Gender, War, and World Order: A Study of Public Opinion 9781501738159

Citation preview

Gender, War, and World Order

a volume in the series

Cornell Studies in Security Affairs Edited by Robert J. Art, Robert Jervis, and Stephen M. Walt A list of titles in this series is available at cornellpress.cornell.edu.

Gender, War, and World Order A Study of Public Opinion

Ric h ar d C . Ei c henberg

Cornell University Press Ithaca and London

Copyright © 2019 by Cornell University All rights reserved. Except for brief quotations in a review, this book, or parts thereof, must not be reproduced in any form without permission in writing from the publisher. For information, address Cornell University Press, Sage House, 512 East State Street, Ithaca, New York 14850. Visit our website at cornellpress.cornell.edu. First published 2019 by Cornell University Press Library of Congress Cataloging-in-Publication Data Names: Eichenberg, Richard C., 1952– author. Title: Gender, war, and world order : a study of public opinion / Richard C. Eichenberg. Description: Ithaca : Cornell University Press, 2019. | Series: Cornell studies in security affairs | Includes bibliographical references and index. Identifiers: LCCN 2018059500 (print) | LCCN 2018060259 (ebook) | ISBN 9781501738159 (e-book pdf) | ISBN 9781501738166 (e-book epub/mobi) | ISBN 9781501738142 | ISBN 9781501738142?(hardcover ;?alk. paper) Subjects: LCSH: War—Public opinion—Sex differences. | Violence—Public opinion— Sex differences. | National security—Public opinion—Sex differences. | Women and war. | Sex differences (Psychology) Classification: LCC U21.2 (ebook) | LCC U21.2 .E43 2019 (print) | DDC 303.6/ 6081—dc23 LC record available at https://lccn.loc.gov/2018059500

For Drusilla K. Brown

Contents

List of Tables and Figures

ix

Acknowledgmentsxiii xvii

List of Abbreviations

Introduction: Gender, War, and World Order

1

1. Hypotheses, Data, and Method

11

2. Threats, Power, War, and Institutions

27

3. The Gendered Politics of Defense Spending

47

4. American Attitudes toward Torture

62

5. Gender Difference in American Public Opinion on the

Use of Military Force

79

6. Gender Difference in Cross-National Perspective

98

7. Global Variation in Gender Difference

116



137

Conclusion: The Shadow of Violence

Appendix153 Notes159 References165 Index177

vii

Tables and Figures

Tables 2.1. Percent rating selected issues as a “very important” threat, 200630 2.2. Percent who agree that “the best way to ensure peace is through military strength,” by ideology and gender, 2004–5 32 2.3. Responses to the statement: “Under some conditions war is necessary to obtain justice,” June–October 2013 36 2.4. Percent who agree that “when vital interests of our country are involved, it is justified to bypass the UN,” 2003–5 41 2.5. Support for sharing the burden of NATO military action, 200844 2.6. Percent who agree/disagree that country should abide by an EU decision to use military force, 2006–7 45 3.1. Average national gender difference in net support for defense spending, 2002–13 55 3.2. Regression analysis of support for defense spending, 2003 58 and 2013 3.3. Gender difference in percent who favor the European Union becoming a superpower “even if this implies greater military expenditures,” 2003–5 60 4.1. Opinions of torture in response to variously worded questions67 4.2. Support for torture by party and gender 71 4.3. Regression analysis of support for torture and presidential approval74

ix

TABLES AND FIGURES

5.1. Support for using force and gender difference in twentyseven historical episodes 83 5.2. Support for using military force for specific types of military action86 88 5.3. Average gender difference by principal policy objective 5.4. The effect of mentioning casualties on support for using force89 5.5. Gender difference during wars in Iraq and Afghanistan 90 5.6. Gender difference and multilateral participation in military actions92 5.7. Regression analysis of support for using military force and 93 resulting gender difference 6.1. Support for the use of force and gender difference in 101 hypothetical circumstances 6.2. Comparison of support for using force and gender difference in two hypothetical situations 102 6.3. Gender difference in support for troops in Afghanistan 104 6.4. Support for sending arms and military supplies to antigovernment groups in Syria 110 6.5. Percent approving air strikes or ground troops if 114 negotiations on Iranian nuclear program fail, 2012 7.1. List of historical episodes, number of survey questions, and 118 number of states surveyed 7.2. Support for using force and gender difference for specific 125 types of military actions 7.3. Support for using force and gender difference by principal 126 policy objective (PPO) 7.4. Support for using military force and gender difference according to which multilateral actor is mentioned in the 127 survey question 7.5. Support for using military force and gender difference by 130 national characteristics 7.6. Regression analysis of support among men and women for using military force and resulting gender difference 132 155 A1. List of sources for data shown in Table 4.1 A2. List of states included in the data collection on comparative support for using military force (chapter 7) 157

Figures 2.1. Gender difference in attitudes toward global involvement, 2002–528 2.2. Gender difference in views of war, June–October 2013 35

x

9781501738142_Text.indb 10

4/9/2019 7:25:01 PM

TABLES AND FIGURES

2.3. The relationship between the political empowerment of women and the size of the gender difference on “war is necessary”37 3.1. Net support for defense spending and gender difference in 49 the US, 1965–2013 3.2. Net support for defense spending and gender difference, 2002–1353 4.1. Number of mentions of Abu Ghraib in New York Times and TV news 64 4.2. Percentage of respondents who say torture is “frequently or 70 sometimes justified,” by gender and party identification 4.3. Combined marginal effects of partisanship and gender on 76 support for torture 5.1. Average absolute gender difference by half decades 85 6.1. Gender difference in support for US drone strikes 107 7.1. Support of women and men for using military force in ten historical episodes 123 7.2. Gender difference in support for using force according to which military forces are mentioned in the question 128

xi

Acknowledgments

Thirty years ago, I published a book on public opinion and national security in Western Europe. The word gender never appeared in that book. This book, in contrast, focuses exclusively on the relationship between gender and attitudes toward security issues. What changed? The answer is both personal and intellectual. I have benefitted from the encouragement, assistance, and insights of a large number of generous people who invested the time to teach me a new subject. The list of my debts is long. First, the personal. Drusilla K. Brown, to whom this book is dedicated, has shown unfailing faith in my efforts as a scholar. Equally important, her work as scholar, mentor, and advocate for justice represents a role model for students of gender politics. I am profoundly grateful for her confidence, guidance, and for our life together. My transition to scholar of gender politics resulted from a number of profound intellectual influences. I had the very good fortune to begin my career under the tutelage of Professor Catherine McArdle Kelleher, who has remained my mentor, colleague, and friend ever since. Catherine was among the first women to receive the PhD in security studies at MIT, in 1967. She was the first employer to pay me for working in political science, opened doors for my first research trip to Europe, and invited me to coauthor my first conference paper and published article. I am extraordinarily grateful for her generosity, good spirit, and friendship. As the references in this book make clear, three scholars have had an important impact on my thinking. About twenty years ago, while preparing for my class on public opinion and foreign policy, I read a superb article by Pamela Conover and Virginia Sapiro (“Gender, Feminist Consciousness,

xiii

ACKNOWLEDGMENTS

and War,” 1993). The article remains one of our discipline’s most important contributions to the study of attitudes toward war. It is a model of theoretical development, survey design, and clear writing, and it convinced me that my future work would fall short were I to continue neglecting the importance of gender in my research. The second profound influence was Joshua Goldstein’s magisterial book War and Gender (2001). Until Goldstein wrote his book, most early scholarship on gender focused on the reasons that women might hold certain opinions, but Goldstein makes the important argument that gender difference in attitudes toward issues of war and peace result in equal or greater part from the socialization and attitudes of men. I owe a huge intellectual debt to more than two hundred students—most of them women—who have taken my class on gender issues in world politics at Tufts University. I am very proud of the fact that every student speaks during every meeting of this class. I learned more from them in the last ten years than I had learned in the previous twenty. Their ideas shaped this book in every way, and I am grateful that they allowed me to learn while pretending to be the teacher. Two of those students contributed directly to the completion of this book. Benya Kraus and Anna Weissman served as my editors on earlier drafts. Both of them read every word and offered suggestions for improvement. To Anna and Benya, a special word of thanks. Richard J. Stoll and I met on our first day of graduate school and have been collaborators ever since. Ric is without question the most cheerful and least complicated collaborator on the planet. He also read every word of the manuscript and offered suggestions for improvement. I am grateful for our collaboration and for his permission to use portions of one of our collaborative articles in chapter 3. Among many other scholars of public opinion and foreign policy who allowed me to learn from them and offered comments on earlier versions of this book, I am especially grateful to Karen Devine, Ole Holsti, Pierangelo Isernia, Mary-Kate Lizotte, and Hans Rattinger. Thanks also to Adam Berinsky for sharing his raw data on gender difference in his database of World War II survey questions. A number of Tufts University students served as my research assistants over the years. Many thanks to the members of this team who collected a large amount of original data: Jennifer Basch, Eric Giordano, Victoria Gilbert, Lily Hartzell, Angela Kachuyevski, Brian Loeb, Tali Paransky, Blair Read, Elizabeth V. Robinson, Laura Schenkein, and Judith Walcott. Tufts University has been my intellectual home for more than thirty years, and I am grateful for the personal and institutional support that I have received. Thanks to Jeff Berry, Kent Portney, and Debbie Schildkraut for listening patiently as I explained the book’s evolution. I also acknowledge the financial support of the Tufts Committee on Faculty Research Awards for released time, research assistance, and support for the costs of manuscript preparation.

xiv

ACKNOWLEDGMENTS

I was fortunate to have the opportunity to present draft portions of the manuscript to several institutions and organizations. My thanks to Steve Brooks and the John Sloan Dickey Center for International Understanding at Dartmouth College, and to Deborah Jordan Brooks, Yusaku Horiuchi, and the Dickey Center fellows for very helpful comments on the project. Thanks also to Rob Paarlberg and Harvard’s Weatherhead Center for International Affairs for inviting me to present portions of the research, and to Tom Scotto and Pierangelo Isernia for inviting me to participate in the European Consortium for Political Research Joint Sessions on public opinion and the use of force. A portion of chapter 3 appeared in earlier form in the Journal of Conflict Resolution 56, no. 2 (2012): 336–39. Earlier versions of chapter 5 appeared in International Security 28, no. 1 (summer 2003): 110–41; and International Stud­ ies Quarterly 60, no. 1 (2016): 138–48. Some data employed in chapter 7 are taken from Janet G. Stotsky, Sakina Shibuya, and Suhaib Kebhaj, “Trends in Gender Equality and Women’s Advancement,” IMF Working Paper 16/21 (Washington, DC: International Monetary Fund). I am grateful to the authors for their permission to present the data here. Roger Haydon at Cornell University Press has been a calm and reassuring editor, and two anonymous reviewers provided thoughtful and detailed suggestions for sharpening the book’s argument and contribution. Ellen Murphy and Jennifer Savran Kelly at Cornell cheerfully answered my many questions as I prepared the book for production. Finally, I am profoundly grateful to the four accomplished offspring who grew up in an academic household and love to tease dad about it. Elizabeth, Tzeidel, Carter, and Madeleine have contributed to this book in many ways, not least by preventing me from taking myself too seriously.

xv

Abbreviations

CI CIA EU GMF GSS ICPSR ISAF ISIS JCPOA

Cooperative internationalism (Wittkopf) Central Intelligence Agency (US) European Union German Marshall Fund General Social Survey Interuniversity Consortium for Political and Social Research International Security Assistance Force (Afghanistan) Islamic State of Iraq and Syria Joint Comprehensive Plan of Action (final Iran nuclear agreement) JPOA Joint Plan of Action (interim Iran nuclear agreement) Militant internationalism (Wittkopf) MI NATO North Atlantic Treaty Organization OECD Organization for Economic Cooperation and Development OLS Ordinary least squares Office of Research, US Department of State ORDOS PPO Principal policy objectives UK United Kingdom UNESCO United Nations Educational, Scientific and Cultural Organization US United States (of America)

xvii

Gender, War, and World Order

Introduction Gender, War, and World Order

I ask three sets of questions in this book. First, are there significant differences in the opinions of men and women on issues of national security? Does gender difference vary across issues, culture, or time? Second, what are the sources of these differences? What hypotheses explain them? Finally, what are the theoretical and political implications of these attitudinal differences? In this book I assemble and analyze comparative survey evidence on the first set of questions, but I also explore the causes and implications of gender difference. Overreliance on single poll questions, often within single conflicts at single points in time, have weakened scholars’ understanding of these questions, and past research frequently concentrated on major wars to the exclusion of the study of other security choices. Prior work has also focused almost exclusively on gender difference in the US. In this book, I marshal an extensive collection of global opinion polls on a variety of global issues, in many countries, in many cases over a substantial number of years. The data cover both wars and lesser conflicts. I also analyze gender difference on fundamental global issues, such as the nature of power and power balance and the role of international institutions, as well as specific policy choices such as defense spending or the employment of torture as an instrument of policy. The data presented in this book represent the most comprehensive documentation of gender difference yet available in the scholarly literature. My analysis of these data substantially expands our knowledge of the extent of gender difference on national security issues. Whereas prior

1

INTRODUCTION

knowledge was largely limited to gender difference in the US, especially during major wars, I show in the chapters to follow that gender difference exists in many countries, especially on the question of using military force. Importantly, I also demonstrate that on some security issues, gender difference is small or nonexistent. Moreover, by comparing the size of gender difference on a variety of security issues, I show that the level of violence associated with military actions and wars appears to be the primary correlate of the size of gender difference. At the aggregate level, it is clear that violence and its consequences are the primary reason for the relative skepticism that women in many countries display when contemplating threatened or actual uses of military force. Gender difference on other global issues, such as the nature of power, the utility of alliances, or the legitimizing function of international institutions, are much smaller or even nonexistent. My results have both immediate and long-run political implications. At present, a number of ongoing conflicts and crises could lead decision makers to contemplate the use of military force. The data reported here indicate that where public opinion on using force divides closely, it is often related to gender. In the US and Europe in particular, a majority of men favor the use of force in some situations, but a majority of women do not. As Conover and Sapiro remarked with respect to public reactions to the Gulf War of 1991, “These gender differences are some of the largest and most consistent in the study of political psychology and are clearly of a magnitude that can have real political significance under the right circumstances” (Conover and Sapiro 1993, 1095). Gender difference both in the US and within the public opinion of allies therefore complicates what is already a difficult task of consensus building. Indeed, in the longer term, the political mobilization of women and the increasing representation of women in political institutions could shift government policies away from forceful approaches to international conflicts. For example, Koch and Fulton (2011) studied twenty-two democracies from 1970 to 2000 and found that the increasing representation of women in legislatures reduced the conflict behavior and defense spending of their countries, a result that is not surprising given the data that I present here. Clayton and Zetterberg (2018) found that large increases in parliamentary gender quotas lead to an increase in spending for health care and a decrease in spending on defense. In a number of studies, Valerie Hudson and her colleagues found that societies characterized by higher gender equality exhibit higher domestic stability and more peaceful international behavior (Hudson et al. 2009; Hudson et al. 2012). In summary, the combination of gender difference in preferences on issues of war and peace and the increasing political mobilization of women may herald a gradual shift away from war and violence as instruments of policy.

