. “Subsidies, Hierarchy and Peers: The Awkward Economics of Higher Education.” Journal of Economic Perspectives (Winter): 13–36. —­—­—. 2004. “Differentiation among US Colleges and Universities.” Review of Industrial Organ­ization 24: 334–354. Winston, Gordon, and Ethan Lewis. 1997. “Physical Capital and Capital Ser­vice Costs in U.S. Colleges and Universities: 1993.” Eastern Economic Journal 23 (Spring): 165–189. Wolff, Edward N. 2012. “The Asset Price Meltdown and the Wealth of the M ­ iddle Class.” National Bureau of Economic Research Working Paper 18559 (November). —­—­—. 2013. “The Asset Price Meltdown and the Wealth of the ­Middle Class.” US2010 Proj­ect (May). Accessed April 29, 2017. https://­s4​.­ad​.b ­ rown​.­edu​/­Projects​ /­Diversity​/­Data​/­Report​/­report05012013​.­pdf. Woo, Jennie H., and Susan P. Choy. 2011. “Merit Aid for Undergraduates: Trends from 1995–96 to 2007–08.” Stats in Brief, U.S. Department of Education (October). Young, Michael. (1958) 1994. The Rise of the Meritocracy. London: Thames and Hudson. Reprint, New Brunswick, NJ: Transaction. Zelizer, Viviana A. 1985. Pricing the Priceless Child: The Changing Social Value of ­Children. Prince­ton, NJ: Prince­ton University Press. Zumeta, William, David W. Breneman, Patrick M. Callan, and Joni E. Finney. 2012. Financing American Higher Education in the Era of Globalization. Cambridge, MA: Harvard Education Press. Zweigenhaft, Richard L. 1993. “Prep School and Public School Gradu­ates of Harvard: A Longitudinal Study of the Accumulation of Social and Cultural Capital.” The Journal of Higher Education 64 (March–­April): 211–225.

424

References

Acknowl­edgments

For her research assistance on this proj­ect I am especially indebted to Danielle Vance-­McMullen, who systematically tracked down and organized data on hundreds of colleges. I am also grateful to Analese Bridges, Mingyuan Hua, Phoebe Long, B. J. Loudermilk, Lisa Sapozhnikov, Amanda Sear, Ryan Smith, Katherine Tamer, and Natalie Trebes. I received helpful comments and suggestions from Anthony Broh, Mark Chaves, Jim Clotfelter, Edward Fiske, James Hearn, Cory Koedel, Thomas LeBien, Frank Levy, Theresa Newman, Joseph Price, Joseph Quinn, James Roberts, Sarah Turner, my colleagues at the Straus Institute, and participants at seminars at the City University of New York, Syracuse University, and the New York Federal Reserve. And, for sharing their insights about higher education over the years, I thank Ronald Ehrenberg, Craufurd Goodwin, and Paula Stephan. I thank all of the above without, of course, implicating them for any errors that remain in what I have written. I also benefited greatly from the conversations I had with administrators and faculty during visits I paid to a number of colleges, including American University, Elon University, High Point University, Howard University, Manhattan College, College of Mount Saint Vincent, Sarah Lawrence College, and the University of North Carolina Chapel Hill. I received valuable data from the Higher Education Research Institute at UCLA, the Council for Aid to Education, and the law schools at Duke University, Harvard University, and New York University. All figures and tables have been newly compiled from previously published and unpublished data. The reference staff of the Goodson Law Library at Duke helped me locate references on dozens of occasions, for which I am very grateful. I received financial support from the Straus Institute for the Advanced Study of Law and Justice at NYU, where I spent the 2013 / 14 year on sabbatical, and from the Center for the Study of Philanthropy and Voluntarism and the Sanford School of Public Policy at Duke University. The views expressed are mine and do not necessarily reflect ­those of any organ­i zation or person.

Index

Academically Adrift (Arum, Roksa), 261, 307 Academic majors: selectivity, effect on, 188–190; majors, STEM, 273–276; majors, business, 276–277; vocational or professional, 341, 363n12; decrease in humanities, 388n46, 388n58 Academic majors, decrease in humanities, 186–187 Academic rigor, 60–61, 201, 256–257, 388n58 Academic stratification, 208, 235, 254, 343, 345 ACT test. See Standardized tests (SAT, ACT) Admissions criteria, 13; history of discrimination and preference, 41–42; open enrollment, 42–43; nonprice rationing and selectivity, 46, 124; “need-­blind” admissions, 166, 349; legacy admissions, 170–172, 349, 384n63; preference for athletes, 172–173, 385n65; underrepresented minority preferences, 173–174, 385n71 Admissions office: baby boom and admissions selectivity, 147–150; choosing or recruiting, 149, 153; Common Application, 155, 170;

