Networks, Knowledge Brokers, and the Public Policymaking Process 3030787540, 9783030787547

Citation preview

Networks, Knowledge Brokers, and the Public Policymaking Process Edited by Matthew S. Weber · Itzhak Yanovitzky

Networks, Knowledge Brokers, and the Public Policymaking Process “This edited collection provides a major contribution our understanding of the use of research evidence in policy making. Weber and Yanovitzky have curated a fascinating set of accounts of social network analysis as a tool for exploring knowledge brokerage and policymaking. The collection helps us to move forward both in terms of research methods for this important emerging area of scholarship and also in terms of our analysis of knowledge brokerage within complex systems. I shall be keeping a copy on my shelf and look forward to sharing it with my students and colleagues in years to come.” —Annette Boaz, Professor of Health and Social Care Policy, London School of Hygiene and Tropical Medicine, UK “Relationships shape what we know and how we share information, and conversely what knowledge and individuals are isolated and excluded from social systems. Knowledge brokers, Networks, and the Policy Process brings together a collection of essays and empirical studies that add much-needed ideas to our understanding of how brokers of knowledge, individual and organizational networks, and the policy process interact. The range of theoretical and analytic approaches examined will help us better navigate evidence use in power structures, nested structures, and politics varied policy areas. This book is a must read for those who have not yet discovered the critical role knowledge brokers and networks play in the many facets of policymaking.” —Kimberly DuMont, Vice President, AIR Equity Initiative, American Institutes for Research (AIR) “Researchers have for some time considered how knowledge is utilized in policymaking, but less is known about the oil that lubricates the transfer of information in the policymaking machinery. In this illuminating volume, Weber and Yanovitzky assemble leading thinkers to consider the role of knowledge brokers in facilitating movements of information through policy networks around various but related topics—education, immigration, nutrition, healthcare, and the timely issue of misinformation. These outstanding scholars provide us with methodological breakthroughs that shed light on types of knowledge brokering, transactions, preferences, and behaviors of network actors in think tanks, the media, research and policymaking. Networks, Knowledge Brokers, and the Public Policymaking Process advances the field not only on the structural issues of networks and

knowledge brokering on different issues, but even on the nature of knowledge on these issues.” —Christopher Lubienski, Professor of Education Policy, Indiana University “Using social network analysis, this book demystifies how research makes it way into public policy and shines a bright light on the knowledge brokers who make it happen. Network analyses enable us to see the complex web of relationships between researchers, policymakers, advocates, think tanks, journalists, and the public that shapes how research is applied in policy. Spanning health and education policy, the chapter authors describe different types of knowledge brokers, ways to identify them in the policy ecosystem, and how to understand their roles in spreading research ideas in policy circles. They also provide keen insights into strategies for building more robust networks that connect research and policy. This is the authoritative text on how to apply network analysis to improving the use of research evidence in policy.” —Vivian Tseng, Senior Vice President, Program William T. Grant Foundation www.wtgrantfoundation.org

Matthew S. Weber · Itzhak Yanovitzky Editors

Networks, Knowledge Brokers, and the Public Policymaking Process

Editors Matthew S. Weber School of Communication and Information Rutgers, The State University of New Jersey New Brunswick, NJ, USA

Itzhak Yanovitzky School of Communication and Information Rutgers, The State University of New Jersey New Brunswick, NJ, USA

ISBN 978-3-030-78754-7 ISBN 978-3-030-78755-4 (eBook) https://doi.org/10.1007/978-3-030-78755-4 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Cover illustration: gettyimages/Viaframe This Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Foreword: Multimodal, Multidimensional, and Multilevel Social Network Systems

Knowledge Brokers, Networks, and the Policy Process is a collection of largely empirical studies that examines how institutional brokers utilize network strategies to help people in their social systems learn about, obtain, and benefit from receiving resources that are potentially available under public policy. The articles vary widely in terms of the social systems and policy domains they study. And, they vary considerably in terms of the theories they test and the analytic techniques they employ. An interesting first chapter written by the editors sets the context and expectations for the volume and a final chapter, written by the same authors, summarizes the findings, offers helpful critiques, and uses this platform to articulate an agenda for future work. The chapters are well written by highly competent scholars in related but different fields and provide an excellent representation of scholarship in this complex and diverse area. If you are interested in aspects of knowledge brokerage, networks, and the policy process or how these three phenomena interact, you will find this an excellent volume on the current state of this genre of research. In what follows I review several network developments that expand our collective ability to examine more complex and intricate networks and network properties. In my view, this is one important way in which this corpus of research can take the next big step forward. Historically, social scientists have studied human attributes, like education, gender, race, and socioeconomic status, to explain how humans behave. An important departure from this tradition occurred in the v

vi

FOREWORD: MULTIMODAL, MULTIDIMENSIONAL, AND …

middle of the twentieth century when scholars began to focus on relationships, the network of connections that people build, maintain, and dissolve with others in their social worlds. Over the years scholars have championed one approach or the other. But, it is not difficult to see that both approaches have merit and can contribute to our understanding of human behavior. In fact, it is easy to make the case that both approaches should be used, and where possible, used together. Over the last two decades scholarship has shifted in this direction, using network models that incorporate human attributes or, alternatively, models that study human attributes in the context of social networks. Importantly, most of the chapters in this book focus on both networks and human attributes. This is a departure from prior traditions and norms as early network scholars studied networks that were restricted to three major network properties. These properties are centered on nodes, relationships (links), and levels. The nodes of a network are the objects that are linked together to create the network. Almost exclusively, early scholars studied unimodal networks that represented only one kind of node, like the networks among students or links that tied together administrators. Current research has moved toward multimodal networks, those that contain multiple kinds of objects like people and data bases or brokers and knowledge objects. Similarly, early studies were limited to studying single relations such as friendship, collaboration, or teamwork relations. These were called unirelational networks. Of course, we know that most complex systems contain multiple kinds of relations that exist among the nodes at the same time. Today, studies are being conducted on several relations simultaneously as multidimensional (or multirelational) networks, such as office, social, and professional relations. In the early days of network research scholars were also only able to study networks that occurred on one level at a time, called unilevel networks. However, we know that many networks operate on multiple levels in the empirical world such as educational systems. In this social system students are nested within classes, classes are nested within schools, schools are nested within districts, and districts are nested within statewide educational systems. These are called multilevel networks. Today we have the ability to capture and analyze the influence of nesting in real world networks, which improves our ability to explore and understand the complexity of multilevel nested networks. When taken together, these developments mean that we are no longer limited to studying single object type networks, with only one relation, at

FOREWORD: MULTIMODAL, MULTIDIMENSIONAL, AND …

vii

only one level. Rather, we can now theorize, operationalize, and analyze brokers, networks, and the policy process with multiple types of nodes, say people and knowledge objects, multiple types of relations, say brokering relation links and knowledge transfer links, and at multiple levels, like classrooms, schools, and school districts. Almost all of the studies in this book focus on large scale social issues, what De la Haye et al. call “a whole-of-system” approach. This approach attempts to capture as much of the working apparatus of the entire system as possible. Needless to say, this is a daunting task. Although the details differ from study to study, most of them report trying to capture how different sets of people are tied together with resources of one form or another. One set constitutes the brokers who are affiliated with the societal institutions. The other set is the people who are being served by these institutions, whether education, health, or some other social service. There is a set of links within the people who are identified as brokers and another set of links within the people who are being served, such as students and/or parents. And there is a set of ties between brokers and recipients. This latter set of links provides the mechanism for transferring the resources from the educational or medical or employment institutions to those who are in need of them. The resources take many different forms including knowledge artifacts, financial assistance, social support, etc. Brokers are key people in helping to transfer these resources, and different strategic practices that make things work as smoothly as possible abound. What does this idealization of the research reported in these chapters show? First, a case can be made that the networks are multimodal, that is, there are more than one type of objects. For example, some people are brokers and some are recipients. But these are not the only possible types of objects. For example, Lawlor et al identify the components of their research as contributor, knowledge objects, and recipients. In this representation, knowledge can be considered another type of network node and formally analyzed as part of the overall network giving researchers an opportunity to see how different knowledge objects are tied to brokers and recipients and influence the outcome of brokering processes. Similarly, Flannigan et al. discuss how brokers facilitate access to research knowledge for educational leaders. Clearly, brokers are one type of node in the network and educational leaders are another nodal type. But research knowledge can also be treated as a nodal type and linked to both brokers and educational leaders to provide the network that ties these all together.

viii

FOREWORD: MULTIMODAL, MULTIDIMENSIONAL, AND …

Second, although most of the studies are unirelational in that they examine only one relation in the networks, there are clearly multiple relations that are both possible and likely. For example, it would be informative to create relationship linkages showing the different types of resources transferred from the institutions to recipients. Third, it is clear that there is at least some level of nesting in the networks studied here, so the real social system is constituted as a multilevel rather than a unilevel network. Although none of the studies incorporated this feature, it is now possible to analyze multilevel network data to capture the influence of different levels on the operation of the network, and this would be a good analytical strategy for future research. A major part of this book focuses on brokers and their efforts to broker knowledge in the context of policy processes. As is the case in the majority of brokerage studies, the papers in this collection by and large assume that brokerage is a positive thing for the people being brokered, what we might identify as a positivity bias. In many cases, including several of the research projects reported in these chapters, that is a reasonable assumption. But, it is important to remember that not all brokers and not all brokerage engagements are beneficial to the people being brokered. Ron Burt’s view of brokers such as those in his studies of bank managers, largely views brokers as exploiters who use the structural holes between others in the network as their opportunity to gain power and material benefit by keeping the holes open and the others disconnected. Another example is the case of the “Cupid Broker,” where a broker deliberately links to two or more others in the network not for the benefit of the others but for the benefit of the broker. These cautionary tales point to the importance of assessing who benefits from the brokering, the broker, or the brokered. Of course, brokering need not be a one-sided affair. There may be situations in which brokering that benefits the broker only is a good thing. And there are brokering circumstances in which keeping others disconnected is a good outcome as in those attempting to keep warring parties separate or preventing the flow of contagious viruses between disparate individuals. And there is no reason not to consider the possibility that both broker and brokered may benefit from brokering activities. A good example of a study that uses all of these multiple features is an article published by Woody Powell and colleagues in the American Journal of Sociology in 2005. It focused on the emergence, evolution, and entrenchment of the biotechnology industry from late 1980 to 2003. Needless to say, this is an extraordinarily large and complex

FOREWORD: MULTIMODAL, MULTIDIMENSIONAL, AND …

ix

social system. They studied the networks among five different nodal types: (1) University biology departments and biotechnology centers, (2) Dedicated Biotechnology Firms (Startups) (3) Venture Capital Firms, (4) Pharmaceutical Companies, and (5) Governmental Regulatory Agencies. The four multiple linkages they studied among these five types of nodes were (1) Research and Development, (2) Finance, (3) Commercialization, and (4) Licensing (largely by government agencies). They also examined data over time between the late 1980s and 2003. By studying five different nodal types and four different relations together they were able to provide a much more complex and integrated analysis and understanding of how the biotechnology industry was launched, transformed, and embedded into society than studying this process as separate nodal networks, based on separate sets of relations and separate network levels. The study of communication and other social networks has grown exponentially during the twenty-first century. Brokerage roles and knowledge brokering processes have become important objects of significant empirical investigation to the role of public policy, as this book amply demonstrates. And policy processes have never been more important in societies around the world than they are during the present era. Knowledge Brokers, Networks, and the Policy Process could not have been published at a better time. And, as described above there is considerable room for future scholarship to grow in this area by theorizing, operationalizing, and analyzing multimodal, multidimensional, and multilevel network models of these important aspects of policy processes. Peter Monge Emeritus Professor of Communication, Annenberg School of Communication Emeritus Professor of Management and Organization Marshall School of Business University of Southern California Los Angeles, CA, USA

Acknowledgements

The editors would like to thank the authors for the enthusiasm, time and energy that went into the chapters in this edited volume. We are particularly grateful for the inspiring conversation that took place at the workshop held at University of Minnesota in September 2019, which helped spark the development of this book. In the process of editing this book we were fortunate to be part of a network of forty one authors, each of whom brought a unique perspective to this work and helped to make this volume complete. Special thanks are due to the University of Minnesota for hosting the workshop and supporting the development of this edited volume. In particular, we acknowledge the support of Dr. Elisia Cohen. In addition to providing additional resources to support the workshop, Dr. Cohen was an active participant in the conversations and helped to push the boundaries of this work. We are also grateful to Dr. Jennifer Watling Neal and Dr. Zachary Neal, both of whom helped with the planning and organization of the workshop, and also contributed to the early conceptualization of this book. Most importantly, we acknowledge the generous support of the William T. Grant Foundation, including Dr. Adam Gamoran, Dr. Vivian Tseng, Dr. Kim DuMont and Dr. Lauren Supplee. The William T. Grant Foundation has encouraged this research and the development of this book since it began during a coffee chat at a conference in 2018.

xi

Contents

Knowledge Brokers, Networks, and the Policymaking Process Matthew S. Weber and Itzhak Yanovitzky

1

Disseminating Evidence to Policymakers: Accounting for Audience Heterogeneity Jonathan Purtle

27

“Being Important” or “Knowing the Important”: Who Is Best Placed to Influence Policy? Kathryn Oliver

49

Integrating Connectionist and Structuralist Social Network Approaches to Understand Education Policy Networks: The Case of the Common Core State Standards and State-Provided Curricular Resources Emily M. Hodge, Susanna L. Benko, and Serena J. Salloum Measuring Issue Preferences, Idea Brokerage, and Research-Use in Policy Networks: A Case Study of the Policy Innovators in Education Network Joseph J. Ferrare, Sarah Galey-Horn, Lorien Jasny, and Laura Carter-Stone

71

101

xiii

xiv

CONTENTS

Broken Bridges: The Role of Brokers in Connecting Educational Leaders Around Research Evidence Kara S. Finnigan, Alan J. Daly, Anita Caduff, and Christina C. Leal An Ego-Network Approach to Understanding Educator and School Ties to Research: From Basic Statistics to Profiles of Capacity Elizabeth N. Farley-Ripple and Ji-Young Yun Mixing Network Analysis and Qualitative Approaches in Educational Practices Mariah Kornbluh A Multi-Level Framework for Understanding Knowledge Sharing in Transnational Immigrant Networks Rosalyn Negrón, Linda Sprague-Martínez, Eduardo Siqueira, and Cristina Brinkerhoff

129

155

183

205

Promoting Healthy Eating: A Whole-of-System Approach Leveraging Social Network Brokers Kayla de la Haye, Sydney Miller, and Thomas W. Valente

239

Brokerage-Centrality Conjugates for Multi-Level Organizational Field Networks: Toward a Blockchain Implementation to Enhance Coordination of Healthcare Delivery Kayo Fujimoto, Camden J. Hallmark, Rebecca L. Mauldin, Jacky Kuo, Connor Smith, Natascha Del Vecchio, Lisa M. Kuhns, John A. Schneider, and Peng Wang

265

Platformed Knowledge Brokerage in Education: Power and Possibilities Jennifer A. Lawlor, J. W. Hammond, Carl Lagoze, Minh Huynh, and Pamela Moss Network Approaches to Misinformation Evaluation and Correction Katherine Ognyanova

315

351

CONTENTS

Closing the Theory–Research Gap in Knowledge Brokerage: Remaining Challenges and Emerging Opportunities Itzhak Yanovitzky and Matthew S. Weber Index

xv

375

393

List of Contributors

Susanna L. Benko Ball State University, Muncie, IN, USA Cristina Brinkerhoff School of Social Work, Boston University, Boston, MA, USA Anita Caduff University of Rochester, Rochester, NY, USA Laura Carter-Stone Vanderbilt University, Nashville, TN, USA Alan J. Daly University of California At San Diego, San Diego, CA, USA Kayla de la Haye Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA Elizabeth N. Farley-Ripple University of Delaware, Newark, DE, USA Joseph J. Ferrare University of Washington Bothell, Bothell, WA, USA Kara S. Finnigan University of Rochester, Rochester, NY, USA Kayo Fujimoto The University of Texas Health Science Center at Houston, Houston, TX, USA Sarah Galey-Horn University of Edinburgh, Edinburgh, UK Camden J. Hallmark The University of Texas Health Science Center at Houston, Houston, TX, USA xvii

xviii

LIST OF CONTRIBUTORS

J. W. Hammond School of Education, University of Michigan, Ann Arbor, MI, USA Emily M. Hodge Montclair State University, Montclair, NJ, USA Minh Huynh School of Education, University of Michigan, Ann Arbor, MI, USA Lorien Jasny University of Exeter, Exeter, UK Mariah Kornbluh Department of Psychology, University of South Carolina, Columbia, SC, USA Lisa M. Kuhns The University of Texas Health Science Center at Houston, Houston, TX, USA; Ann & Robert H. Lurie Children’s Hospital of Chicago, Chicago, IL, USA Jacky Kuo The University of Texas Health Science Center at Houston, Houston, TX, USA Carl Lagoze School of Information, University of Michigan, Ann Arbor, MI, USA Jennifer A. Lawlor School of Information, University of Michigan, Ann Arbor, MI, USA Christina C. Leal University of California At San Diego, San Diego, CA, USA Rebecca L. Mauldin The University of Texas at Arlington, Arlington, TX, USA Sydney Miller Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA Pamela Moss School of Education, University of Michigan, Ann Arbor, MI, USA Rosalyn Negrón Anthropology, University of Massachusetts, Boston, MA, USA Katherine Ognyanova Rutgers University, New Brunswick, NJ, USA

LIST OF CONTRIBUTORS

xix

Kathryn Oliver Public Health and Policy, London School of Hygiene and Tropical Medicine, London, UK Jonathan Purtle Dornsife School of Public Health, Drexel University, Philadelphia, PA, USA Serena J. Salloum Ball State University, Muncie, IN, USA John A. Schneider The University of Texas Health Science Center at Houston, Houston, TX, USA; University of Chicago, Chicago, IL, USA Eduardo Siqueira School for the Massachusetts, Boston, MA, USA

Environment,

University

of

Connor Smith Houston, TX, USA Linda Sprague-Martínez School of Social Work, Boston University, Boston, MA, USA Thomas W. Valente Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA Natascha Del Vecchio University of Chicago, Chicago, IL, USA Peng Wang The University of Texas Health Science Center at Houston, Houston, TX, USA; Centre for Transformative Innovation, Swinburne University of Technology, Melbourne, Australia Matthew S. Weber Rutgers University, New Brunswick, NJ, USA Itzhak Yanovitzky Rutgers University, New Brunswick, NJ, USA; Department of Communication, Rutgers University, New Brunswick, NJ, USA Ji-Young Yun Johns Hopkins University, Baltimore, MD, USA

List of Figures

Disseminating Evidence to Policymakers: Accounting for Audience Heterogeneity Fig. 1 Fig. 2

Trustworthy sources of research, stratified by ideology City policymakers’ perceptions of factors that have very strong effects on health disparities, stratified by ideology

33 37

Integrating Connectionist and Structuralist Social Network Approaches to Understand Education Policy Networks: The Case of the Common Core State Standards and State-Provided Curricular Resources Fig. 1

Sociogram of ELA Resource Providers (Note Circles represent SEAs. White circles represent SEAs that adopted CCSS; black circles represent SEAs that did not adopt CCSS. Gray squares represent intermediary organizations. Node size denotes level of influence. Line thickness denotes strength of tie, and arrows indicate directionality)

79

xxi

xxii Fig. 2

Fig. 3

LIST OF FIGURES

Sociogram of sponsoring organizations to messages about elements of close reading (Note White circles indicate resources’ sponsoring organizations. Black squares indicate specific instructional messages about close reading. Tie strength notes the number of times a particular organization sponsored a resource in the sample expressing a message about close reading. The node size of close reading messages [black squares] indicates the number of resources expressing that message [larger nodes indicate that more resources expressed a particular message about how teachers should enact close reading]) Sociogram of resources to messages about elements of close reading (Note White circles indicate individual resources with messages about close reading. Black squares indicate specific instructional messages about close reading. Tie strength indicates resources that were duplicated in the sample [e.g., 12 of the 31 resources with messages about close reading were the Publisher’s Criteria]. The node size of close reading messages [black squares] indicates the number of resources expressing that message was present [larger nodes indicate that more resources expressed a particular message about how teachers should enact close reading])

85

86

Measuring Issue Preferences, Idea Brokerage, and Research-Use in Policy Networks: A Case Study of the Policy Innovators in Education Network Fig. 1 Fig. 2

Fig. 3

Affiliation network of PIE members and their policy preferences (member node labels suppressed) First two dimensions of correspondence analysis plot of research use behaviors by policy topics (emphasis on policy topics) First two dimensions of correspondence analysis plot of research use behaviors by policy topics (emphasis on research use behaviors)

110

119

120

Broken Bridges: The Role of Brokers in Connecting Educational Leaders Around Research Evidence Fig. 1

Brokerage roles

134

LIST OF FIGURES

Fig. 2 Fig. 3 Fig. 4

Social network maps of the leadership district for research evidence, data use, and expertise in Years 1 and 3 Formal roles bridged by area superintendents as liaisons for research evidence Area superintendents as research evidence liaisons for principals

xxiii

142 145 146

An Ego-Network Approach to Understanding Educator and School Ties to Research: From Basic Statistics to Profiles of Capacity Fig. Fig. Fig. Fig. Fig.