2

GENDER, WAR, AND WORLD ORDER

The Plan of the Book and Summary of Findings In the pages that follow, I present evidence on the extent of gender difference in public opinion across a large number of security issues, countries, and time periods. In chapter 1, I review five sets of hypotheses that seek to explain the magnitude and variation of gender difference on security issues. I also offer thoughts about what specific evidence would constitute a confirmation or refutation of each set of hypotheses. Put briefly, the essentialist hypothesis predicts that gender difference will be large and constant across cultures, issues, and time. Hypotheses relating gender difference to economic change and the political mobilization of women suggest two things: that large gender difference will characterize issues of pragmatic self-interest (such as a preference for social spending over defense spending) and that the magnitude of gender difference will correlate with the extent of the economic and political mobilization of women. A third set of hypotheses argues that gender difference on security issues is embedded in a broader difference of worldviews that results from the socialization of boys to the masculine norms required of soldiers. This hypothesis predicts that gender difference will characterize views of power, power balance, the acceptability of war, and support for international institutions. Fourth, research on perceptions of threat, risk, and fear of violence predicts that gender difference arises primarily in reaction to the threat or actual use of violent military force. Finally, evidence from several cross-national studies suggests that a society’s geopolitical position mediates gender difference. In addition to a detailed explication of these hypotheses, chapter 1 describes the data that I analyze and briefly discusses methodological issues. Chapter 2, directly addresses the hypothesis that gender difference arises from a broader difference in worldviews. Specifically, I examine gender difference in attitudes toward international involvement, threats, the utility of power and balance of power, the acceptability of war, and multilateral institutions and their legitimizing functions. I find little evidence to support the hypothesis of a difference in worldviews. International involvement is uncontroversial in most states and evokes no gender difference. Concerning general attitudes toward multilateral organizations, it is only in the US that there is a substantial gender difference (in support for the UN). When the issue turns to the role of the UN, NATO, or the European Union in legitimizing or collectivizing military actions, women are less likely to express support than men are—the opposite of what one would expect, and this pattern suggests that the use of military force is the issue that divides the genders. Multilateral endorsement or participation is secondary. Finally, gender difference is not universal across issues or nations. Cross-nationally, gender difference is most consistently prominent in the US across a number of issues, which should alert us to the possibility that gender politics in the US have some unique qualities. 3

INTRODUCTION

Second, the fact that gender difference varies cross-nationally and across issues casts doubt on the essentialist hypothesis. I explore the pragmatic politics of gender difference in chapter 3, which describes the historical evolution of men and women’s opinions of defense spending in the US and compares gender difference in the US to that in Europe since 2002. In general, women are moderately less supportive of defense spending in both the US and Europe, but the evidence also indicates that the primary factor underlying gender difference is beliefs about war and military force. Gender difference on defense spending also varies substantially. In some years, difference is large, but in many more years, it is small or even positive (that is, women favor defense more than men). However, the largest gender difference on defense spending occurred in 2003 and 2004, the first two years of a very controversial war in Iraq, providing further evidence that the negative reaction to the defense budget among women arose more from war than by the pragmatic considerations of budgetary politics. Chapter 4 documents gender difference on a subject that has provoked a great deal of controversy in the US: the use of torture as a policy tool in the war against terror. To frame the issue, I ask if partisanship or gender difference more strongly influence opinions of torture. The findings suggest that, like other opinions on national security issues, opinions of torture are structured above all by partisanship, but this polarization is accompanied by substantial gender difference that at times equals or even supersedes partisanship. The results also reinforce an important argument made by Sapiro: that the effect of gender on political attitudes is context specific (2002; see also Sapiro 2003). Many theories of gender difference on national security issues aspire to a universal or indeed essentialist explanation, but the results in this chapter show that gender is important at specific times under specific circumstances. In chapter 5, I present my analysis of a substantial new data collection on gender difference in US public support for using military force in twenty-six historical episodes since 1980. I find that fewer women supported the use of military force for any purpose. In all but one historical case, women were more critical of using force. Second, the magnitude of this gender difference varies considerably. This raises the question of what factors explain such variation. Third, gender difference reveals no upward secular trend and shows no correlation with the partisanship of the president. Fourth, in most historical conflicts, women demonstrate more sensitivity to humanitarian objectives and to the loss of human life. However, an exception to this pattern occurred during the wars in Iraq and Afghanistan: a large gender difference existed at the beginning of both wars, but it narrowed during the course of the wars because men reduced their support more rapidly than women did. Fifth, despite the presence of gender difference, no categorical template captures the views of men and women. An average of 44 percent

4

GENDER, WAR, AND WORLD ORDER

of women supported the use of force since 1982, and an average of 42 percent of men opposed it. In summary, gender difference varies in response to specific contexts and to specific military actions. As is true in the findings in other chapters, I conclude that the essentialist model is inconsistent with the substantial variation in gender difference, and several findings in the chapter add further evidence that violence and war evoke the largest gender difference. I analyze gender difference in a cross-national comparison of more than forty countries in chapter 6. I examine two sets of survey questions: hypothetical questions about deploying troops, and questions about using force in three specific historical situations. Gender difference in the hypothetical scenarios is small. However, when actual combat is the focus of the hypothetical question, gender difference increases, with women more opposed. In contrast, women show higher support for using force to further human rights. When I focus on specific historical cases in which military force has been used or threatened, gender difference is larger—in some cases substantially larger and politically significant. In fact, in questions concerning deployment and maintenance of troops in Afghanistan, use of drone strikes, and potential use of force against Iran, gender difference is substantial and statistically significant. In many countries gender polarization complicates a precarious national majority or plurality in favor of military action. Majorities of men favor military action, while women oppose it. Finally, in surveys concerning the provision of military assistance to regime opponents in the Syrian civil war, there is evidence that local circumstance and context affects the size of gender difference. In some countries (Jordan), women are supportive of assisting Syrian rebels, while in others (Egypt) they are less so. The findings in this section therefore suggest that national context mediates the extent of gender difference. In chapter 7, I present a cross-national analysis of gender difference in ten historical episodes since 1990. I based the analysis on a unique new data collection of opinions concerning the use of military force in sixty-two countries, beginning with the Gulf War of 1991 and ending with reactions to Russian intervention in Ukraine in 2015. The analysis demonstrates that gender difference varies substantially across societies and within a number of international conflicts. These findings once again cast doubt on any theory that would predict constant gender difference, most importantly any essentialist, biological explanation. There are many commonalities in the correlates of men’s and women’s attitudes toward the use of force, but the direction of gender difference is generally that women are less supportive of using military force than men. Gender difference varies positively with economic growth and measures of gender equality, which suggests that it results from changes in the political circumstances, economic structures, and attitudes that accompany economic development. Nonetheless, while it is true that gender difference is higher on average in more

5

INTRODUCTION

economically developed societies with higher levels of gender equality, gender difference concerning the use of force is also large in several societies with lower levels of gender equality (Brazil, Saudi Arabia, and Turkey). Finally, a comparison of the correlates of gender difference in the US to gender difference within US allies and within other societies indicates a number of commonalities. Humanitarian interventions lower gender difference, and violent actions such as air strikes increase gender difference. Among the notable cross-national differences, gender polarization outside the US is lower for multilateral missions because women react more positively than men do. In the US, both men and women display increased support for military action if it involves multilateral missions. Thus, gender difference arising from a divide between more “realist” men and “liberal” women is not a universal phenomenon. Taken together, the data presented in this book point to the centrality of military force, violence, and war as the most important variables affecting gender difference. This is true at both the aggregate level (percentage of women and men who support using force) and the individual level (the impact of gender, controlling for other variables that influence individual attitudes). A second prominent finding is that the magnitude of gender difference on security issues correlates with the economic development and the level of gender equality in a society. Third, while it is true that large gender difference is present in many countries, gender polarization is most consistent across the widest range of issues in the US. In the concluding chapter, I return to these findings. I also make suggestions for additional research to uncover the causal process that makes women more skeptical of the use of military force relative to men. Finally, I discuss the implications of my findings for the politics of national security in the Atlantic alliance and the impact of increasing gender equality on global security.

The Evidence Gap in the Study of the Gender Gap In their study of gender difference in US citizen reactions to the 1990–91 Persian Gulf crisis and war, Pamela Conover and Virginia Sapiro analyzed a number of survey questions dealing with hypothetical national security policies as well as concrete questions involving the use of military force and its consequences. The results showed that gender difference on the hypothetical questions was weak or nonexistent. However, when the analysis turned to specific questions concerning the use of force against Iraq and the civilian and military casualties that could result, the difference became large indeed. Conover and Sapiro concluded, “When we moved from the abstract to the concrete—from hypothetical wars to the Gulf War—the distance separating women and men grew, and on every measure, women reacted more negatively. These gender differences are some of the largest

6

GENDER, WAR, AND WORLD ORDER

and most consistent in the study of political psychology and are clearly of a magnitude that can have real political significance under the right circumstances” (Conover and Sapiro 1993, 1095). Less than ten years later, as NATO warplanes attacked Serbia, the Chris­ tian Science Monitor reported that gender difference in US public opinion concerning the war over Kosovo was far smaller than in the past: “As debate persists in America over how much to use force, fewer women are ‘doves’ ” (Marks 1999, 1). After September 11, 2001, the Council on Foreign Relations conducted a survey on defense issues and reported “women’s opinion on defense policies has been transformed” because their views of defense spending and missile defense (among other issues) now closely resembled those of men (Council on Foreign Relations 2001). However, in yet another reversal of patterns, studies of citizen reaction to the wars in Iraq and Afghanistan found large gender differences (Eichenberg 2003; Huddy, Feldman, and Cassese 2009). A Pew Research Center poll in 2013 also found a substantial gender difference in Americans’ approval of drone strikes as part of the “war against terror” (Stokes 2013). What explains these gyrations in the size of the gender difference on issues of war and peace? One might speculate that the 1991 Gulf War was unique, a dramatic, highly publicized occurrence accompanied by considerable discussion of potential casualties and an extremely polarized political leadership. That public opinion would also polarize in such a context is not surprising. The Kosovo War, in contrast, began with President William Clinton’s stated intention to avoid using ground troops, thus lessening the fear of casualties. One might also speculate (as did Marks 1999) that the war over Kosovo involved humanitarian and other issues that convinced women of the moral necessity of using force to halt the atrocities carried out by Serbian forces in Kosovo—many of which were directed against women. Further, while the attacks of September 11 understandably brought near unanimity to the views of US citizens and leaders in the short term, the actual response of the US government over the next ten years caused a substantial polarization of public opinion concerning the wars in Iraq and Afghanistan and the use of military force more generally. However plausible such speculation, there is no basis in the social science literature for favoring one or the other of these explanations because scholarly knowledge of gender difference on national security issues remains incomplete. One obstacle to generalization is a tendency to focus on single survey questions (as in the examples above). We know from the literature on public opinion that a single survey question is a misleading guide to the public’s preferences. Survey respondents are sensitive to variations in question wording, so responses may change in reaction to even minor differences in wording, and these variations may have gendered implications (Zaller and Feldman 1992). For example, some scholars have speculated that women are more likely to support military actions

7

INTRODUCTION

undertaken or endorsed by multilateral institutions, but testing this argument requires comparison of questions concerning unilateral and multilateral military actions. Research also requires multiple questions to assess the extent of gender difference on other issues, such as sensitivity to casualties in military operations, support for defense spending, or support for torture. Second, existing scholarship focuses almost exclusively on major wars, such as World War II, Korea, Vietnam, and the first Gulf War—all highly salient, violent, and costly struggles. Far less research analyzes limited conflicts with more limited costs and objectives. For example, considerable survey data exist on the US public reaction to proposed or actual interventions during the 1980s in El Salvador, Libya, Nicaragua, and Lebanon, to the wars in the Balkans in the 1990s, and more recently to the wars in Libya and Syria. Exploitation of data from these cases seems particularly important given the finding in some research that gender difference may be smaller in the case of interventions in civil wars, for peacekeeping operations in the aftermath of civil wars, or for interventions that are explicitly humanitarian (Eichenberg 2003; Brooks and Valentino 2011). As Nincic and Nincic observed in their study, “Our research encompassed only very major instances of military intervention, and what applies at one extreme of the continuum may not apply to less dramatic and costly cases of resort to force” (2002, 565). Of course, major wars are important, and earlier wars may provide a useful template for studying the more recent wars in Afghanistan and Iraq. However, the considerable survey data amassed during these wars remain underexploited. Pollsters in the US conducted well over four hundred surveys during the wars in Afghanistan and Iraq, but only a few studies have addressed the issue of gender and support for using force (Huddy et al. 2005; Huddy, Feldman, and Cassese 2009; Gelpi, Feaver, and Reifler 2005–6). This is curious, since some previous research finds that the mention of casualties in surveys increases gender difference (Conover and Sapiro 1993; Eichenberg 2003). Surveys conducted during more than ten years of war in Iraq and Afghanistan provide the basis for a more exhaustive analysis of whether gender difference increases with the mention of casualties and whether gender difference increases as casualties mount over time. More generally, because the wars in Afghanistan and Iraq occurred after the attacks of September 11, 2001, a more extensive study of the data may provide some insight into the question of whether gender difference underwent a change after the attacks on the US or during the ensuing “war against terror.” Third, while it is certainly possible that opposing attitudes toward war are the primary cause of gender difference, this remains a hypothesis to test against competing hypotheses. For example, some scholars have argued that gender difference on war issues is in fact part of a broader difference

8

GENDER, WAR, AND WORLD ORDER

in worldview that resembles the liberal-realist divide in international relations theory (Tickner 1988; Wolford and Johnston 2000). If so, gender difference in attitudes toward international involvement, alliances, military power, balance of power, and international institutions should also be evident, but few existing studies examine these issues. Fourth, the scholarly literature exhibits an almost exclusive focus on gender difference in the US, but the American experience may not be generalizable. The US has been the dominant global power since the end of the Cold War and has engaged in many military interventions, most of them contentious in domestic politics. Further, political conflict over gender issues has been a prominent feature of American politics more generally. Indeed, some scholars have argued that gender difference in the US now rivals partisanship as the most important domestic cleavage (Erikson, MacKuen, and Stimson 2002, 168–76). Although cross-national studies of gender difference on security issues are few, those that do exist produce conflicting results. In some countries, gender difference is large (Togeby 1994), but in others it is small or nonexistent (Arian 1996; Tessler and Warriner 1997). Moreover, most research outside the US context presents case studies of individual countries on a single issue, such as sending British or Canadian troops to Afghanistan (Boucher 2010; Clements 2013). In short, there is a need for truly cross-national analysis that compares gender difference on a wider range of security issues for a larger number of countries. Comparison is necessary for an additional reason: many hypotheses that seek to explain gender difference require evaluation in broad cross-cultural comparison. This is especially true of hypotheses that imply universal gender difference independent of cultural, political, economic, and geopolitical circumstance. For example, the so-called essentialist hypothesis argues that gender difference is rooted in biological difference and thus predicts large and unvarying gender difference across cultures, political systems, and global position. Other theories requiring comparative inquiry include the argument that gender difference arises from changes in society, especially changes in the access of women to higher education and increasing levels of labor force participation (Inglehart and Norris 2003; Iversen and Rosenbluth 2006). A test of these theories requires comparison of societies that vary in culture, level of economic development, and geopolitical position. Fifth, existing research lacks analysis over time. This is an important limitation because of the inherently historical nature of several hypotheses about gender difference. For example, several scholars hypothesize that the emergence of gender difference results from the gradual political mobilization of women that accompanies the increasing access of women to higher education and labor markets (Brandes 1994; Inglehart and Norris 2003; Iversen and Rosenbluth 2006). Inglehart and Norris associate this process with a shift to the left in women’s policy preferences, including a preference for more pacifist policies, but confirmation requires historical evidence to

9

INTRODUCTION

evaluate the argument more fully (2003, 92). Similarly, reasoning from the fact that a gender difference in partisan identification and voting in the US arose and widened in the 1980s (Kaufmann and Petrocik 1999; Erikson, MacKuen, and Stimson 2002: 168–76), we might hypothesize that gender difference on security issues will show a similar pattern. Perhaps women support military actions undertaken by Democratic presidents, and men support the actions of Republican presidents. Note that these two hypotheses yield different predictions: if Inglehart and Norris are right, we would expect a secular increase in the magnitude of gender difference on security issues over time, but the partisan hypothesis predicts variation according to which party occupies the presidency. Without historical data, however, we cannot know which is closer to the truth. Finally, in this book I try to improve on the ad hoc quality of theorizing that characterizes much of the scholarship on gender difference and international security. Especially in the early literature, the discovery of gender difference that had previously escaped the attention of scholars often led to speculation that amounted to a flight into stereotypes, especially speculation that women are simply more “caring and nurturing” than men. To many feminists (as well as others), this stereotype dismissed the possibility that women’s views were arrived at through rational calculation and life experience, and they ignored altogether the obvious point that gender difference is also a function of men’s opinions. Of course, specific hypotheses based on socially constructed gender roles have been formalized and tested (Elder and Greene 2007; Brooks and Valentino 2011), but a fuller assessment of this and other hypotheses is needed. Although I do not offer a single formalized causal model of gender difference in this book, I do attempt to assess the extent to which the weight of the evidence in available survey data is consistent with existing hypotheses.