name-­brand colleges selectivity, 163–164; early decision, 168, 219, 384n49–50; wait list and student intention, 169, 384n53; reliance on standardized test scores, 215 Affirmative action: Regents of the University of California v. Bakke, 173; Grutter v. Bollinger, 174; Hopwood v. Texas, 174 Agnes Scott College, liberal arts emphasis, 185 Alabama A&M University, ­family income of students, 69 American Association of University Professors (AAUP), 47, 57, 137–139 American Graffiti, 95 Amer­ic­ a’s Best Colleges, 103, 375n59–60 Amherst College, 11, 108, 164, 339; top high school students at, 61–62; single sex to coed, 95; unresponsive supply in admissions, 123; mission statement, 164; U.S. News ranking and ac­cep­tance rate, 164–165; ­family income and selectivity, 242, 397n39; use of resources by students, 346; highly selective, 377n20 Angell, Robert Cooley, 256, 399n3

Animal House, 257 Appalachian State University, 30 Armstrong, Elizabeth, 199, 259 Arum, Richard, 261, 307 Association of American Colleges, 58, 59 Astin, Alexander, 233, 234, 255 Athletics, 75; student activity or commercial enterprise, 71–74; football, 154; graduation rates of recruited athletes, 173; tool to attract students, 201, 400n9; spectator sports, 202–203. See also National Collegiate Athletic Association (NCAA) Atlantic, 228 Augusta College, 30 Barron’s, 44 Bates College, liberal arts emphasis, 185 Baumol, William, 58 Berea College: U.S. News ranking, 105, 375n62; Washington Monthly ranking, 108 Berry College, 238 Big Sort, The (Bishop), 297 Binstock, Jeanne, 20 Bishop, Bill, 297 Boston College, 20–22, 60, 341–342; theology requirement, 21, 363n18; vocational or professional majors, 21–22; enrollment management, 154; majors offered, 341, 363n24; affluence and admissions, 343 Bowdoin College, 34; increase in average income of nonwhite students, 250; high school study time of freshmen, 280 Bowen, Howard, 119–120 Bowen, William, 6, 7, 172–173 Brandeis University, 397n39; religious origin, 16, 35; liberal arts emphasis, 184–185; highly selective, 377n20 Brewster, Kingman, 56, 369n14 Brigham Young University, religious origin, 16

428 I N D E X

Brint, Steven, 35, 119 Brown University, 11; single sex to coed, 94; selectivity of, 165; U.S. News ranking and ac­cep­tance rate, 165; no “need-­blind” admissions, 166; high school study time of freshmen, 280; geographic shift away from Northeast, 372n18 Brown v. Board of Education, 42, 89 Bucknell University, liberal arts emphasis, 185 California Institute of Technology (Cal Tech), 34, 61; tuition subsidies, 142; highly selective, 164, 377n20; national market, 220; ­family income and selectivity, 397n39 Calvin College, ­family income, 240 Campus residential community, 29–30, 364n44 Carleton College, 191; ­family income and selectivity, 69, 396n39 Car ­ne­g ie Foundation, 105, 375n62 Catholic University, 60 Chetty, Raj, 243, 328–329, 343 Chronicle of Higher Education: guns and unrest in 1979, 53–54, 56; financial crisis, 57–58 Citadel, The, 29, 72, 117; Robert E. Lee, 29 City College of New York (CCNY), 42, 55 Claremont McKenna College, expectation to join a fraternity or sorority, 278 Coleman, James, 119 College Board, “Student Search Ser­v ice,” 152–153, 155, 382n10 College experiences, 7–8; appeal to fun aspects, 200, 200–201, 260; campus life, 30, 196–197; upgraded facilities, 197–198; spectator sports, 202; builder of ­human capital, 257, 344–345; schoolwork or other interests, 259–260, 400n13; time spent studying, 266. See also Athletics

College gradu­ates: critical thinking skills, 307; graduation rates, 307–308, 310; personal development and quality of life, 308–309; peer effects, 309; ­labor market and earnings, 310, 311–313; mea ­sure­ment of outcomes, 310, 315–317, 325–326; law school admission, 313, 314, 321–­323; PhD earned, 313, 314, 317–­321, 405n27; prestigious postgraduate scholarship, 313, 315, 322; Who’s Who listings, 313; business leadership, MBA or law school, 314 College industry: impact of, 353–354; elite colleges, concentration of wealthy students, 9–10, 13, 351–352; decentralized, geo­graph­i­cally anchored, 12; need to attract students, 39, 152; nonprofit status, 40; as a market, 43–44, 346–347; Institutional aims, 50–51; financial crisis, 57–59; part church and part car dealership, 116; pluralism, 298–299; segmented market, 338, 350; market changes, 339. See also In­equality; National market hypothesis College industry, oddities: unique or unusual aspects of, 44; asymmetric information about colleges, 44–45, 367n99; importance of physical campus and buildings, 45; nonprice rationing and selectivity, 46; price discrimination, 46; positional competition, 47, 104, 120, 164; tenure, 47 College industry, types: church-­l ike, mission-­d riven, 117–118; financial viability, survival, 118–119; well-­ endowed institutions, 119–120, 122–123 College of Charleston, 60 College of Mount Saint Vincent, 178; enrollment management, 176 College of William and Mary, 31, 180–181 Collegiate Learning Assessment, 307