1 2 3 4 5

Overview of survey of evidence in education Example item from network portion of SEE-S Distribution of reported resources Four profile MLPA solution for level 1 (educators) Two-profile MLPA solution for level 2 (schools)

160 161 166 170 172

A Multi-Level Framework for Understanding Knowledge Sharing in Transnational Immigrant Networks Fig. 1

Fig. 2

Distribution of agreement about “American traits” in a transnational network based on cultural consensus analysis The distribution of anxiety and cultural consonance in one ego-net

227 230

Promoting Healthy Eating: A Whole-of-System Approach Leveraging Social Network Brokers Fig. 1 Fig. 2

Key brokerage points to implement change in community food systems From Fig. 1 of McGlashan et al. (2018), “Diagrams of the Shape Up Somerville (SUS) and Romp & Chomp (R&C) steering committee networks,” representing the discussion relationships during the community-based childhood obesity prevention interventions (Key: Blue = respondents from the steering committee, White = non-respondent consenting steering committee members, Gray = non-consenting steering committee members, and Red = other nominated contacts external to the steering committee)

242

254

xxiv Fig. 3

LIST OF FIGURES

From Fig. 1 of McGlashan et al. (2019), “Conceptualization of a steering committee social network overlaid on the causal loop diagram to create a multilevel structure” (Key: The blue nodes and network represent the steering committee collaboration network, with black ties representing members’ actions on risk factors labeled in the causal loop diagram)

255

Brokerage-Centrality Conjugates for Multi-Level Organizational Field Networks: Toward a Blockchain Implementation to Enhance Coordination of Healthcare Delivery Fig. 1

Fig. 2 Fig. 3 Fig. 4 Fig. 5

A typology of social mechanisms to conceptualize organizational brokerage behavior (Note Modification of Fig. 1.1 [Hedström & Swedberg, 1998, p. 11] and Fig. 1 [Hedström & Ylikoski, 2010, p. 23]) Multi-level collaboration network in PrEP care delivery system Multi-level collaboration network with nodal size as indicating within-level degree Multi-level collaboration network with nodal size as indicating bridging degree Hybrid P2P network architecture for PrEP care delivery system

273 278 291 291 304

Platformed Knowledge Brokerage in Education: Power and Possibilities Fig. 1

Circles represent the traditional three-actor triadic brokerage relationship; the square represents a platform to which all actors may have a relationship, creating a two-mode network

322

Network Approaches to Misinformation Evaluation and Correction Fig. 1 Fig. 2 Fig. 3 Fig. 4

The role of social networks at each stage of our interaction with misinformation Political and correction condition in the social correction experiment Perceived accuracy by correction type Perceived accuracy by message type

356 364 365 366

List of Tables

Disseminating Evidence to Policymakers: Accounting for Audience Heterogeneity Table 1 Table 2

Primary sources that policymakers turn to for behavioral health research to inform policy decisions Important attributes of behavioral health research

30 35

“Being Important” or “Knowing the Important”: Who Is Best Placed to Influence Policy? Table Table Table Table

1 2 3 4

Characteristics of network sample Characteristics of authorities Power Hubs Influence Hubs

56 57 59 59

Integrating Connectionist and Structuralist Social Network Approaches to Understand Education Policy Networks: The Case of the Common Core State Standards and State-Provided Curricular Resources Table 1 Table 2 Table 3

SEAs and organizations most commonly named as sponsors of CCSS resources Resource category, type, and emphasis for all resources MRQAP regression model

77 81 90

xxv

xxvi

LIST OF TABLES

Measuring Issue Preferences, Idea Brokerage, and Research-Use in Policy Networks: A Case Study of the Policy Innovators in Education Network Table 1 Table 2 Table 3

Distribution of types of evidence cited in PIE members’ publicly available policy briefs/reports Results of ERGM analysis of PIE network members’ policy preferences PIE members participating in the most 4-cycle brokerage chains

112 114 116

Broken Bridges: The Role of Brokers in Connecting Educational Leaders Around Research Evidence Table 1 Table 2 Table 3

Whole network measures for the research evidence, data use, and expertise networks in Years 1 and 3 Average brokerage role measures for each leadership role group for years 1 and 3 Number of percentage of brokerage roles among area superintendents

140 143 144

An Ego-Network Approach to Understanding Educator and School Ties to Research: From Basic Statistics to Profiles of Capacity Table 1 Table 2 Table 3 Table 4

Multi-level categorization of resources for accessing research-based information ENA size, composition, and heterogeneity statistics Most frequently nominated resources Distribution of profiles across schools

165 167 168 171

A Multi-Level Framework for Understanding Knowledge Sharing in Transnational Immigrant Networks Table 1 Table 2 Table 3 Table 4

Sample demographic characteristics Brazilian immigrant health and the composition of their Ego-networks (n = 30) Dominican immigrant health and the composition of their Ego-networks (n = 28) Comparison of main findings for Brazilian and Dominican Ego-networks

217 218 220 224

LIST OF TABLES

xxvii

Brokerage-Centrality Conjugates for Multi-Level Organizational Field Networks: Toward a Blockchain Implementation to Enhance Coordination of Healthcare Delivery Table 1 Table 2 Table 3

Typology of brokerage-centrality conjugates: (Non-PrEP/PrEP) providers as brokers ERGM specification for multi-level networks RGM results

280 288 293

Platformed Knowledge Brokerage in Education: Power and Possibilities Table Table Table Table Table Table

1 2 3 4 5 6

Brokerage types, Gould and Fernandez (1989) Overview of platform cases Users and brokerage types Summary of knowledge objects Summary of knowledge organization Functions for engaging with objects and other users

318 324 328 330 334 339

Knowledge Brokers, Networks, and the Policymaking Process Matthew S. Weber and Itzhak Yanovitzky

The public policy process is a complicated labyrinth of competing actors, interests, and agendas. Policymaking occurs in an ecosystem where policymakers, advocates, think tanks, the public, journalists, and researchers engage and interact in order to craft new policies. From this perspective, interest in knowledge brokerage as a mechanism for impacting the policymaking processes has grown in recent years. Knowledge brokers are key intermediaries who facilitate the exchange of knowledge between individuals or organizations who do not already have direct relationships or established mechanisms for connecting with one another (Lomas, 2000; Ward et al., 2009). In theory, knowledge brokers are positioned to connect actors, including policymakers and practitioners, and can be particularly influential in the context of acquiring, interpreting, and using evidence to support arguments for or against adopting proposed policies or recommended practices.

M. S. Weber (B) · I. Yanovitzky Rutgers University, New Brunswick, NJ, USA e-mail: [email protected] I. Yanovitzky e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 M. S. Weber and I. Yanovitzky (eds.), Networks, Knowledge Brokers, and the Public Policymaking Process, https://doi.org/10.1007/978-3-030-78755-4_1

1

2

M. S. WEBER AND I. YANOVITZKY

This book is about knowledge brokerage and its potential to impact public policymaking and practice through the lens of social network analysis. In the following chapters, we balance a number of different epistemological perspectives, research domains, and methodological approaches in order to present a holistic perspective of knowledge brokerage. As a focal point, we advocate for a networks perspective in part because of the power of network analysis to unpack both the structural aspects of knowledge and information exchange, as well as the ability of network analysis to capture the social context and social interactions that impact those interactions (Oliver & Faul, 2018). Throughout, we make the case that a networks perspective provides a critical avenue for investigating the conditions and actions that can promote more frequent and informed use of research in policy and practice settings.

Knowledge Brokerage and Use of Research Evidence Knowledge brokers occupy a critical role in bringing research evidence into a policymaking ecosystem. Research evidence (i.e., empirical findings derived from systematic research methods and analyses) has significant potential to improve both public policy and professional practice. It is rarely the only, or even the most important, form of knowledge considered in these settings, yet it is frequently present and routinely invoked when decisions are made and actions are justified. As such, persistent gaps between what research shows to be effective and the actual policies and practices that are adopted and implemented may be due to how research is used in decision-making processes rather than whether research evidence is used at all. The fundamental challenge of improving the use of research evidence (URE) in policy and practice extends beyond the effective translation and transfer of scientific knowledge to promoting an informed URE and facilitating its infusion into decision-making routines. A central conundrum for many policy and practice fields, including the ones represented in this edited volume, is how this may be accomplished. Our point of departure is the growing interest across academic disciplines in creating robust mechanisms for improving knowledge brokerage and URE in policy and practice. Historically, research on this topic was motivated by the “two communities” metaphor or the notion that scientists and policymakers (but also practitioners) occupy separate communities, with distinct languages, values, and reward system, with little or no

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

3

meaningful opportunities to interact with one another (Bogenschneider & Corbett, 2010). Accordingly, considerable investments were made in making relevant research evidence more accessible to policymakers and practitioners as well as in programs and interventions designed to facilitate direct interactions between scientists and policymakers or practitioners. Some of these approaches, such as research-practice partnerships, show significant promise, as current thinking has evolved to recognize that, at least in the policy domain, URE unfolds within a complex web of relationships, settings, and contexts (Tseng, 2012). These complex webs, referred to as policy ecosystems, comprise a space within which actors both inside and outside of government (e.g., policymakers, bureaucrats, advocates and interest groups, think tanks, scientists, journalists, and ordinary citizens) interact in complex and dynamic ways to craft and enact public policies. Consequently, URE in policymaking looks nothing like the orderly, systematic, and calculated process that is often envisioned and/or prescribed in research. One important implication of this is that leveraging existing pathways and mechanisms through which research routinely makes its way into policy decision-making processes may not always lead to an increase in the likelihood that research is utilized. To more fully appreciate the importance of knowledge brokers to promoting evidence-based policymaking, consider the following example taken from a study that examined URE in the context of the formulation of federal policies to address the epidemic of childhood obesity in the United States (see Yanovitzky & Weber, 2020).. Rates of childhood obesity had already reached an alarming level when the U.S. Surgeon General issued the now famous “Call to Action to Prevent and Decrease Overweight and Obesity” in 2000. In the wake of the Surgeon General’s call to action, and following the subsequent push for remedies by a broad coalition of policy actors and interest groups, the period from 2000 to 2014 saw intense legislative efforts in the U.S. Congress which produced policy solutions to the problem, including reforming school nutrition and physical activity guidelines, regulating consumption of sugary snacks and drinks, and mitigating the harmful effects of advertising and marketing of food directly to children (Brescoll et al., 2008). A document analysis of Congressional legislative activity during this period revealed 224 congressional bills, 190 committee hearings and reports, and 372 records of floor debates focused on the issue of childhood obesity-related policies (Yanovitzky & Weber, 2020). These document texts were coded to extract information about the scope, type,

4

M. S. WEBER AND I. YANOVITZKY

and timing of research evidence use in the formulation of these policies, including all sources and suppliers of research evidence. Network analysis was employed next to map and analyze nodes (actors) and ties (relationships) that were highly instrumental in introducing and facilitating URE in this context. For instance, on March 26, 2009, Representative Joe Baca (D-CA) chaired a hearing of the Subcommittee on Department Operations, Oversight Nutrition, and Forestry of the Committee on Agriculture in the House of Representatives. In his prepared opening remarks, and in an effort to quantify the scope and magnitude of the obesity problem, Baca noted, “The problem of obesity plagues all Americans, and I state all Americans, either directly or indirectly. Statistics indicate more than half of our population is considered obese. That, in and of itself—is a shocking number.” The statistic included a reference in the prepared and filed remarks, and cited a study published in Health Affairs (Thorpe et al., 2004). To date, this particular study, which details the relationship between obesity-related illnesses and rising medical costs, has been cited more than 500 times. Baca repeated the same statistics twice in his opening remarks, and three other Congressional representatives who were present went on to reference the same statistic and citation in subsequent hearings that year. Thorpe, who had written the original study, had testified before the U.S. Senate Committee on Health, Education, Labor, and Pensions Hearing in 2008, citing the same piece of research (Prevention and Public Health: The Key to Transforming our Sickcare System, 2008). William Dietz, Director, Division of Nutrition, Physical Activity, and Obesity National Center for Chronic Disease Prevention and Health Promotion, also testified to the Subcommittees on Health, and Oversight and Investigations of the Committee on Energy and Commerce in the United States House of Representatives in December of 2009, again reiterating the same statistic (Innovative Childhood Obesity Practices, 2009). However, a network analysis of legislators connected to this particular piece of research evidence flagged Baca as a knowledge broker, with a critical role regarding drawing other legislators’ attention to this particular statistical fact and ensuring its reuse by leveraging his leadership position and connections to other legislators who are active in this policy space. This specific example illustrates one path by which knowledge brokers are able to routinely influence URE in the policymaking process. Knowledge brokerage is often described as the iterative process of translating,

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

5

synthesizing, disseminating, and exchanging research evidence to inform the decisions and actions of practitioners and policymakers (Lomas, 2000; Ward et al., 2009). Knowledge brokers therefore play several crucial knowledge translation and transfer roles, including knowledge management (providing users with research-based insights tailored to their unique settings and needs), liaison (facilitating direct contacts and collaboration between producers and users of research), and building users’ capacity to access, evaluate, and implement research-based knowledge (Bornbaum et al., 2015). Beyond knowledge transfer, knowledge brokers also have an important role in building relationships and mobilizing for actions that can improve the likelihood that relevant and credible research evidence is considered and discussed in the course of decision-making processes (Yanovitzky & Weber, 2019). Knowledge brokers can help to promote mutual understanding between diverse stakeholders and key decision-makers (Lomas, 2000), and help users navigate barriers to effective URE (Neal et al., 2015). To do this, knowledge brokers must acquire a diverse set of skills and competencies related to evidence gathering, critical assessment of evidence, and mediation (Meyer, 2010; Ward et al., 2009). In addition, a complementary view on knowledge brokerage emphasizes the central network position that knowledge brokers occupy regarding the flow and exchange of information within a network of actors. This conception of knowledge brokerage has a rich history in social network theory (Burt, 2001). Through this lens, brokers are viewed as organizations or individuals who bridge structural holes, defined as a gap between two actors with complementary resources or information. Knowledge brokers “use their in-between vantage points to spot old ideas that can be used in new places, new ways and new combinations” (Hargadon & Sutton, 2000). A more recent conception of knowledge brokerage in this tradition recognizes the value of decoupling a brokerage position from brokering activity (Obstfeld et al., 2014). That is, knowledge brokerage can occur in a wide variety of structural contexts, including closed, dense networks, and that opportunities to broker knowledge are not entirely contingent on structural holes; brokers can (and do) broker knowledge between actors who are already connected (Gould & Fernandez, 1989). Further, opportunities to broker knowledge do not automatically imply motivation to broker knowledge. Even when a given structural pattern provides opportunity for some kind of brokerage, the intent and intensity of brokering will vary as a function of brokers’

6

M. S. WEBER AND I. YANOVITZKY

goals and intentions. Thus, knowledge brokerage is more complex than simply considering an actor’s network position. Knowledge brokers may occupy specific network positions, but brokerage roles are actually fluid and change over time as a function of actions taken by brokers and other actors in the network (Fritsch & Kauffeld-Monz, 2008; Gould & Fernandez, 1989; Obstfeld et al., 2014).

Knowledge Brokerage in Policy and Practice Settings In recent decades, scholars across diverse academic disciplines and fields of practice have turned an eye to knowledge brokerage. Serious efforts to theorize and study knowledge brokerage have emerged in sociology, political science, education, public health, criminal justice, and communication and information sciences, to name a few. A major thrust of this work involves situating knowledge brokerage in the context of dynamic processes, relationships, and routines. In general, policy-focused scholarship seeks to position knowledge brokerage relative to the policy process (agenda-setting, policy formulation, policy implementation, and policy evaluation) and critical points of entry into the process (e.g., policy windows, see Kingdon, 1993). This work stretches across multiple levels of policymaking (local, state, national, and international) and considers a multitude of diverse actors (both inside and outside of government) who are active in the policy ecosystem and their complex relationships in an effort to identify and leverage knowledge brokers. Extending from this body of work, practice-oriented knowledge brokerage scholarship is research that focuses on a primary concern of successfully building knowledge brokers’ capacity, crafting explicit brokerage roles, and infusing knowledge brokerage into existing systems and decision-making routines, particularly those that touch on problems of practice. Practice-oriented work is geared toward the development and testing of knowledge brokering interventions. As a consequence, knowledge brokerage research in the policy domain tends to be relationshipfocused whereas practice-oriented knowledge brokerage research places greater emphasis on the role and functions of knowledge brokers. The complexity of knowledge brokerage as a subject of research is particularly apparent in fields where policy and practice are intertwined. The three we chose to feature here—health, education, and communication—are at the forefront of knowledge brokerage scholarship and

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

7

application, although equally influential work is being done in other fields (e.g., criminal justice, environmental sciences, etc.). Further, we concur with recent calls for a greater cross-disciplinary synergy to clarify the roles, functions, and activities, as well as the processes and mechanisms that can support effective knowledge brokerage (Hering, 2016; Robins et al., 2012). We are further convinced that social network analysis is well equipped to support this effort by providing robust tools for mapping, understanding, and subsequently improving the flow of research evidence into policy, and for understanding and interrogating the broader context of research evidence use.

Knowledge Brokers in Health and Medicine Knowledge brokerage research is well established in the domains of health and medicine. In these domains, research tends to be practice-oriented and knowledge brokers are often defined explicitly as information professionals with strong relation-building skills who work to promote engagement in evidence-informed decision-making (Robeson et al., 2008). The job of the knowledge broker is to become an expert within a given domain, facilitate relationships, and educate others with regards to the importance of particular sets of research (Dobbins et al., 2004, 2009; Straus et al., 2013; Ungar et al., 2015). In health domains, knowledge brokers are well established as actors who translate research into practice in order to advance evidence-based decision-making (Innvaer et al., 2002; Jordan et al., 2009). Within hospitals and other healthcare organizations, knowledge brokerage is often formalized in job roles. Many health institutions have formalized the role of knowledge brokers as a way of facilitating the translation of research evidence into practice. The American Association for the Advancement of Science defines knowledge broker as an emerging profession, and it is even possible to search for knowledge broker jobs on popular job boards such as LinkedIn. Formal brokerage roles are particularly important in medical practice, where it is well established that patient health is directly impacted by the effectiveness of implemented policies (Lavis et al., 2003b). Thus, research implementation scholars focus on knowledge brokers as both individuals (Zook, 2004) and groups or organizations (Hargadon, 1998, 2002; Lavis et al., 2003a). Examples of organizations that serve as knowledge brokers would include advocacy groups and nonprofits that work with

8

M. S. WEBER AND I. YANOVITZKY

partners to bring research into practice settings. Testing the effectiveness of knowledge brokers in practice, Forsetlund and Bjorndal (2002) trained knowledge brokers working in Norwegian healthcare to intervene in medical training by purposefully informing colleagues about new research relevant to care protocols. Their work helped to demonstrate the potential of knowledge brokers to impact and improve health outcomes. Extant research on knowledge brokerage in the context of health policy shows that individuals in knowledge brokerage roles are able to effectively translate research into practice, but are also tasked with enabling socialization to aide in collaborative processes in healthcare practice (Bate & Robert, 2002; Jansson et al., 2010). The healthcare sector has often been characterized by silos of information and tribes or cliques of practitioners. Therefore, knowledge brokers’ contributions often extend beyond research translation to breaking down barriers and facilitating the exchange of information. This means that individuals in knowledge broker roles must overcome personal differences, cultural differences, and other types of conflicts that typically occur in interpersonal relationships (Long et al., 2012, 2013). Part of the process of knowledge brokerage thus involves socialization and building personal connections in order to facilitate knowledge exchange (Long et al., 2012). The emphasis on socialization echoes earlier research on brokerage, which focuses on the importance of a knowledge broker’s capacity for establishing relationships and forging connection with a diverse network of peers (Burt, 2005). Knowledge brokers are responsible for establishing relationships with users, and for insuring that those relationships are trusting and positive in order to facilitate research implementation (Gravois Lee & Garvin, 2003; Roy et al., 2003). In turn, this means that knowledge brokers must be aware of the challenges facing end users, and the environmental conditions that may impact end users’ day-to-day work environment. Knowledge brokers are expected to be experts in multiple domains and to have a well-reasoned strategy for forming connections and enabling knowledge transfer (Lyons et al., 2006). Further, research shows that in order to be effective, knowledge brokers must take strategic action to enhance their ability to be successful in their brokerage activity (Chambers et al., 2010; Conklin et al., 2007). Dobbins et al. (2009) accordingly note that the knowledge broker position is challenging due to the multiple demands that are placed on the position beyond the basic function of research translation.

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

9

Focusing on this challenge, Dobbins et al. (2005, 2009) conducted a series of studies in which she worked with colleagues to implement an intervention in three public health units in Canada. The intervention focused on training individuals to work as knowledge brokers with the goal of improving evidence-informed decision-making. Findings from the study varied, but researchers noted that the communities of interaction between practitioners became more tightly knit when the practice-based purpose-trained knowledge brokers intervened (Yousefi-Nooraie et al., 2014, 2015). In sum, in health policy and medical practice knowledge brokers are tasked with responsibility for establishing relationships and creating conditions that are favorable to research engagement and implementation. The theme of translating research into practice is common across domains. The role of knowledge broker is not as well established in other areas of research, meaning that fields such as education and communication focus more on the practice of brokering knowledge and less on the specific job function of a knowledge broker.

Knowledge Brokers in Education Similar to research in health, knowledge brokers in education occupy a critical role translating research into practice. Unlike health, where there is a recognized need for knowledge brokers as professionals, knowledge brokerage in education is primarily seen as a mechanism for closing the research-practice gap (e.g., Flaspohler et al., 2012; Tseng, 2012). Underscoring the significance of knowledge brokerage in this context, a recent study found that nearly two thirds of educators surveyed could not find a path to engage with research through their existing social networks (Neal et al., 2019). When research is utilized in education policy at the district or school level, it tends to be utilized in a superficial way and knowledge brokerage is seen as one tool for promoting a more meaningful and engaged URE in educational practice (Finnigan et al., 2013). Establishing a mechanism of knowledge brokering that is integrated with educators’ professional routines and practices is seen as having a significant potential to improve URE in decision-making processes that take place within school systems (Brown & Zhang, 2017). Access to research evidence is central to improving outcomes within school districts. In turn, research shows a disconnect exists between

10

M. S. WEBER AND I. YANOVITZKY

district offices and school principals, and suggests that knowledge brokers could help to bridge that gap and improve the URE within districts (Daly, Finnigan, Jordan, et al., 2014). Formally, school principals and superintendents often dictate how research evidence is utilized in schools, for instance by repackaging data into manageable packets of information that align with established expectations within a given district (Coburn et al., 2009). This alignment with established expectations limits the ability of educators to implement research evidence in a way that deviates from expected norms. Knowledge brokers serve a critical role in facilitating collaborations among key actors within an education system, including educators, researchers, and policymakers (Coburn et al., 2013). Common brokerage activities include working to establish research alliances or research-practice partnerships that more formally connect educators and researchers. At a more focused level, educator-targeted strategies such as data coaching have also been used by knowledge brokers to build educators’ capacity to collect and use evidence in practice (Huguet et al., 2014). There is a well documented tendency of educators working at the classroom level to seek knowledge via outside channels (Daly, Finnigan, Moolenaar, et al., 2014). This means that rather than seeking knowledge or research evidence from traditional channels within a school or district, educators often engage informal networks of peers or peer organizations to this end. Within schools, these informal structures and relationships that enable the flow and exchange of evidence are often overlooked despite their clear importance in informing URE in practice (FarleyRipple & Buttram, 2014). This finding is echoed in further studies that show the conception of evidence use in research does not fit comfortably with actual practice in schools (Farley-Ripple & Cho, 2014; Finnigan et al., 2013), leading to the use of more informal channels. When an educator has a connection with someone who is in a structural position that enables them to facilitate research exchange, there is a significant increase in the likelihood that an educator will be able to engage with a researcher (Neal et al., 2019); that structural position does not necessarily equate to a formal position within the education system. A broad body of scholarship points to a need for improved knowledge brokerage activity within education. There is also evidence that success in knowledge brokering activities can be beneficial for educators. For

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

11

example, research shows that when members of a school’s staff seek information about school programs and practices and are successful, the path of knowledge transfer aligns with what is traditionally known as knowledge brokerage (Neal et al., 2015). In other words, the notion of a key actor translating research between otherwise disconnected parties holds weight in the education sector, but an investment of time and resources is needed to make that process more prevalent. The concept of knowledge brokering is critical to future gains in education policy, although the practice of the knowledge brokerage role in policymaking clearly differs from what has been seen in health and medicine.