10

chapter 1

Hypotheses, Data, and Method

Previous scholarship shows that gender difference on national security exists on some issues, at some times, and in some countries. However, explaining gender difference has proven more difficult. As I observed in the introduction, explanations tend to be ad hoc—reacting to the most recent major event and relying on fragmentary polls on single issues, historical episodes, countries, or points in time. In this chapter, I bring some conceptual organization to explanations of gender difference by reviewing and evaluating several groups of hypotheses.

The Essentialist View A large body of scholarship on gender difference evaluates the putative effect of biological sex or socialization to emotions, norms, and roles based on biological sex (especially the fact that women bear children and have the largest responsibility for their nurture and survival). Hypotheses that emphasize biological sex often derive from evolutionary biology, especially the fact that natural selection favored larger, physically stronger males who could successfully compete—fight—for territory, mates, and food. Women, in contrast, bore children and provided the nurture that ensured survival of the group (Goldstein 2001, 128–83; Hudson et al. 2008–9). Related hypotheses based on motherhood carry this argument further, arguing that “women’s unique role as nurturers and primary caregivers endows mothers with a greater respect for life and makes them more empathetic and caring toward others” (Brooks and Valentino 2011, 272). Goldstein summarizes the argument more directly: “[Motherhood] bests suits [women] to give life, not take it. Women are more likely to oppose war, and more likely to find alternatives to violence in resolving conflicts” (2001, 43). Hypotheses such as these are highly contested (see Goldstein 2001, 128–82; and Reiter 2015 for a review of evidence). Most scholars now discount them for an important reason: hypotheses based on biological difference

11

CHAPTER ONE

predict large and invariant gender differences across time, issues, and cultures, but existing scholarship demonstrates that gender difference varies across all of these dimensions (Reiter 2015, 1311). For example, research has largely disconfirmed the hypothesis that mothers or fathers have distinct views on security issues. That is, although women in general show lower support for defense spending and the use of force, mothers are not different from other women (Conover and Sapiro 1993; Elder and Greene, 2007). In addition, in studies of US opinion, scholars find large gender difference on some global issues and conflicts, but small difference on others, and gender difference varies over time (Conover and Sapiro 1993, 1086–95; Burris 2008; Eichenberg 2003). As noted above, there is less cross-national research, but the limited evidence suggests variation rather than uniformity. Evidence from the wealthier Western democracies shows gender difference, but in other countries, the difference is modest. For example, Tessler and his colleagues studied attitudes toward diplomacy, military action, and territorial compromise in Egypt, Israel, Palestine, Kuwait, Jordan, and Lebanon and found very small gender differences (Tessler and Warriner 1997; Tessler, Nachtwey, and Grant 1999). In summary, the variation in gender difference over time and across societies with differing cultures and strategic concerns suggests that factors other than biological sex determine gender difference. Nonetheless, a final assessment of the degree of constancy or variation in gender difference must remain tentative, because the number of historical and cross-national studies is small. In addition, to my knowledge, no cross-national study evaluates identical issues across an identical period of time.

Economic Development and the Political Mobilization of Women A second set of hypotheses about gender difference in attitudes toward global issues emphasizes the attitudinal changes that result from the cultural, economic, and political transformations associated with the transition to industrial and postindustrial societies. Scholars have identified two trends that have significant implications for gender difference in the politics of national security. The first is the increasing access of women to higher education and to the labor market outside the home. As late as the 1980s, women represented less than half of enrolled students in postsecondary education in Western Europe and exactly half in the United States. In some countries, it was much lower (for example, approximately 38 percent in the United Kingdom and the Netherlands). By 2009, in contrast, well over half of postsecondary students in Europe and the United States were women, and in some countries, the increase was prodigious (for example, from about 38 percent to substantially over 50 percent in the United Kingdom and the Netherlands). In Turkey, only a quarter of postsecondary students

12

GENDER, WAR, AND WORLD ORDER

in 1980 were women; by 2009, the figure had reached 43 percent (UNESCO Institute for Statistics 2015; 2017). Similar changes occurred in labor markets. In the 1970s, less than 50 percent of women were active in the labor force in most countries for which data are available. The exception is Sweden, where exactly half of women had joined the labor force. In the US, France, Germany, Italy, and the Netherlands, an average of 37 percent of women participated in the labor market during the 1970s (Federal Reserve Bank of St Louis 2017; OECD 2017). This translated into an average of 38 percent of the total labor force. By 2010, however, more than half of women worked outside the home in many countries, and women represented more than 45 percent of the labor force in most countries of Europe and in the United States (OECD 2017). Inglehart and Norris provide the most comprehensive analysis of the political implications of these changes. These scholars focus on the emergence of “psychological autonomy” and shifting policy preferences among women (2003, 19–26). For example, as societies shift from rural, agricultural production to urban, industrial production, fertility rates decline and the primacy of extended families is replaced by the nuclear family, both of which reduces the caregiving burdens of women. Reinforcing this shift is an increase in the public provision of services previously provided by women in the home, especially child and elder care and health care. As these changes take place, women take advantage of the expanded availability of public education and access to the paid labor force. Finally, industrial and postindustrial production are accompanied by a shift in values, the most important of which is the decline of traditional forms of authority, including religious authority and patriarchy, and their replacement by secular, rationalist authority and increasing support for gender equality. Together, these changes reduce the burdens of women in the home, erode the traditional values that marginalized women’s political voice, and provide educational opportunities and increased access to paid work. These changes have substantial political consequences. For example, the increase in the labor force participation of women has led some scholars to argue that gender difference in policy preferences arises from the distinct material needs of men and women rather than innate characteristics or differing conceptions of national security. As women enter the workforce in growing numbers, their reliance on social services grows as well. For example, in an early study, Shapiro and Mahajan found significant gender differences on public spending issues in the US: “Women were more supportive of a guaranteed annual income, wage-price controls, equalizing wealth, guaranteeing jobs, government-provided health care, student loans, and rationing to deal with scarcer goods” (Shapiro and Mahajan 1986, 51). Gilens (1988) made a similar finding in a study of the role of policy preferences in explaining gender difference in President Reagan’s job approval ratings. Strikingly, Gilens found that divided gender preferences on the

13

CHAPTER ONE

issues of defense spending and social spending produced the strongest correlation with Reagan’s approval ratings. Moreover, these effects were stronger than partisanship in explaining Reagan’s approval. As a result, Gilens asserts, “gender differences in the evaluation of politicians will extend beyond President Reagan and are likely to appear whenever military or social welfare issues figure prominently in the public’s assessment” (1988, 45). Recent scholarship replicates these findings beyond the 1980s. For example, Richard Stoll and I found that women in the US were about 8 percentage points less likely to support increased defense spending during the period 1965–2007 (Eichenberg and Stoll 2012). Analyzing a variety of policy items from US election studies over the period 2000 to 2004, Crowder-Meyer found large, consistent gender differences on defense issues (including defense spending) and social welfare issues. Furthermore, Crowder-Meyer shows that men and women differ both in their prioritization of these issues and in their propensity to condition their voting behavior on these issues. Men are more likely to give defense a higher priority and to base their evaluation of candidates on the issue. Women, in contrast, rank social welfare higher and are more likely to condition candidate evaluations on the issue (Crowder-Meyer 2007). Similarly, Kaufman and Petrocik studied the impact of gender difference in policy attitudes on both party identification and the vote in the 1992 and 1996 presidential elections. The results show that social spending and defense-related attitudes are strong correlates of party identification and voting in 1992 and 1996 (along with other policy attitudes). There is less research on gender differences in policy preferences outside the US, but the evidence that does exist suggests that women’s higher relative preference for social service programs is widespread. For example, analyzing a question from the World Values Survey that asks if “government should take more responsibility to ensure that everyone is provided for,” Inglehart and Norris find that “women are overwhelmingly more favorable to an active role for the state”(Inglehart and Norris 2003, 83). Similarly, Iversen and Rosenbluth studied a sample of ten OECD democracies and found that “women everywhere want the government to take a more active role in public employment creation,” in part to support the “partial socialization of family work,” but also to increase women’s employment prospects outside the home (2006, 18). Change in labor markets also helps to explain an additional pattern in industrial and postindustrial societies: women have shifted their partisan loyalties to the left, and there is some evidence that they have become more pacifist in orientation. For example, Inglehart and Norris show that women born in the middle years of the twentieth century showed a slight preference for parties of the right. Over time, however, there has been a leftward shift, with younger women in many European countries now slightly more likely to prefer parties of the left (Inglehart and Norris 2003, 83–88). The

14

GENDER, WAR, AND WORLD ORDER

same pattern is evident in the United States and other Western societies (Kaufmann and Petrocik 1999; Iversen and Rosenbluth 2006). The fact that younger generations of women have shifted most distinctly in their political orientation has obvious implications for the future politics of security. As younger generations of women replace their elders, there should be increasing gender difference in policy views.1 Gender difference in policy preferences amplifies the increased political mobilization and independence of women. For example, one of the most consistent findings of political behavior research in the US is that engagement and participation increase with income and education: the higher the level of educational attainment and material resources, the higher the rate of political engagement (Burns, Schlozman, and Verba 2001; Inglehart and Norris 2003, 102–4). Second, traditional patterns of patriarchy are weakened as women gain what Inglehart and Norris refer to as “economic and psychological autonomy” (2003, 90). To the extent that women have or develop distinct policy preferences, educated and employed women are more likely to express them. As Iverson and Rosenbluth put it, under conditions of traditional patriarchy “family members are assumed to have more or less identical preferences.” However, as women become more independent through higher education and paid employment (and the increasing incidence of divorce), “we have to treat family members as individuals with distinct and potentially conflicting preferences” (Iversen and Rosenbluth 2006, 2).2 Further, as societies undergo the social, economic, and cultural changes of economic development, the policy preferences of women differentiate from those of men. Specifically, the expansion in women’s labor force participation increases their support for social service spending that socializes family work (Iversen and Rosenbluth 2006, 12–13). Finally, some scholars argue that women’s shift to the left in ideology, partisanship, and policy preferences includes an increase in pacifist policy views. For example, Inglehart and Norris observe that women’s preference for social programs and spending is accompanied by a preference for “pacifism in the deployment of military force” (2003, 92). Interestingly, in their study of Israel and several Arab countries, Tessler and Warriner find that, although gender does not correlate with citizen views of how to resolve the Middle East conflict, support for gender equality has a very strong effect: Specifically, Israelis, Egyptians, Palestinians, and Kuwaitis who are more supportive of equality between women and men are also more favorably disposed toward diplomacy and compromise in the Arab-Israeli dispute, whereas those who are less supportive of gender equality are in each case less likely to favor resolving the conflict on this basis. The relationship is highly statistically significant in all four instances, indicating that, in this sense at least, there is a clear connection between feminism and issues of gender on the one hand, and concerns related to international conflict on the other. Moreover, again, the dissimilar characteristics of the four societies in

15

CHAPTER ONE

which the same relationship has been observed strongly suggests that this relationship is of broad applicability. (1997, 275)

Whether these findings have “broad applicability” is not clear cut, however. On the one hand, Tessler and Warriner may be right: a shift in societal attitudes toward gender equality—a shift that would include changing attitudes among men—should reduce the significance of the relationship between gender and political attitudes and thus reduce differences in policy preferences due to gender alone. On the other hand, the research of Inglehart, Norris, and others suggests that an increase in gender equality leads to the differentiation of women’s policy preferences from those of men, which should increase gender difference in attitudes. Which view is correct is a matter for research. In any case, there are three implications of these changes for the study of gendered perspectives on global issues. First, gender difference should be rooted above all in pragmatic self-interest, especially the preference for social programs over national defense. Second, gender difference on all issues should vary cross-nationally with the level of economic development—or more specifically with the increase in access to education and labor force participation that leads to an increase in the “psychological autonomy” of women and a differentiation of household preferences. Finally, we must consider the alternative hypothesis that a societal shift toward greater gender equality may be associated with lower—not higher—gender difference in security attitudes.

Threat, Risk, and Violence Conover and Sapiro report an interesting finding in their study of gender differences during the Persian Gulf crisis and war of 1990–91. Women in the US were more likely to exhibit a “fear of war” and to express what the authors call “isolationist” sentiments, that is, they were more likely to agree that “this country would be better off if we just stayed home and did not concern ourselves with problems in other parts of the world” (Conover and Sapiro 1993, 1088–91). Berinsky finds a similar gender difference in the US in support for proposals to assist Britain and France prior to US entry into World War II, versus the option of “staying out” (Berinsky 2009, 53–54). While puzzling perhaps for students of international relations or political behavior, these results are in keeping with scholarship that investigates the relationship among gender, threat perceptions, anxiety, and perceptions of risk. Specifically, there is substantial evidence that women perceive higher threat and risk emanating from their environments than do men in the same environments (Gustafson 1998). There are several patterns in the literature. First, women are more likely to express vulnerability to violence 16

GENDER, WAR, AND WORLD ORDER

(Hollander 2001; Jackson 2009), to fear victimization (Gustafson 1998), and to perceive external threats (Hurwitz and Smithey 1998; Huddy et al. 2002; Huddy, Feldman, and Cassese 2009). Second, exactly why women are more likely to express vulnerability or fear of victimization remains unclear. Hollander (2001) argues that the greater vulnerability of women is a social construct: women perceive vulnerability because society teaches them to feel vulnerable. Gustafson (1998) acknowledges the social construction of vulnerability, but he observes that feelings of vulnerability also arise from lived experience, including differential exposure to the risk of violence. Third, women’s feelings of vulnerability and the fear of falling victim to all types of crime correlate strongly with fear of sexual assault. In fact, the fear of sexual assault correlates with seemingly unrelated phenomena, such as the threat of nuclear attack or terrorism (Huddy, Feldman, and Cassese 2009). The latter association is not surprising given the increasing amount of scholarship that documents the high levels of violence—especially sexual violence—that is directed against women in wartime (Hudson et al., 2008–9; Cohen, Green, and Wood, 2013; Cohen 2016). However, a fourth pattern in the scholarly literature is that women are less likely to favor a forceful or violent reaction to threats. For example, women in the US felt more threatened than men by terrorism after September 11, 2001, but they were less likely than men to endorse retaliatory measures, such as the initiation of the war in Afghanistan (Huddy et al., 2005; Huddy, Feldman, and Cassese 2009) and the war in Iraq (Eichenberg 2003). The reason is that women are also more likely to experience anxiety at the prospect of military retaliation, and anxiety increases perceptions of risk, uncertainty, and loss of control: “Women express higher levels of anxiety and perceive greater risks associated with war and terrorism” (Huddy et al. 2005, 594–95). Following Eysenck (1992), Albertson and Gadarian define anxiety as an “ ‘unpleasant and aversive state’ with the purpose of detecting threat and danger in the environment” (2015, 8). Anxious individuals cope by seeking more information, among other reactions (Albertson and Gadarian 2015, 10). Furthermore, “anxious individuals are motivated to reduce anxiety, leading to a preference for safe options. . . . These psychological consequences of anxiety lead us to predict that individuals who feel anxious will be inclined to oppose risky overseas military action” (Huddy, Feldman, and Cassese 2009, 291–92; emphasis added). In a study of attitudes in the US toward the use of military force, Devine found precisely the same relationship: individuals who placed a high value on personal safety and avoided situations in which their safety might be threatened were also less likely to support the use of military force (2013). She concludes that avoiding the use of force was akin to the “avoidance behavior” that individuals employ to ensure their personal safety: “The question may be eliciting a focus on believing in practicing avoidant behaviour: not sending troops would be a version of international avoidant behaviour” (Devine 2013, 17).