Colorado State University, football stadium, 197 Columbia University: protest and building sit-in, 55; deficits in 1970, 57; top high school students at, 62; share of undergraduates, 123–124; charitable gifts, 128; U.S. News ranking and ac­cep­tance rate, 165; production of distinguished alumni, 313; per student income, 371n47; geographic shift away from Northeast, 372n18 Comparative Guide to American Colleges (Cass and Birnbaum), 60, 103, 209, 370n28 Comparison categories: entering students’ standardized test scores, 51; public vs. private (not-­for-­profit), 51; status as a HBCU, 51 Comparison criteria, 25–27, 209–210, 364n33, 364n35–36; low-SAT public colleges, 27–30; commercial spectator sports, 30; high-SAT public colleges, 31, 365n47; lowest-SAT private colleges, 32–33; very-high-SAT private colleges, 33–35, 122–123, 377n20–22; Freshman Survey, 48–50 Comparison criteria, faculty: teaching loads, 190–191; tenured and non-­ tenured faculty, 192–193, 195–196, 389n66, 389n69; ratio of instructors to students, 193–195 Comparison criteria, majors offered, 189–190; low-SAT public colleges, 388n51; lowest-SAT private colleges, 388n52; very-high-SAT private colleges, 388n53–54, 388n56–57 Comparison criteria, quality, 207–208 Comparison criteria, statement of purpose, 386n14; low-SAT public colleges, 180; high-SAT public colleges, 180–181; lowest-SAT private colleges, 181, 183; very-high-SAT private colleges, 184

I NDE X 429

Competition: definition of, 5; “positional” competition, 120 Competition, colleges, 75, 175, 340–341; name-­brand or less selective institutions, 11–12; selectivity and academic standing, 60–62, 64–66; selectivity and college-­educated ­fathers, 66–67; selectivity and ­family income, 68–69; selectivity and school resources, 69; admission spots, supply of, 123–124; private college marketing, 152, 382n7; private college marketing, merit scholarships, 153–154. See also Faculty; U.S. News and World Report Competition, students: admission criteria, 13–14; affluence and admissions, 82–83, 123, 214, 255, 280, 398n55; admission to high-ranking schools, 175, 340; parental help and attitude, 213–215, 218, 391n13, 392n17, 393n33; strategic be­hav­ior for highranking schools, 215–216; “hypercredentialing,” 218, 272, 349; admissions counselors, 218–219; early decision, 219; high school grade inflation, 224, 393n40; high school study time, 264; academic stratification, 343–344; parental help, activities, 392n32; application boot camp, 393n34 Conant, James, 63 Conroy, Pat, 29, 72, 354 Cornell University: protests, 55; public university, 108; charitable gifts, 128; production of distinguished alumni, 313; per student income, 371n47; geographic shift away from Northeast, 372n18 Council for Aid to Education, 128–129 Dale, Stacy, 332 Dartmouth College: single sex to coed, 94; U.S. News ranking and ac­cep­tance rate, 165; geographic shift away from

430 I N D E X

Northeast, 372n18; highly selective, 377n20; ­family income and selectivity, 397n39 Davidson College, 34–35; single sex to coed, 94; highly selective, 164; decline in religious identification, 184, 387n31; national market, 220; high school study time of freshmen, 280 Demerath Peter, 213, 218, 272 Demographic trends, 372n22; number of eigh­teen-year-olds, 83–84; geographic shift away from Northeast, 85–86, 109, 372n18; racial / ethnic composition, 87–93, 372n19, 372n21; racial integration, 89; foreign students, 92, 373n26; female students and single sex colleges, 94–95; stability in number of freshmen, 121–122 Dickens, Charles, 306 Diminished Democracy (Skocpol), 300 Discounting, 46, 157–158, 165–166, 174–175, 197, 340 Diversity, definition of, 5 Donations. See Philanthropy Du Bois, W. E. B., 36 Duffy, Elizabeth, 148 Duke University, 55, 336–338; mission statement, 119; tuition increases, 124; spending issues and competition, 126; charitable gifts, 128; faculty compensation, 139–140; merit scholarships and discounting, 165; legacy admissions, 172; Au Bon Pain, 197; students’ hypothesis, 260; law school admission, 314–315, 319, 323 Eastern Kentucky University, 29–30 Eastern Michigan University, 180 Eckerd College, 34 Economic efficiency, 330–331, 334–335, 337; se­lection bias in determining returns from attending a college, 331–332, 406n31; choice of college,