Knowledge Brokers in Communication In recent years, research on knowledge brokerage has shifted to focus on the communicative role of brokers in effectively communicating scientific research findings to external audiences. Communication research on this topic shifts the focus of inquiry from the structure of the network to the nature of the knowledge brokerage activity, as well as the context of knowledge brokerage. This means that the role of a knowledge broker is increasingly complex, and ought to be defined both in terms of what brokers do as well as the broker’s structural position (Ward et al., 2009). From a communication perspective, knowledge brokers can thus be viewed as occupying a continuum of roles, ranging from passive to active forms of brokerage. Much of the extant work on communication and knowledge brokerage has focused on science communication and media. To that end, scholars in science communication, have looked at how individuals help advance scientific research by strategically facilitating the exchange of knowledge (Meyer, 2010), noting that the strategy and role evolves based on the political context and the research evidence. In related work focused on understanding the connection between journalists and research evidence, scholars demonstrated that journalists have the capacity to serve as knowledge brokers based on the context of their job role (Nisbet & Fahy, 2015). That work, conducted in the context of climate change research, showed how journalists navigate between political biases and audience demands to communicate complex research evidence in digestible components. Indeed, knowledge brokers who work to communicate the meaning of research, translating scientific knowledge into practice, do not simply

12

M. S. WEBER AND I. YANOVITZKY

broker an interaction between two actors. Rather, knowledge brokers add meaning and engage with both the senders and receivers of communicated knowledge (Meyer, 2010). While the broker may add meaning, the way in which that meaning is conveyed also matters. The context and agency of a knowledge broker is central to the role, as is the way in which knowledge is communicated. Returning to the media, professionals such as journalists walk a fine line in translating available research evidence into language that is accessible to public audiences and policymakers. The translation of research evidence into practice matters, and the framing of the issue has a significant impact on how audiences come to understand key issues (Nisbet, 2009). In recent work looking at the complexity of knowledge brokers, scholars theorized that the role of knowledge brokerage is more nuanced than a binary of either being a knowledge broker or not. Rather, Yanovitzky and Weber (2019) illustrate that knowledge brokers create awareness of knowledge, help to make knowledge accessible, create engagement with knowledge, link disconnected parties, and mobilize knowledge into action. Each of these steps of knowledge brokerage involves an effect on the flow of knowledge between parties, as well as an effect on outcomes including policy actions. Moreover, this approach places the nature of the communication at the center of the knowledge brokerage process.

Knowledge Brokerage and Social Network Analysis The intersection of social network analysis and knowledge brokerage in policymaking is an emerging area of interest to many fields of practice and there is good reason to believe that the study of knowledge brokerage would benefit from opportunities to interface with research conducted across disciplinary boundaries. For example, in recent years, massive public investments in translational health research resulted in this field making significant strides regarding the design, implementation, and evaluation of systems and strategies for facilitating evidence-based policymaking (Kuo et al., 2015). In addition, new valuable insights concerning the flow and exchange of knowledge and research evidence through networks of individuals, groups, and organizations has emerged from research conducted within the fields of communication and information

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

13

(Sullivan et al., 2013), social work (Palinkas et al., 2011), and education (Daly, Finnigan, Jordan, et al., 2014), and psychology (Neal et al., 2015). Across domains, from health to education to policy to communication, there is a common focus on the role of the knowledge broker in connecting actors and facilitating the use of knowledge. Recent work in education and health emphasizes the complex nature of knowledge brokerage, and complementary to this communication scholarship emphasizes both the context and content of brokerage activity. Research on knowledge brokerage consistently points to the importance of understanding both the structure and nature of brokerage activity. Social network analysis provides a lens for delving deeper into myriad nuances of networks that bring together policy advocates and practitioners in their day-to-day efforts to broker evidence into policymaking processes. Further, this approach provides a common thread for understanding the challenges of policymaking and knowledge brokerage across domains. Scholars studying policymaking and associated practice-based activity recognize that social network analysis is a key method and theoretical approach for examining and improving policymaking processes (Rhodes, 2008). A network perspective on knowledge brokerage is advantageous because it bridges disciplinary divides and places the emphasis on the activity of knowledge brokerage as opposed to the specific role (Neal et al., 2015). There is an increasing awareness that knowledge brokering is performed by media and organizations, as well as by individuals such as policymakers—and social network analysis provides a common approach to understanding these different levels of brokerage. Social network analysis provides a means to mapping and analyzing the flow of knowledge brokerage and research evidence (Contandriopoulos et al., 2010; Lavis et al., 2003b). Recent work has sought to advance this approach, focusing on descriptive analyses of the factors that influence the movement of research evidence through networks of policy actors (Shearer et al., 2014). Further, as the methodological underpinnings of social network analysis have advanced, there have been numerous calls for the application of new methods in network analysis to the study of policymaking processes (Lubell et al., 2012; Robins et al., 2012). While the type, scope, and nature of the evidence engaged by policymakers and practitioners vary across fields, the process by which evidence is acquired, interpreted, and brokered, and the key challenges experienced, are likely similar (Best & Holmes, 2010). When focusing on social network analysis, a common thread exists in the study of knowledge

14

M. S. WEBER AND I. YANOVITZKY

brokerage and brokering in networks that can be applied across domains, giving researchers a common language for understanding, comparing, and improving the use of research evidence. Simultaneously, recent thinking on knowledge brokering and URE in policymaking processes recognizes that a linear decision-making approach to understanding the ways in which policymakers engage with research and other forms of knowledge may fail to fully account for the complex and fluid nature of URE in these settings (Gough & Boaz, 2017). Present work consistently shows that the use of evidence in knowledge brokering processes remains limited (Coburn & Turner, 2012; Farley-Ripple, 2012; Honig & Coburn, 2007), and while some work has focused on the ways in which policymakers engage with knowledge and research, the flow of evidence into policymaking remains understudied. Thus, as noted, there is also a need to examine the flow of research and knowledge through networks of knowledge brokers in order to understand the process by which research and knowledge move from knowledge to action (Neal et al., 2015; Yanovitzky & Weber, 2019).

Explicit Network Measurement of Brokerage Activity In the domain of social network analysis, there is a long tradition of studying brokerage, but it is only recently that this work has been brought into the domains of education, health, and communication to explicitly examine processes of knowledge brokerage. Network scholars have for decades considered the role of key actors in bridging gaps that exist between distinct groups. Brokerage analysis, for instance, focuses on using network metrics to look at the different roles that an individual might occupy—from bridge to consultant to gatekeeper. Other measures have been developed to capture influencers, or to measure the degree to which an individual engages with others. But the pure networks measure of brokerage explicitly captures the degree to which an individual serves as the shortest connection between two otherwise disconnected groups. Early work at the intersection of social network analysis and policy research shows clear connections between core social network concepts and strategic acts of knowledge brokerage (Fritsch & Kauffeld-Monz, 2008; Scott & Hofmeyer, 2007; Waring et al., 2013). There are well established measures within social network analysis that provide a starting point for examining knowledge brokerage activities. Consider,

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

15

for instance, basic descriptive network measures, concepts such as structural holes, and higher order concepts such as exponential random graph modeling. Traditional network measures. At a fundamental level, basic network measures of centrality, degree centrality, and density help to define the characteristics of the network. In addition, these core measures help to explain the potential for brokerage activity to occur (Zack, 2000). Even at a descriptive level, network measures of structure are thus useful in assessing both the role of individual actors and the nature of the overall network. Structural holes. Moving beyond basic network measures, structural holes are characteristics of the network that derive their value-potential through the separation of non-redundant sources of information (Burt, 1992). Actors in brokerage roles bridge these holes, deriving power through control of access to alternative opinions in the network, early access to new opinions and methods of practice, and the ability to move ideas between groups (Burt, 2005). The greater the proliferation of structural holes within networks, the greater the ability of organizational actors is to negotiate the transfer of information and benefit from the control over information movement (Burt, 1992). Structural holes and brokerage opportunities can be precisely measured in order to understand brokerage activity that either has occurred or is likely to occur in a network. Exponential random graph modeling. Advances over the past decade in the field through exponential random graph modeling and associated approach have further advanced the modeling of brokerage as a complex phenomenon (e.g., Robins et al., 2012). Graph models provide a mechanism for understanding the probabilistic likelihood that brokerage activity will occur in a given network configuration. In sum, there are well established methods within network analysis that provide a baseline for studying knowledge brokerage.

Social Network Analysis, Knowledge Brokerage, and Research Evidence The chapters included in this volume establish clear processes for implementing network typologies, network terminology, and knowledge brokerage in policymaking. In addition, they permit the comparison, assessment, and delineation of social network approaches to knowledge brokerage in a variety of contexts, and provide a suite of useful

16

M. S. WEBER AND I. YANOVITZKY

social network analysis tools for collecting, mapping, and analyzing knowledge brokerage. Finally, the chapters that follow alert scholars to important research design and measurement considerations for guiding rigorous, theory-informed evaluations of knowledge brokerage from a social network analysis approach. Following this introduction, Purtle’s chapter (Chapter “Disseminating Evidence to Policymakers: Accounting for Audience Heterogeneity ”) aims to situate knowledge brokerage in the context of broader efforts to disseminate research evidence to policymakers. There are already a variety of strategies and approaches that are routinely used for dissemination, some of which are more effective than others, and this chapter highlights the tailoring of strategies to match audience heterogeneity as key for effective dissemination efforts. Oliver (Chapter ““Being Important” or “Knowing the Important”: Who is Best Placed to Influence Policy?”) next connects knowledge brokerage to this larger effort by introducing a hub-authority model of knowledge brokerage that is specific to policy and sensitive to the role that power plays in knowledge brokerage. By calling specific attention to the role of power Oliver helps to expand our understanding of the link between knowledge brokering and URE in the policymaking context. Following these opening chapters, the discussion moves to consider the application and context of knowledge brokering activity. Focusing broadly on education policy, Hodge, Benko, and Salloum (Chapter “Integrating Connectionist and Structuralist Social Network Approaches to Understand Education Policy Networks: The Case of the Common Core State Standards and State-Provided Curricular Resources”) examine knowledge brokerage in the context of implementing common core policies and demonstrate the utility of document analysis when paired with social network analysis. Continuing the theme of policy formation and policy networks, Farrare and colleagues (Chapter “Measuring Issue Preferences, Idea Brokerage, and Research-Use in Policy Networks: A Case Study of the Policy Innovators in Education Network”) home in on URE in education policymaking networks, and introduce the reader to two-mode network analysis. Their chapter examines an important case study of policy advocates engaged in contemporary education reform. Continuing the discussion of knowledge brokerage in the context of education reform, Finnegan and colleagues (Chapter “Broken Bridges: The Role of Brokers in Connecting Educational Leaders around Research Evidence”) focus on the specific types of brokerage roles occupied by different actors,

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

17

including a discussion of the role of educational leaders as influential knowledge brokers. Moving to consider knowledge brokerage in the context of education practice, Farley-Ripple and Yun (Chapter “An Ego-Network Approach to Understanding Educator and School Ties to Research: From Basic Statistics to Profiles of Capacity”) explores school-based decision-making networks using ego-network analysis to detect knowledge brokerage of research. Their work uses large-scale surveys to examine ego-networks of research evidence engagement. Next, Kornbluh (Chapter “Mixing Network Analysis and Qualitative Approaches in Educational Practices”) examines the dissemination of evidence-based practices among students engaged in youth-led participatory action research. The chapter also advances methodology, presenting an approach for integrating social network analysis and qualitative research in order to more completely understand brokerage processes in the context of research-practice partnerships. Negron and colleagues (Chapter “A Multi-Level Framework for Understanding Knowledge Sharing in Transnational Immigrant Networks”) subsequently broaden the lens on network analysis of knowledge brokerage through a rich and in-depth ethnographic examination of policymaking and healthcare networks in immigrant communities. Their work draws on a network perspective, but introduces the reader to qualitative approaches to network analysis. Shifting to knowledge brokerage in health and medicine contexts, de la Haye and colleagues (Chapter “Promoting healthy eating: A whole-of system approach leveraging social network brokers”) unpack efforts to promote healthy eating through a whole network analysis and systems approach to understanding policymaking efforts. Focusing on medical interventions, Fujimoto and colleagues (Chapter “Brokerage-Centrality Conjugates for Multi-Level Organizational Field Networks: Toward a Blockchain Implementation to Enhance Coordination of Healthcare Delivery”) go on to introduce the reader to exponential graph modeling and uses data from their work on HIV prevention to examine implementation sciences practices through the lens of social network analysis. Their work contributes to the emerging field of health systems science in the domain of population health and social determinants of health by providing a network paradigm for organizational research. Further, they introduce the idea of blockchain and envision knowledge brokerage in power dynamics of the future in the era of digital transformation. Lawlor and colleagues (Chapter “Platformed Knowledge Brokerage in

18

M. S. WEBER AND I. YANOVITZKY

Education: Power and Possibilities ”) expand the focus on knowledge brokerage to include the role of technological platforms. Their work expands the definition of knowledge brokers to non-human agents and analyzes the role of knowledge repositories in facilitating the brokerage of and dissemination of key information relevant to policymaking processes. In closing the book, Ognyanova expands the lens of discussion in this volume to focus on the increasing role of misinformation in information ecosystems. Ognyanova’s work (Chapter “Network Approaches to Misinformation Evaluation and Correction”) discusses mechanisms of social network analysis as a means of understanding and potentially correcting for the prominence of misinformation in a given network. Finally, Yanovitzky and Weber (Chapter “Closing the Theory-Research Gap in Knowledge Brokerage: Remaining Challenges and Emerging Opportunities”) close the book with a look towards the future of knowledge brokerage research in the context of URE. The final chapter considers specific opportunities to advance social network analysis in this domain, including the development of knowledge brokerage interventions to test specific mechanisms and effects. In sum, this edited volume engages with critical issues of social network analysis and knowledge brokerage in policymaking contexts, focusing on the ways in which knowledge and research are utilized, as well as how research informs policy and practice across domains, including communication, health, and education, although this work is equally relevant to other fields with central interest in the role of research evidence in decision-making processes. Collectively, the contributions included in this volume represent a serious effort to uncover the factors and conditions that enable effective brokering of research into policy and practice through the systematic and rigorous application of social network analysis. As such, they advance theory and research on knowledge brokerage in significant ways, including advancing methodology, and also have significant potential to inform the development, implementation, and evaluation of effective knowledge brokerage interventions to facilitate URE in policy and practice.

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

19

References Bate, S. P., & Robert, G. (2002). Knowledge management and communities of practice in the private sector: Lessons for modernizing the National Health Service in England and Wales. Public Administration, 80. https://doi.org/ 10.1111/1467-9299.00322. Best, A., & Holmes, B. (2010). Systems thinking, knowledge and action: Towards better models and methods. Evidence & Policy: A Journal of Research, Debate and Practice, 6(2), 145–159. https://doi.org/10.1332/174 426410X502284. Bogenschneider, K., & Corbett, T. J. (2010). Family policy: Becoming a field of inquiry and subfield of social policy. Journal of Marriage and Family, 72(3), 783–803. https://doi.org/10.1111/j.1741-3737.2010.00730.x. Bornbaum, C. C., Kornas, K., Peirson, L., & Rosella, L. C. (2015, 2015/11/20). Exploring the function and effectiveness of knowledge brokers as facilitators of knowledge translation in health-related settings: A systematic review and thematic analysis. Implementation Science, 10(1), 162. https:// doi.org/10.1186/s13012-015-0351-9. Brescoll, V. L., Kersh, R., & Brownell, K. D. (2008). Assessing the feasibility and impact of federal childhood obesity policies. The Annals of the American Academy of Political and Social Science, 615(1), 178–194. Brown, C., & Zhang, D. (2017). How can school leaders establish evidenceinformed schools: An analysis of the effectiveness of potential school policy levers. Educational Management Administration & Leadership, 45(3), 382– 401. Burt, R. S. (1992). Structural holes: The social structure of competition. Harvard University Press. Burt, R. S. (2001). Structural holes versus network closure as social capital. In N. Lin, K. Cook, & R. S. Burt (Eds.), Social capital: Theory and research (pp. 31–56). Routledge. Burt, R. S. (2005). Brokerage and closure: An introduction to social capital. Oxford University Press. Chambers, L. W., Luesby, D., Brookman, C., Harris, M., & Lusk, E. (2010). The seniors health research transfer network knowledge network model: system-wide implementation for health and healthcare of seniors. Healthcare Management Forum, 23. https://doi.org/10.1016/j.hcmf.2010.01.001. Coburn, C. E., Honig, M. I., & Stein, M. K. (2009). What’s the evidence on districts’ use of evidence. In J. D. Bransford, D. J. Stipek, N. J. Vye, L. M. Gomez, & D. Lam (Eds.), The role of research in educational improvement (pp. 67–86). Harvard Education Press. Coburn, C. E., Penuel, W. R., & Geil, K. E. (2013). Practice partnerships: A strategy for leveraging research for educational improvement in school districts. William T. Grant Foundation.

20

M. S. WEBER AND I. YANOVITZKY

Coburn, C. E., & Turner, E. O. (2012, Feburary 1). The practice of data use: An introduction. American Journal of Education, 118(2), 99–111. https:// doi.org/10.1086/663272. Conklin, J., Stolee, P., Luesby, D., Sharratt, M. T., & Chambers, L. W. (2007). Enhancing service delivery capacity through knowledge exchange: The Seniors Health Research Transfer Network. Healthcare Management Forum, 20. https://doi.org/10.1016/s0840-4704(10)60087-7. Contandriopoulos, D., Lemire, M., Denis, J.-L., & Tremblay, É. (2010, December 1). Knowledge exchange processes in organizations and policy arenas: A narrative systematic review of the literature. The Milbank Quarterly, 88(4), 444–483. https://doi.org/10.1111/j.1468-0009.2010.00608.x. Daly, A. J., Finnigan, K. S., Jordan, S., Moolenaar, N. M., & Che, J. (2014, March 1). Misalignment and perverse incentives: Examining the politics of district leaders as brokers in the use of research evidence. Educational Policy, 28(2), 145–174. https://doi.org/10.1177/0895904813513149. Daly, A. J., Finnigan, K. S., Moolenaar, N. M., & Che, J. (2014). The critical role of brokers in the access and use of evidence at the school and district level. In K. S. Finnigan & A. J. Daly (Eds.), Using research evidence in education: From the schoolhouse door to Capitol Hill (pp. 13–31). Springer International Publishing. https://doi.org/10.1007/978-3-319-04690-7_3. Dobbins, M., Davies, B., Danseco, E., Edwards, N., & Virani, T. (2005). Changing nursing practice: Evaluating the usefulness of a best-practice guideline implementation toolkit. Nurse Leadership (Tor Ont), 18. https://doi.org/ 10.12927/cjnl.2005.17034. Dobbins, M., DeCorby, K., & Twiddy, T. (2004). A knowledge transfer strategy for public health decision makers. Worldviews Evid Based Nurs, 1. https:// doi.org/10.1111/j.1741-6787.2004.t01-1-04009.x. Dobbins, M., Robeson, P., Ciliska, D., Hanna, S., Cameron, R., O’Mara, L., DeCorby, K., & Mercer, S. (2009). A description of a knowledge broker role implemented as part of a randomized controlled trial evaluating three knowledge translation strategies [journal article]. Implementation Science, 4(1), 23. https://doi.org/10.1186/1748-5908-4-23. Farley-Ripple, E., & Cho, V. (2014). Depth of use: How district decision-makers did and did not engage with evidence. In A. Bowers, A. Shobo, & B. Barnett (Eds.), Using data in schools to inform leadership and decision making (pp. 39– 66). Information Age Publishing. Farley-Ripple, E. N. (2012). Research use in school district central office decision making: A case study. Educational Management Administration & Leadership, 40(6), 786–806. https://doi.org/10.1177/1741143212456912. Farley-Ripple, E. N., & Buttram, J. L. (2014). Developing collaborative data use through professional learning communities: Early lessons from Delaware. Studies in Educational Evaluation, 42, 41–53. https://doi.org/10.1016/j. stueduc.2013.09.006.