17

CHAPTER ONE

Because women are more likely to feel vulnerable and threatened, the avoidance of violent responses should be more prominent among women. In summary, there is substantial theory and evidence to suggest that the central issue differentiating the views of women and men are violence and the risk of violence. Specifically, we would expect the threat or actual use of violent military force to produce the largest gender difference in public opinion. However, we might also expect to find that women are more supportive of military interventions that are designed to mitigate the effects of violence, as some studies of citizen support for humanitarian intervention have suggested (Brooks and Valentino 2011; Eichenberg 2003).

Socialization and the Worldviews of Women and Men Each of the three theoretical perspectives reviewed above yields specific hypotheses about gender difference on global issues. The essentialist perspective, rooted as it is in biological sex, predicts large and invariant difference across cultures, time, issues, and strategic situations. Hypotheses based on economic development and the political mobilization of women focus more narrowly on pragmatic concerns, especially the preference of women for social over national security spending. In addition, economic development produces increased labor force participation and educational attainment among women, and these two changes contribute to changes in the political preferences and political autonomy of women. The implication is that gender difference will correlate cross-nationally with the level of economic development. Finally, a focus on risk and violence predicts that gender difference will be evident primarily on the question of employing violent military force and perhaps on military actions designed to mitigate the human suffering caused by violence. A fourth, broader perspective argues that gender difference is the result of the differential socialization of boys and girls into competing worldviews that resemble the familiar realist and liberal frameworks of international relations theory (Wohlford and Johnston 2000). Indeed, the argument of scholars who articulate this view is that society socializes boys to norms that resemble the tenets of realism—an individualist, competitive, power-oriented worldview— while girls experience socialization to liberal norms that emphasize connection, community, cooperation, and an aversion to violence. The implication of this argument is that gender difference should align on those issues that most differentiate realist and liberal thought. The clearest articulation of this argument appears in the scholarship of Tickner (1988; 1992; 1997) and Goldstein (2001). Tickner argues that Masculinity and politics have a long and close association. Characteristics associated with “manliness,” such as toughness, courage, power, independence, and even physical strength, have, throughout history, been most

18

GENDER, WAR, AND WORLD ORDER

valued in the conduct of politics, particularly international politics. Frequently, manliness has also been associated with violence and the use of force, a type of behavior that, when conducted in the international arena, has been valorized and applauded in the name of defending one’s country. (1992, 6)

Given the dominance of men in both political practice and intellectual discourse, Tickner argues that it is not surprising that the analytical and prescriptive paradigms of international relations theory are also masculinized. Tickner sees parallels between the hierarchical distinctions in power, emotion, and reason that characterize social constructions of gender and the realist discourse of international relations: “The construction of this discourse and the way in which we are taught to think about international politics closely parallel the way in which we are socialized into understanding gender differences” (1992, 9). In her critique and reformulation of Hans Morganthau’s classic statement of realist theory, Tickner focuses in particular on what she considers the masculinized conception of power (Tickner 1997; Morganthau 1948). In Tickner’s view, Morganthau’s core concept and prescription of the national interest defined in terms of power reflects a masculine conception rooted in dominance or efforts to prevent dominance. Drawing on the works of Nancy Hartsock (1983) Hannah Arendt (1970), and Jane Jaquette (1984), Tickner asserts that women’s conceptions of power emphasize energy, capacity and potential (Hartsock), “the human ability to act in concert,” or “action which is taken in connection with others who share similar concerns” (Arendt). She also cites Jane Jacquette’s argument that “since women have had less access to the instruments of coercion, women have been more apt to rely on power as persuasion” and coalition building (Tickner 1997, 434). In other words, Tickner is arguing that society socializes men to an understanding of power that resembles realism, while women are socialized to a more cooperative worldview that resembles liberalism. Thus, Tickner argues that our intellectual paradigms reflect broader societal norms and expectations, but this raises a question: What is the specific content of this socialization that is relevant to gendered views of global issues? Since most citizens do not read the works of international relations scholars, where do these norms originate? Goldstein (2001) provides an answer in his encyclopedic synthesis of social science research on the relationship between war and gender roles. Goldstein reverses the familiar theoretical argument that domestic injustice—including gender inequality—causes war. He argues in contrast that gender roles are the result of what he calls the “war system,” which he defines as “the interrelated ways that societies organize themselves to participate in potential and actual wars . . . [including] military spending and attitudes toward war, in addition to standing military forces and actual fighting” (2001, 3). Because 19

CHAPTER ONE

nation-states are aware of a latent, persistent “shadow” of war—”like a patient with cancer in remission”—they must prepare for self-defense (2001, 3). In Goldstein’s view, this task presents nation-states with a dilemma. Human beings do not take naturally to fighting and killing. Indeed, the violence, chaos, and brutality of war are frightening and repulsive to most. As a result, “cultures mold males into warriors by attaching to ‘manhood’ or ‘masculinity’ those qualities that make good warriors . . . gender identity becomes a tool with which societies induce men to fight.”3 Further, there is a striking cross-cultural consistency in the existence of “manhood rituals” in which boys are required to display warriorlike qualities to achieve the status of “men,” including the qualities of bravery, endurance, strength, skill, and honor. Boys and men are also taught to suppress emotions (fear, sadness) that might interfere with the warrior’s obligations. The result is a set of socialization norms in which maleness equates with physical strength, willingness to fight, and an emotional stoicism, while society socializes women to represent the contrast of the feminine “other.” To be a warrior (male) is to be strong, assertive, courageous, and willing to engage in violence. To be feminine is to be the opposite—weak, compliant, and unwilling to fight. Finally, drawing on difference feminists such as Gilligan (1982), Goldstein observes, “Men and women think differently about their separateness or connection with other people. . . . Boys construct social relationships in terms of autonomous individuals, interacting according to formal rules, whereas girls construct social relationships as networks of connection.” Further, men compete “to be alone” at the top of a hierarchy, whereas women seek to be at the center of a web (Goldstein 2001, 46). The preference for hierarchy and individualism is reinforced by socialization to the model of the masculine, individualist warrior (2001, 280). A recent study suggests that Goldstein is correct to focus attention on the socialization of men. Wood and Ramirez examined the relationship between belief in gender equality and support among women and men for the use of military force (2017). They found that higher support for gender equality among women has no impact on their support for using military force—they are already less supportive than men. However, the effect of higher support for gender equality is much more pronounced among men. Wood and Ramirez reasoned that the impact on men was greater because “men’s support for the use of force should decline if they cease to adopt these traditionally hypermasculine attitudes when their beliefs toward gender equality become more progressive. This is because support for gender equality typically develops alongside a broader set of worldviews and behaviors that are associated with men adopting nontraditional feminine roles” (2017, 5). Wohlford and Johnston applied precisely this reasoning in their study of public support for international institutions in the US (2000). Controlling

20

GENDER, WAR, AND WORLD ORDER

for partisanship, education, race, and income, they found that gender was among the strongest correlates of support for global institutions. Similarly, Brooks and Valentino articulate a hypothesis that relates the “consensus orientation” of women to gender difference on the question of military intervention: “A large body of psychological research shows that women place greater value on group relationships than do men and that women are more likely to favor cooperation and compromise over aggression as a means for settling disagreement” (Brooks and Valentino 2011, 273, citing Bystydzienski 1993; and Beutel and Marini 1995). Brooks and Valentino do not hypothesize that this makes women more likely to oppose all war. Instead, they are more likely to support it when “consensus-driven organizations like the U.N. have approved the war,” and their survey experiment confirms that this is the case in a sample of US citizens (2011, 273, 278–80). These observations on connection and cooperation are reminiscent of both normative and empirical versions of liberal theories of international relations, in particular, the argument of liberal theorists that international institutions mitigate the pathologies of an international system composed of autonomous sovereign states in a competitive balance of power. For example, in a 1917 peroration against the balance of power, President Woodrow Wilson argued, “There must be not a balance of power but a community of power. . . . When all unite to act in the same sense and with the same purpose, all act in the common interest and are free to live their own lives under a common protection” (cited in Craig and George 1983, 52). Virginia Woolf echoes these same liberal, cosmopolitan themes in Three Guineas, in which she observed, “As a woman, I have no country. As a woman, I want no country. As a woman, my country is the whole world” (as cited in Goldstein 2001, 44). Finally, two additional considerations might underlie greater support among women for international institutions and their support for multilateral decisions and actions. First, multilateral actions collectivize the human and financial cost of war. Thus, if women are sensitive to potential casualties in war, the pooling of effort with others reduces the risk to the lives of a single country’s soldiers. Similarly, to the extent that women are wary of the financial costs of war on pragmatic grounds (it threatens social and other programs of higher value to women), military actions that share the costs should be more acceptable. Second, as I noted above, women are more sensitive to the risk of violence, and multilateral actions usually delay the onset of violence because they require a substantial period of consensus building. For that reason, multilateral actions may appeal disproportionately to women who prefer to use violence only as a last resort. Conover and Sapiro argue, “The point is not that women learn early in life never to engage in conflict nor use violence, but rather that they learn to put off the use of violence until later in the course of a conflict than do men, to escalate its use more slowly, and to be more emotionally upset by it” (Conover and

21

CHAPTER ONE

Sapiro 1993, 1096). The same reasoning might explain the higher level of support among women for international institutions more generally (Wolford and Johnston 2000). Taken together, these ideas provide a theoretically coherent argument both for the existence of gender difference on global issues and for the specification of which issues should evoke the largest difference. If society socializes boys and girls into competing understandings of power, competition, the prospects for institutional cooperation, and the inevitability of war and violence, we would expect to find that gender difference characterizes attitudes in all of these domains. Specifically, we would expect to find gender differences on three sets of issues: the virtue of national power and power balancing; the role of international institutions in managing international conflict; and the acceptance of war and violence as means for resolving conflicts.

National Characteristics and International Contexts Sapiro (2002; 2003) reminds us that contextual factors mediate the extent to which gender difference emerges in political attitudes and behavior. For example, as noted above, gender difference may be unique to the US context because social, cultural, and political issues with gender implications are both salient and polarizing in domestic political debates. Geopolitical context may also mediate gender difference. The US clearly towers over other states in terms of military power, and it has frequently taken the lead in military actions during and after the Cold War. This prominent global role may heighten gender difference. Similarly, evidence shows that societies in intractable conflicts polarized less on security issues by gender or any other social division. This appears to be the case in Israel and Palestine, for example (Tessler and Warriner 1997). Thus, as I analyze gender difference on security issues, I will be attentive to the possibility that national characteristics and contextual factors condition gender difference.

Data and Method Each of the hypotheses described in previous sections point to specific issues and concerns on which gender difference should be large. Unfortunately, there is no single opinion survey to employ in an integrated, comparative evaluation of these hypotheses. The arguments reviewed above cover a large number of global issues, from fundamental beliefs about power and war to specific questions about international institutions, defense spending, war and violence, and the objectives of military actions. However, with the exception of the United States and Western Europe,

22

GENDER, WAR, AND WORLD ORDER

there are no surveys that query respondents on all of these issues. Indeed, surveys that include countries other than the US and Europe typically ask just a few questions about global issues, and the wording of the questions frequently differ. Moreover, some hypotheses about gender difference are inherently historical, but few surveys exist that would allow us to trace the evolution of gender difference over time. This is a particular problem for global opinion surveys, which exist only for the recent past. Finally, some hypotheses require analysis of individuals’ socialization or life experience, and these sorts of question are rarely included in surveys about national security or world affairs. For example, to study hypotheses about the gendered nature of risk perceptions or masculinized soldierly norms requires analysis of surveys that include questions about childhood socialization, but such questions are not included in global opinion surveys or even surveys in individual countries. Lacking a single, comparative survey covering a variety of issues, I take a different approach in this book.4 First, I analyze a number of available surveys to study gender difference on the full range of policy issues related to international security, including the nature of power and power balance, the legitimizing role of international institutions, defense spending, and the use of violent military force. Within each chapter, I draw on multiple surveys to assess the extent to which the data are most consistent with the hypotheses described above. For example, if gender difference arises primarily from differing spending preferences of women and men, then this should be evident by comparing opinions of defense and social spending to opinions on other security issues. Similarly, if the most salient difference between women and men lies in perceptions of risk and vulnerability, then gender difference in opinions should be most evident on questions concerning war, violence, and the use of military force. Moreover, if gender difference results from socialization to a broader, gendered view of world affairs, then gender difference should be most visible for specific elements of this worldview: opinions of the utility of power, balance of power, the acceptability of war, and the legitimacy of international institutions. In short, my assessment of the plausibility of each hypothesis relies on an evaluation of the weight of the evidence across multiple surveys on a number of issues, in many countries, for as much of recent history as possible. This approach provides a much broader examination of evidence than could be achieved using a single survey in one or several countries on a limited set of issues at a single point in time. I collected the data that I employ from several sources. In some cases, I analyze data at the individual level and assess the magnitude and significance of gender difference while controlling for other factors that also influence security opinions (such as ideology, age, or level of education). Among the more valuable sources for analysis at the individual level are the Trans­ atlantic Trends surveys conducted by the German Marshall Fund of the

23

CHAPTER ONE

United States in as many as fourteen countries from 2002 to 2014 and the Pew Global Attitudes Surveys of global public opinion (more detail on each of the surveys employed is provided in the appendix). In addition, I draw on my own original data collections. These include a compendium of gender breakdowns in surveys on defense spending in the US (presented in chapter 3), torture (presented in chapter 4), and American and global opinions on the use of military force in a number of historical episodes since 1980 (presented in chapters 5 through 7). Taken together, these data provide the most comprehensive comparative, historical analysis of gender difference in the existing literature.

Sex and Gender Throughout this book, I employ the term gender rather than sex to avoid any presumption that observed variations in the opinions of women and men are due to innate differences arising from biological sex. As I argued above, many hypotheses concerning difference in the views of women and men are rooted in a broader construction of gender that arises from the norms, roles, and expectations that society imposes on “women” and “men.” Additional hypotheses predict difference as a function of the pragmatic self-interest, socialization, or lived experience of women and men. The findings that I report show clearly that difference in the views of women and men are not a result of biological sex. They are variable and context dependent. Moreover, the number of factors that mediate gender difference is large. They include socially constructed differences in perspective, as well as material factors such as pragmatic self-interest, lived experience, political culture and environment, and strategic situation.5

Men Have Gender Too The scholarly literature on gender and politics often employs a term that I avoid: the gender “gap.” I avoid the term because it has two subtle but pernicious effects. First, the term normalizes the views of men as the “correct” policy and narrows scholarly attention to a search for the reasons for the “divergent” views of women. Just as the terms missile gap or achievement gap suggest that one nation or group is ahead and one behind, the term gender gap often leads to a discussion of why the views of women are different. Especially in the early literature on gender difference in opinions of war and peace issues, this often led to ad hoc speculation about the reasons for this supposed divergence—especially essentialist speculation that women are more “caring and nurturing.”

24

GENDER, WAR, AND WORLD ORDER

Most important, the second effect of defining gender difference in terms of the divergent views of women was to focus theory and research on women to the exclusion of theoretical thinking about men’s policy views. An obvious point: the quantity calculated as a gender “gap” or “difference” is composed of two elements—the views of men and those of women. Nonetheless, until recently, there has been less focus on understanding one part of this identity: why men express specific preferences. The theoretical turning point came in the work of Goldstein (2001), who concluded his synthesis of a massive amount of scholarly literature with the argument that gendered war roles are rooted above all in the socialization of boys and men to the martial virtues of the soldier. Similarly, Hollander (2001) concluded from her focus group research that men are socialized to a belief that they are invulnerable to risk and violence than others. Kaufmann and Petrocik (1999) concluded in their analysis of voting in US elections that changing partisan alignments were due largely to a migration of men away from the Democratic Party. Finally, Wood and Ramirez (2017) found that gender equal attitudes have a larger impact on men than on women because gender equal attitudes erode the martial, masculine values taught to boys as part of the socialization process. All of these examples reinforce the fact that a singular focus on women is misplaced.