effect on earnings, 332–333, 406n36; effect on admissions, studies of, 344–345 Economic model of enrollment, 399n2 Economic segregation, 229 Educated for Romance (Holland, Eisenhart), 258 Education system, socioeconomic status (SES) and gaps in achievement, 230–231 Ehrenberg, Ronald, 210 Eisenhart, Margaret, 258 Elite colleges, 344–345; income of entering students, 13; benefit of income disparity, 114; nonwhite middle-­and upper-­income admissions, 249–250, 397n46, 398n47; study time of students, 266; production of distinguished alumni, 313; disproportional success of gradu­ates, 315–317, 346; resources per student and productivity, 334 Elon University: decline in religious identification, 182; pragmatic education, 183; donut break and “turning 21” ceremony, 197 Endowments, 134–137; stock market effect, 13, 114–115, 127–129, 133–134, 378n29; dwindling endowment income, 58; capital campaign, 128; investment risk and portfolio size, 135, 380n45; per-­student endowment, 135–137. See also Philanthropy Enrollment management, 154–155, 157–158, 382n19; price structure and merit scholarships, 156–157; need-­ based financial aid, 162, 383n37; merit scholarships as a tool, 225–226 Expenditures and assets, increasing differences, 136, 380n53 Eymonerie, Maryse, 58 Faculty: compensation, comparison of, 70, 137, 341–343, 381n57; public and private research universities compen-

sation, 139–140, 381n58; teaching loads decline, 190–191; research interests primary, 261 Falwell, Jerry, 282 Fayetteville State University, prior names indicate HBCU history, 37 Financial investment, college as, 209, 391n1–2 Fiske, Edward, 152 Fiske Guide to Colleges, 9, 44, 209 Florida Memorial University, 37; mission like less selective white public colleges, 186 Forbes, 180 Fordham University, 20 For-­profit colleges, 16 Franklin and Marshall College, 384n51; colleges’ strategic be­hav­ior for ranking, 169 Fraternities and sororities, 30, 363n15; Illinois State University, 20; Tennessee Wesleyan College, 24; underage drinking, 199–200 Freeman, Richard, 77 ­Free Speech Movement, 54 Freshman, The, 257 Freshman Survey, 47–51, 61, 67, 68, 219, 367n102; En­g lish as a second language, 92–93; grade inflation, 224; HBCU and socioeconomic stratification, 230; high SES students to elite colleges, 234–236; ­family income data, 235–242, 396n38; private high school attendance, 246–248, 397n45; diversity, 249–250, 397n46; blacks’ choice of HBCU or other college, 252–253; high school study time, 262–266, 400n17– 24; high school time spent socializing, 266–268; high school time on TV, 268–269; high school working for pay, 269–270; high school time spent talking to teachers, 270–271; high school community

I NDE X 431

Freshman Survey (continued) ser­v ice time, 270–272; ambitions, 272–274; majors, STEM, 273–276; majors, business, 276–277; high school consumption of alcohol, 277–278; expectation to join a fraternity or sorority, 278–279, 402n27; identity of Protestant and Catholic universities, 283; religious preference, 283–286, 402n5, 402n7; religious observance, 290–292; po­l iti­cal views, 293–294, 296, 297–298 Gallup Branding Index, 180 Geiger, Roger, 39, 121 Georgetown University, 20; U.S. News ranking, 105 Georgia Institute of Technology (Georgia Tech), 60, 180–181, 242, 397n40; public research university, 31; HOPE scholarship, 101 Gettysburg College, expectation to join a fraternity or sorority, 278 Goldberg, Idana, 148 Goucher College, liberal arts emphasis, 185 Government support, 102, 109; Financial aid, loans, grants and tax credits, 97–98, 374n47; states’ role in tuition, 99–101; states’ role in scholarships, 101; states’ role in scholarships, Georgia’s HOPE scholarships, 101, 351; in­equality and public policy, 351–352. See also Legislation Griffith, Amanda L., 221 Grinnell College, 60 Hamilton, Laura, 199, 259 Hamilton College, f­ amily income and selectivity, 396n39 Hansen, W. Lee, 99 Harvard University, 23; admission criteria and religion, 42–43; Vietnam War protest, 55; SAT and application, 63;