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

21

Finnigan, K. S., Daly, A. J., & Che, J. (2013). Systemwide reform in districts under pressure: The role of social networks in defining, acquiring, using, and diffusing research evidence. Journal of Educational Administration, 51(4), 476–497. https://doi.org/10.1108/09578231311325668. Flaspohler, P. D., Meehan, C., Maras, M. A., & Keller, K. E. (2012). Ready, willing, and able: Developing a support system to promote implementation of school-based prevention programs. American Journal of Community Psychology, 50(3–4), 428–444. Forsetlund, L., & Bjorndal, A. (2002). Identifying barriers to the use of research faced by public health physicians in Norway and developing an intervention to reduce them. Journal of Health Services Research & Policy, 7 . https://doi. org/10.1258/1355819021927629. Fritsch, M., & Kauffeld-Monz, M. (2008). The impact of network structure on knowledge transfer: An application of social network analysis in the context of regional innovation networks. The Annals of Regional Science, 44(1), 21. https://doi.org/10.1007/s00168-008-0245-8. Gough, D., & Boaz, A. (2017). Politics and practices of knowledge production and use. Evidence & Policy: A Journal of Research, Debate and Practice, 13(3), 397–400. https://doi.org/10.1332/174426417X15008911642971. Gould, R., & Fernandez, R. (1989). Structures of mediation: A formal approach to brokerage in transaction networks. Sociological Methodology, 19, 89–126. https://doi.org/10.2307/270949. Gravois Lee, R., & Garvin, T. (2003). Moving from information transfer to information exchange in health and health care. Social Science & Medicine, 56. https://doi.org/10.1016/s0277-9536(02)00045-x. Hargadon, A. (1998). Firms as knowledge brokers: Lessons in pursuing continuous innovation. California Management Review, 40. https://doi.org/10. 2307/41165951. Hargadon, A. (2002). Brokering knowledge: Linking learning and innovation. Research in Organizational behavior, 24. https://doi.org/10.1016/s01913085(02)24003-4. Hargadon, A., & Sutton, R. I. (2000). Building an innovation factory. Harvard Business Review, 78(3), 157–157. Hering, J. G. (2016, March 1). Do we need “more research” or better implementation through knowledge brokering? Sustainability Science, 11(2), 363–369. https://doi.org/10.1007/s11625-015-0314-8. Honig, M. I., & Coburn, C. (2007, July 1). Evidence-based decision making in school district central offices: Toward a policy and research agenda. Educational Policy, 22(4), 578–608. https://doi.org/10.1177/089590480 7307067.

22

M. S. WEBER AND I. YANOVITZKY

Huguet, A., Marsh, J. A., & Farrell, C. (2014). Building teachers’ data-use capacity: Insights from strong and developing coaches. Education Policy Analysis Archives, 22. https://doi.org/10.14507/epaa.v22n52.2014. Innovative Childhood Obesity Practices, Uniteds States House of Representatives. (2009). (110th). Innvaer, S., Vist, G., Trommald, M., & Oxman, A. (2002). Health policymakers’ perceptions of their use of evidence: a systematic review. Journal of Health Services Research & Policy, 7 . https://doi.org/10.1258/135581902 320432778. Jansson, S. M., Benoit, C., Casey, L., Phillips, R., & Burns, D. (2010). In for the long haul: Knowledge translation between academic and nonprofit organizations. Qualitative Health Research, 20. https://doi.org/10.1177/104973 2309349808. Jordan, M. E., Lanham, H. J., Crabtree, B. F., Nutting, P. A., Miller, W. L., Stange, K. C., & McDaniel, R. R. (2009). The role of conversation in health care interventions: Enabling sensemaking and learning. Implementation Science, 4. https://doi.org/10.1186/1748-5908-4-15. Kingdon, J. W. (1993). How do issues get on public policy agendas. Sociology and the Public Agenda, 8(1), 40–53. Kuo, T., Gase, L. N., Inkelas, M., The Population, H., & Policy, W. (2015, December 1). Dissemination, implementation, and improvement science research in population health: Opportunities for public health and CTSAs. Clinical and Translational Science, 8(6), 807–813. https://doi.org/10. 1111/cts.12313. Lavis, J., Robertson, D., Woodside, J., McLeod, C., & Abelson, J. (2003a). How can research organizations more effectively transfer research knowledge to decision makers? The Milbank Quarterly, 81. https://doi.org/10.1111/ 1468-0009.t01-1-00052. Lavis, J., Robertson, D., Woodside, J. M., McLeod, C. B., & Abelson, J. (2003b, June 1). How can research organizations more effectively transfer research knowledge to decision makers? The Milbank Quarterly, 81(2), 221–248. https://doi.org/10.1111/1468-0009.t01-1-00052. Lomas, J. (2000). Essay: Using ‘linkage and exchange’ to move research into policy at a canadian foundation. Health Affairs, 19(3), 236–240. https:// doi.org/10.1377/hlthaff.19.3.236. Long, J. C., Cunningham, F. C., & Braithwaite, J. (2012). Network structure and the role of key players in a translational cancer research network: A study protocol. BMJ Open, 2. https://doi.org/10.1136/bmjopen-2012-001434. Long, J. C., Cunningham, F. C., & Braithwaite, J. (2013). Bridges, brokers and boundary spanners in collaborative networks: A systematic review [journal article]. BMC Health Services Research, 13(1), 158. https://doi.org/10. 1186/1472-6963-13-158.

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

23

Lubell, M., Scholz, J., Berardo, R., & Robins, G. (2012, August 8). Testing policy theory with statistical models of networks. Policy Studies Journal, 40(3), 351–374. https://doi.org/10.1111/j.1541-0072.2012.00457.x. Lyons, R., Warner, G., Langille, L., & Phillips, S. J. (2006). Piloting knowledge brokers to promote integrated stroke care in Atlantic Canada. In Evidence in action, acting on evidence: A casebook of health services and policy research knowledge translation stories. Meyer, M. (2010). The rise of the knowledge broker. Science Communication, 32(1), 118–127. Neal, J. W., Neal, Z. P., Kornbluh, M., Mills, K. J., & Lawlor, J. A. (2015, December 1). Brokering the research–practice gap: A typology. American Journal of Community Psychology, 56(3), 422–435. https://doi.org/10. 1007/s10464-015-9745-8. Neal, J. W., Neal, Z. P., Mills, K. J., Lawlor, J. A., & McAlindon, K. (2019, October 1). What types of brokerage bridge the research-practice gap? The case of public school educators. Social Networks, 59, 41–49. https://doi.org/ 10.1016/j.socnet.2019.05.006. Nisbet, M. C. (2009, March 1). Communicating climate change: Why frames matter for public engagement. Environment: Science and Policy for Sustainable Development, 51(2), 12–23. https://doi.org/10.3200/ENVT.51.2.12-23. Nisbet, M. C., & Fahy, D. (2015, March 1). The need for knowledge-based journalism in politicized science debates. The Annals of the American Academy of Political and Social Science, 658(1), 223–234. https://doi.org/10.1177/ 0002716214559887. Obstfeld, D., Borgatti, S. P., & Davis, J. (2014). Brokerage as a process: Decoupling third party action from social network structure. In Contemporary perspectives on organizational social networks (Vol. 40, pp. 135–159). Emerald Group Publishing Limited. https://doi.org/10.1108/S0733-558 X(2014)0000040007. Ogden, C. L., Carroll, M. D., Kit, B. K., & Flegal, K. M. (2014). Prevalence of childhood and adult obesity in the United States, 2011–2012. JAMA, 311(8), 806–814. Oliver, K., & Faul, M. V. (2018). Networks and network analysis in evidence, policy and practice. Evidence & Policy: A Journal of Research, Debate and Practice, 14(3), 369–379. https://doi.org/10.1332/174426418X15314037 224597. Palinkas, L. A., Holloway, I. W., Rice, E., Fuentes, D., Wu, Q., & Chamberlain, P. (2011, September 9). Social networks and implementation of evidence-based practices in public youth-serving systems: A mixed-methods study. Implementation Science, 6(1), 113. https://doi.org/10.1186/17485908-6-113.

24

M. S. WEBER AND I. YANOVITZKY

Prevention and Public Health: The Key to Transforming our Sickcare System, Unites States Senate. (2008). (110th Congress). Rhodes, R. A. W. (2008). Policy network analysis. In R. E. Goodin, M. Moran, & M. Rein (Eds.), The Oxford handbook of public policy (pp. 425–447). Oxford University Press. Robeson, P., Dobbins, M., & DeCorby, K. (2008). Life as a knowledge broker in public health. Journal of the Canadian Health Libraries Association, 29. https://doi.org/10.5596/c08-025. Robins, G., Lewis, J. M., & Wang, P. (2012). Statistical network analysis for analyzing policy networks. Policy Studies Journal, 40(3), 375–401. Roy, M., Parent, R., & Desmarais, L. (2003). Knowledge networking: A strategy to improve workplace health and safety knowledge transfer. Electronic Journal on Knowledge Management, 1. Scott, C., & Hofmeyer, A. (2007). Networks and social capital: A relational approach to primary healthcare reform. Health Research Policy and Systems, 5(1), 9. https://doi.org/10.1186/1478-4505-5-9. Shearer, J. C., Dion, M., & Lavis, J. (2014, October 30). Exchanging and using research evidence in health policy networks: A statistical network analysis. Implementation Science, 9(1), 126. https://doi.org/10.1186/s13012-0140126-8. Straus, S., Tetroe, J., & Graham, I. D. (2013). Knowledge translation in health care: Moving from evidence to practice. Wiley. Sullivan, S., Pierce, C. S., Leonardi, P. M., & Contractor, N. (2013, January 1). Explaining idea sharing mechanisms: Linking diversity and network factors to explore creative teams. Academy of Management Proceedings, 2013(1), 16068. https://doi.org/10.5465/ambpp.2013.16068abstract. Thorpe, K. E., Florence, C. S., Howard, D. H., & Joski, P. (2004, January 1). The impact of obesity on rising medical spending. Health Affairs, 23(Suppl1), W4–480-W484–486. https://doi.org/10.1377/hlthaff.W4.480. Tseng, V. (2012). The uses of research in policy and practice. Society for Research in Child Development Washington. Ungar, M., McGrath, P., Black, D., Sketris, I., Whitman, S., & Liebenberg, L. (2015, January 12). Contribution of participatory action research to knowledge mobilization in mental health services for children and families. Qualitative Social Work. https://doi.org/10.1177/1473325014566842. Ward, V., House, A., & Hamer, S. (2009). Knowledge brokering: The missing link in the evidence to action chain? Evidence & Policy: A Journal of Research, Debate and Practice, 5(3), 267–279. Waring, J., Currie, G., Crompton, A., & Bishop, S. (2013). An exploratory study of knowledge brokering in hospital settings: Facilitating knowledge sharing and learning for patient safety? Social Science & Medicine, 98, 79–86. https:// doi.org/10.1016/j.socscimed.2013.08.037.

KNOWLEDGE BROKERS, NETWORKS, AND THE POLICYMAKING PROCESS

25

Yanovitzky, I., & Weber, M. (2020). Analysing use of evidence in public policymaking processes: A theory-grounded content analysis methodology. Evidence & Policy: A Journal of Research, Debate and Practice, 16(1), 65–82. https:// doi.org/10.1332/174426418X15378680726175. Yanovitzky, I., & Weber, M. S. (2019). News media as knowledge brokers in public policymaking processes. Communication Theory, 29(2), 191–212. https://doi.org/10.1093/ct/qty023. Yousefi-Nooraie, R., Dobbins, M., & Marin, A. (2014). Social and organizational factors affecting implementation of evidence-informed practice in a public health department in Ontario: A network modelling approach. Implement Science, 9. https://doi.org/10.1186/1748-5908-9-29. Yousefi-Nooraie, R., Dobbins, M., Marin, A., Hanneman, R., & Lohfeld, L. (2015). The evolution of social networks through the implementation of evidence-informed decision-making interventions: A longitudinal analysis of three public health units in Canada [journal article]. Implementation Science, 10(1), 166. https://doi.org/10.1186/s13012-015-0355-5. Zack, M. H. (2000). Researching organizational systems using social network analysis. In Proceedings of the 33rd Annual Hawaii International Conference on System Sciences. Zook, M. A. (2004). The knowledge brokers: Venture capitalists, tacit knowledge and regional development. International Journal of Urban and Regional Research, 28. https://doi.org/10.1111/j.0309-1317.2004.00540.x.

Disseminating Evidence to Policymakers: Accounting for Audience Heterogeneity Jonathan Purtle

In June 1982, House Bill 608 was going to the Illinois State Senate for a vote. The bill required that children age four and younger wear seatbelts, and was based on a rapidly growing body of research evidence. One week before the Senate vote, two psychologists at DePaul University—Leonard Jason and Thomas Rose—randomly selected half of the State Senators and mailed them an evidence summary about child motor vehicle safety (Jason & Rose, 1984). They did not mail anything to the other half of Senators. When the bill came to a vote, the Senators who were mailed the evidence summary were significantly more likely to vote “yea” on the bill than Senators who were not. Evidence was sent to policymakers, an evidence-based policy was passed as a result, and lives of children in Illinois were saved. This case study casts the potential of disseminating evidence to policymakers in a resplendent light. But a lot has changed since 1982. Today, policymakers operate in the midst of a digital deluge of credible information, misinformation, and disinformation—and it is increasingly challenging to discern one from the

J. Purtle (B) Dornsife School of Public Health, Drexel University, Philadelphia, PA, USA e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 M. S. Weber and I. Yanovitzky (eds.), Networks, Knowledge Brokers, and the Public Policymaking Process, https://doi.org/10.1007/978-3-030-78755-4_2

27

28

J. PURTLE

others (Iyengar & Massey, 2019; Scheufele & Krause, 2019). As such, it seems naïve to think that the simple provision of evidence could affect a policymaker’s behavior (Cairney & Oliver, 2017; Oliver et al., 2014). However, I believe that the dissemination of research evidence can still play an influential role in policymaking processes. Dissemination is by no means a panacea to the challenges of evidence-informed policymaking, but it can have an impact if dissemination activities are informed by theory and data and if the right messenger delivers the right message to the right policymaker at the right time (Purtle, Nelson et al., 2020). A key to the success of dissemination efforts is recognizing heterogeneity within policymaker audiences and tailoring messages to account for these differences (Hirsh et al., 2012; Noar et al., 2007; Purtle, Lê-Scherban et al., 2018; Slater, 1996). Three bodies of research support the notion that the deliberate dissemination of evidence to policymakers can have impacts. First is research related to marketing and advertising. Informed by theories of consumer psychology, the marketing and advertising industry’s use of sophisticated and data-driven messaging techniques has been tremendously successful at persuading consumers and influencing their behavior (Haugtvedt et al., 2018; Kosinski et al., 2013; Liu-Thompkins, 2019; Matz et al., 2017). Second is research from communication science. Informed by framing theory, experiments show that public support for evidence-supported policies can be cultivated by packaging evidence in particular ways for different audiences (Gollust et al., 2017; Kennedy-Hendricks et al., 2016; Lynch & Gollust, 2010; E. McGinty et al., 2018; E. E. McGinty et al., 2013; Niederdeppe et al., 2015; Young et al., 2016). Third, is research from the fields of political science and implementation science. Field experiments conducted with policymakers (Baekgaard et al., 2017; Brownson et al., 2011; Butler & Nickerson, 2011; Grossman & Michelitch, 2018; Jason & Rose, 1984; Levine, 2020a, 2020b; Lyons et al., 2013; Niederdeppe et al., 2016; Nyhan & Reifler, 2015) and health care professionals (Doctor et al., 2018; Meisel et al., 2016; Sacarny et al., 2018) show that the strategic packaging and delivery of information can affect policymaker and practitioner behavior. This chapter focuses on the dissemination of evidence to policymakers and has two main sections. I first present empirical evidence to illustrate heterogeneity among policymakers in terms of the sources they turn to for evidence, the sources they perceive as reliable, the attributes of evidence they perceive as most important, and their baseline knowledge

DISSEMINATING EVIDENCE TO POLICYMAKERS …

29

and attitudes. I then review three broad types of strategies that can help account for audience heterogeneity and enhance the effectiveness of dissemination efforts: audience segmentation analysis, message tailoring, and framing. The chapter touches on these topics in moderate depth and interested readers may consult additional reviews about the dissemination of evidence to policymakers and other audiences (Bogenschneider & Corbett, 2011; Brownson et al., 2018; McCormack et al., 2013; Purtle et al., 2020; Purtle et al., 2020). I use the language of “dissemination” throughout the chapter, a concept from the field of implementation science. The National Institutes of Health (NIH) define dissemination research as “the scientific study of targeted distribution of information and intervention materials to a specific public health or clinical practice audience.” For the purposes of this chapter, public policymakers are the target “public health practice audience.” This chapter is narrowly focused on dissemination strategies that asynchronously push evidence to policymakers and not other complementary—and potentially more effective—strategies that encourage the pull of research findings by policymakers, cultivate public demand for evidence-supported policies, or foster coalition building (Lavis et al., 2006).

Evidence to Inform the Dissemination of Evidence to Different Policymaker Audiences As colleagues and I describe in our review of approaches to “data-driven dissemination” (Purtle et al., 2020), empirical evidence is critical to making sound decisions about how research evidence is packaged and delivered to policymakers. In this section I present data from surveys that colleagues and I conducted of elected and administrative public policymakers at different levels of government. I present data to illustrate four domains that should be considered when designing and executing dissemination strategies that target policymakers: (a) the sources they turn to for evidence, (b) the sources they perceive as reliable brokers of evidence, (c) the attributes of evidence they perceive as most important, and (d) their baseline knowledge and attitudes about issues that are the focus of dissemination efforts. For each of these domains I highlight heterogeneity between policymakers with different characteristics. The survey data focus on evidence related to behavioral health (i.e., mental health and substance use) and health disparities, but the findings are likely generalizable to other health and social issues.

30

J. PURTLE

What Sources Do Policymakers Turn to for Research Evidence? Knowing the sources that policymakers turn to for research is important because it can indicate which knowledge brokers are best positioned to deliver research evidence to policymakers (Brownson et al., 2018). Table 1 presents results from five surveys of different types of policymakers (elected and administrative) at different levels of government (state and county) conducted between 2017 and 2020. In all of the surveys, policymakers were asked a very similar question about the sources that they turn to for behavioral health research. The question asked: Table 1 Primary sources that policymakers turn to for behavioral health research to inform policy decisions Turn for behavioral Turn for behavioral health research health research in general specifically related to youth

Patient advocacy organizations Legislative staff/Staff within agency Professional assistance organizations Provider advocacy organizations University researchers Industry

State Legislators N = 475 2017a

State Mental Health Agency Officials, Senior-Level N = 43 2017b

%

State Mental Health Agency Officials, Seniorand Mid-Level N = 221 2019c %

County Mental Health Agency Officials, Seniorand Mid-Level N = 113 2020d %

State Substance Use Agency Officials, Seniorand Mid-Level 2020 N = 42e %

County Substance Use Agency Officials, Seniorand Mid-Level 2020 N = 80f %

% 53.0

16.3

33.9

38.9

4.8

16.3

51.0

60.5

38.0

18.6

33.3

23.8

38.0

79.1

52.5

45.1

81.0

71.3

30.0

16.3

31.2

44.2

4.8

15.0

27.0

55.8

57.0

43.4

71.4

46.3

11.0

2.3

1.8

1.8

2.4

3.8

Response rates: a = 16.4%, b = 84.0%, c = 33.7%, d = 25.4%, e = 37.8%, f = 16.9%

DISSEMINATING EVIDENCE TO POLICYMAKERS …

31

“If you were going to seek out behavioral health research to make a policy decision, who would you turn to?” and instructed respondents to select three sources from a list. Comparing responses to this question between the different types of policymakers highlights heterogeneity in the evidence-seeking behaviors of policymakers. I first compare the samples of state legislators and senior-level state mental health agency policymakers, both surveyed in 2017 (Purtle, Dodson et al., 2018b; Purtle et al., 2019). These two types of policymakers differ dramatically in their content expertise about behavioral health issues and the entities to which they are most directly accountable (e.g., constituents and political party for legislators; governor and state legislature for senior-level mental health agency policymakers) (Purtle et al., 2020). Among legislators, the source most frequently turned to for behavioral health research was mental health patient advocacy organizations (53%)—such as the National Alliance on Mental Illness—but only 16% of mental health agency policymakers turned to this source (Purtle, Dodson, Nelson et al., 2018; Purtle et al., 2019). In contrast, only 38% of legislators turned to professional assistance organizations—such as the National Conference of State Legislatures—for behavioral health research, while this was the primary source for mental health agency officials, with 79% turning to professional assistance organizations such as the National Association of State Mental Health Program Directors. About half as many legislators turned to universities for behavioral health research than mental health agency policymakers did (27 vs. 56%). I now shift to comparing administrative policymakers in mental health to substance use agencies. While mental health and substance use agencies are often considered under the same umbrella of “behavioral health,” the mental health and substance use agencies are often administratively separate and our data suggest that policymakers in these two types of agencies might turn to different sources for research evidence. We find that mental health agency officials at state and county levels turn to similar sources; and that these sources are different than those turned to by substance use agency officials at state and county levels. A primary source of research among mental health agency officials at state and county levels is mental health patient advocacy organizations (34 and 39%, respectively), such as the National Alliance on Mental Illness, as well as provider advocacy organizations (31 and 44%, respectively), such as the American Psychological Association. Substance use agency officials at both state and county levels, however, do not frequently turn to such advocacy organizations.