Statistical Significance and Political Significance The question of statistical significance versus substantive significance in survey analysis deserves elaboration. Because the survey samples for many countries and groups of countries numbers several thousand respondents, even small percentage gender differences may be statistically significant. However, statistical significance alone does not translate into political significance. For example, I show in chapter 2 that opinions toward international involvement—essentially a question on isolationism—show gender differences of about 5 percentage points in the United States, Western Europe, and Eastern Europe (and about the same within each of several countries within these groups). This difference is large enough to qualify as statistically significant given the large sample size. However, these differences occur around an average above 70 percent of both genders who favor international involvement. This is a robust consensus that characterizes opinions across all categories of ideology, education, and political engagement. Given the near-unanimous consensus in favor of international involvements, small gender differences are unlikely to have much political impact. Thus, as I describe gender differences on this and other questions in the chapters to follow, I take care to assess the likely political significance of any gender divide that emerges.

25

CHAPTER ONE

When opinions divide more closely, the level of statistical significance is more relevant. Following Page and Shapiro (1991), I treat a gender difference of at least 6 to 8 percentage points as both statistically and substantively meaningful when the sample size is sufficiently large to warrant this inference.6 I employ the terms significant and statistically significant interchangeably when describing these differences. In addition, when possible I also report the significance of gender while controlling for other variables, such as ideology or education. However, as I noted above, I am less interested in the question of whether every gender difference on every question is statistically significant. Rather, I seek to discover if the weight of the evidence over many questions indicates a consistent pattern of gender difference on specific security issues or in specific countries or groups of countries. One of my goals in this book is to make it accessible to the widest possible audience. For that reason, I keep statistical analysis and jargon to a minimum. My goal is to describe as comprehensively as possible the extent to which the views of women and men differ in language that is as clear as possible. Of course, that does not mean that I can ignore the fact that gender is not the only variable that influences attitudes toward national security. Indeed, the literature on public opinion and security policy makes it clear that partisanship and other variables are important correlates of attitudes (Holsti 2004; Everts and Isernia 2015; Eichenberg and Stoll 2017). Partisanship is a particularly strong organizing device. Citizens form their views on security issues in part based on their own partisanship and in part by following the cues of party leaders (Berinsky 2009). Other variables, such as basic attitudes toward war and level of education, may also affect opinions on some specific security issues (Bartels 1994; Wittkopf 1990). In the chapters to follow, I present some analyses at the individual level with controls for these and other variables. In some cases, I present these regression results in their entirety, while in others I report them in a footnote. My hope is that the presentation strikes a balance between statistical precision and simplicity of expression.

26

chapter 2

Threats, Power, War, and Institutions

Scholars often find gender difference in public opinion on issues of war and military force, but an important unanswered question is whether the difference is part of a broader divergence in the attitudes of women and men, perhaps reflecting a more general contrast in worldviews. In addition, most of the available evidence on gender difference is limited to studies of the American public, which may be unique in its gender polarization, views of national security, or both. I find little evidence to support the hypothesis of differing worldviews. International involvement is uncontroversial in most states and evokes no gender difference. Concerning general attitudes toward multilateral organizations, it is only in the US that there is a substantial gender difference (in support for the UN). When the issue turns to the role of the UN, NATO, or the European Union in legitimizing or collectivizing military actions, women are less likely to express support than men are—the opposite of what one would expect. This pattern suggests that the specific issue of using military force is the principal source of division. Multilateral endorsement or participation is secondary. Finally, gender difference is not universal across issues or nations. Cross-nationally, gender difference is most consistently prominent in the US across a number of issues, which should alert us to the possibility that gender politics in the US have some unique qualities. Second, the fact that gender difference varies cross-nationally and across issues casts doubt on the essentialist hypothesis.

International Involvement and Threats Some previous research suggests that women are less likely to support international involvements than are men, a difference we might attribute to the greater risk aversion of women (Conover and Sapiro 1993). However, this study was limited to attitudes in the United States during the crisis that preceded the Persian Gulf War of 1991, a time when perceptions of risk were understandably high. Figure 2.1 reports gender difference in answers 27

80

CHAPTER TWO

69

76

66

40

60

66

81

78

0

20

Percent "Active Part"

70

74

Eastern Europe

USA Men

Turkey

Western Europe

Women

Figure 2.1.  Gender difference in attitudes toward global involvement, 2002–5: “Do you think it will be best for the future of [country] if we take an active part in world affairs or if we stay out of world affairs?” Percent “active part.” Note: Number of responses: USA (3,242), Western Europe (20,382), Eastern Europe (5,047), and Turkey (2,309). Western Europe includes France, Germany, Italy, Netherlands, Portugal, Spain, and the United Kingdom. Eastern Europe includes Bulgaria, Poland, Romania, and Slovakia. Source: German Marshall Fund, Transatlantic Trends, 2002–5. The appendix contains a complete list of Transatlantic Trends surveys with archival links.

to a very similar question posed in eleven countries in the German Marshall Fund’s Transatlantic Trends from 2002 through 2005. The question asked: “Do you think it will be best for the future of [country] if we take an active part in world affairs or if we stay out of world affairs?” Although the potential risks of global involvement were also high during this period—it included the years after September 11, 2001, and the invasions of Afghanistan and Iraq—figure 2.1 shows a very high cross-national consensus among both men and women in favor of international involvement. Well over 60 percent of women and 70 percent of men in all countries favor involvement, and there is virtually no variation in this consensus by ideological orientation, by generational cohort, or in any specific country. Among those with higher education within both genders, there is even greater support for international involvement (more than 75 percent), and this consensus is very stable over time.1 On the question of international involvement, then, there is no controversy anywhere, let alone a gendered one. Of course, it may be that the

28

GENDER, WAR, AND WORLD ORDER

question is so anodyne as to be of limited utility. In the aftermath of September 11, 2001, international involvement of one kind or another may have appeared unavoidable. On the other hand, that seems precisely the point. The involvements debated and undertaken from 2001 to 2005 included the invasions of Afghanistan and Iraq, and most of the countries analyzed here deployed troops to one or both of these theaters. In spite of this context, the data show that the question of involvement in global affairs did not evoke controversy. Moreover, the data cast doubt on one specific hypothesis linking gender to global affairs. The fact that global involvement does not evoke gender difference suggests that it does not arise from a generalized aversion to international risks. A repeated finding of social science research is that women perceive their environment as more threatening than men. We would expect this tendency to be particularly evident during the last twenty years, a period characterized by major terrorist attacks in many parts of the world, two major wars, and many smaller conflicts and insurgencies. Testing this proposition, however, is not as easy as one would like. In the case of the Transatlantic Trends surveys, there are a variety of questions on threats over many years, but the wording of these questions suggests to respondents that one phenomenon or another (terrorism, climate change) is indeed a threat—it is simply a question of whether it is a “very important” or “somewhat important” threat. On most issues of international security (terrorism, Iranian nuclear program), the responses yield very high percentages indicating that the problem is indeed perceived as a very or somewhat important threat. On most of these questions, gender differences are very small. However, there are some differences in the intensity with which respondents perceive threats (the percentage who find a threat “very important” rather than “somewhat important”). A selection of these threats is shown in table 2.1, where responses are arranged not by the level of perceived threat (terrorism and Iranian nuclear weapons are the highest) but by the size of gender difference. The table makes clear that “violence and instability in Iraq” evokes the largest gender difference, but only in the United States and Western Europe. Gender difference is also fairly large for the perceived threat from the “spread of disease,” especially in the US and Western Europe. Significantly, there is not a general pattern in which women feel more threatened by all international problems. Indeed, women feel less threatened by Islamic fundamentalism. In addition, gender difference in threat perception varies considerably across regions as well as issues. In the United States and Western Europe, there are some gender differences, and these are generally larger than in Eastern Europe and Turkey (where women often feel less threatened than men are).2 Why are gender differences largest on the issue of the threat from “violence and instability in Iraq” and secondarily from the threat of the “spread of disease/avian flu?” One hypothesis is that the wording of these two

29

Table 2.1.  Percent rating selected issues as a “very important” threat, 2006 Men (%)

Women (%)

Gender difference (women–men) (%)

N

Violence/instability in Iraq United States Western Europe Eastern Europe Turkey

54 50 38 38

75 65 45 37

21 15 7 –1

550 3,556 1,891 437

Spread of disease/ avian flu United States Western Europe Eastern Europe Turkey

47 42 38 48

57 51 44 42

10 9 6 –6

545 3,583 1,963 451

Iran acquiring nuclear weapons United States Western Europe Eastern Europe Turkey

80 69 55 29

83 74 62 32

3 5 7 3

549 3,573 1,909 434

International terrorism United States Western Europe Eastern Europe Turkey

79 76 68 61

84 81 69 54

5 5 1 –7

546 3,594 1,989 459

Global economic downturn United States Western Europe Eastern Europe Turkey

51 55 41 55

60 59 41 45

9 4 0 –10

541 3,554 1,888 447

Islamic fundamentalism United States Western Europe Eastern Europe Turkey

63 64 49 27

59 68 48 23

–4 4 –1 –4

534 3,485 1,757 412

Note: See figure 2.1 for countries included in Western and Eastern Europe. Source: German Marshall Fund, Transatlantic Trends, 2006.

9781501738142_Text.indb 30

4/9/2019 7:25:02 PM

GENDER, WAR, AND WORLD ORDER

threats most directly evokes physical harm to individuals. “Violence” in Iraq or “disease” seem much more personally immediate than the threat from an “economic downturn” or even “terrorism” in the abstract. The plausibility of this hypothesis reinforces research findings reviewed earlier that suggest that gender differences are highest on issues that directly imply violence. In any case, taken together with the results on international involvement discussed in the previous section, I conclude that it is not so much involvement in world affairs that evokes gender difference, but rather the threat of violence and harm.

Military Power and Balance Of Power I expect gender difference in assessments of the utility of military power to be high. I reviewed three hypotheses in chapter 1 that suggest why this should be the case. First, gender difference in existing research is generally high on issues of force and violence, and military power is the instrument of violence. Second, just as liberal theorists criticize balance of power as the path to an insoluble security dilemma or the cause of conflict itself, feminists criticize hierarchies of power and dominance. Liberal and feminist theorists also criticize military power and balance of power for the same reason: peace or stability based on national military power impedes the creation of community. Finally, military power is expensive. To the extent that gender difference on security issues is due to pragmatic considerations, we would expect military power to raise the specter of a guns/butter trade-off and thus increase gender difference. The question displayed in table 2.2 bluntly evokes the question of the utility of military power. The question asks respondents to agree or disagree with the statement that “the best way to ensure peace is through military strength.” Several things in table 2.2 stand out. First, there is a chasm in the responses between the US and Turkey on the one hand, and Western and Eastern Europe on the other. In the former, a majority of the population agrees that military strength ensures peace, but in the latter, this view is a distinct minority. Second, these views correlate strongly with ideological self-placement among both men and women—the right supports military strength more than the left. Nonetheless, given the overwhelming skepticism of Europeans toward military strength, this polarization may not be politically meaningful, as it is a view shared on all sides of the political spectrum. Finally, it is only in the United States that there is a gender difference in the population as a whole, and it is significant both statistically and politically. In the US, a majority of men agrees that military strength ensures peace, but a majority of women disagrees.3 Further, as the bottom of table 2.2 shows, including party identification complicates the gender difference but does not remove it. Military strength is most highly valued by

31

CHAPTER TWO

Table 2.2.  Percent who agree that “the best way to ensure peace is through military strength,” by ideology and gender, 2004–5 Ideology

Left

Gender Men Women (%) (%)

Center

Right

Total

Men Women (%) (%)

Men Women (%) (%)

Men Women (%) (%)

N

United States

34

25

60

43

72

61

55

44

2172

Western Europe

20

18

29

23

41

33

30

25

14221

Eastern Europe

22

19

26

15

32

27

27

21

4049

Turkey

44

49

51

57

64

59

57

54

2,309

Party identification

United States

Democrat

Independent

Republican

Total

Men Women

Men Women

Men Women

Men Women

40

32

50

43

77

64

55

44

N 2,174

Note: See figure 2.1 for countries included in Western and Eastern Europe. Ideological selfplacement is a seven-point scale (1 [extreme left] through 7 [extreme right]), here collapsed into a three-point scale (1–3=Left 4=Center 5–7= Right). Party identification in the United States does not include “leaned” independents. Source: German Marshall Fund, Transatlantic Trends, 2004–5.

both men and women in the Republican Party, but there is a gender division within the latter as well. The overall result in the US is a society divided in two ways on the issue of military power: both by party and gender. A second evaluation of power occurs in a question about “superpowers.” Over the last twenty years, there has been a great deal of discussion in Europe about the US becoming a “hyper power” and the need to balance US power by enhancing European military capabilities (Posen 2004; 2006; Peters 2014). Transatlantic Trends pursued this question during 2002–5 by asking the following: “In thinking about international affairs, which statement comes closer to your position about the United States and the European Union: • The US should remain the only superpower • The European Union should become a superpower, like the United States • No country should be a superpower [volunteered]” Prior research would lead us to expect one of two gender patterns in the response to this question. First, because of the hypothesized relative antipathy of women to power and hierarchy, we would expect women to favor the “no country should be a superpower” response. Secondarily, we might also expect women to favor the “Europe as a superpower” response, for it

32

GENDER, WAR, AND WORLD ORDER

at least suggests that Europe should balance the dominant power. Indeed, the reference to the European Union might be attractive to women because it refers to a supranational entity, which suggests community rather than competition. The responses to this question indicate that neither pattern is present.4 In fact, gender differences are close to nonexistent in all countries but the US. American opinions on this issue reveal division, but gender plays a relatively minor role. It is true that American women are slightly less favorable toward the idea of the US as a single superpower. Nonetheless, polarization by party and ideology dwarfs the gender difference. For example, on the left of the political spectrum in the US, only 26 percent of respondents believe in “unipolarity.” On the political right, in contrast, it is 60 percent. These ideological divisions do not exist in other countries. In contrast, in Europe there is a strong consensus in favor of the European Union becoming a superpower, while in Turkey there is a division between this point of view and the view that no country should be a superpower. What is important for my purpose here is the finding that there is virtually no gender difference of consequence on this issue in Europe or Turkey. Thus, although some hypotheses on gender difference would point to opinions of power as a likely fulcrum of gender cleavage, I find evidence for the hypothesis only in the US, and even there ideology overshadows gender difference. In summary, to the extent that questions of power and balance of power are central to the differing views of liberals and realists, there is no evidence that these issues underlie gender difference.

The Acceptability of War Much of the evidence for gender difference occurs in studies of attitudes toward war and other violent conflicts, but almost all of this evidence comes from the US. As a result, it is difficult to tell if the gender difference is a uniquely American phenomenon, reflecting the US global role and perhaps the uniquely polarized nature of gender politics in the US. It is therefore useful to compare the views of men and women in as many countries as possible. The Halifax International Security Forum and the German Marshall Fund recently fielded a question that attempts to measure support for the proposition that war is sometimes necessary as an instrument of policy. The question asks: “Please tell me whether you agree or disagree with the following—Under some conditions war is necessary to obtain justice.” The question is not without some weaknesses. The mention of “justice” is of particular concern, especially in the environment after September 11, 2001, when respondents might interpret the question as specifically inspired by the attacks on the US. Yet research by Hurwitz and Peffley (in

33

CHAPTER TWO

the US) has shown that a basic measure of the “morality of war” is one of the “core values” that has a strong influence on a number of other security attitudes (1987). Moreover, this “war is necessary” question was not formulated in reaction to recent events. Its origins date to the 1930s as part of the research program of L. L. Thurstone, a social psychologist who was a pioneer in the development of attitude scales on a variety of topics (1929; 1931; 1959). During the 1930s, Thurstone and his students invested considerable effort to develop a “pacifism” scale, and the “war is sometimes necessary” question was one of twenty-three items from which the scale was constructed (Droba 1931; Peterson 1931). In fact, the item is still in use by psychologists who study attitudes toward war (Kuterovac 2000; Jones-Wiley, Restori, and Lee 2007), and it resembles similar attitudinal questions on the use of military force that are strongly correlated with attitudes toward national security in other studies (Hurwitz and Peffley 1987; Bartels 1994). In short, the question is attractive because it seeks to measure a basic attitude toward war that is independent of time and circumstance. The question has other virtues as well. The first is the blunt invocation of “war,” a welcome contrast to survey questions that often employ abstractions, such as “military action.” Second, the measure from the two organizations mentioned above is available in identical form in twenty-nine countries, and the GMF question is available for more than ten years in as many as sixteen countries. As a result, comparative analysis can now extend prior work on the US. Third, the question subtly invokes the ambivalence that most citizens have about policy choice (Zaller and Feldman 1992). War may be necessary, but only “under some conditions.” The research question is whether women and men resolve this ambivalence in different ways. A fourth virtue of the question is that it has proven to be a very robust predictor of opinions on other security issues, which increases confidence that it measures fundamental attitudes toward military force (Everts and Isernia 2015). In summary, whatever doubts one might have about the wording of the question, it seems to measure a fundamental toleration or rejection of war as an instrument of policy. Figure 2.2 summarizes the size of the gender difference on the “war is necessary” question. There are several striking features of the results. First, the gender differences are large. The average difference is 13 percentage points across all the countries shown, and the difference exceeds the threshold of statistical significance in twenty-one of the twenty-nine countries shown in the graphic. This is far larger than any gender difference reviewed in earlier sections of this chapter. It is also the only question producing consistent gender difference across a large number of countries. Clearly, gender difference on the question of whether war is a necessary instrument of policy is not unique to citizens of the US superpower. The issue of war divides the genders in most countries.