432 I N D E X

share of undergraduates, 122–124; tuition increases, 124; charitable gifts, 128; tuition subsidies, 142; U.S. News ranking and ac­cep­tance rate, 165; legacy admissions, 171; ­family income and selectivity, 243; founding and religion, 283, 352; production of distinguished alumni, 313; Rhodes Scholars, 313, 405n19; law school admission, 314–315, 323; students’ ­family income, 371n47; geographic shift away from Northeast, 372n18; highly selective, 377n20 Harvey Mudd College, 61; highly selective, 377n20; ­family income and selectivity, 397n39 Haverford College: single sex to coed, 95; highly selective, 377n20; ­family income and selectivity, 397n39 Herfindahl index for religious homo­ geneity, 288, 403n9 Higher Education Research Institute, 47 High Point University, 198–199, 390n76, 390n77 Historically black college or university (HBCU), 25, 36–37; Florida Memorial College, Baptist, 25; More­house College, 36; Tuskegee University, 36; Oakwood University, Seventh Day Adventists, 37; tuition dependence, 161; mission like less selective white public colleges, 185–186; national market hypothesis, 221; loss of high-­a chieving black students, 224, 251–253, 394n41, 398n48–49; ­family income, 238–240, 244, 248; private high school attendance, 251; high school time on TV, 269; least religiously diverse, 288–290; most likely to attend religious ser­v ices, 290; liberal po­l iti­cal views, 293 History and development of colleges: private, religious, and privileged, 37–39, 366n72; states’ role, 39–41;

admissions criteria, 41–43. See also Morrill Act of 1862 Hoekstra, Mark, 331 Holland, Dorothy, 258 Holy Cross, College of the, 20 HOPE Scholarship, 101, 159, 351, 393n36 Howard Payne University, 32, 117, 365n51; religious mission, 182–183; pragmatic education, 183 Howard University: low assets per student, 144; ­family income, 240 Hoxby, Caroline, 26–27, 64, 211–213, 219–220, 226, 331, 332–334, 346, 351, 406n32, 406n38 Huckleberry Finn (Twain), 50 Illinois State University, 18–20, 341; vocational or professional majors, 19–20; affluence and admissions, 343 Income disparity, 145–146. See also Endowments; Faculty; Students Income distribution, 76, 371n4, 371n5, 371n6, 378n28; population and, 3, 10–11, 13, 339, 362n7; college earnings advantage, 78, 80–82, 347, 372n7; effect on college market, 79, 108, 114–115, 372n10; affluence and admissions, 82–83, 343 Indiana State University, enrollment management, 157–158 In­equality: definition of, 5–6; elite colleges, concentration of wealthy students, 9–10; in­equality dividend, 115–116, 338–339; race, socioeconomic status, choice, 348. See also Income distribution Intergenerational mobility, definition of, 362n8 Iowa State University, 60 Iowa Wesleyan University, f­ amily income, 238, 240 “Isomorphism,” definition of, 17 Ivory Tower, The, 257

Jacob, Brian, 201, 210, 391n6 Jane Austen in Scarsdale (Cohen), 214 Jefferson, Thomas, 354 Jencks, Christopher, 255, 352 Johns Hopkins University, 157; ­family income and selectivity, 69, 397n39; charitable gifts, 128; per student income, 371n47 Kalamazoo College, liberal arts emphasis, 185 Karabel, Jerome, 42, 233 Kent State University, 52, 55 Kerr, Clark, 177 Khan, Shamus, 253 King, Martin Luther, Jr., 36 Krueger, Alan B., 332 Land grant colleges, 31; Tuskegee University, 36. See also Morrill Act of 1862 Lareau, Annette, 213 Lee, Robert E. See Citadel, The Lee University, 32, 182, 386n21 Legislation, 347; G.I. Bill (1944), 40, 97; Morrill Act of 1862, 40; Morrill Act of 1890, 40; National Defense Education Act (1957), 40, 97; Civil Rights Act of 1964, 54; Voting Rights Act of 1965, 54; Higher Education Act of 1965 amendments, 97; Public Health Ser ­v ice Act, 367n84 Lehigh University, liberal arts emphasis, 185 Liberty University, 282; conservative po­l iti­cal views, 295 Lovejoy’s College Guide, 103 Lowell, Lawrence, 42 Madonna College, 186 Majors, vocational or professional: overlapping, 22; few vocational at top-ranked schools, 35

I NDE X 433

Market orientation, 203; amenities and student ser­v ices, 178, 200–201; grade inflation, 178; liberal arts role, 178, 181; adjuncts and other nonregular instructors, 179 Marshall Scholarship, 324–­325 Mas­sa­chu­setts Institute of Technology (MIT), 34; share of undergraduates, 124; U.S. News ranking and ac­cep­tance rate, 165; per student income, 371n47; highly selective, 377n20 Matthew effect, 113, 126–127, 134, 146, 161, 338 McCall, Brian, 201, 210, 391n6 McNamara, Robert, 55 Meritocracy, 94, 213, 254–255. See also Young, Michael Merit scholarships. See Tuition Merton, Robert, 113, 376n2 Middlebury College: ­family income, 236; ­family income and selectivity, 397n39 Midwestern University, 199–200 Military colleges, 12, 29, 365n47; conservative po­l iti­c al views, 293 Mills, C. Wright, 25 Mills College, parental income and college selectivity, 65 Monks, James, 210 More­house College, 36 Morgan State University, f­ amily income and selectivity, 69 Morison, Samuel Eliot, 283, 352 Morrill Act of 1862, 16, 39–40 Morrill Act of 1890, 16, 40 Morse, Robert, 103 Mount Holyoke College: ­family income and selectivity, 69, 397n39; liberal arts emphasis, 185 Mount St. Mary’s College of California, 183