32

J. PURTLE

Only about 5% of state substance use agency officials turn to them as do about 15% of county substance use agency officials. Professional assistance organizations, such as the National Association of State Alcohol and Drug Abuse Directors, are more of a primary source among substance use agency officials at state (81%) and county (71%) levels than mental health agency officials at these levels (53 and 45%, respectively). Who Do Policymakers Perceive as Reliable Sources of Research Evidence? Perceptions of source reliability relate to the extent to which policymakers view an entity as a consistently credible and trustworthy source of research evidence. It is possible for a policymaker to perceive an entity as very reliable but not regularly turn to it for evidence because they do not think that the evidence provided will be relevant to their needs (Dunn & Laing, 2017). Figure 1 compares two different types of elected policymakers (mayoral officials and state legislators) who were surveyed four years apart. The data show consistent patterns in the perceived reliability of research evidence from different sources and consistent variations in these perceptions by the ideology of the policymaker. The mayoral official survey was fielded in 2016, completed by 230 U.S. city mayors and deputy mayors, and had a response rate of 30.3% (Purtle, Henson, et al., 2018). The state legislator survey was fielded in 2012, completed by 841 legislators, and had a response rate of 46.0% (Brownson et al., 2016; Purtle et al., 2016). In both surveys, policymakers rated “the reliability of research information” from different sources on 5-point scales (1 = very unreliable, 5 = very reliable). These ratings are dichotomized in Fig. 1 with ratings of 4 or 5 classified as “reliable.” In both surveys policymakers also rated their social and fiscal ideology on 7-point Likert scales. The sum of these scores was used to create an aggregate ideology score (range = 2–14) and respondents with scores of 2–5 were classified as liberal, scores of 6–8 were classified as moderate, and scores of 9–12 were classified as conservative. Figure 1 compares proportion of policymakers in the surveys that rated each source as reliable stratified by ideology, with data displayed separately for the mayoral official and legislator official samples. Among both mayoral officials and state legislators, universities were most frequently identified as a reliable source of research, but a significantly larger proportion of liberals than conservatives identified them

DISSEMINATING EVIDENCE TO POLICYMAKERS …

33

Panel A. U.S. City Mayors and Senior Staff, 2016, N = 230 94% 86% 75% 67% 48%

27% 28% 25%

22% 24%

University*

44%

43%

37%

Philanthropies*

Industry*

Liberal

19% 17%

13% 10%

ConsƟtuents

Moderate

Advocacy groups* ConservaƟve

6%

The media

* Indicates one or more differences by ideology are significant at p ≤ .05, chi-squares test Panel B. U.S. State Legislators, 2012, N = 841 88% 74% 60%

54% 45%

39% 38%

33% 21%

15% 16%

15% 8% 6% 4%

Not asked University*

Philanthropies

Industry*

Liberal

ConsƟtuents*

Moderate

Advocacy groups* ConservaƟve

The media

* Indicates one or more differences by ideology are significant at p ≤ .05, chi-squares test

Fig. 1

Trustworthy sources of research, stratified by ideology

as such. A significantly larger proportion of liberals than conservatives also identified advocacy organizations as a reliable source (i.e., 44 vs. 17% among mayoral officials; 33 vs. 15% among state legislators). In contrast, a substantially smaller proportion of liberals and moderates than conservatives identified industry as a reliable source. Among mayoral officials, industry was identified as reliable by 43% of conservatives but only 22% of liberals and 24% of moderates. A similar pattern was observed among

34

J. PURTLE

state legislators, with 45% of conservatives identifying industry as a reliable source compared to 15% of liberals and 16% of moderates. What Do Policymakers Perceive as the Most Important Attributes of Evidence? Understanding what policymakers want from evidence that is disseminated to them can help ensure that evidence summaries include these attributes. In turn, this can enhance the extent to which evidence is perceived as relevant and thus increase the likelihood evidence is engaged with, cognitively processed, and informs policy decisions (Petty & Cacioppo, 1986; Purtle et al., 2020). Table 2 presents data from the aforementioned surveys of elected and administrative policymakers at state and county levels conducted between 2017 and 2020. In all of the surveys, policymakers were asked “If you were to receive behavioral health research, how important would it be that the research have each of the following characteristics?” on 5-point scales (1 = not important, 5 = very important). These ratings are dichotomized in Table 2 with ratings of 4 or 5 classified as “important.” Across all of the policymaker samples, the relevance of evidence to the population that the policymaker serves, the inclusion of cost-effectiveness and budget impact data, and brevity were consistently perceived as important attributes of behavioral health research. Of note, across all of the samples, research being “delivered by someone I know or respect” was rated as less important than other attributes of evidence. This suggests that the content of disseminated evidence might be more important than the source from which it is delivered. The findings presented in Table 2 are generally consistent with those observed in the aforementioned survey of mayoral officials (Purtle et al., 2017) and 2012 survey of state legislators (Purtle et al., 2016), in which the same questions were asked but not explicitly focused on behavioral health research. Unlike the survey findings related to the sources that policymakers turn to for behavioral health research (Table 1), in which there was meaningful variation in responses between the different types of policymakers, there is much more homogeneity between policymakers in terms of the attributes of evidence they perceive as important. The same was observed in the 2017 survey of state legislators, when respondents were stratified by their political party affiliation (Purtle, Dodson, Nelson et al., 2018) and the 2016 survey of mayoral officials and 2012 survey of state legislators,

DISSEMINATING EVIDENCE TO POLICYMAKERS …

Table 2

35

Important attributes of behavioral health research

Relevant to constituents/state/county residents Provides data on cost-effectiveness/ budget impact Presented in a brief, concise way Tells a story of how an issue affects constituents/residents of my state/county Delivered by someone I know or respect

Turn for behavioral health research in general

Turn for behavioral health research, specifically related to youth

State State Legislators Mental N = Health 475 Agency 2017a Officials, SeniorLevel N = 4 3 2017b

State County State County Mental Mental Substance Substance Health Health Use Use Agency Agency Agency Agency Officials, Officials, Officials, Officials, SeniorSeniorSeniorSeniorand and and and MidMidMidMid-Level Level Level Level 2020 N = N = 2020 N = 80f 221 113 N = 4 2019c 2020d 2e % % % %

%

%

76.4

93.0

92.3

78.8

90.2

93.7

82.1

86.0

87.4

48.7

87.8

90.1

81.5

81.0

85.1

92.1

87.8

91.2

67.7

73.8

67.8

70.8

87.8

71.3

67.7

40.5

43.4

38.1

58.6

33.8

Response rates: a = 16.4%, b = 84.0%, c = 33.7%, d = 25.4%, e = 37.8%, f = 16.9%

when policymakers were stratified by their ideology (Purtle et al., 2018a; Purtle et al., 2017). Taken together, these findings suggest that there is not substantial heterogeneity among policymakers in terms of what they want from research evidence and that local relevance, economic data, and brevity are of paramount importance.

36

J. PURTLE

How Do Knowledge and Attitudes About Evidence Vary Among Policymakers? For any issue that is the focus of a dissemination effort, there is variation in policymakers’ baseline knowledge and attitudes related to the issue. It is important to understand baseline knowledge and attitudes, and how they vary among policymakers with different characteristics, because doing so can inform decisions about what evidence is selected and how it is framed for different audiences. Across issues and types of policymakers, colleagues and I have consistently found that knowledge and attitudes about evidence varies significantly by policymaker ideology. This is illustrated in Fig. 2 with data from the aforementioned 2016 survey of mayoral officials and data from a 2016 survey of 305 U.S. city health commissioners and their senior staff (Purtle, Henson et al., 2018). In both surveys, policymakers indicated the extent to which they thought that different factors “affect differences in health between socially advantaged and disadvantaged groups” (i.e., contribute to health disparities) on an 11-point scale (0 = no effect, 10 = very strong effect). These rating are dichotomized in Fig. 2 with ratings of 8, 9, and 10 classified as “very strong effect.” The ideology of these policymakers was operationalized using the aforementioned approach. Figure 2 shows the proportion of policymakers that perceive each factor as having a very strong effect on health disparities stratified by ideology, with data displayed separately for the mayoral official and health commissioner official samples. Among both mayoral and health commissioner officials, a significantly larger proportion of liberals than conservatives perceived each factor as having a very strong effect on health disparities. The relative magnitude of these liberal-conservative differences was similar in both samples, with the largest difference being the perceived effect of housing on health disparities. It is noteworthy that there was significant ideological heterogeneity in attitudes about health disparities among health commissioner officials because these are administrative (i.e., not elected) policymakers who have content expertise on public health issues. The fact that these ideological differences exist signals that an administrative policymaker’s ideological orientation cannot be separated from their professional policymaking capacity. In prior research colleagues and I have found similar ideological differences in attitudes about the existence and fairness of health disparities (Purtle, Henson et al., 2018) and state legislators’ knowledge about

37

DISSEMINATING EVIDENCE TO POLICYMAKERS …

Panel A. U.S. City Mayors and Senior Staff, 2016, N = 230 23% Housing

51% 67% 56%

EducaƟon

67% 66% 51%

Income

73% 86% 34%

Stress

50% 52% ConservaƟve

Moderate

Liberal

Panel B. U.S. City Health Commissioners and Senior Staff, 2016, N = 305 35% Housing

67% 81% 67%

EducaƟon

78% 82% 60%

Income

88% 89% 53%

Stress

66% 75% ConservaƟve

Moderate

Liberal

Fig. 2 City policymakers’ perceptions of factors that have very strong effects on health disparities, stratified by ideology

38

J. PURTLE

the impacts of state behavioral health parity laws (Nelson & Purtle, 2020; Purtle, Le-Scherban et al., 2019) and attitudes about evidence related to adverse childhood experiences (Purtle, Lê-Scherban, Wang et al., 2019).

Strategies to Account for Audience Heterogeneity When Disseminating Evidence to Policymakers The data presented above illustrate that policymakers are a heterogenous population and suggest that it is often important to disseminate evidence in different ways to policymakers with different characteristics. Three broad types of strategies—audience segmentation analysis, message tailoring, and framing—can help translate such data into concrete decisions about how research evidence is packaged for and delivered to different types of policymakers. Audience Segmentation Analysis The purpose of audience segmentation analysis is to identify sub-groups within a population who have similar knowledge, attitudes, and behaviors related to an issue (Kreuter & Bernhardt, 2009; Slater, 1996). By identifying these sub-groups, messages can be tailored for individuals within different audience segments. There are two broad approaches to audience segmentation—demographic separation and empirical clustering (Smith, 2017)—that can be applied to policymakers as well as other public health practice audiences and the general public. With demographic separation, a population of policymakers is divided into segments based on demographic (e.g., political party, ideology, and gender) or professional (e.g., elected or administrative, agency type) characteristics. The survey data presented in the previous section are examples of demographic separation approaches to audience segmentation. A strength of demographic separation is that segmenting variables, or proxies for them, are readily observable. This makes it possible to know which segment an individual policymaker belongs to and deliver segmenttailored dissemination materials accordingly. A weakness of demographic separation, however, is that it assumes that there is meaningful homogeneity among policymakers within each segment. This is not always true and, as a result, demographic separation can fail to identify the most meaningful sub-groups of policymakers within a population (Smith, 2017).

DISSEMINATING EVIDENCE TO POLICYMAKERS …

39

With empirical clustering approaches to audience segmentation, statistical techniques (e.g., latent class analysis, k-means clustering) are used to identify segments by identifying relationships between multiple variables—such as knowledge and attitudes about evidence, preferences for receiving evidence, and policymaking behaviors. A strength of empirical clustering is that the use of multiple variables can produce a more nuanced and precise understanding of audience segments than demographic separation (Smith, 2017). A weakness of empirical clustering, however, is that its utility hinges upon the ability to use non-latent audience characteristics (e.g., demographics) to predict latent segment membership and deliver the appropriately tailored dissemination materials. Empirical clustering approaches have been used to identify audience segments among the general public that vary in terms of how they think about issues such as climate change (Arbuckle et al., 2017; Hine et al., 2014; Maibach et al., 2011; Nisbet et al., 2011; Poortinga & Darnton, 2016) and health equity (Bye et al., 2016). Colleagues and I used latent class analysis to analyze data from the aforementioned 2017 survey of state legislators to identify audience segments of legislators that vary in their thinking and policymaking behaviors related to behavioral health issues (Purtle, Lê-Scherban et al., 2018). We identified a large (47% of the sample) segment of legislators that we named “Budget-Oriented Skeptics with Stigma” because they were characterized by being strongly influenced by budget considerations, not thinking that behavioral health treatments were effective, and having high levels of mental illness stigma (as well as ideological conservativism). We also identified a small (24% of the sample) segment of legislators who we named “Action-Oriented Supporters ” because they were characterized by introducing behavioral health bills, perceiving behavioral health issues as policy priorities, and being strongly influenced by research evidence. Message Tailoring Tailoring entails manipulating messages so that they are aligned with the personal attributes of individual message recipients. Tailored messages are generally more effective than “one-size-fits-all” messages (Kreuter et al., 2013; Noar et al., 2007). Dijkstra (2008) provides an overview of three types of tailoring strategies that are relevant to the dissemination of research evidence to policymakers: personalization, adaptation, and feedback.

40

J. PURTLE

Personalization involves tailoring messages so that they include recognizable aspects of the recipient (e.g., their name) (Liu-Thompkins, 2019). For example, messages for policymakers can be personalized to include attributes such as their legislative district number, committees of membership, and name of agency, sub-division, and professional title. Personalization can enhance the effectiveness of dissemination materials by signaling that the content is relevant, which can, in turn, increase the likelihood of message recipients engage with the information. For example, an advertising experiment found that simply adding the message recipient’s first name to the subject line of an e-mail increased the probability of the e-mail being opened by 20% and decreased the probability of unsubscribing from future e-mails by 31% (Sahni et al., 2018). While personalization can promote engagement with, and cognitive processing of, messages it can be counterproductive if it prompts information privacy concerns or overtly signals that the messenger is making an attempt at persuasion (Friestad & Wright, 1994; Petty & Cacioppo, 1986). Adaptation entails tailoring the content of messages to increase their congruence with the knowledge, attitudes, and general worldview of recipients (Dijkstra, 2008). For example, in reference to the audience segmentation analysis of legislators described above, messages adapted for legislators in the Budget-Oriented Skeptics with Stigma segment could be adapted to emphasize the cost–benefits of investments in evidencebased behavioral health treatments, contain details about the effectiveness of these treatments, and use words and phrases that do not amplify stigma toward people with mental illness. Like with empirical clustering approaches to audience segmentation, the utility of adaptation is contingent upon the ability to match individuals with appropriately adapted messages. Social media and other online forums in which people leave “digital footprints” have dramatically increased the ability to do this (Kosinski et al., 2013). For elected policymakers, databases such as Quorum—which centralizes data on elected officials’ social media posts, newsletters to constituents, and other artifacts of policy behavior—can be used to inform the adaptation and delivery of messages for elected policymakers with different characteristics. Feedback-tailored messages include data about the message recipient’s behavior. According to feedback intervention theory (Kluger & DeNisi, 1996), messages that include negative feedback (e.g., include data about the number of unsupported claims a policymaker made about a specific issue and the potential political consequences of these claims) and also

DISSEMINATING EVIDENCE TO POLICYMAKERS …

41

include recommendations for improvement (e.g., evidence-based statements that could counter the unsupported claims) can be persuasive because they are perceived as relevant and increase motivation (Nyhan & Reifler, 2015). Feedback-tailored dissemination strategies have been shown to affect the behaviors of health care providers (Doctor et al., 2018; Sacarny et al., 2018) and similar strategies could be used with policymakers. Framing Framing involves selectively emphasizing some aspects of an issue, but not others, with the goal of influencing specific knowledge, attitudes, or behaviors among a target audience (Druckman & Lupia, 2017; Milkman & Berger, 2014). Framing can be a valuable strategy when disseminating evidence to elected officials because they rely heavily on heuristics (i.e., cognitive shortcuts) to make decisions (Vis, 2019). Strategic framing can exploit this and infuse evidence into decision-making processes by framing evidence in ways that fit within the dominant heuristics that policymakers use to think about health and social issues. Framing is related to the concept of message adaption, described above, but broader in the sense that evidence can be deliberately framed in a certain way even if audience segmentation and message tailoring are not conducted (i.e., all members of an audience receive an identical message in which an issue is strategically framed the same way). Framing decisions can be deliberate, but all information is framed when communicated. For example, the inclusion of evidence about cost-effectiveness in a policy brief frames the issue in monetary terms and the inclusion of evidence about disparities frames the issue in terms of equity and justice. The inclusion of brief narratives (i.e., stories) in evidence summaries can also play a powerful role in framing an issue (Frank et al., 2015). Numerous experiments conducted with the general public have used narratives to manipulate the framing of evidence and cultivate policy support for different issues (Gollust et al., 2017; Kennedy-Hendricks et al., 2016; Lynch & Gollust, 2010; E. McGinty et al., 2018; E. E. McGinty et al., 2013; Niederdeppe et al., 2015; Young et al., 2016). Ideally, decisions about how to frame evidence for policymakers are informed by empirical data—such as from public opinion experiments or surveys of policymakers—and theory about the mechanisms through which a specific frame would produce the desired effect. For example, to inform the

42

J. PURTLE

framing of evidence about mental health issues for policymakers, Corrigan and Watson (2003) synthesized findings from social psychology research to create a theoretical framework for how policymakers make mental health policy decisions.

Conclusion As the chapters in this book illustrate, evidence-use in policymaking is an extremely complex phenomenon. Even if executed with absolute precision (i.e., the right messenger delivers the right message to the right policymaker at the right time) the impact of disseminating evidence to policymakers—in isolation from other knowledge translation efforts— would likely be marginal. That said, if strategically used by knowledge brokers as part of a larger knowledge translation system, dissemination can be useful tool for promoting the use of research evidence in policymaking.

References Arbuckle, J., Tyndall, J., Morton, L., & Hobbs, J. (2017). Climate change typologies and audience segmentation among Corn Belt farmers. Journal of Soil and Water Conservation, 72(3), 205–214. https://doi.org/10.2489/ jswc.72.3.205. Baekgaard, M., Christensen, J., Dahlmann, C. M., Mathiasen, A., & Petersen, N. B. G. (2017). The role of evidence in politics: Motivated reasoning and persuasion among politicians. British Journal of Political Science, 1–24. https://doi.org/10.1017/S0007123417000084. Bogenschneider, K., & Corbett, T. J. (2011). Evidence-based policymaking: Insights from policy-minded researchers and research-minded policymakers. Routledge. Brownson, R. C., Dodson, E. A., Kerner, J. F., & Moreland-Russell, S. (2016). Framing research for state policymakers who place a priority on cancer. Cancer Causes & Control, 27 (8), 1035–1041. https://doi.org/10.1007/s10552016-0771-0. Brownson, R. C., Dodson, E. A., Stamatakis, K. A., Casey, C. M., Elliott, M. B., Luke, D. A., Wintrode, C. G., & Kreuter, M. W. (2011). Communicating evidence-based information on cancer prevention to state-level policy makers. Journal of the National Cancer Institute, 103(4), 306–316. https://doi.org/ 10.1093/jnci/djq529. Brownson, R. C., Eyler, A. A., Harris, J. K., Moore, J. B., & Tabak, R. G. (2018). Research full report: Getting the word out: New approaches for disseminating public health science. Journal of Public Health Management and Practice, 24(2), 102. https://doi.org/10.1097/PHH.0000000000000673.

DISSEMINATING EVIDENCE TO POLICYMAKERS …

43

Butler, D. M., & Nickerson, D. W. (2011). Can learning constituency opinion affect how legislators vote? Results from a field experiment. Quarterly Journal of Political Science, 6(1), 55–83. https://doi.org/10.1561/100.00011019. Bye, L., Ghirardelli, A., & Fontes, A. (2016). Promoting health equity and population health: How Americans’ views differ. Health Affairs, 35(11), 1982–1990. https://doi.org/10.1377/hlthaff.2016.0730. Cairney, P., & Oliver, K. (2017). Evidence-based policymaking is not like evidence-based medicine, so how far should you go to bridge the divide between evidence and policy? Health Research Policy and Systems, 15(1), 1–11. https://doi.org/10.1186/s12961-017-0192-x. Corrigan, P. W., & Watson, A. C. (2003). Factors that explain how policy makers distribute resources to mental health services. Psychiatric Services, 54(4), 501– 507. https://doi.org/10.1176/appi.ps.54.4.501. Dijkstra, A. (2008). The psychology of tailoring-ingredients in computertailored persuasion. Social and Personality Psychology Compass, 2(2), 765–784. https://doi.org/10.1111/j.1751-9004.2008.00081.x. Doctor, J. N., Nguyen, A., Lev, R., Lucas, J., Knight, T., Zhao, H., & Menchine, M. (2018). Opioid prescribing decreases after learning of a patient’s fatal overdose. Science, 361(6402), 588–590. https://doi.org/10.1126/science. aat4595. Druckman, J. N., & Lupia, A. (2017). Using frames to make scientific communication more effective. The Oxford Handbook of the Science of Science Communication, 243–252. https://doi.org/10.1093/oxfordhb/978019049 7620.013.38. Dunn, G., & Laing, M. (2017). Policy-makers perspectives on credibility, relevance and legitimacy (CRELE). Environmental Science & Policy, 76, 146–152. https://doi.org/10.1016/j.envsci.2017.07.005. Frank, L. B., Murphy, S. T., Chatterjee, J. S., Moran, M. B., & BaezcondeGarbanati, L. (2015). Telling stories, saving lives: Creating narrative health messages. Health Communication, 30(2), 154–163. https://doi.org/10. 1080/10410236.2014.974126. Friestad, M., & Wright, P. (1994). The persuasion knowledge model: How people cope with persuasion attempts. Journal of Consumer Research, 21(1), 1–31. https://doi.org/10.1086/209380. Gollust, S. E., Barry, C. L., & Niederdeppe, J. (2017). Partisan responses to public health messages: Motivated reasoning and sugary drink taxes. Journal of Health Politics, Policy and Law, 42(6), 1005–1037. https://doi.org/10. 1215/03616878-4193606. Grossman, G., & Michelitch, K. (2018). Information dissemination, competitive pressure, and politician performance between elections: A field experiment in Uganda. American Political Science Review, 112(2), 280–301. https://doi. org/10.1017/S0003055417000648.

44

J. PURTLE

Haugtvedt, C. P., Herr, P. M., & Kardes, F. R. (2018). Handbook of consumer psychology. Routledge. Hine, D. W., Reser, J. P., Morrison, M., Phillips, W. J., Nunn, P., & Cooksey, R. (2014). Audience segmentation and climate change communication: Conceptual and methodological considerations. Wiley Interdisciplinary Reviews: Climate Change, 5(4), 441–459. https://doi.org/10.1002/wcc.279. Hirsh, J. B., Kang, S. K., & Bodenhausen, G. V. (2012). Personalized persuasion: Tailoring persuasive appeals to recipients’ personality traits. Psychological Science, 23(6), 578–581. https://doi.org/10.1177/0956797611436349. Iyengar, S., & Massey, D. S. (2019). Scientific communication in a post-truth society. Proceedings of the National Academy of Sciences, 116(16), 7656–7661. https://doi.org/10.1073/pnas.1805868115. Jason, L. A., & Rose, T. (1984). Influencing the passage of child passenger restraint legislation. American Journal of Community Psychology, 12(4), 485– 494. https://doi.org/10.1007/BF00896507. Kennedy-Hendricks, A., McGinty, E. E., & Barry, C. L. (2016). Effects of competing narratives on public perceptions of opioid pain reliever addiction during pregnancy. Journal of Health Politics, Policy and Law, 41(5), 873–916. https://doi.org/10.1215/03616878-3632230. Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin, 119(2), 254. https://doi.org/10. 1037/0033-2909.119.2.254. Kosinski, M., Stillwell, D., & Graepel, T. (2013). Private traits and attributes are predictable from digital records of human behavior. Proceedings of the National Academy of Sciences, 110(15), 5802–5805. https://doi.org/10. 1073/pnas.1218772110. Kreuter, M. W., & Bernhardt, J. M. (2009). Reframing the dissemination challenge: A marketing and distribution perspective. American Journal of Public Health, 99(12), 2123–2127. https://doi.org/10.2105/AJPH.2008.155218. Kreuter, M. W., Farrell, D. W., Olevitch, L. R., & Brennan, L. K. (2013). Tailoring health messages: Customizing communication with computer technology. Routledge. Lavis, J. N., Lomas, J., Hamid, M., & Sewankambo, N. K. (2006). Assessing country-level efforts to link research to action. Bulletin of the World Health Organization, 84(8), 620–628. https://doi.org/10.2471/blt.06.030312. Levine, A. S. (2020a). Why do practitioners want to connect with researchers? Evidence from a field experiment. PS: Political Science & Politics, 1–6. https://doi.org/10.1017/S1049096520000840. Levine, A. S. (2020b). Why do practitioners want to connect with researchers? Evidence from a field experiment. https://www.r4impact.org/sites/default/ files/interaction_paper_revised_0_0.pdf.