34

GENDER, WAR, AND WORLD ORDER

-23

Belgium Sweden Australia Canada Germany Mexico Hungary France Argentina Brazil Netherlands South Africa Japan South Korea Portugal Spain Italy Slovakia India Poland United Kingdom Bulgaria Russia USA Indonesia Saudi Arabia China Romania Turkey -25

-22

-21

-19 -19 -19

-20

-18

-17

-16 -16 -16 -16

-15

-14

-15

-13 -13

-12 -12

-11 -11 -11

-7

-10

-6 -6

-5 -5

-4 -4 -4 -5

0

Gender difference (women - men)

Figure 2.2.  Gender difference in views of war, June–October 2013: “Please tell me whether you agree or disagree with the following—Under some conditions war is necessary to obtain justice.” (Gender difference in percent agree.) Sources: German Marshall Fund, Transatlantic Trends, 2013; and Halifax International Security Forum, Global World Affairs Survey, 2013 (survey by IPSOS).

The Transatlantic Trends data for this question are available for the period 2003 through 2013 (not shown). They demonstrate that gender difference on the necessity of war question is large and stable over time in Western and Eastern Europe. In Western Europe, the difference averages 14 percentage points and is never less than 12 percentage points. In Eastern Europe, it averages 10 percentage points and is never less than 7 percentage points. There is more variation in the US and Turkish figures, but there are occasionally very large differences and the overall average is a substantial gender difference. In summary, on the fundamental question of whether war is “sometimes necessary,” a significant gender difference characterizes most countries for which data are available. Evidence from Europe, Turkey, and the US also show that the difference characterizes most years. Nonetheless, a notable finding in the previous graphic is that there is some cross-national variation in the magnitude of gender difference, from a low of 4 percentage points in Romania, Turkey, and China to more than 20 percentage points in Belgium, Sweden, and Australia (figure 2.2). The

35

CHAPTER TWO

gender difference also varies somewhat over time in the US. This variation casts doubt on the essentialist, biological hypothesis. As I observed in chapter 1, evidence in support of the essentialist hypothesis would consist of large and unvarying gender difference across different political, cultural, and strategic contexts. Moreover, if we focus on the level of acceptance of war rather than on gender difference, as shown in table 2.3, we see that majorities of men state that war is necessary in only nine of the twenty-nine countries, and in five countries, a majority of women finds that war is sometimes necessary. Clearly, the views of both men and women on the necessity of war varies considerably, suggesting that factors other than

Table 2.3.  Responses to the statement: “Under some conditions war is necessary to obtain justice,” June–October 2013 Men

China USA United Kingdom India Saudi Arabia Australia Canada Netherlands Indonesia Sweden South Africa Mexico Poland Belgium South Korea France Turkey Brazil Russia Germany Hungary Bulgaria Portugal Japan Romania Slovakia Argentina Italy Spain

Women

Agree (%)

Disagree (%)

Agree (%)

Disagree (%)

Gender difference (%)

82 76 68 66 63 63 58 51 50 49 49 47 46 43 42 40 40 38 37 37 36 35 31 31 30 29 28 27 23

18 24 32 34 37 37 42 49 50 51 51 53 54 57 58 60 60 62 63 63 64 66 69 69 70 71 72 73 77

78 70 57 55 58 42 39 35 45 27 33 28 35 20 28 23 36 22 31 18 18 27 18 16 26 17 12 15 10

22 30 43 45 42 58 61 65 55 73 67 72 65 80 72 77 64 78 69 82 82 73 82 84 74 83 88 85 90

–4 –5 –10 –12 –5 –21 –18 –16 –5 –21 –16 –19 –12 –24 –14 –17 –4 –16 –6 –19 –17 –8 –13 –15 –4 –13 –16 –11 –13

Sources: German Marshall Fund, Transatlantic Trends, 2013; and Halifax International Security Forum, Global World Affairs Survey, 2013 (survey by IPSOS).

36

9781501738142_Text.indb 36

4/9/2019 7:25:03 PM

GENDER, WAR, AND WORLD ORDER

25

gender per se underlie the gender difference that results. Put differently, while there is certainly a tendency for women to express less agreement with the argument that war is necessary, the magnitude of this difference is affected by factors other than—or in addition to—gender alone. I explored what these factors might be in a regression analysis with the magnitude of gender difference as the dependent variable. Figure 2.3 displays the variable that emerged as the most strongly correlated with gender difference: an index of the political empowerment of women that is part of the annual World Economic Forum’s annual “gender gap” report (World Economic Forum 2014). This measure represents the ratio of women to men in three categories: parliamentary officeholders, minister-level positions, and total years in the office of chief executive (president or prime minister). Thus, the scatter gram in figure 2.3 shows that gender differences on war

Belgium Sweden

Canada

Mexico

Germany

Hungary France

15

Brazil

Argentina

Japan South Korea

Portugal Italy Poland

Netherlands

South Africa

Spain United Kingdom

India

10

Slovakia

Bulgaria

5

Gender difference (absolute value)

20

Australia

Russia USA Saudi Arabia Indonesia Romania China

0

Turkey

.1

.2 .3 Index of political empowerment of women

.4

.5

Figure 2.3.  The relationship between the political empowerment of women and the size of the gender difference on “war is necessary.” Note: The horizontal axis displays an index of the political empowerment of women from the annual World Economic Forum “gender gap” report. The measure combines the ratio of women to men in three categories: parliamentary officeholders, minister-level positions, and total years in the office of chief executive (president or prime minister). The vertical axis displays the gender difference in agreement with the statement that “war is sometimes necessary to obtain justice.” Sources: German Marshall Fund, Transatlantic Trends, 2013; and Halifax International Security Forum, Global World Affairs Survey, 2013 (survey by IPSOS). Index of political empowerment is from World Economic Forum (2014).

37

CHAPTER TWO

correlate most strongly with a variable that measures the extent of actual progress in achieving political gender equality. A second variable—the level of women’s educational attainment relative to men, is also strongly correlated with gender difference on war.5 What the regression analysis therefore shows is that the differentiation of women’s views of war from those of men increases as the cognitive skills of women and the political representation of women increase. There are two possible interpretations of this correlation. The first is that the policy preferences of women change as society changes, as Inglehart and Norris argue in their theory of economic development and attitudes toward gender equality (2003). As noted in chapter 1, Inglehart and Norris argue that the increasing educational attainments of women and their increasing participation in the labor force contribute to an increase in the “psychological autonomy” of women and therefore to an increasing differentiation of their policy preferences from those of men. The second possibility is that the policy views of women had long been different from men, but the expression of those views awaited the political opportunity to do so—or perhaps awaited the presence of female candidates and officeholders whose example and views serve to mobilize those of women in general (Burns, Schlozman, and Verba 2001). Whatever the case, the size of the gender difference in attitudes toward war adds an important complication to the domestic politics of national security in many countries. Although I do not show the results here for reasons of space, further analysis of the “war justice” question in the Trans­ atlantic Trends data show that gender complicates the traditional ideological polarization that scholars consider the most important domestic cleavage on issues of war and peace. The most significant division is in Western Europe, where there is a center-right majority toleration for war among men, but there is not a similar toleration among women on the center and right. Indeed, the gender divisions on the center and right of the spectrum in Western Europe are deep—women on the right have opinions that more closely resemble those of both men and women on the left of the ideological spectrum than men on the center and right. The significance of gender for opinions of “war and justice” is further confirmed in an analysis with controls for variables that figure prominently in theories explaining attitudes toward war (including support for US global leadership, support for the NATO alliance, left-right ideology, level of education, age, level of political engagement, external threats, and controls for year and country of sampling). Gender differences remain significant in these analyses in almost every country (Eichenberg and Stoll 2015). In fact, in some countries (Germany), gender is by far the strongest correlate of attitudes toward war. In summary, not only is there evidence that there are gender differences on fundamental attitudes toward war, there is also evidence that these differences remain an important influence on opinions on security policy that

38

GENDER, WAR, AND WORLD ORDER

equal, and sometimes surpass, the importance of the ideology that has traditionally been the main focus of scholarship in the field.

The Importance of Attitudes toward War There is additional importance to the finding that fundamental attitudes toward war polarize significantly by gender: it implies that gender ultimately influences opinions on many other security policy issues, because opinions of war and the use of military force are a significant correlate of these security policy opinions. Indeed, in past research, attitudes toward war and military force are the most consistent predictor of opinions on a host of security issues. Hurwitz and Peffley, who distinguish between three types of attitudes, conducted the most fundamental analysis of this type: core values (morality of war, ethnocentrism), international postures (internationalism, militarism), and opinions on specific issues (1987). In their view, the core value “morality of war” is causally prior to citizens’ preferred international posture, which in turn helps citizens form attitudes on specific policy choices, such as defense spending. In fact, opinions of defense spending correlate strongly with a militarist posture, which in turn correlates strongly with views of the morality of war. Thus, Hurwitz and Peffley essentially argue that fundamental attitudes toward war and military force are a primary determinant of defense spending preferences as well as preferences on other security policy issues (1987, 197). Similarly, Bartels analyzed the impact of ideology and a variety of foreign policy attitudes on Americans’ support for defense spending in 1992. The results were unequivocal: “The magnitudes of the various parameter estimates clearly suggest that the dominant factor in producing support for defense spending in 1992 was a general willingness ‘to use military force to solve international problems’ . . . it is the dominant determinant of defense spending preferences in every specification, regardless of which other variables are included” (Bartels 1994, 481). Richard Stoll and I analyzed the impact of the “war is sometimes necessary” question on support for increased defense spending in thirteen countries covered by Transatlan­ tic Trends. We concluded that beliefs about the necessity of war were the most important correlate of support for defense spending (2015). Finally, Everts and Isernia analyzed the impact of American and European attitudes on the necessity of war on support for using force in Iraq and Afghanistan and found a very strong relationship (2015, 122–34; see also Reifler et al. 2014). In summary, basic attitudes toward war and military force act as filters, aiding citizens in arriving at opinions on specific issues, such as military spending, the importance of alliance commitments, or committing military forces in specific circumstances. Because gender correlates very strongly

39

CHAPTER TWO

with attitudes toward war, it therefore has an indirect impact on opinions of these issues. In other cases, gender may have a direct impact as well.

Multilateral Institutions and Legitimacy Citizens’ opinions about the value and legitimizing role of multilateral institutions offer a particularly useful opportunity to discern the difference between men and women concerning military action. Previous research offers some evidence that women offer higher support for international institutions and are more likely to support military actions if international or multilateral institutions endorse them. In addition, the logic of burden sharing might suggest that women would also be more likely to support military actions carried out by or with the sanction of the NATO alliance or European Union. In this section, I evaluate these hypotheses. g e nera l assessmen ts o f th e u n i ted nat i o ns The United Nations has been the focus of a substantial number of global polls. Beginning in 2002, the Pew Global Attitudes Survey asked the following question in as many as thirty-nine countries: “Please tell me if you have a very favorable, somewhat favorable, somewhat unfavorable or very unfavorable opinion of . . . the United Nations (Pew Global Attitudes 2013). From 2003 through 2006, Transatlantic Trends asked, “Would you say your overall opinion of the United Nations (UN) is very favorable, mostly favorable, mostly unfavorable, or very unfavorable?” Although not terribly specific, the question does allow a rare comparative evaluation of the hypothesis that women express higher support for international institutions. A list of the gender difference on these questions for all countries produces a lengthy table, but they can be easily summarized: 64 percent of men and 65 percent of women express a favorable view of the UN, so the average gender difference is very small indeed. Further, there are fifteen countries in which men are slightly more favorable toward the UN than women are. Gender difference on the UN question is also very small in the Transatlantic Trends surveys. The notable exception is the US, where women are more favorable to the UN by 9 to 17 percentage points—a significant difference. In fact, although general population opinions of the UN in the US are favorable, they are polarized both by partisanship and by gender. For example, about 80 percent of Democrats (men and women) are favorable to the UN, but among Republicans, there is a significant gender split. Among Republican women, 53 percent express favorable views of the UN, compared to only 31 percent of Republican men. In summary, while opinions from around the globe describe largely favorable attitudes toward the UN and show no evidence of gender difference, in the US opinions divide by both party and gender. Nonetheless,

40

GENDER, WAR, AND WORLD ORDER

there is little comparative evidence for the hypothesis that women show significantly higher support for international institutions. General favorability toward the UN is one thing, but perhaps the more important question is whether favorable attitudes translate into a willingness to accept the UN’s injunctions or whether it is justifiable to ignore the UN altogether. In 2003–5, Transatlantic Trends asked precisely this question: “Please tell me whether you agree or disagree with each of the following: When vital interests of our country are involved, it is justified to bypass the UN.” From the literature reviewed above, a plausible hypothesis is that women are more likely to reject this statement because it contradicts the presumed consensus-building function of the United Nations, and it rejects the cost and risk sharing that UN actions or mandates can offer. However convincing the hypothesis, table 2.4 shows little evidence to support it. The table shows that overall agreement with this sentiment is

Table 2.4.  Percent who agree that “when vital interests of our country are involved, it is justified to bypass the UN,” 2003–5

Men (%)

Women (%)

Total (%)

Gender difference (women–men) (%)

USA Western Europe Eastern Europe Turkey

67 47 58 78

56 49 55 76

61 48 57 77

–11 2 –3 –2

3,085 19,200 4,233 1,814

USA United Kingdom Spain Poland Turkey Slovakia Portugal Italy Netherlands France Germany

67 58 49 54 78 64 46 40 52 44 39

56 50 44 50 76 63 49 44 58 51 49

61 54 44 52 77 63 47 42 55 46 44

–11 –8 –5 –4 –2 –1 3 4 6 7 10

3,085 2,857 1,993 2,451 2,026 1,781 2,716 2,929 2,846 2,943 2,917

Ideology Gender United Kingdom USA Germany Netherlands

Left

Center

N

Right

Men

Women

Men

Women

Men

Women

N

47 50 29 38

39 37 41 50

60 69 38 53

53 53 47 56

63 85 50 57

58 68 51 58

2,565 2,657 2,800 2,845

Note: See table 2.2 for definition of ideology and figure 2.1 for list of countries included in Western and Eastern Europe. Source: German Marshall Fund, Transatlantic Trends, 2003–5.