434 I N D E X

National Assessment of Adult Literacy, 261 National Association of College and University Business Officers (NACUBO), 135 National Collegiate Athletic Association (NCAA), 202; NCAA Division I-­A , 73–74 National market hypothesis, 211–212, 219–221, 226–227, 393n35–36; selectivity of elite schools, 235–236 New Depression in Higher Education, The, 59, 126 Newman Guide to Choosing a Catholic College, 282, 283 New York University (NYU): charitable gifts, 128; law school admission, 314–315, 323 Nixon, Richard M., 56 Normal schools. See Illinois State University Northeastern University, strategic be­hav­ior for U.S. News ranking, 169–170 Northern Michigan University, post-­ college economic success, 330 Northwestern University, 250; faculty compensation, 139–140; national market, 220; ­family income, 240 Notre Dame, University of, religious origin, 16 Oakland University, ­family income, 238 Oakwood University, 37, 186, 388n45 Oglethorpe University, 33, 183 Ohio State University, 28–30, 364n38; protests ­a fter Kent State, 55 Oklahoma State University, 127 Oral Roberts University: religious origin, 16; conservative po­l iti­cal views, 295 Organ­ization for Economic Co-­operation and Development (OECD), 10 Origins, historical diversity, 16–17

Oseguera, Leticia, 234 Overeducated American, The (Freeman), 77 Paired Comparison Method, 406n37 Paying for the Party (Armstrong, Hamilton), 199–200, 259 Pell Grants, 46, 98, 102, 162, 241, 374n46 Pennsylvania State University, preference for athletes, 385n65 Philanthropy, 133; largest donations to universities, 127–128; contributions to best- ­endowed universities, 129–132, 379n36–38 Piketty, Thomas, 134 Po­l iti­cal views: evangelical Protestantism and conservative politics, 295; self-­segregation, 297; pluralism, 298–299; college experience and learning opportunities, 299–300 Prestige, 15–16; academic quality and, 25 Price discrimination, 46, 156–157, 162–163, 166, 169, 175 Prince­ton Review, 260 Prince­ton University, 6, 8, 55, 60; endowment, 8; admissions criteria and religion, 42–43; deficits in 1970, 57; SAT and application, 63; ­family income and selectivity, 69, 242, 243, 397n39; single sex to coed, 94; share of undergraduates, 122–123; U.S. News ranking and ac­cep­tance rate, 165; legacy admissions, 171; ­family income, 236; production of distinguished alumni, 313; post-­college economic success, 330; per student income, 371n47; geographic shift away from Northeast, 372n18 Proj­ect Talent, 65–66, 201; college enrollment and SES, 231, 395n19 Public universities: unequal funding, 16; comparison to private colleges, 25; financial squeeze, 115. See also Morrill Act of 1890

Purdue University: enrollment management, 155–156; need-­based financial aid, 161 Pye, Kenneth, 126, 378n26 Racial discrimination and segregation, 42, 54, 87, 89, 254, 337, 348, 353 Rankings of colleges, 102–103, 109; Washington Monthly, 108, 338. See also U.S. News and World Report Reed College, non-­cooperation with U.S. News ranking, 375n71; ­family income and selectivity, 397n39 Regent University, 282; conservative po­l iti­cal views, 295 Religion: role in history of American higher education, 16, 38, 283; comparison category, lowest-SAT private colleges, 32–33, 365n54; top-­ranked schools secular, 35; re­sis­tance to market orientation, 179, 181; Madonna College, Catholic, 186; Catholic high schools, 246–248, 397n43; college guides for par­t ic­u ­lar religions, 282; evangelical Protestantism and conservative politics, 282; decline nationwide mirrored by college students, 287, 403n12; homogeneity or diversity, 287–289; Catholic and evangelical colleges homogeneous, 289–290, 403n10; Jewish students at most selective private colleges, 292 Religious affiliation: Boston College, Jesuit, theology requirement, 20–21; Tennessee Wesleyan College, Methodist, 23, 363n26; Howard Payne University, Baptist, 32; Lee University, Church of God (Pentecostal), 32; Davidson College, Presbyterian, 35; Wake Forest University, Baptist, 35; Florida Memorial College, Baptist, 37; Oakwood University, Seventh Day Adventist, 37; Southern Adventist