DISSEMINATING EVIDENCE TO POLICYMAKERS …

45

Liu-Thompkins, Y. (2019). A decade of online advertising research: What we learned and what we need to know. Journal of Advertising, 48(1), 1–13. https://doi.org/10.1080/00913367.2018.1556138. Lynch, J., & Gollust, S. E. (2010). Playing fair: Fairness beliefs and health policy preferences in the United States. Journal of Health Politics, Policy and Law, 35(6), 849–887. https://doi.org/10.1215/03616878-2010-032. Lyons, R. A., Kendrick, D., Towner, E. M., Coupland, C., Hayes, M., Christie, N., Sleney, J., Jones, S., Kimberlee, R., & Rodgers, S. E. (2013). The advocacy for pedestrian safety study: Cluster randomised trial evaluating a political advocacy approach to reduce pedestrian injuries in deprived communities. PLoS one, 8(4), e60158. https://doi.org/10.1371/journal.pone.0060158. Maibach, E. W., Leiserowitz, A., Roser-Renouf, C., & Mertz, C. K. (2011). Identifying like-minded audiences for global warming public engagement campaigns: An audience segmentation analysis and tool development. PloS one, 6(3), e17571.https://doi.org/10.1371/journal.pone.0017571. Matz, S. C., Kosinski, M., Nave, G., & Stillwell, D. J. (2017). Psychological targeting as an effective approach to digital mass persuasion. Proceedings of the National Academy of Sciences, 114(48), 12714–12719. https://doi.org/ 10.1073/pnas.1710966114. McCormack, L., Sheridan, S., Lewis, M., Boudewyns, V., Melvin, C. L., Kistler, C., Lux, L. J., Cullen, K., & Lohr, K. N. (2013). Communication and dissemination strategies to facilitate the use of health-related evidence. Database of Abstracts of Reviews of Effects (DARE): Quality-assessed Reviews [Internet]. Centre for Reviews and Dissemination (UK). McGinty, E., Pescosolido, B., & Goldman, H. (2018). Communicating about mental illness and violence: Balancing increased support for services and stigma. Journal of Health Policy, Politics and Law, 43(2), 185–228. https:// doi.org/10.1215/03616878-4303507. McGinty, E. E., Webster, D. W., & Barry, C. L. (2013). Effects of news media messages about mass shootings on attitudes toward persons with serious mental illness and public support for gun control policies. American Journal of Psychiatry, 170(5), 494–501. https://doi.org/10.1176/appi.ajp.2013.130 10014. Meisel, Z. F., Metlay, J. P., Sinnenberg, L., Kilaru, A. S., Grossestreuer, A., Barg, F. K., Shofer, F. S., Rhodes, K. V., & Perrone, J. (2016). A randomized trial testing the effect of narrative vignettes versus guideline summaries on provider response to a professional organization clinical policy for safe opioid prescribing. Annals of Emergency Medicine, 68(6), 719–728. https://doi.org/ 10.1016/j.annemergmed.2016.03.007. Milkman, K. L., & Berger, J. (2014). The science of sharing and the sharing of science. Proceedings of the National Academy of Sciences, 111(Suppl. 4), 13642–13649. https://doi.org/10.1073/pnas.1317511111.

46

J. PURTLE

Nelson, K. L., & Purtle, J. (2020). Factors associated with state legislators’ support for opioid use disorder parity laws. International Journal of Drug Policy, 82, 102792. Niederdeppe, J., Roh, S., & Dreisbach, C. (2016). How narrative focus and a statistical map shape health policy support among state legislators. Health Communication, 31(2), 242–255. https://doi.org/10.1080/104 10236.2014.998913. Niederdeppe, J., Roh, S., & Shapiro, M. A. (2015). Acknowledging individual responsibility while emphasizing social determinants in narratives to promote obesity-reducing public policy: A randomized experiment. PloS one, 10(2), e0117565. https://doi.org/10.1371/journal.pone.0117565. Nisbet, M. C., Maibach, E., & Leiserowitz, A. (2011). Framing peak petroleum as a public health problem: Audience research and participatory engagement in the United States. American Journal of Public Health, 101(9), 1620–1626. https://doi.org/10.2105/AJPH.2011.300230. Noar, S. M., Benac, C. N., & Harris, M. S. (2007). Does tailoring matter? Meta-analytic review of tailored print health behavior change interventions. Psychological Bulletin, 133(4), 673. https://doi.org/10.1037/0033-2909. 133.4.673. Nyhan, B., & Reifler, J. (2015). The effect of fact-checking on elites: A field experiment on US state legislators. American Journal of Political Science, 59(3), 628–640. Oliver, K., Lorenc, T., & Innvær, S. (2014). New directions in evidence-based policy research: A critical analysis of the literature. Health Research Policy and Systems, 12(1), 34. https://doi.org/10.1186/1478-4505-12-34. Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In Communication and persuasion (pp. 1–24). Springer. Poortinga, W., & Darnton, A. (2016). Segmenting for sustainability: The development of a sustainability segmentation model from a Welsh sample. Journal of Environmental Psychology, 45, 221–232. https://doi.org/10.1016/j.jenvp. 2016.01.009. Purtle, J., Dodson, E. A., & Brownson, R. C. (2016). Uses of research evidence by State legislators who prioritize behavioral health issues. Psychiatric Services, 67 (12), 1355–1361. https://doi.org/10.1176/appi.ps.201500443. Purtle, J., Dodson, E., & Brownson, R. (2018a). Political party, ideology, and variations in research dissemination preferences and research use practices among US state legislators. Paper presented at the Implementation Science. Purtle, J., Dodson, E., Brownson, R. (2018b). Political party, ideology, and variations in research dissemination preferences and research use practices among US state legislators. Implementation Science, 13(Suppl. 3), A39.

DISSEMINATING EVIDENCE TO POLICYMAKERS …

47

Purtle, J., Dodson, E. A., Nelson, K., Meisel, Z. F., & Brownson, R. C. (2018). Legislators’ sources of behavioral health research and preferences for dissemination: Variations by political party. Psychiatric Services, 69(10), 1105–1108. https://doi.org/10.1176/appi.ps.201800153. Purtle, J., Henson, R. M., Carroll-Scott, A., Kolker, J., Joshi, R., & Diez Roux, A. V. (2018). US mayors’ and health commissioners’ opinions about health disparities in their cities. American Journal of Public Health, 108(5), 634–641. https://doi.org/10.2105/AJPH.2017.304298. Purtle, J., Henson, R. M., Carroll-Scott, A., Kolker, J., & Roux, A. D. (2017). US mayors’ evidence dissemination preferences: Towards evidence-based city policies. Paper presented at the 10th Annual Conference on the Science of Dissemination and Implementation. Purtle, J., Lê-Scherban, F., Nelson, K. L., Shattuck, P. T., Proctor, E. K., & Brownson, R. C. (2019). State mental health agency officials’ preferences for and sources of behavioral health research. Psychological Services. https://doi. org/10.1037/ser0000364. Purtle, J., Lê-Scherban, F., Wang, X., Brown, E., & Chilton, M. (2019). State legislators’ opinions about adverse childhood experiences as risk factors for adult behavioral health conditions. Psychiatric Services, 70(10), 894–900. https://doi.org/10.1176/appi.ps.201900175. Purtle, J., Lê-Scherban, F., Wang, X., Shattuck, P. T., Proctor, E. K., & Brownson, R. C. (2018). Audience segmentation to disseminate behavioral health evidence to legislators: An empirical clustering analysis. Implementation Science, 13(1), 121. https://doi.org/10.1186/s13012-018-0816-8. Purtle, J., Le-Scherban, F., Wang, X., Shattuck, P. T., Proctor, E. K., & Brownson, R. C. (2019). State legislators’ support for behavioral health parity laws: The influence of mutable and fixed factors at multiple levels. The Milbank Quarterly, 97 (4), 1200–1232. https://doi.org/10.1111/1468-0009.12431. Purtle, J., Marzalik, J. S., Halfond, R. W., Bufka, L. F., Teachman, B. A., & Aarons, G. A. (2020). Toward the data-driven dissemination of findings from psychological science. American Psychologist, 75(8), 1052. https://doi.org/ 10.1037/amp0000721. Purtle, J., Nelson, K. L., Bruns, E. J., & Hoagwood, K. E. (2020). Dissemination strategies to accelerate the policy impact of children’s mental health services research. Psychiatric Services, appi. ps. 201900527. https://doi.org/ 10.1176/appi.ps.201900527. Sacarny, A., Barnett, M. L., Le, J., Tetkoski, F., Yokum, D., & Agrawal, S. (2018). Effect of peer comparison letters for high-volume primary care prescribers of quetiapine in older and disabled adults: A randomized clinical trial. JAMA Psychiatry, 75(10), 1003–1011. https://doi.org/10.1001/jam apsychiatry.2018.1867.

48

J. PURTLE

Sahni, N. S., Wheeler, S. C., & Chintagunta, P. (2018). Personalization in email marketing: The role of noninformative advertising content. Marketing Science, 37 (2), 236–258. https://doi.org/10.1287/mksc.2017.1066. Scheufele, D. A., & Krause, N. M. (2019). Science audiences, misinformation, and fake news. Proceedings of the National Academy of Sciences, 116(16), 7662–7669. https://doi.org/10.1073/pnas.1805871115. Slater, M. D. (1996). Theory and method in health audience segmentation. Journal of Health Communication, 1(3), 267–284. https://doi.org/10. 1080/108107396128059. Smith, R. A. (2017). Audience Segmentation Techniques. Vis, B. (2019). Heuristics and political elites’ judgment and decision-making. Political Studies Review, 17 (1), 41–52. https://doi.org/10.1177/147892 9917750311. Young, R., Hinnant, A., & Leshner, G. (2016). Individual and social determinants of obesity in strategic health messages: Interaction with political ideology. Health Communication, 31(7), 903–910. https://doi.org/10. 1080/10410236.2015.1018699.

“Being Important” or “Knowing the Important”: Who Is Best Placed to Influence Policy? Kathryn Oliver

Those wishing to influence policy and practice are often advised to work through intermediaries (Bednarek et al., 2016; Farrell et al., 2019; Ward, 2017). This is because they are better placed to understand and influence the machinery of the policy process (Oliver et al., 2013), hold credibility with decision-makers (Cvitanovic et al., 2016) and/or maintain relationships in both evidence-producing and evidence-using communities (Oliver & Faul, 2018). For example, researchers may wish to connect with brokers on the edge of the policy world who can speak for them in policy discussions, or they may wish to identify political or professional opinion leaders and influence them directly. But how should researchers attempt to do this? It would be a pity if significant efforts were directed to building relationships with people who, ultimately, were not influential or able to change policy or practice. How can researchers be sure they are connecting with the right people?

K. Oliver (B) Public Health and Policy, London School of Hygiene and Tropical Medicine, London, UK e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 M. S. Weber and I. Yanovitzky (eds.), Networks, Knowledge Brokers, and the Public Policymaking Process, https://doi.org/10.1007/978-3-030-78755-4_3

49

50

K. OLIVER

Much has been written about what makes a person a “key player” in a network or community (Ballester et al., 2006; Borgatti, 2006; Long et al., 2013). For some political scientists, power is seen as a resource or a possession—something which is bestowed on an individual by virtue of holding certain characteristics (being male or wealthy, for example) or roles (holding an executive position or being a clinician) (Coleman, 1988; Dahl, 1961). Here, power is largely defined as authority (executive responsibility), which cannot be delegated and is connected with decisionmaking bodies and roles. Small-world and leadership studies draw on this definition of power, which tends to explain power through the possession of sets of personal characteristics (Brass, 1984; Cronin, 2011; Strauss, 1962). However, others view power and influence as broader concepts than merely a set of ranking criteria. Lukes (1974) argues that we should not just look at which decisions are made and by whom, but at the other faces of power: decisions not made, how agendas are set and reinforced, and how debates and issues are themselves framed. Here, power is conceptualized as a dominating force, or at least as a relational one (Morriss, 2006). If one approaches power as an enacted interaction, we can begin to imagine power in terms of strategies, rather than characteristics. As Scott argues, “social power is a form of causation that has its effect in and through social relations (Isaac, 1992, see also Isaac, 1987)” (Scott, 2001). Increasingly, researchers have emphasized the importance of interactions between actors, rather than actor attributes, to explain both formal and informal types of power (Balkundi & Kilduff, 2005; Freeman, 1991). French and Raven (1956) listed sources of social power as: capacity to coerce, ability to reward, incumbency in a position of authority, recognized expertise, and referent power (esteem or charisma) (French & Raven, 1956). These relate to ideas concerning leadership, such as charismatic leadership (Weber, 1947); described as a “gift” of personality, in the context of a religious leader, with “a mission believed to be embodied in him.” Its success depends on followers having faith in a leader (Bryman, 1993). While described in a religious context this phenomenon can clearly also apply in political leadership. This has been applied to more everyday settings, in Scott’s description of “office charisma,” where “any occupant of a particular position must have certain personal qualities and such an occupant may therefore, be able to demand a degree of personal allegiance.” In terms of making health policy, the personality and credibility of a leader may affect policy content and success.

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

51

Moving beyond questions of personality and character, Treadway described how political skill allows people to achieve influence by “combin[ing] social astuteness with the capacity to adjust their behavior to different and changing situational demands in a manner that appears to be sincere, inspires support and trust, and effectively influences and controls the responses of others” (Ferris et al., 2007; Treadway et al., 2013). Politically skilled individuals are therefore able to present their intentions to others, and make these intentions attractive (Fiske & Taylor, 1984; Spence, 1974; Treadway et al., 2013). Treadway describes this as “one’s expertise, charisma, and goodwill” which resides in informal relationships. For researchers wishing to influence policy and practice, this all implies that—in line with the current evidence about evidence use (Oliver & Boaz, 2019)—those who are most likely to be of use to them are those who are embedded in social processes through which policy and practice already take place. Identifying key players is not about simply looking at hierarchical organograms and emailing those at the top of the tree; rather, it is about understanding the complex web of relationships which underpin the machinery of policy (Oliver & Faul, 2018; Klijn & Koppenjan, 2000).

Use of Network Analysis to Study Power and Influence Fortunately, network analysis offers a robust analytical approach to identifying these individuals or cliques. Influence has been studied by network analysts for several decades. One way of conceptualizing power influence was proposed by Burt (1992), in his theory of structural holes. This theory suggests that individuals attain influence through bridging gaps in social communities which others are not able to, in effect becoming the gatekeeper, or “broker” for resources (whether knowledge, information, social capital, or some other asset). Gould and Fernandez (1989) take a similarly positional approach to identifying influential actors, in their typology of brokerage structures. This describes five structures, or types of relationships, which indicate that individual actors are in a position to “broker” assets. One can then analyze a social network to identify how frequently these structures appear, or how frequently actors are part of these structures. To analyze transactional networks, such as financial exchange, these microstructures are useful to explore social

52

K. OLIVER

interactions. For more intangible assets, such as social capital, trust, and influence, however, these measures are very vulnerable to both missing data, and to variation in how these properties are viewed and attributed by network members. How should one use structural measures to analyze relationships which may be interpreted differently by network members? One approach is to use reputational approaches to collect perceptions about powerful actors (Lewis, 2006; Pfeffer & Salancik, 1974). Lewis carried out several network studies using both positional and reputational approaches to test the hypothesis that the high status of the medical profession in health policy-making has declined, and to unpack patterns of connectedness between influential individuals and groups, and their personal and positional resources. Lewis concluded that structurally it appeared that medicine was not formally represented in power networks, but when individual ties were examined using scores for homophily, centrality, and betweenness, clinicians were central to informal networks influencing health policy Lewis, 2006). Such reputational measures assume that those asked are knowledgeable about power, that they are willing to divulge what they know, and that the process of asking does not create the phenomenon (e.g., in that it might make people rank their acquaintances for the first time) (Pfeffer, 1981). Although such measures have been criticized (Kadushin, 1968) they have shown high agreement across informants and have correlated with other indicators of power (Pfeffer, 1981; Pfeffer & Salancik, 1974). However, both power and influence are in part, attributed properties; people cannot be powerful in a vacuum, but rely on others accepting and validating their position. When considering reputational networks, there are methodological issues to consider. Most rely on asking individuals (known as egos) to nominate others (known as alters) within their social network. Reputational approaches in general may be subject to recall bias, assume that respondents are knowledgeable about who is powerful/influential/a source of information, assume that respondents will answer exhaustively and truthfully, and finally that the act of asking does not create the relationship in the respondent’s mind (Tichy et al., 1979). Asking people to nominate influential alters, for example, is an inherently subjective, interpretive exercise. People’s opinions will differ; people will have partial understandings of the domains which alters are trying to influence, and may or may not have a realistic view of what it takes. They may nominate only those they know, or those they wish to be connected to. This subjectivity and partiality is unlikely to be

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

53

randomly distributed, however. Being able to accurately identify other powerful and influential actors may be a characteristic of powerful and influential actors. However, some research suggests that dominant actors tend to have better recall and a greater knowledge of the network than peripheral actors (Freeman et al., 1987; Krackhardt, 1990) although there is conflicting evidence about this. Previous measures have shown high levels of agreement between participants, however, and have also been correlated with other indicators of power (Krackhardt, 1990). To summarize, many researchers (substantive and methodological) wish to identify influential members of networks. This would be especially beneficial for those concerned with evidence use in policy and practice, because many researchers have only a partial understanding of the social communities they are aiming to influence. Previous work suggests that those with an accurate picture of the social network are both more likely to be influential, and better able to manipulate network structure Leik (1992). The challenge is to explain both the mechanisms by which a network exerts a structural effect on an individuals’ behavior or views, and the likely strength and predictability of such an effect (Marsden, 1983). These assume that social proximity equate to social influence, an idea which has its roots in psychology and behavioral studies (e.g., Simmel, 1898). Hubs and Authorities Taking alters’ connections into account is a helpful way of bridging the theoretical gap between individual and whole network-level measures described by Cook (1983). Hubs and Authorities scores are a type of centrality measure, derived from eigenvector centrality, where each node has a centrality which is determined by the whole network structure. Hubs and Authorities analysis takes this one step further as a natural generalization of eigenvector centrality. Two scores are generated using an iterative updating procedure for each node. Each node is assigned an initial authority weight (Ai ) and an initial Hub weight (Hi ). The vector of all Authority weights is denoted by A = (a1 , a2 , a3 , … an ) and the vector of all Hub weights by H = (h1 , h2 , h3 , … hn ). These vectors are set originally at 1, and re-calculated after applying operations A = At h and H = Aa. The scores are then normalized, and the procedure is repeated. After a sufficient number of iterations, the vectors converge to the principal eigenvectors of matrices At A and AAt (Kleinberg, 1999).

54

K. OLIVER

Using Hubs and Authorities captures centrality in directed networks in an intuitively appealing manner. Power and influence are both reputed and attributed social properties; one cannot be influential if alone. These measures described above present an interesting potential development for the power and influence literature. Hubs and Authorities centrality identifies both “those everyone thinks are important” (the Authorities) and “those who know everyone” (Hubs) arguably as important in policymaking as the latter, for understanding how power operates in their community. By separating “knowledge of important actors” from “recognized as an important actors” we can start to think about power and influence as an attributed property (like centrality, or being male) and as a strategic ability to identify the right people to influence. To our knowledge, Hubs and Authorities has not yet been used on social data. Eigenvectors and derived measures are vulnerable to missing data, perhaps explaining this surprising gap. In essence, this calculation allows us to “weight” the votes of those who are more knowledgeable about social structure. In turn, this offers a way to skew the analysis in favor of those with more accurate insight into social structures. I use this approach to explore the following questions: • Who is good at identifying important actors? (i.e., who are the Hubs?) • Are any of these people also important? (Are any Hubs also Authorities?) • How do these individuals exert power?

Methods I draw on survey and interview data from a multi-modal (power, influence and evidence use) network of 225 policy and research actors in a large urban area in the United Kingdom (Oliver, 2012). The network data was drawn from a survey of 152 individuals working on public health in a large city in the United Kingdom. The city included ten local authorities (LA), one umbrella authority, and ten National Health Service (NHS) trusts, with one regional public health authority). Using three waves of snowballing sampling, respondents were asked to identify influential and powerful alters (up to seven), and those who were sources of evidence and information for public health policy-making. Interviews were up to

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

55

one hour, semi-structured and recorded, transcribed, and analyzed using a thematic approach. Hubs and Authorities analysis produces two scores for each node for each network, between −1 and 1. The Authorities scores were calculated and those actors scoring the highest in each category were listed. To show the range of values and actors identified, the top twenty Authorities for each network have been shown here. Each network is clearly dominated by a small set of actors who score very highly, after which the values tail off rapidly. A cut-off point was decided for each network to identify this set of actors (marked in blue); where possible, this was imposed where there was a relatively large step down in values. However, as this is an arbitrary decision, these scores were treated as indicative rather than definitional. Complementing these data, I use accounts from semi-structured interviews with network members, which aimed to gather respondents’ reasons for nominations, their definitions of powerful and influential actors, and the characteristics of power and influence. Qualitative data were collated into frameworks and summarized in text and tabular form in each section for convenience. Where appropriate, illustrative quotes have been added to clarify or support a conclusion drawn in the text. To preserve anonymity of respondents, I have given network members and interviewees pseudonyms corresponding with their sex, and an indicative label consisting of their sector (health, local authority), level (chief executive; mid-level manager).