41

9781501738142_Text.indb 41

4/9/2019 7:25:03 PM

CHAPTER TWO

surprisingly strong in Western and Eastern Europe, which had shown very high UN favorability ratings in the surveys discussed in the previous section. Nonetheless, more than a majority of Eastern Europeans and a sizable percentage of Western Europeans are prepared to bypass the UN. Citizens in the US and Turkey are less favorable to the UN to begin with, so they are understandably more prepared to bypass the institution. There are significant gender differences in only five countries: the US, United Kingdom, France, Germany, and the Netherlands, but they are not all in the same direction. In the US and UK, women conform to the hypothesis: they are less willing to bypass the UN, especially those on the left of the political spectrum (shown at the bottom of table 2.4). In France, Germany, and the Netherlands, by contrast, women are more likely to favor bypassing the UN, a pattern also most pronounced on the political left. Why should women in these three countries display attitudes that are opposite to those of women in the other two? The answer likely lies in the context in which citizens appraised possible UN action in the debate that led to the Iraq War in 2003. In the US and the UK, governments were pushing hard for a UN resolution that would authorize a coalition to use force in Iraq, but the US government in particular made clear that it would go to war in any case (thus bypassing the UN). This was an action that women on the left in the US and UK rejected. In France and Germany, by contrast, governments had made clear in 2002 that they would not participate in a war against Iraq under any circumstances, including the eventuality of a UN resolution. In this case, women on the left in France, Germany, and the Netherlands also gravitated toward the “antiwar” position by endorsing the view that their country should bypass the UN if it passed a resolution endorsing intervention in Iraq. In short, in these five countries, one can interpret the attitudes of women (in particular) as more opposed to the use of force relative to men, but this opposition found expression in different ways depending on the political context in each country. The important point is that the gender difference arises from opposition to the use of force. Attitudes toward the UN appear secondary. A similar result occurred in responses to a Transatlantic Trends question in the 2005. The question asked respondents to agree or disagree with the following statement: “The use of military force is more legitimate when the United Nations (UN) approves it.” The hypothesis would be that women are more likely to agree with this statement, but with the exception of the United States, the opposite is actually true: men are slightly more likely to agree. In Germany and Italy, the difference is very large; about 70 percent of men agree with the statement, but among women, the figure is 55 percent.6 As was true with the question on bypassing the United Nations, in these two countries approval by the United Nations is not enough to overcome skepticism among women about the use of military force. War—not international institutions—is the issue.

42

GENDER, WAR, AND WORLD ORDER

t h e u ni ted nati o n s, th e nato a lli a nc e , and legiti m ac y A similar pattern is revealed in a more concrete set of questions in 2003 that probed respondents’ sensitivity to the participation of the UN or the NATO alliance in a hypothetical attack to “eliminate weapons of mass destruction” in North Korea or Iran. Formulated in the shadow of events leading up to the war in Iraq, three variants of the question were offered to different portions of the sample: one variant described an attack by the United States alone, a second described an attack by the NATO alliance, and a third described an attack by the United Nations. Following the hypothesis that I have been pursuing here, the expectation is that mention of the multilateral options would increase the support of women and reduce gender difference. The responses to this question demonstrate the opposite.7 It is true that support for military action is higher among both men and women in the multilateral versions of the question (an action led by NATO or the UN). However, men are far more responsive to the multilateral cue than women are. This is particularly clear in the case of Western and Eastern Europeans. Less than a majority of men and women there favor an attack carried out by the United States alone, but when the question mentions NATO or the UN as the agent of the attack, support increases to a majority of men in both cases. Among women, however, the increase is much less substantial. As a result, there is a majority in favor of a military action carried out by NATO or the UN among men in Western and Eastern Europe, but among women there is never a majority. Although women are somewhat influenced by the multilateral cue, they remain less supportive of a military attack. These findings occur despite the fact that women exhibit higher support for the NATO alliance than men. For the period 2002–14, 60 to 70 percent of both genders in all countries surveyed in Transatlantic Trends considered NATO “essential to our country’s security.” The figures for women average 4 percentage points higher, and there are no countries in which women show lower support for NATO. In two countries (France and the US), women are more supportive of the NATO alliance by a significant margin. I noted above that multilateral military actions have the advantage of sharing human risk and financial cost, perhaps one reason that women might display higher support for military actions that have multilateral endorsement or participation. The questions on military action against Iran and North Korea reviewed immediately above cast some doubt on this thesis, but those questions do not explicitly mention burden-sharing considerations. Table 2.5 summarizes two questions designed specifically to test the proposition that all members of the NATO alliance should share in the provision of the troops or the financial costs of military action. The first question simply elicits agreement or disagreement with the proposition that “all

43

CHAPTER TWO

NATO member countries should contribute troops if the NATO alliance decides to take military action.” The second question queries agreement with the statement that “all NATO member countries should share in the financial costs of a NATO military action even when they do not contribute troops.” Table 2.5 shows that the responses to these questions are generally positive about the norm of burden sharing. However, men are more positive than women are. In fact, in Germany, Italy, Portugal, and Spain, robust majorities of men agree with the statements, but women divide closely. Thus, although it is true that both men and women display higher support for these norms than they do in response to questions about specific military actions, there is no indication that the prospect of sharing the risk and cost of NATO actions reduces the difference between men and women in support for these actions. Once again, it appears that the important consideration underlying gender difference is women’s greater aversion to the use of force and not the presence or absence of international institutions.

Table 2.5.  Support for sharing the burden of NATO military action, 2008 “To what extent do you tend to agree or disagree: that all NATO member countries should contribute troops if the NATO alliance decides to take military action?” “To what extent do you tend to agree or disagree: that all NATO member countries should share in the financial costs of a NATO military action even when they do not contribute troops?” Agree all members should contribute to financial cost

Agree all members should contribute troops

Men (%)

Women (%)

Gender difference (women–men) (%)

United States Western Europe Eastern Europe Turkey

83 70 56 37

81 62 45 22

–2 –8 –11 –15

84 73 57 33

81 61 46 23

–3 –12 –11 –10

1,002 7,016 4,002 997

Germany Portugal Spain Italy Slovakia Bulgaria Poland Netherlands United Kingdom

65 74 63 58 44 46 64 83 85

48 64 48 45 31 39 52 82 80

–17 –10 –15 –13 –13 –7 –12 –1 –5

71 73 68 59 44 46 62 87 85

54 57 54 45 32 36 53 78 76

–17 –16 –14 –14 –12 –10 –9 –9 –9

1,010 1,005 998 999 1,031 970 987 1,025 1,000

Men (%)

Women (%)

Gender difference (women–men) (%)

N

Note: Countries shown individually are those for which the gender difference on one or both questions is statistically significant. See figure 2.1 for countries included in Western and Eastern Europe. Source: German Marshall Fund, Transatlantic Trends, 2008.

44

9781501738142_Text.indb 44

4/9/2019 7:25:03 PM

GENDER, WAR, AND WORLD ORDER

th e use o f mi li ta ry fo rce by th e e u ro p e a n  u ni o n Finally, it is useful to examine a question that measures reactions to a military action carried out by an important multilateral organization: the European Union. The Transatlantic Trends surveys in 2006 and 2007 included the following question designed in part to probe citizen support for an integrated approach to European security: “Some say that in order for the European Union to assume a greater international role it needs to do certain things. To what extent do you agree with the following? If the European Union should decide to use military force, [our country] should abide by that decision, even if [our country] disagrees.” Table 2.6 shows that a majority or strong minority of men in all but one country (Slovakia) would support a European Union decision to use force. Among women, in contrast, there is strong support in only three countries. In most countries, the gender difference is one of the largest reported in this chapter. Further, regression experiments (not shown) demonstrate that gender remains a very significant predictor in the presence of controls for left-right ideology, the belief that war is necessary, support for a strong EU global role, and the belief that economic power is more important than military power. Why are the differences so large? One explanation is that the question combines two considerations that evoke gendered responses: the first is military action, and the second is the process of European integration. We Table 2.6.  Percent who agree/disagree that country should abide by an EU decision to use military force, 2006–7 “Some say that in order for the European Union to assume a greater international role it needs to do certain things. To what extent do you agree with the following? If the European Union should decide to use military force, [country] should abide by that decision, even if [country] disagrees.” Men

Germany Portugal Italy France Spain Bulgaria Netherlands Poland Slovakia United Kingdom Turkey Romania

Women

Agree (%)

Disagree (%)

Agree (%)

Disagree (%)

Agree gender difference (women–men) (%)

44 60 55 46 54 48 54 54 36 44 53 57

55 36 44 53 45 43 45 39 58 51 37 34

22 38 37 30 38 34 41 43 25 35 44 50

77 53 60 68 59 51 57 44 63 62 29 37

–22 –22 –18 –16 –16 –14 –13 –11 –11 –9 –9 –7

N 1,921 1,901 1,948 2,018 2,028 1,772 1,978 1,820 1,472 1,913 1,412 1,636

Source: German Marshall Fund, Transatlantic Trends, 2006–7.

45

9781501738142_Text.indb 45

4/9/2019 7:25:03 PM

CHAPTER TWO

know that women are slightly more skeptical of European integration than men are, presumably because of fears that market integration will lead to cuts in social programs (Nelson and Guth 2000). Here we see that a supranational decision to use military force evokes an even stronger skepticism. The one thing we do not see is any indication that the relative aversion of women to the use of military force disappears when the proposed action would occur jointly as the result of what is essentially a supranational European decision. As was true in prior sections of this chapter, the important issue for women seems to be opposition to the use of force and not considerations of multilateralism.8

Threats, Power, War, and Institutions The data reviewed in this chapter suggest that, as Wittkopf once concluded with respect to American public opinion in general, international involvement itself does not polarize opinion along gender lines. Rather it is the means and consequences of involvement (Wittkopf 1990). International involvement is everywhere uncontroversial and evokes no gender difference. On the threat of violence, however, there are gender differences, and there are indications that military power divides the genders. The surveys presented in this chapter cast doubt on the hypothesis that a more generalized liberal vision of international relations underlies the views of women. Only in the US is there a substantial gender difference in UN favorability. When the issue turns to the role of the UN, NATO, or the European Union in legitimizing or collectivizing military actions, women are less likely to express support than are men—the opposite that one would expect. These results suggest that the question of using military force is the issue that drives gender difference. Multilateral endorsement or participation is secondary. Finally, an important finding to this point is that gender differences are not universal across issues or nations. Cross-nationally, gender difference is most consistently prominent in the US across a range of issues, which should alert us to the possibility that gender politics in the US have some unique qualities. Second, the fact that gender difference varies cross-nationally and across issues further undermines the plausibility of the essentialist hypothesis.

46

chapter 3

The Gendered Politics of Defense Spending

In chapter 1, I reviewed evidence that gender difference on national security issues arises from the fact that men place a higher priority on defense spending. There is also evidence that women place a higher priority on social spending. In some studies, these differing priorities on defense and social spending appear to be the key issue driving gender difference in assessments of presidential approval and even voting behavior (Gilens 1988; Kaufmann and Petrocik 1999). However, almost all of this research focuses on citizen preferences in the US, and even in the US, the research includes data for only one or two years. What is lacking is a cross-national analysis that examines gender difference on spending issues over a substantial number of years.

Support for Defense Spending in the United States, 1965–2013 The largest challenge to evaluating the historical evolution of gender difference on the issue of defense spending is the task of assembling a time series of gender breakdowns. As I noted in the introductory chapter, there has been little time-series research on gender politics generally. The major reason is that survey organizations rarely publish historical gender breakdowns. The data exist, but scholars must retrieve them from a number of print and electronic sources that in some cases require reprocessing from older data storage formats. For study of gender difference in the US, there are two candidate time-series. The Gallup Organization has been asking an identically worded question on defense spending since 1969, and the General Social Survey (GSS) has asked a very similar question since 1973, although the utility of the latter is limited by a number of gaps and by the shift to bi-yearly surveys since 1994. I chose the Gallup question as the “core series” for this chapter, because the series begins earlier than GSS and because Gallup asks the question in identical form almost yearly. It is also helpful that Gallup fields the question during the first half of the year in most years since 1980. As Wlezien has shown (1996), public opinion on defense spending 47

CHAPTER THREE

correlates most strongly with the information that is available at specific phases of the budgetary process (presidential requests early in the year and appropriations later in the year). Using a question administered in the first half of the year ensures that respondents have been exposed to the publicity and debate that surrounds the announcement of the president’s budget in February. When the Gallup question was not available for a particular year, I sought first to substitute the GSS variant—the most similar available alternative. When the GSS question was not available, I employed questions from other survey organizations with wording as similar to the Gallup question as possible. Details on each data point appear in the appendix. The exact wording of the Gallup question is as follows: There is much discussion as to the amount of money the government in Washington should spend for national defense and military purposes. How do you feel about this: do you think we are spending too little, too much, or about the right amount?

Wlezien has argued that the interesting political question is the balance of citizen preferences concerning spending change, that is, the percentage favoring increases and decreases in spending (1996). I therefore operationalize support for defense spending as net support, computed as follows: Net support = % too little/(% too little + % too much) * 100 By employing this measure of net support, I focus on what I call crystallized opinion—an active stand on the issue of change in the defense budget. The measure is essentially the percentage of crystallized opinion that prefers an increase.1 Figure 3.1 displays the evolution of net support for defense spending among men and women in the US. The top half of the figure shows the trends in net support, while the bottom half shows the yearly gender difference (women net support—men net support), with a negative difference indicating that women show less support for defense spending. Two features of the graphics stand out. First, the top half of the graphic indicates that women and men represent what Page and Shapiro have called “parallel publics” (1991, 294). Page and Shapiro found that any difference in the opinions of population subgroups, once established, changes little over time. Instead, opinions—including those of women and men—move largely in tandem (Page and Shapiro 1991, 294). Page and Shapiro speculate that the parallel movements of opinion are attributable to the fact that all segments of the population respond to a similar mix of information at the same time (295). This hypothesis receives support in the work of Enns and Kellstedt, who find—surprisingly—that citizens of both high and low

48

Net support for increased defense spending among men and women (USA) 1965-2013

80

net support (%)

60

40

20

0 1960

1970

1980

1990 Men

2000

2010

Women

Gender difference in net support for increased defense spending (USA)

-20

gender difference (women - men) -10 0

10

1965-2013

1960

1970

1980

1990

2000

2013

Figure 3.1.  Net support for defense spending and gender difference in the US, 1965–2013. Sources: 1965–2007, Gallup Poll, as reported in Eichenberg and Stoll 2012; German Marshall Fund, Transatlantic Trends, 2008–13.