I NDE X 435

Religious affiliation (continued) University, 182; Trevecca Nazarene University, 182; William Carey University, Baptist, 182 Resource allocation, 10, 345 Reynolds, O. Edgar, 65 Rhodes, James, 52 Rhodes Scholars, 324 Rice University: highly selective, 377n20; ­family income and selectivity, 397n39 Riesman, David, 179, 181, 255, 352, 362n3 Robertson, Pat, 282 Roksa, Josipa, 261 Rothstein, Donna S., 221 Rutgers University, 55 Saint Leo University, 183 Sarah Lawrence College, 34; non-­ cooperation with U.S. News ranking and aftermath, 107–108, 375n69; highly selective, 377n20; f­ amily income and selectivity, 397n39 SAT (Scholastic Aptitude Test). See Standardized tests (SAT, ACT) Savannah State University, mission like less selective white public colleges, 186 Science, 113 Shepherd College, 30 Shulman, James, 172–173 Skocpol, Theda, 300 Slate, 127–128 Smith College, SAT and application, 63 Socioeconomic status (SES): college enrollment rates, 231–232; two-­and four-­year college divide, 232; effect on choice of college, 232–233; effect on admissions, studies of, 233–235, 395n24; expansion of college-­going, 233–235; effect on elite colleges, 234–236, 242–244; ­family income, 236–241, 396n33–36; effect on selectivity over time, 240–243, 248, 348; tax returns to determine ­family

436 I N D E X

income, 243; parental education, 245–246; high school working for pay, 269; zip code–­level data on ­family income, 397n41. See also Historically black college or university (HBCU) Sotomayor, Sonia, 6–7, 44, 261, 344, 354 Southern Adventist University, 182 Southern Methodist University, U.S. News ranking and ac­cep­tance rate, 165 Sports. See Athletics Standardization, conformity and differentiation among colleges, 17–18, 362n6, 362n7 Standardized tests (SAT, ACT), 363n10; Illinois State University scores, 19, 341; Boston College scores, 21, 341; Tennessee Wesleyan College, 24, 342; comparisons, 25–26; comparison type, low-SAT public colleges, 27–29; comparison type, high-SAT public colleges, 31; comparison type, lowestSAT private colleges, 32–33; comparison type, very-high-SAT private colleges, 33–35, 122–123; origin and history, 62–64; national use by admissions offices, 211–213; student strategy, test-­preparation courses, 216–217, 392n23; student strategy, special accommodation, 217–218, 392n28; scholastic segregation, 221–223, 393n38–39; f­ amily income, parents’ education, and scores, 231 Stanford University, 11; top high school students at, 61; share of undergraduates, 122–123; charitable gifts, 128; U.S. News ranking and ac­cep­tance rate, 165; ­family income and selectivity, 243; use of resources by students, 346; per student income, 371n47; highly selective, 377n20 Stange, Kevin, 201, 210, 391n6 State University of New York Stony Brook, 242, 397n40

Statistical Abstract of the United States, 76 Steiner, Peter, 57 St. John’s College, 34; highly selective, 377n20 Stratification, college resources, 69–71 Stratification, definition of, 75 Stratification, students: academic readiness, 59–66; economic backgrounds, 64–69, 348–349, 350–351, 371n47; strongest students matched to high resource colleges, 345–346 Students: peer exposure, 8–9, 229; stability in number of freshmen, 121–122; expenditures per student, 126–127, 141–142; donations per student, 132–133, 379n39–40; per-­student endowment, 135–136; ratio of students to faculty, 142; tuition subsidies and other income sources, 142–143, 381n62; net assets per student, 143–144; baby boom and admission selectivity, 147–148; need-­based financial aid, 162–163, 376n8, 383n37; proximity advantage to elite colleges, 221, 393n37; financial aid role in decision, 225–226; typology of, 258, 399n3; critical thinking skills, 261; impact of subenvironments, 307, 404n2; college enrollment rates, 395n17. See also College experiences; College gradu­a tes; Competition, students; Freshman Survey Survey of Earned Doctorates, 314 Swarthmore College, 55, 61; tuition increases, 124; liberal arts emphasis, 184; national market, 220; highly selective, 377n20; ­family income and selectivity, 397n39 Teachers colleges: definition of, 20. See also Illinois State University