Results One hundred and fifty-two actors were contacted to take part in the survey. Responses were received from 123 actors (response rate 80.90%). Twenty-six actors declined, four left their jobs, and 14 were unreachable. Seventy-five responses were useable (participation rate 49%). The sample was split between National Health Service and related originations (33%), local authorities (36%), voluntary and third sector (10%), and intelligence/evidence-producing organizations (17%) (see Table 1). Reponses were evenly split between executives in the NHS and local authority, with a similar number of responses received from academics and researchers. Nearly half the academics (or around a quarter of the overall sample) sampled or nominated were clinically trained. Fifty-eight percent of the sample were male, with public health professionals having the

56

K. OLIVER

Table 1

Characteristics of network sample

Job type Public health professional Other types of clinicians NHS executive or Director Public health intelligence staff Council executive or councillor Managers, officers, other staff Academic or researcher Charity director Central government staff/MP Unknown Total

% male

% medics

39 83 62 69 76 52 61 42 62 0 58

68 100 23 6 9 6 44 0 15 0 26%

Total (%) 31 (14) 6 (3) 26 (12) 16 (7) 33 (15) 50 (22) 36 (16) 12 (5) 13 (6) 2 (1) 225

smallest proportion of males. Job types found within the whole network sample are described in Table 1. In total, the power network comprises 171 nodes (51 respondents, 171 nominations in total of 40 actors), and the influence network 262 nodes (61 respondents, 229 nominations in total of 131 actors). Of those, the most frequently nominated included: the regional statutory authority for public health, two chief executives from local authorities (none from NHS trusts), two of the ten possible local Directors of Public Health— officers tasked locally with delivering public health policy, and three policy managers. Who Are the Important Actors and How Do We Know Them? The Authorities analysis for the power network is highly clustered, identifying only five main actors, with a lower-scoring second set of actors (mainly public health professionals) appearing before a rapid decline in values. Similarly, the influence network is dominated by seven actors. There is a high degree of overlap between the network Authorities, although the influence network includes more mid-level managers as authorities, and fewer executives. For ease of interpretation, Table 2 presents indegree, Authority status (i.e., over the threshold), and executive position for the key actors in each network. What this tells us is that most of the Power Authorities identify individuals who correspond to traditional interpretations of power in the

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

Table 2

57

Characteristics of authorities

Actor

Job type

In-degree (power/influence)

Power authority

Influence authority

Executive position

Emma

Director of public health Policy manager, NHS Director of public health Chief executive, LA Chief executive, LA Policy manager, NHS Director of public health Policy manager, LA

22/19

x

x

x

18/15

x

x

14/12

x

x

13/5

x

8/6

x

Alistair Pat Arthur Patrick Evan Heidi David

x x

4/7

x

6/6

x

3/5

x

x

literature; those who are directly responsible and accountable for the activity of an organization, and often explicitly defined in terms of characteristics of powerful individuals—having access to more opportunities, having budgets at their disposal, being recognized by other senior political leaders. They included key decision-making bodies as Chief executives, or as managers, including Emma, who was. [the[ professional leader for public health in Greater Manchester, by virtue of the fact of her being the Director of Public Health for the North West. And there is, you know, a lot of deference and due respect to Emma for that reason. (Alistair, policy manager, NHS)

Key actors were also described as having personal characteristics such as being charismatic, credible, or sensible. Terms such as “gravitas” and “charisma” were often used to describe influential and powerful actors. This personal aspect of policy-making seemed to be both a characteristic of powerful people—in the sense of forceful people being able to drive things forward—and of exerting influence: The Chief exec previously… was very influential, very influential because she had a particularly controlling style of being a chief exec… she had

58

K. OLIVER

quite significant … because of having some control over what went to board meetings, the format in which it was presented, even perhaps reinterpreting the way that decisions had been or the outcome of decisions and reinterpreting them in a particular way that it was a different outcome. (Charlie, council manager)

In addition to holding formal executive positions and having particular personal characteristics, power and influence could also be associated with particular, and particularly effective, ways of working. Sam, an exacademic and NHS adjunct, described mid-level managers as channeling influence, specifically naming three policy managers: Alistair, Evan, and David. Their influence was expressed in the sense that “policies would happen.” Alistair (the second-ranked Authority for both power and influence) managed a regular meeting of all local Directors of Public Health. He instituted a more formalized monthly meeting for the ten DsPH across the conurbation and linked them to other important organizations across the conurbation. Alistair’s influence was therefore attributed to his role within this inter-organizational network, and because of his gatekeeping role with regard to the DsPH and other public health experts. His power and influence was of a different order from the DsPH. He was not “at their elevated level” but did successfully “corral them” (Archie, policy manager, local authority). As John, a DPH explained: I suppose each of the ten DPHs in their own lead area, erm would have a degree, a degree of power. They end up carrying out [the work] through Alistair though. (John, DPH, medic)

Similarly, an NHS manager was described influential because of his personal style and abilities: And somebody like Evan is a pivot. Evan works with all the groups in the Association of Greater Manchester PCTs… so he… has this astonishingly adept diplomatic manner… Something’s happened recently and I said to him you should be in the FCO [Foreign Office]. And he’s really good at assessing, engaging, what will… get us through to where we want to get to. Now, I think some of that is… clearly an individual matter, but I don’t think it’s all. It’s all that, you know, in any kind of organisation or group you have people who do that kind of stuff…. I’m just full of admiration for how he handles it all. (Ada, DPH)

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

59

Powerful and influential actors (Authorities), were therefore formally holders of executive roles, personally seen to be forceful or charismatic, or able to connect important organizations through skilful management. Who Can Accurately Identify the Important Actors? The Hubs analysis identified a different set of people from the Authorities analysis, mainly middle managers in the NHS and local authorities (see Tables 3 and 4). Policy managers were a significant group in the Hubs analyses. This group carried out tasks such as finding information, drafting of reports Table 3

Power Hubs

Actor

Job type

Sector

James Noah Thomas Evan Luke Daniel Archie Maria Madison Alistair

Policy manager Evidence-producer Public health professional Policy manager Public health professional Public health professional Policy manager Public health professional Evidence-producer Policy manager

LA NHS-assoc NHS/LA NHS NHS NHS/LA LA NHS NHS-assoc NHS-assoc

Table 4

Medic

x

x

Power Hubs score 0.288 0.286 0.281 0.279 0.276 0.238 0.234 0.229 0.227 0.198

Influence Hubs

Actor

Job type

Sector

Evan Alistair Arthur Oliver Madison Judy Daniel Charlie

Policy manager Policy manager Chief Executive Public health professional Evidence-producer Evidence-producer Public health professional Policy manager

NHS NHS-assoc NHS NHS/LA NHS-assoc Uni NHS/LA LA

Medic

x x

Influence Hubs score 0.402 0.309 0.256 0.240 0.217 0.214 0.211 0.211

60

K. OLIVER

and presentations, and being general “go-to” person for public health. One described himself and his colleagues as: A bunch of figures in an entrepreneurial role people who don’t have any formal power, but they are given a mandate to do stuff… what they have to try and do is bring different constituencies together and encourage people to do something that might benefit the organisation, but might not benefit another organisation. (Charlie, council manager)

Personal and professional characteristics- of a different type than for powerful Authorities were also described for this group, such as being able to chair a meeting effectively, being able to ask awkward questions, or helping a group “reach a decision.” This was put down to just “being pleasant, sometimes.” (Charlie, manager) which could perhaps be understood as having good diplomatic skills. These Hubs describe the powerful and influential in insightful terms. Powerful people were potentially influential, if they “acted like a leader” and were forceful, or had strong personalities. One manager described his perception of power and influence as being: …based on myths about people’s effectiveness and behaviour in stories…Your influence and your reputation - that’s affected by how sociable or… well, you can prove yourself to be effective in different ways. Some of that is just might just be a balance sheet, or a set of outcomes, indicators, but I don’t reckon it really is often. I reckon it’s more whether you come across as credible and people, as I say, I think it’s like your stories that get passed round organisations. (Charlie, council manager)

In this context, reputation and stories about actors become both proof of and predictors for an actor’s influence. This aspect of leadership and decision-making was not always seen in a positive light, however—perhaps particularly by those who felt or were less powerful. One actor, not identified as a power or influence Authority, and clearly not involved in major decision-making, despite being in a potentially influential role, felt that: You get key players, it’s like any team, you get the alpha and beta types, you get leaders, natural leaders, come to the fore, and they’re not always the most expected or the most desirable, frankly… There’s a massive underlying agenda which is around personality and power (Matthew, ex-manager, NHS).

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

61

Hubs, therefore, are individuals able to identify powerful and influential actors, and in some cases, possess the managerial and personal skills to manage and work with these individuals effectively to achieve policy change. For both successful and unsuccessful Hubs, personality and character were felt to be important aspects of the influencing process. Power Through Agency or Structure? Three individuals, all mid-level managers, were consistently identified as both Hubs and Authorities: Alistair, Evan, and David. As previously described (Oliver et al., 2013), these actors used four main strategies to influence the public health policy process, including controlling decisionmaking organizations, controlling policy content, managing policymakers directly, and using network structures. They were all involved in the creation and manipulation of governance structures, providing policy content, building leadership support around their proposals, and controlling meetings as decision-making arenas. Being close to powerful groups allowed individual managers to exert a range of influences, including writing the agenda for the meetings, providing policy content, and providing experts to attend the meetings. This clearly allowed power to be exerted in an operational sense, through directly filtering what business is done by the decision-makers (particularly for the first three strategies.) What of the fourth strategy: Using network structures? Being able to recognize and create important relational ties, and to maximize the use of those ties in achieving policy goals. This was explicitly discussed by those in bridging roles, predominantly the policy managers, who frequently discussed their roles in terms of relationships, building up or cultivating relationships and even defining roles for external actors: Me, Alistair and Evan, we’re running this place, in the core group… we know where power centres are, we know how far to nudge, we know how to attach an idea to [his chief exec}… that’ll make her look good in AGMA [the Association of Greater Manchester Authorities] Chief Execs’ [meeting]. (David, Council Officer)

The strategies used by the policy managers all required skills in relational working to be effective. Charlie described his job as bringing people together to try and have a conversation, or to work out relationships between different parts of the policy machinery and facilitate those.

62

K. OLIVER

Archie also described his job as making links between people, facilitating conversations between different groups, acknowledging that this made him influential. Another Hub also explained her role in relational terms: We can’t just sit in an office and dream things up… I think a lot of people forget that that’s how things work in the real world, is through relationships and it does take time to build relationships, to build trust, and so you know, reorganisations that lose lots of people mean you just have to start all over again because that is how it, that is how the world works, that’s how you get things done.

However, not all actors agreed that relational working was a good thing. Several actors voiced concerns that this way of doing things could seem manipulative or underhanded. One Director of Public Health claimed to be influenced only by the “strength of an idea,” actively rejecting the idea that relational working was important. Others in the network described him as a liability likely to bring a policy proposal into disrepute. When this was followed up, participants who described him in this sense explained that he was unable to present effectively, could not command respect and so was not influential—despite clearly being well-informed and intelligent. Perhaps he was less influential precisely because he did not grasp the importance of interpersonal relationships. Whether palatable or not, both the qualitative and network data suggest that those able to identify, mobilize and use ties are well-placed to wield influence and power across the machinery of policy. Alistair articulated this clearly: The third category of… influence is people who make the system work. And that’s why, so I would have, you know, people like Evan as central to the… relationship. He… doesn’t have any power invested in him as a as an individual, um and he doesn’t have any accountability as such because he’s accountable to other people but he, and I would myself, put myself in this category as well, we just try and make the system work… if there wasn’t some of those fixed points in this system er such as Evan and myself and others I’m not sure that you know… that the system would hang together very effectively.…

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

63

Discussion Researchers can struggle to identify influential actors in organizations who can help them influence policy and practice. Hubs and Authorities analyses (alongside other network measures) can help us identify both those widely considered to be influential, and those good at spotting important people. This study suggests that the former (Authorities) should only be engaged with if they are also Hubs who have a better understanding of the policy machinery and the network structure surrounding it. Hubs and Authorities identified overlapping but different sets of actors. Bonacich centrality identifies those with advantageous network connections; but Hubs and Authorities measures move beyond this as a proxy measure of how good people are at describing their social context, not just benefitting from it. Authorities were mostly those identified as key players through standard centrality measures. Hubs were mainly not identified as powerful and influential. Power in the literature is said to be associated with holding particular positions in important organizations, or having access to particular expertise (such as clinical training). Holding executive roles did not appear to be connected with being powerful or influential. Although some were nominated, their importance appeared to be due to personal style, reputation or skills, which corresponds with existing evidence from the literature (Krackhardt, 1990; Krackhardt & Hanson, 1993; Mehra et al., 2001). There has been a long debate over whether network perception is a characteristic of powerful actors or not (Krackhardt, 1990; Oh & Kilduff, 2008; Simpson & Borch, 2005). Some connect being influential with having a good understanding of one’s own network position; others suggest that network perception is more accurate in those with less power (Simpson & Borch, 2005). This study supports the view that those with more accurate perception of social structure were not more powerful in general, but an accurate perception of the informal influence network can itself be a base of power in the organization (Krackhardt, 1990) and can facilitate the leader’s ability to forge successful coalitions (Janicik & Larrick, 2005). The study extends these insights to hypothesize that the accuracy of network perception is indeed higher in those without formal authority (mainly the Hubs) but that these actors can nevertheless be more influential in policy networks. For the set of individuals identified as both Hubs and Authorities—that is, commonly recognized as powerful and influential, and as being good

64

K. OLIVER

at identifying those who others agreed were powerful and influential— these policy managers were characterized as being good at relationships; maintaining relationships, identifying important connections and creating them. Work on social power by Scott (2001), which distinguishes between formal and informal power, and work on political skills (Fiske & Taylor, 1984; Scott, 2001; Treadway et al., 2013) support this interpretation. They wielded power not through inherent qualifications or position— rather through other people’s perceptions of them, and their use of interpersonal and political knowledge which (a) helped them to achieve and (b) exploit these brokerage positions. As conceived by Burt and Lukes, power is the control of a desirable resource (Lukes, 1974) and an actor gains power by being a required intermediate (Burt, 2003), whereby powerful actors attempt to increase other’s dependence on them. Merely being a Hub does not guarantee influence over powerful actors; accurate identification of these individuals does not necessarily correspond to being able to influence them. A combination of ability to recognize important relationships, and the skills to build and exploit those relationships appears to be important Rowley & Baum (2008). For researchers, the lesson here is to connect with those who act in these roles. We can hypothesize that people such as Alistair, David, and Evan are essential to the effective functioning of any policy community (certainly, this was their estimation). Researchers would be well-advised, if this is true, to try and identify these individuals rather than building relationships with the Authority-type individuals, who may not share the skills and interests in relationship building. However, they should be aware that in general, the academics in this data set did not share the common perceptions of others about the network structure. One peripheral academic made a few mistakes about who people worked for, identifying the council leader as the Chief Executive, or working for the wrong borough, for example. The same actor identified a senior academic as influential because “he’s sort of got one foot in both camps so he’s an academic but he’s also very involved in you know local policy”—and yet no one else nominated him at all, indicating that this perception was not shared by actors from the policy world. If researchers are not able to build accurate mental maps of the policy communities they are trying to influence, they are unlikely to succeed. Another lesson which researchers may wish to take from this study concerns the fact that policy managers had no public health expertise and yet were considered influential. One possible explanation for this is

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

65

that public health policy is a false construction; public health may be just one part of the whole body of public policy. The policy managers may have been influential in this wider body of policy, and therefore, by default, also influential here. This would in turn mean that the influence of DsPH outside public health policy may be less. Again, this implies that researchers would be better off working through those who have social networks in place and have strategies to build and maintain networks.

Limitations Hubs and Authorities measures are vulnerable to missing data, like all eigenvector-derived measures. However, both sets of measures were recalculated after removal of non-respondents without significant change. Four Authorities did not respond, and thus were excluded from the Hubs analysis and we cannot comment on the accuracy of their mental maps of the community. Respondents may also refrain from nominating themselves which may also skew the analysis. The most common application for Hubs and Authorities, and indeed what it was designed for, is the ranking of webpages, from which data about ties is usually collected via webcrawlers. This maximizes data collection, meaning that this sociological application may be more vulnerable to misinterpretation than the usual application of Hubs and Authorities measures. As stated above, the thresholds for the Authorities and Hubs sets were a matter of judgment. To avoid over-emphasizing this cut-off, the results for each set have been discussed in the light of the top twenty actors, with especial reference to the dominating top set, rather than using the set as a unit of analysis. The scores are indicative rather than definitive and should be read as such. However, the inclusion of David as an influence Authority and Alistair as a power Hub are both worth mentioning. As they are both policy managers, and their scores were at the lowest end of the top set, the narrative about the dominance of the policy managers may have given extra weight to the decision to include them in this “top set.” However, there was a step down in the scores subsequent to theirs. This could be seen as a case where the network and qualitative data were used in conjunction to support a hypothesis in the other. Those actors fresh in people’s memory may be most likely to be nominated, and as always there is a risk of recall bias. Power and influence are both problematic concepts in themselves. Respondents interpreted influence as “is influential,” “influences me,” “influences the system”—and

66

K. OLIVER

probably in more diverse ways. Some people described “influence” as a way of effecting change; other answered the literal question posed, which was “influences my views.” For this reason, the definitions and characteristics of powerful and influential people were analyzed together, with similarities and differences highlighted. This is perhaps reflected in the common comment that people found “doing the survey hard or challenging.” However, by comparing the qualitative and network data some of these definitions were pulled out. The core group were well known, and consistently nominated. Some studies suggest that their answers will be more accurate (Krackhardt & Hanson, 1993). As there is a high consistency among answers in the core, this adds strength to the network findings despite the relatively low response rate (Grannis, 2009). In conclusion, this novel application of Hubs and Authorities analysis to social data allows us to identify both highly nominated, and highly perceptive social network members. In conjunction with the qualitative data, these data suggest that those able to perceive network structure accurately are also well-able to manipulate and control policy processes. For those wishing to influence the policy process, including researchers, this suggests two possible actions. First, researchers could seek to become so familiar with the social structure of a policy community that they are themselves able to undertake the work of creating, maintaining and exploiting important relationships. Second, and perhaps more feasibly, researchers have the option of working to identify those who are already embedded in these communities, and seek to collaborate with them as they support the policy system.

References Balkundi, P., & Kilduff, M. (2005). The ties that lead: A social network approach to leadership. Leadership Quarterly. https://doi.org/10.1016/j.leaqua.2005. 09.004. Ballester, C., Calvó-Armengol, A., & Zenou, Y. (2006). Who’s who in networks. Wanted: The key player. Econometrica. https://doi.org/10.1111/j.14680262.2006.00709.x. Bednarek, A. T., Shouse, B., Hudson, C. G., et al. (2016). Science-policy intermediaries from a practitioner’s perspective: The Lenfest Ocean Program experience. Science and Public Policy. https://doi.org/10.1093/scipol/scv008. Borgatti, S. P. (2006). Identifying sets of key players in a social network. Computational and Mathematical Organization Theory, 12(1), 21–34. https://doi. org/10.1007/s10588-006-7084-x.

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

67

Brass, D. J. (1984). Being in the right place: A structural analysis of influence in an individual organization. Administrative Science Quarterly, 29(4), 518–539. Bryman, A. (1993). Charismatic leadership in business organizations: Some neglected issues. The Leadership Quarterly, 4(3–4), 289–304. Burt, R. S. (1992). Structural holes. Harvard university press. Chicago. Burt, R. S. (2003). The social structure of competition. In R. Cross, A. Parker, & L. Sasson (Eds.), Networks in the knowledge economy (pp. 13–56). Cambridge University Press. Coleman, J. S. (1988). Social capital in the creation of human capital social capital in the creation of human capital’ AJS Volume 94 Supplement S95– S120 S95. Source: American Journal of Sociology. Cook, K. S., Emerson, R. M., Gillmore, M. R., & Yamagishi, T. (1983). The distribution of power in exchange networks: Theory and experimental results. American Journal of Sociology, 89(2), 275–305. Cronin, B. (2011). Networks of corporate power revisited. In Procedia—Social and Behavioral Sciences. https://doi.org/10.1016/j.sbspro.2011.01.007. Cvitanovic, C., McDonald, J., & Hobday, A. J. (2016). From science to action: Principles for undertaking environmental research that enables knowledge exchange and evidence-based decision-making. Journal of Environmental Management. https://doi.org/10.1016/j.jenvman.2016.09.038. Dahl, R. A.(1961). Who governs? Democracy and power in an American City. The Journal of Politics. By Robert A. Dahl (pp. xii, 355. $6.75). Yale University Press. https://doi.org/10.2307/2127730. Farrell, C. C., Harrison, C., & Coburn, C. E. (2019). “What the hell is this, and who the hell are you?” Role and identity negotiation in research-practice partnerships. AERA Open, 5(2), 233285841984959. https://doi.org/10.1177/ 2332858419849595 (Sage Publications). Ferris, G. R., Treadway, D. C., Perrewé, P. L., Brouer, R. L., Douglas, C., & Lux, S. (2007). Political skill in organizations. Journal of Management, 33, 290–320. Fiske, S. T., & Taylor, S. E. (1984). Social cognition. Reading, Mass, AddisonWesley Pubishing Company. MLA (7th ed.) Freeman, C. (1991). Networks of innovators: A synthesis of research issues. Research Policy. https://doi.org/10.1016/0048-7333(91)90072-X. Freeman, L. C., Romney, A. K., & Freeman, S. C. (1987). Cognitive structure and informant accuracy. American anthropologist, 89(2), 310–325. Chicago. French, J., & Raven, B. (1956). The bases of social power. In D. Cartwright (Ed.), Studies in social power (pp. 150–167). Institute for Social Research. Gould, R. V., & Fernandez, R. M. (1989). Structures of mediation: A formal approach to brokerage in transaction networks. Sociological methodology, 89– 126. Grannis, R. (2009). From the ground up. Princeton University Press.