CHAPTER THREE

political sophistication respond to information in a similar fashion: “The subgroups generally changed opinion at the same time, in the same direction, and to about the same extent . . . for predominantly the same reasons” (2008, 433; see also Soroka and Wlezien 2008). In both of these works, the authors conclude that a substantial difference may characterize the base preferences of different population subgroups, but the magnitude of the difference changes little over time. Nonetheless, the question is whether the lower net support among women and the occasionally large gender differences shown in the bottom of figure 3.1 arise from differential responses to the same stimuli or—perhaps less likely—from different correlates of net support among women and men. In fact, although the opinions of men and women move in parallel directions, they do not always change by the same magnitude. As a result, there is substantial variation in the magnitude of gender differences (bottom half of figure 3.1). The average absolute gender difference in net support is 7.3 percentage points, slightly higher than the difference required to establish statistical significance between two survey subsamples, but the gap reaches more than 10 percentage points in eight years and more than 20 points in two years. In addition, as expected, men show higher support for defense spending on average. Support among men is higher in about 70 percent of the years in the series. In years when men display higher support for defense, they do so by almost 9 percentage points. When women show higher support, it is by a lesser amount (4.8 percentage points). The variability of gender difference has important implications. First, the data show that gender difference is not categorical. Women do not always show less support for defense spending. What this means is that existing evidence does not support explanations for gender differences that are rooted in essentialist hypotheses. This is an important finding and one that highlights the importance of analyzing time-series data. If essentialist explanations were true, gender differences would be uniformly large and would vary little over time. Second, the variations in gender difference presumably relate to contextual factors that move the opinions of men and women in different directions—or by different degrees—in particular circumstances. Based on the evidence reviewed above, there is reason to believe that these differences are rooted in the spending priorities of men and women or differences in their reactions to war. Richard Stoll and I studied these variations in the net support of men and women in a series of regression analyses (Eichenberg and Stoll 2012). With one exception, we found that men and women react to similar stimuli in similar ways. Specifically, both men and women react strongly to the most recent change in the defense budget itself, to the casualties of war (especially during the Vietnam War), and to a trade-off of health care spending in favor of defense. In only one case did women react more negatively to a “guns-butter” trade-off: in 1982, women reacted much more negatively to

50

GENDER, WAR, AND WORLD ORDER

the large increases in Reagan defense spending in relation to health care spending (Eichenberg and Stoll 2012, 340–42). Of course, this does not mean that men and women support defense spending in equal measure. As noted above, women on average show significantly less support for defense spending. However, it does mean that the yearly variations in gender difference remain close to the “normal” average range, so the important question becomes this: what determines this base level of difference in the first place? Bartels provides one answer (1994). He estimated the impact of ideology and a variety of foreign policy attitudes on Americans’ support for defense spending in 1992. The results were unequivocal: “The magnitudes of the various parameter estimates clearly suggest that the dominant factor in producing support for defense spending in 1992 was a general willingness ‘to use military force to solve international problems.’ . . . it is the dominant determinant of defense spending preferences in every specification, regardless of which other variables are included ”(Bartels 1994, 481). What is more, when Bartels turned his attention to support for spending in the Cold War years of 1982–84, he found the same result: support for defense spending correlated most strongly with a general willingness to use force rather than with other factors. He concludes, “One implication of these results is that, even in the Cold War era, defense spending preferences were determined more by a predisposition to favor or oppose the use of force in the international arena than by either general political ideology or attitudes toward the Soviet Union per se”(1994, 485). Put differently, Bartels’s findings suggest that defense-spending preferences correlate more with long-held attitudes toward war and military force than with short-term variation in threats or estimates of adversaries. The implication of Bartels’s findings is that the base gender difference in support for defense spending is rooted in a gender difference toward military force and war. That is, although the size of the gender difference on defense spending may vary somewhat from year to year, the fact that women generally display less support for defense spending likely results from the fact that they also show less support for military force and war. Recall that in chapter 2, I reviewed surveys on the necessity of war that demonstrated that women consistently show less support for the view that war is sometimes necessary (figure 2.2). Those differences were significant and consistent over a decade. Further, Richard Stoll and I have shown that this gender difference is highly significant after controlling for a number of variables that affect opinions of war. In the US, support for the “necessity of war” correlates most strongly with support for increased defense spending, and support for war is in turn strongly related to gender—women are less likely to believe that war is sometimes a necessary instrument of policy (Eichenberg and Stoll 2015). In summary, the evidence from over forty years of polling on defense spending issues in the US demonstrates two

51

CHAPTER THREE

things. First, women in the US on average display less support for defense spending. Second, this gender difference is rooted in the fact that men and women have fundamentally different attitudes toward war.

Comparing European and American Attitudes Research on opinions of defense spending outside the US is rare.2 For students of gender difference, this is unfortunate, because the US may be unique in many ways. First, the US is a superpower that has been involved in many wars during the period studied in the previous section, and attitudes toward war are highly gendered, thus contributing to gender difference on defense spending issues. Second, the welfare state is at the core of the ideological and partisan polarization that characterizes American politics, while in Europe there is a broad, cross-party consensus surrounding the major pillars of the welfare state. For this reason, a gendered “guns-butter” debate may be more likely to arise in the US. On the other hand, since 2008 European economies have experienced slow economic growth and pressure on budgets because of the financial and sovereign debt crises that have hindered the continent’s economies. At the same time, instability and conflict in Europe (Ukraine) and the Middle East increased pressure on European governments to increase defense spending. Under such circumstances, a gendered sensitivity to trade-offs might appear in Europe as well. For all of these reasons, a comparative assessment of gender difference on the defense budget offers a particularly useful opportunity to explore the pragmatic, budgetary roots of gender difference. The Transatlantic Trends series contains a question about defense spending in seven years: 2002, 2003, 2004, 2008, and 2011–13. The questions are variants on the familiar form of survey measures on government spending: “Do you think defense spending should be increased, kept the same, or decreased?”3 In the following analysis, I operationalize support for defense spending using the same net support measure that I employed for the US in the previous section: Net support = %increase/ (%increase + %decrease) * 100 Net support for defense is thus the percentage who support an increase in spending relative to the total percentage who want a change from the current level of the defense budget. Figure 3.2 displays the evolution of this net support measure (top) together with the gender difference, that is, net support among women minus net support among men (bottom). The top half of figure 3.2 shows that net support for defense spending among both men and women is higher in the US and Turkey than in Europe (although in 2002 and 2003

52

Net support for defense spending among men and women 2002-2013

Western Europe

USA 80

76 76

60

49

40

43

40

31

45

44

28

2002

2003

2004

49

2008

45 31

24

20 0

53

51 40

2011

2012

2013

35

2002

40

46

41

2003

2004

Eastern Europe 80

79

73

79

30 31

2008

2011

23 22

2012

29 30

2013

Turkey 75 77

76 64

60

54

40

50

54

47

41

34

36

32

39 27

46 37

34

2002

2003

2004

2008

2011

2012

2013

2002

Men: net support (%)

2003

43

40

29

23

20 0

42

32

2004

2008

2011

2012

2013

Women: net support (%)

Gender difference in net support for defense spending 2002-2013 Western Europe

USA 20 10 0

5

-4

-4

-10

gender difference (%)

4

0

-20

-18

1

1

-5

-1

-3 -9

-14

-30

-26

2002

2003

2004

2008

2011

2012

2013

2002

2003

2004

Eastern Europe

2008

2011

2012

2013

Turkey

20

13

18 10

10

1

0 -10

-6

-3

-20

-3

-5

-5

-5

2011

2012

2013

-4

-20

-30 2002

2003

2004

2008

2002

2003

2004

2008

2011

2012

2013

Figure 3.2.  Net support for defense spending and gender difference, 2002–13. Source: German Marshall Fund, Transatlantic Trends for years indicated. The survey was not conducted in Turkey prior to 2004.

CHAPTER THREE

support in Eastern Europe approached that of the US and Turkey). Second, as was true in the longer historical series for the US described in the previous section, the opinions of men and women appear to move in parallel— although the direction of movement is not the same in each regional group. In the US and Europe, support for defense was higher in the earlier 2000s but declined through midcentury, perhaps in response to frustration with the war in Iraq or (in the US) to the increase in defense spending after 2001. Beginning in 2008, however, support began to increase somewhat in the US, while in both Western and Eastern Europe it declined sharply in reaction to the financial crisis that began in 2008. Turkey, by contrast, shows an entirely different pattern: in 2004, net support for defense was quite low, but it increased steadily through 2013. It may be that Turkish citizens were reacting to the instabilities of the Middle East region, especially after 2011 and continuing through 2013 as the conflict in Syria created instability on Turkish borders. The signal characteristic of the gender differences in the bottom of the graphic is their variability. In the US and Europe, women are on average less supportive of defense spending, but the magnitude of the difference is larger in the US. In Turkey, women show higher support for defense. The gender difference also varies considerably over time: in some years, the differences are large, in others they are quite small, and in some years, women actually support defense spending more than men. Finally, there is considerable cross-national variability, as displayed in table 3.1. The average gender difference is statistically meaningful in the first six countries listed in the table, but in the others, it is insignificant or positive. In Italy and Turkey, the positive difference is very close to significant. Nonetheless, despite this variation, table 3.1 also shows that there was a very large and significant gender difference in all but three countries in at least one year. The largest difference on average occurs in 2003 and 2004, when the average gender difference across these countries was –8.0 percent. In other years, it averaged less than 3 percent. This suggests that an important factor influencing gender difference was the war in Iraq, which began in 2003 and deteriorated into widespread violence during 2004. However, it seems unlikely that actual spending for the war was the driving factor. The gender difference is large in Germany and France in 2004, but these countries had no troops in Iraq and therefore no war-related spending. Thus, it seems likely that defense spending itself is not the primary driver of attitudes toward defense spending. Rather, they appear more strongly related to attitudes toward war in general or the Iraq War in particular. This interpretation receives additional support from the fact that gender difference in attitudes toward the necessity of war discussed in the previous chapter was largest during the period 2003–5, especially in Eastern and Western Europe. In the next section, I will present further evidence that the support of women for defense spending was significantly lower

54

GENDER, WAR, AND WORLD ORDER

Table 3.1.  Average national gender difference in net support for defense spending, 2002–13

South Korea Sweden Bulgaria USA Poland Netherlands Germany Slovakia Spain Romania United Kingdom France Russia Portugal Italy Turkey Overall

Average gender difference (women–men)

Largest gender difference (women–men)

–14 –11 –11 –9 –8 –7 –6 –3 –2 –2 –1 0 +2 +3 +6 +8 –3

–14 –24 –22 –26 –16 –19 –22 –25 –10 –9 –11 –15 +2 –6 –4 –4 –14

Year of largest gender difference 2012 2013 2012 2012 2013 2004 2003 2004 2004 2013 2003 2008 2012 2011 2004 2011

Note: Data for Russia and South Korea are for 2012 only. Source: German Marshall Fund, Transatlantic Trends, 2002–13.

than men during 2003 and 2004, while in other years the difference was insignificant. Finally, it is interesting to note in table 3.1 that the largest gender difference in some countries occurs in 2012 or 2013, which were years of highly polarized budgetary debates and cuts in spending (US) or economic stress on public budgets generally (Europe). In summary, the largest gender difference emerged either in reaction to war or to fiscal debates that raised concerns about budget cuts given the dominance of austerity as the solution to economic troubles. In other words, gender difference appears related to concerns about war, welfare, or both. The Transatlantic Trends survey asked about social spending programs in three years: 2002, 2003, and 2013.4 Briefly summarized, the responses to these questions show that citizens’ support of social programs is nearly universal in Europe and the US. An average of more than 80 percent of both men and women favor an increase in social spending, compared to the average of 40 percent who favor an increase in defense spending. In addition, although women show slightly stronger support for social spending than men, the difference in most countries is quite small and statistically insignificant. Thus, the consensus in support of social spending is so large that the small gender difference is not likely to be politically significant. Of course, an isolated question on social spending does not raise the prospect of a trade-off in favor of defense spending, and the summary percentages may mask a trade-off calculus among individuals. That is, it is possible that individuals who favor increased social spending also favor

55

CHAPTER THREE

less for defense. However, my analysis shows that this is not the case. In fact, there is virtually no correlation between the two: net support for defense spending is approximately the same for men and women who want to increase social spending, and a preference for cutting social spending does not correlate with a desire to increase defense. In summary, although it is true that social spending is everywhere more popular than defense spending, there is no evidence at the individual level that opinions of social spending condition support for defense. Nor is there any evidence that a trade-off pattern is more characteristic of women.

Regression Model: War, Welfare, and Defense Spending The evidence presented thus far in this chapter suggests that fundamental attitudes toward war and military force most heavily influence support for defense spending and gender difference in support. This suggests that attitudes toward war and military force are more important than pragmatic considerations that predict gender difference from differing social spending priorities of men and women. In this section I evaluate this hypothesis in a regression model of support for defense spending for the years in which survey measures of support for both defense and social spending are available (2003 and 2013) together with a number of important variables of theoretical interest. The dependent variable in the analysis is support for defense spending, ranging from the lowest value for respondents who want to cut defense spending to the highest value for those who want to increase defense. The regression analysis extends and refines the model that Richard Stoll and I reported for five years between 2004 and 2013. Our model specified support for increased defense spending as a function of attitudes toward war and military power, left-right ideology, gender, support of US and EU global leadership, opinions of NATO, and external threat (Eichenberg and Stoll 2015). Here I extend the model for 2003 and 2013 by examining the impact of attitudes toward social spending and by asking if the impact of “guns/butter” considerations is higher for women than for men. Specifically, the model of support for defense spending includes the following variables. First, I specify two measures of attitudes toward war and military force: the belief that war is sometimes necessary (discussed in the previous chapter), and a variable expressing agreement that “economic power is more important than military power.” I expect the “war is necessary’ variable to have a positive impact on support for defense and the “economic power is more important” variable to have a negative influence. Second, I specify a variable measuring respondents’ left-right ideological position and a variable representing gender (female=1). I expect ideology to have a positive impact on support for defense, with those on the right

56

GENDER, WAR, AND WORLD ORDER

more supportive of defense spending. Based on the negative gender differences presented above, I expect gender to have a negative influence, with women showing less support for defense. Third, I specify support for US global leadership as a dummy variable that takes a value of 1 if the respondent believes that “strong US global leadership” is desirable. I also specify a similar variable for EU global leadership. I expect that support for strong US leadership will show a positive association with support for defense spending based on previous research, although a negative association might emerge if respondents follow a “free riding” logic. That is, support for US global leadership might lead respondents to the conclusion that their country need not spend more on defense. In the case of EU leadership, I expect a positive relationship on the assumption that a “strong” EU global role requires investment in military capabilities. However, a negative relationship might emerge if respondents see the EU’s global role in nonmilitary terms, as some scholars have argued (Kagan 2003, 53–55). Fourth, I examine the “guns-butter” dynamic by specifying two variables. First, I specify dummy variables representing those who favor a decrease and those who favor an increase in social spending. If trade-off considerations exist in public opinion, I would expect a positive coefficient for the decrease social spending variable and a negative coefficient for the increase social spending variable. That is, a trade-off pattern exists if those who are less favorable to social spending are also more positive toward defense and those who favor an increase in social spending are also less favorable to defense. Finally, to analyze the impact of external threat on support for defense spending, I specify a dummy variable representing respondents who believe that “China is a military threat” (versus those who believe China is not a threat). This is the only external threat variable available for both 2003 and 2013 in the Transatlantic Trends dataset, but other research has shown that a variety of threat measures do not alter the relationship between the variables specified here and support for defense spending (Eichenberg and Stoll 2015; Bartels 1994). Table 3.2 displays the results. They reveal that support for defense spending in these two years correlates more strongly with basic beliefs about war, military power, and respondent ideology than with gender or trade-off considerations. Beliefs about the necessity of war and respondent ideology are by far the most consistently significant influences on support for defense spending. That is, support for defense spending is highest among those who believe war is necessary and among those on the ideological right. The belief that economic power is more important has a negative impact, but only in Western Europe. Similarly, support for US and EU global leadership has a positive and negative impact, respectively, but the effect is not uniform across countries and regions. Support for US leadership has a positive

57

CHAPTER THREE

Table 3.2.  Regression analysis of support for defense spending, 2003 and 2013 (1) USA

Agree war is sometimes necessary = 1 Agree economic power is more important than military = 1 Left-right ideology (3 point scale) Gender (female = 1) US global leadership is desirable = 1 EU global leadership desirable = 1 China is military threat = 1 Favor decrease in social spending Favor increase in social spending Interaction: female × favor decrease in social spending Interaction: female × favor increase in social spending Constant Observations

0.321*** (0.092) –0.185* (0.081) 0.224*** (0.044) 0.017 (0.010) 0.135 (0.113) –0.235** (0.088) 0.249*** (0.070) –0.083 (0.158) –0.006 (0.103) 0.061 (0.262) –0.023 (0.143) 1.497*** (0.205) 1,081

(2) Western Europe

(3) Eastern Europe

0.228*** (0.048) –0.199*** (0.039) 0.219*** (0.020) –0.009 (0.070) 0.119*** (0.032) –0.011 (0.051) 0.112*** (0.017) 0.006 (0.081) –0.060 (0.070) –0.023 (0.105) 0.004 (0.066) 1.464*** (0.161)

0.206*** (0.038) 0.054 (0.104) 0.004 (0.010) 0.089 (0.056) 0.069** (0.025) 0.048 (0.078) 0.044 (0.047) –0.482* (0.188) 0.207 (0.111) –0.084 (0.150) –0.117 (0.128) 0.362** (0.110)

8,403

1,770

(4) Turkey

0.209* (0.101) 0.051 (0.128) 0.152* (0.059) 0.217 (0.147) –0.151 (0.133) –0.244* (0.124) 0.007 (0.111) –0.282 (0.205) 0.734*** (0.137) –0.391 (0.266) –0.325 (0.207) 0.786*** (0.210) 600

Note: Ordered probit regression with robust standard errors in parentheses. For Western and Eastern Europe, standard errors are clustered by country. Equations include a control for year of survey and country of survey (for Western and Eastern Europe). *** p