Technology, influence of, transportation and communication, 95–96, 109 Tennessee Wesleyan College, 342; religious orientation, 23; majors offered, 23–24; affluence and admissions, 343 Texas Christian University, 60 Tobin, James, 63, 95 Tocqueville, Alexis de, 298 Towson University (Towson State), 30, 180 Trevecca Nazarene University, 182 Truman, Harry S., 65 Truman Scholarship, 325–­326 Tuition, 363n14; Illinois State University, 20; merit scholarships, 46, 153–154, 158; increases exceeding inflation, 124–125; dependence on, 150–151, 159–161; merit scholarships and discounting, 158–159, 165, 349; foreign students, 161, 383n35; in-­state or out-­of-­state rates, 161; name-­brand rates exceed inflation, 166–167 Tulane University: merit scholarships and discounting, 165; expectation to join a fraternity or sorority, 278 Turner, Sarah, 351 Tuskegee University, 36 Twain, Mark, 50 Two-­year colleges, 16 University of Alabama Huntsville, enrollment management, 155 University of Arizona, 56 University of California, Berkeley, 54–55, 374n49 University of California, Davis (UC Davis), 31, 181 University of California, Los Angeles (UCLA), 47, 127 University of California, San Diego, 127, 181; student time spent studying, 201–202

I NDE X 437

University of California, San Francisco, 127 University of California, Santa Cruz, 60 University of California system: college selectivity and socioeconomic bias, 66; state’s contribution, 99; out-­of-­state tuition, 161; study of academic activity time, 260; production of distinguished alumni, 313 University of Chicago: share of undergraduates, 124; production of distinguished alumni, 313; per student income, 371n47 University of Florida, 31; Gator Nation, 181 University of Georgia, 36, 365n48; HOPE scholarship, 101, 159; U.S. News ranking and ac­cep­tance rate, 165 University of Illinois (Champaign-­ Urbana), 18; faculty compensation, 139–140 University of Mas­sa­chu­setts at Amherst, 48, 240; ­family income, 240 University of Miami, 382n16; enrollment management, 154–155; sports clubs and student organ­i zation increase, 197 University of Miami (Ohio), preference for athletes, 385n65 University of Michigan: top high school students at, 61; class bias in admissions, 64–65, 370n36–37; production of distinguished alumni, 313; preference for athletes, 385n65 University of Missouri, dorm with waterpark, 197 University of New Haven, 33, 183 University of North Carolina, Chapel Hill, 99, 126; Carolina Covenant, 101; faculty compensation, 139–140; preference for athletes, 385n65 University of Oxford, 127

438 I N D E X

University of Pennsylvania, 55; charitable gifts, 128; production of distinguished alumni, 313; geographic shift away from Northeast, 372n18 University of Rochester, 250; merit scholarships and discounting, 165 University of South Carolina, f­ amily income, 238–240 University of Southern California, charitable gifts, 128 University of Southern Mississippi, 30, 180 University of V ­ irginia: single sex to coed, 94; Access UVa, 101 University of Washington, 31, 181 University of Wisconsin, 55 Unrest and protests of 1970s, 54–57 Upward mobility, 327–328, 353, 406n28; low-­income student enrollment, 328–­329, 406n30; post-­college economic success, 330, 406n31 U.S. News and World Report, 9, 13, 102–107, 116, 375n60, 375n67; competition and rankings, 120, 339–340, 384n55; effect of merit scholarships on rankings, 156; positional competition metric, 164–165; colleges’ strategic be­hav­ior, 167–170, 180; teaching staff influence, 190, 389n59; effect of change in rank, 210; premier ranking authority, 210; law school rankings, 315, 405n25 Vanderbilt University, 36; illustrative choice of college, effect on earnings, 332 Vassar College, SAT and application, 63 Wabash College, parental income and college selectivity, 65 Wake Forest University, 35; decline in religious identification, 184 Wall Street Journal, 157 Washington, Booker T., 36

Washington and Lee University, parental income and college selectivity, 65 Washington Monthly, 108, 338 Washington Post, 107 Weisbrod, Burton, 99 Wellesley College, 34; SAT and application, 63; tuition increases, 124; marketing to reach prospective students, 153 Wesleyan College, 55; ­family income, 236; ­family income and selectivity, 397n39 Wicksteed, Philip, 118 William Carey University, 182 Williams College: parental income and college selectivity, 65; single sex to coed, 94; marketing to reach prospective students, 153; ­family income and selectivity, 397n39 Winston, Gordon, 116–118

Winston-­Salem State University: ­family income and selectivity, 69; ­family income, 240 ­Women, schoolwork or other interests, 259 Yale University, 56; admission criteria and religion, 42–43; SAT and application, 63; parental income and college selectivity, 65; single sex to coed, 94; share of undergraduates, 122–124; charitable gifts, 128; U.S. News ranking and ac­cep­tance rate, 165; legacy admissions, 171; production of distinguished alumni, 313; per student income, 371n47; geographic shift away from Northeast, 372n18; highly selective, 377n20 York College of Pennsylvania, 33, 183 Young, Michael, 353

I NDE X 439