68

K. OLIVER

Janicik, G. A., & Larrick, R. P. (2005). Social network schemas and the learning of incomplete networks. Journal of Personality and Social Psychology, 88(2), 348. Kadushin, C. (1968). Power, influence and social circles: A new methodology for studying opinion makers. American Sociological Review, 685–699. Kleinberg, J. M. (1999). Authoritative sources in a hyperlinked environment. Journal of the ACM . https://doi.org/10.1145/324133.324140. Klijn, E. H., & Koppenjan, J. F. M. (2000). Public management and policy networks. Public Management: An International Journal of Research and Theory. https://doi.org/10.1080/14719030000000007. Krackhardt, D. (1990). Assessing the political landscape: Structure, cognition, and power in organizations. Administrative Science Quarterly, 342–369. Chicago. Krackhardt, D., & Hanson, J. R. (1993). Informal networks: The company behind the charts. Harvard Business Review, 71(4), 104–111. Leik, R. K. (1992). New directions for network exchange theory: Strategic manipulation of network linkages. Social Networks. https://doi.org/10. 1016/0378-8733(92)90007-T. Lewis, J. M. (2006). Being around and knowing the players: networks of influence in health policy. Social science & medicine, 62(9), 2125–2136. Chicago. Long, J. C., Cunningham, F. C., Carswell, P., et al. (2013). Who are the key players in a new translational research network? BMC Health Services Research, 13(1), 338. https://doi.org/10.1186/1472-6963-13-338. Lukes, S. (1974). Power: A Radical View. Macmillan Press. Marsden, P. V. (1983). Restricted access in networks and models of power. American Journal of Sociology, 88(4), 686–717. Chicago. Mehra, A., Kilduff, M., & Brass, D. J. (2001). The social networks of high and low self-monitors: Implications for workplace performance. Administrative Science Quarterly, 46(1), 121–146. Morriss, P. (2006). Steven Lukes on the concept of power. Political Studies Review, 4(2), 124–135. https://doi.org/10.1111/j.1478-9299.2006.000 104.x. Oh, H., & Kilduff, M. (2008). The ripple effect of personality on social structure: Self-monitoring origins of network brokerage. Journal of Applied psychology, 93(5), 1155. Oliver, K. A. (2012). Evaluating power, influence and evidence-use in public health policy-making: A social network analysis (Doctoral dissertation, University of Manchester). Oliver, K., & Boaz, A. (2019). Transforming evidence for policy and practice: Creating space for new conversations. Palgrave Communications, 5(1), 1–10.

“BEING IMPORTANT” OR “KNOWING THE IMPORTANT” …

69

Oliver, K., & Faul, M. V. (2018). Networks and network analysis in evidence, policy and practice. Evidence and Policy, 14(3), 369–379. https://doi.org/ 10.1332/174426418X15314037224597. Oliver, K., De Vocht, F., Money, A., et al. (2013). Who runs public health? A Mixed-Methods Study Combining Qualitative and Network Analyses. Journal of Public Health. https://doi.org/10.1093/pubmed/fdt039 (United Kingdom). Pfeffer, J. (1981). Understanding the role of power in decision making. Classics of organization theory, 3, 404–423. Pfeffer, J., & Salancik, G. R. (1974). Organizational decision making as a political process: The case of a university budget. Administrative Science Quarterly, 135–151. Rowley, T. J., & Baum, J. A. C. (2008). The dynamics of network strategies and positions. Advances in Strategic Management. https://doi.org/10.1016/ S0742-3322(08)25018-7. Scott J. C. (2001). Power: Key Concepts. Polity Press. England. Simmel, G. (1898). Comment les formes sociales se maintiennent. L’Année Sociologique, 1, 71–107. Simpson, B., & Borch, C. (2005). Does power affect perception in social networks? Two arguments and an experimental test. Social Psychology Quarterly, 68(3), 278–287. Spence, A. M. (1974). Market signaling: Informational transfer in hiring and related screening processes. Cambridge, MA: Harvard University Press. Strauss, G. (1962). Tactics of lateral relationship: The purchasing agent. Administrative Science Quarterly. https://doi.org/10.2307/2390853. Tichy, N. M., Tushman, M. L., & Fombrun, C. (1979). Social network analysis for organizations. Academy of Management Review, 4(4), 507–519. Treadway, D. C., Breland, J. W., Williams, L. M., Cho, J., Yang, J. & Ferris, G. R. (2013). Social influence and interpersonal power in organizations: Roles of performance and political skill in two studies. Journal of Management, 39(6), 1529–1553. Ward, V. (2017). Why, whose, what and how? A framework for knowledge mobilisers. Evidence and Policy. https://doi.org/10.1332/174426416X14 634763278725. Weber, M. (1947). Theory of Social and Economic Organization. Chapter: “The Nature of Charismatic Authority and its Routinization”. Translated by A. R. Anderson and Talcott Parsons.

Integrating Connectionist and Structuralist Social Network Approaches to Understand Education Policy Networks: The Case of the Common Core State Standards and State-Provided Curricular Resources Emily M. Hodge, Susanna L. Benko, and Serena J. Salloum

Within education, social network analysis (SNA) has often been applied to the study of communication networks within schools and districts, whether these are connections between students (Cappella et al., 2013), teachers (Coburn & Russell, 2008), or teachers and leaders (Daly & Finnigan, 2011). Further, these studies generally view network ties as connections that facilitate the flow of information between individuals.

E. M. Hodge (B) Montclair State University, Montclair, NJ, USA e-mail: [email protected] S. L. Benko · S. J. Salloum Ball State University, Muncie, IN, USA e-mail: [email protected] S. J. Salloum e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 M. S. Weber and I. Yanovitzky (eds.), Networks, Knowledge Brokers, and the Public Policymaking Process, https://doi.org/10.1007/978-3-030-78755-4_4

71

72

E. M. HODGE ET AL.

However, SNA can be used to answer different types of questions motivated by distinct theoretical traditions and views of network function. Borgatti and Foster (2003) proposed a typology of network studies based on whether they take a connectionist view of networks, or a structuralist view of networks, as well as whether they explain similarity or difference. If considering similarity, a connectionist view of a network places more emphasis on an individual’s agency, with decisions shaped by access to information or other resources. From this perspective, nodes are most often individuals, and ties between individuals are considered pipes, carrying knowledge or information that shape decisions. When practices diffuse across a network, it is because there are stronger, more coherent, information-carrying relationships between people (Borgatti & Foster, 2003, p. 1003). In contrast, a structuralist perspective considers ties between nodes as girders, similar to structural beams in a building, that support and hold nodes in particular positions within the larger network. In this view, the entities the nodes represent will behave similarly or develop similar characteristics when they have similar positions in the network. Borgatti and Foster (2003) use the example of two people who are central to an advice network—both individuals will become tired of answering the phone because their central network position means that their phone rings frequently. In other words, these individuals develop the same characteristic (not wanting to answer the phone), not because they have communicated or intentionally imitated each other, but because their position in the network led them to a similar behavior. This distinction parallels the structure/agency debate in sociology about the extent to which individuals exercise agency over their decisions, with a connectionist view falling towards the agency side and a structuralist view towards the structure side of this debate (Coburn, 2016; Thornton et al., 2012). However, whether viewed from a connectionist or structuralist perspective, networks can explain more than just similarity in decisionmaking—network studies can also be used to explain difference from both of these points of view. From a connectionist view, the difference in an actor’s performance is explained as a result of access to the resources their alters (nodes connected to the actor) possess. A connectionist perspective places more emphasis on agency because the emphasis is about how actors use resources to which they are connected. From a structuralist point of view, differences in actors’ performance are a result of their position

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

73

within the network structure. Central actors will often receive benefits that peripheral actors do not have due to their network position, and will leverage those structural ties for further benefit. Many of the existing studies in education use a connectionist approach to the study of networks: nodes are individuals, and the ties between them represent advice networks of information about an educational reform (e.g., Coburn & Russell, 2008; Daly & Finnigan, 2010, 2011; Finnigan & Daly, 2013; Moolenaar et al., 2010). These studies provide valuable insight into the conditions under which educational reforms can be implemented in ways associated with meaningful school improvement. However, less common in education research literature are studies in which nodes take the form of organizations or units of governance (e.g., states) involved in educational policy formation, implementation, and diffusion. These studies can explain similarity or difference and can view ties as either mechanisms for knowledge dissemination or as structural similarities. From both connectionist and structuralist perspectives, SNA studies at the organization level can illuminate information flow and decision-making relevant to educational policy at a broader level of analysis than SNA studies of individuals. In the sections below, we describe three inter-related studies of the implementation of the Common Core State Standards (CCSS) that illustrate three distinct uses of SNA. The first two studies fall under a connectionist view of SNA. The initial study uses SNA to visualize a largescale education policy network: the organizational sponsors of resources provided by state educational agencies to support new English language arts (ELA) standards. Here, nodes are state educational agencies (SEAs) and the states/organizations sponsoring the resources found on SEA websites, and ties represent information about instruction for new standards. The second study uses SNA as an approach for visualizing coded qualitative data—in this case, the topics, standards, and messages from influential organizations contained in CCSS resources about how teachers “should” enact the contested practice of close reading. Building on the connectionist idea, this study identifies the ideas conveyed through those connections, asking “What is flowing through the pipes?” In this two-mode network, nodes are (1) the individual resources or the organizations who created them and (2) topics, standards, and instructional messages included within those resources. Ties mean that a resource contains a particular topic, standard, or instructional message.

74

E. M. HODGE ET AL.

Unlike the first two studies, the final study takes a structuralist perspective on the potential of SNA to explain state decision-making, examining the state attributes associated with states turning to shared organizations for resources. In this view of networks as structures, network position can lead to network actors behaving in similar ways. Here, we analyze states’ network position in the core or the periphery of the state-provided resources network to understand the extent to which their position explains their shared organizational ties. Nodes are SEAs, and ties are considered girders that hold SEAs in particular network positions. This study also examines a number of other state attributes, as well as variables representing isomorphic change (DiMaggio & Powell, 1983), which Borgatti and Foster (2003) describe as investigating questions of diffusion and similarity from a structuralist perspective. Taken together, these studies offer three distinct applications of SNA to the study of largescale educational policy implementation.

Mapping the National Landscape of State-Provided Curricular Resources: A Connectionist View of Information Flow Between Organizations and States The first application of SNA to education policy discussed in this chapter is its use in visualizing a largescale education policy network: the national landscape of state-provided instructional resources to support new educational standards for K–12 schools (Hodge et al., 2014). Because of the relatively decentralized approach to education in the U.S., states signing onto the same set of standards in 2010 with the Common Core State Standards was a significant change in the U.S. education policy landscape. While standards-based reform became institutionalized as a reform approach across the globe (Hargreaves et al., 2010; Volante, 2012), its progress was relatively slow in the U.S. as states developed their own standards in the 1990s and instituted assessments in the 2000s. The CCSS was a set of common standards coordinated by two membership groups: the National Governors Association and the Council of Chief State School Officers. Policy entrepreneurs’ arguments for the standards focused on how they would bring more coherence to the curriculum landscape, as resources and curricular materials could be developed for a cross-state marketplace (Kornhaber et al., 2014). Other taglines for

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

75

the standards were “fewer, clearer, higher” (Rothman, 2011), “college and career readiness,” and “rigor for all” (Hodge, 2015), and that these higher expectations would be put into place by “zip code” (Kornhaber et al., 2014). Many states initially signed on to the CCSS in 2010. Subsequently, new organizations formed with the goal of supporting CCSS implementation, and states had an opportunity to look more broadly for curriculum materials and online curricular resources (Rowan & White, 2021). Concurrent with the CCSS, the federal Race to the Top (RTTT) competition offered states millions of dollars as part of the American Recovery and Reinvestment Act of 2008, enacted in the wake of the financial crisis. To score highly, states were required to adopt collegeand career-ready standards and teacher evaluations including measures of student growth. Even just applying to RTTT led to states marshaling new inter-organizational networks (Russell et al., 2014). We take the perspective that instructional and professional resources can serve as tools for teacher learning (Davis & Krajcik, 2005; Drake et al., 2014; Remillard, 2005) and that states are key governance units in supporting teachers (Massell, 1998). In this study, we wanted to understand where states directed teachers for information and the type of information states provided. We created a database of 2023 resources for secondary English language arts (ELA) provided on the SEA websites of all 50 states and Washington, D.C. between August 2015 and March 2016. The team developed a set of inclusion and exclusion criteria for SEA webpages and resources on those pages based on their presumed audience (secondary ELA teachers). We downloaded each resource on the included webpages and logged each one in a matrix, coding it along multiple dimensions, including resource type, content-area emphasis with ELA, sponsoring organization/state, etc. We converted our matrix into an edgelist, then used UCINET and NetDraw (Borgatti, 2002; Borgatti et al., 2002) to visualize the network of state-provided resources in a directed network, in which the tie extends from the state providing a resource on its website to the organization or state that had created the resource.

Findings We archived 2023 resources on 121 SEA webpages from 50 states and Washington, D.C. Resources were sponsored by all states (51) and 262 organizations, for a total of 313 resource-sponsoring entities. While many

76

E. M. HODGE ET AL.

organizations and states were represented, almost three-quarters (72.5%) of those entities were named by only one state as a resource sponsor. There was a relatively small set of organizations linked to by multiple states. In other words, many organizations were in the periphery of the network because only one state had linked to their resources, while a smaller set of states and organizations linked to each other’s resources and were therefore located in the center of the sociogram. The full network displayed in a classic core-periphery shape: the core-periphery correlation statistic between our network and an ideal core-periphery network was 0.9, demonstrating a strong core-periphery structure. A concern of this project was identifying the states and organizations active in the post-CCSS instructional landscape, particularly those that had the most power at the national level in terms of where states sought resources. Thus, we ranked organizations by their indegree centrality, as shown in Table 1, to illuminate the organizations to which the highest number of states had directed teachers. These organizations were a mixture of policy organizations, including the sponsors of the CCSS initiative, disciplinary professional organizations, and resource-providing organizations. States’ patterns of organizational ties did not seem to clearly correspond to CCSS status. While most states adopted the CCSS, not all did—at the time of data collection, seven states never adopted or had repealed the CCSS. However, non-CCSS states like Alaska, Virginia, Texas, Nebraska, and Oklahoma were connected to the larger network, though often indirectly through a shared tie to a more general organization (such as PBS Learning Media). Alaska was the only state with ties to organizations that explicitly focused on the CCSS. The network is illustrated in Fig. 1. Network position and connections often facilitate access to information, but there may be reasons related to policy or politics that could explain network decision-making. Hypothetically, states winning RTTT could have used that money for creating their own resources, thus making them less likely to turn to outside organizations. In contrast, states adopting the CCSS may have more opportunities to link externally, either to resources from other CCSS-adopting states or CCSS organizations. We tabulated the proportion of internal resources in each state, looking for potential patterns in whether or not states had won RTTT or adopted the CCSS in the proportion of internally generated resources. No clear patterns emerged from RTTT status. Although the average number of

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

77

Table 1 SEAs and organizations most commonly named as sponsors of CCSS resources

1 2 3 4 5 6 7 8 9 10 11 12 13 14

15 16 17 18 19 20 21 22

Organization/SEA

Number of SEAs Linking to Org/SEA

Percent of SEAs Linking to Org/SEA (%)

Council of Chief State School Officers National Governors Association Student Achievement Partners International Literacy Association Achieve National Council of Teachers of English Council for Great City Schools Public Broadcasting Service Teaching Channel National Association of State Boards of Education New York State Department of Education The Hunt Institute Engage New York Association for Supervision and Curriculum Development Vermont Writing Collaborative Reading Rockets National Writing Project National Education Association Literacy Design Collaborative Delaware Department of Education America Achieves Kansas State Department of Education

30

58.8

25

49.0

24

47.1

17

33.3

16 16

31.4 31.4

15

29.4

14

27.5

14 13

27.5 23.5

9

17.6

9 9 8

17.6 17.6 15.7

7

13.7

6 6 6

11.8 11.8 11.8

6

11.8

6

11.8

6 6

11.8 11.8

(continued)

78

E. M. HODGE ET AL.

Table 1

23 24 25 26

27 28

(continued)

Organization/SEA

Number of SEAs Linking to Org/SEA

Percent of SEAs Linking to Org/SEA (%)

LearnZillion OER Commons New York City Department of Education North Carolina Department of Public Instruction MetaMetrics Louisiana Department of Education

6 5 5

11.8 9.8 9.8

5

9.8

5 5

9.8 9.8

Note The count and percent represent the number and proportion, respectively, of state educational agencies that provided a link to a resource sponsored by particular state or organization. Fifty-one SEAs are included (including Washington, D.C.)

internal ties was similar across CCSS and non-CCSS states, the average number of external ties varied by CCSS status—CCSS-adopting states had more links to external organizations. The second part of the study used descriptive analysis to understand broad content features of the resources, moving towards identifying the information carried by the “pipes” that the ties represent. In other words, if resources carry information about instruction, what types of information do they contain? To get a broad sense of the content of state-provided resources, we coded along three dimensions. The first dimension was resource category, which we defined by how the resource could be used: practical resources were able to be directly applied to classrooms, whereas conceptual resources provided information about standards that readers would have to translate into classroom action. Resources could also be coded as both conceptual and practical, or other. The second was a more fine-grained classification of resource type, including articles, curriculum guidelines, instructional aids (small and large), professional development (PD), standards documents, and student work. Two resource types, collection and homepage, were developed to acknowledge the limitations of web-based data, in which states would link to an organizational homepage or another webpage containing many other resources that we did not code, in order to set limits on our data collection as including only the resources to which SEAs had directly

Fig. 1 Sociogram of ELA Resource Providers (Note Circles represent SEAs. White circles represent SEAs that adopted CCSS; black circles represent SEAs that did not adopt CCSS. Gray squares represent intermediary organizations. Node size denotes level of influence. Line thickness denotes strength of tie, and arrows indicate directionality)

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

79

80

E. M. HODGE ET AL.

linked. Finally, we coded for the area within ELA that resources focused on, to get a sense of whether resources emphasized reading, writing, and/or speaking and listening. A “general ELA” code was used when resources emphasized more than one area, and a “non-ELA” code was developed for resources that did not focus on ELA specifically. Resource coding revealed over half of state-provided, secondary ELA resources were conceptual (53.8%), meaning that they provided information about the CCSS. Practical resources that could be directly used in the classroom composed 17.5% of the sample. In terms of resource type, the largest categories were collections (24.2%), meaning links to pages with other collections of links/resources; curriculum guidelines (15.4%); PD (15%); homepages (11.5%); and standards documents (9%, or 180 out of 2000 codable resources). The final dimension was emphasis within ELA. Nearly half of resources (44%) focused on ELA in general, rather than a specific area within ELA like reading or writing. In addition, surprisingly, over a third of resources on SEA webpages specifically labeled as being for ELA teachers did not focus on ELA, but on more general concerns like content-area literacy. The results are shown in Table 2. While we did not test relationships between state attributes and resources in a formal way, we did assess differences by CCSS and RTTT status. For resource coding, the average proportion of resources provided by CCSS and RTTT states varied in a few areas. CCSS-adopting states provided a higher proportion of PD resources (17%) than non-CCSS states (1.9%), as well as large instructional aids like unit plans. RTTTwinning states also provided more curriculum guidelines than those that did not win RTTT (21.9% vs. 9.8%).

Discussion This study takes a connectionist perspective, viewing “ties as pipes” to understand how information about standards flowed between states, organizations, and, potentially, teachers, through the information contained in instructional resources (Borgatti & Foster, 2003). Theories of centrality often applied to network studies predict that individuals located towards the center of the network (because they are most often sought out for advice) are the most influential nodes in a network (Freeman, 1979). Applying this idea to the study of state-provided standards resources uses a similar logic: nodes are the organizations/states sponsoring curricular documents, and the organizations/states to which the highest number

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

Table 2 Resource category, type, and emphasis for all resources

81

Total Resources N = 2000

Resource Category Practical Conceptual Both Other Resource Type Article/Report Collection Curriculum Guidelines Homepage Instructional Aid—Small Instructional Aid—Large Professional Development Standards Document Student Work Other Resource Emphasis Reading Writing Listening/Speaking General English Language Arts Non-English Language Arts

Frequency

Percent

350 1075 367 208

17.5 53.8 18.4 10.4

154 484 299 229 143 114 308 180 33 56

7.7 24.2 15.0 11.5 7.1 5.7 15.4 9.0 1.7 2.8

235 160 20 895

11.8 8.0 1.0 44.8

690

34.5

of states turn are the most influential in terms of determining what the CCSS may look like in classroom instruction. Those close to the center of the network, and thus most closely connected to the most central organizations, have access to a set of information about standards that states on the periphery or disconnected from the network do not have. However, we do not assume that the information about standards disseminated by central organizations is in line with disciplinary norms or even with standards; indeed, our work suggests some misalignment between the instructional recommendations of organizations in the center and the periphery (Benko et al., 2020; Hodge et al., 2020). Additionally, states provided a variety of resource types and seemed to pursue a variety of strategies for supporting new standards based on their network

82

E. M. HODGE ET AL.

connections; curiously, these strategies did not seem to correspond with attributes like standards adoption in predictable ways. A core concern of this study was identifying the organizations most influential in making recommendations about what the CCSS meant for classroom instruction. Beyond identifying organizations, we also wanted to learn more about the information flowing between organizations and states through the instructional resources. Further, we wanted to learn about possible differences in instructional messages between professional organizations like the National Council of Teachers of English and the Literacy Research Association and organizations formed specifically to support the CCSS.

Identifying Instructional Messages Flowing Between Organizations and States: Qualitative and Social Network Analysis of Resource Content from Influential Organizations Recall that a connectionist view of networks sees ties as pipes or pathways through which information or other resources flow. In line with this view, our second study examined resources to identify the information about instruction flowing between organizations and states. In other words, this study asks about the ideas conveyed through those connections, or “What is flowing through the pipes?” To better understand the content of information moving between organizations and states in the resources and potential differences in instructional messages across influential organizations, we returned to our matrix and ranked the organizations providing state ELA resources according to their indegree centrality to identify organizations to which the highest numbers of SEAs had linked. We selected the “top 10” most central organizations and identified all resources in our database sponsored by one or more of those organizations for further coding. Our goals were twofold. First, we aimed to describe the contents of state-provided instructional resources created by the most influential organizations in the national network of state-provided resources for ELA in greater details. Second, we aimed to take a deep look at how the resources from different organizations described how teachers should enact the practice of “close reading,” a contested topic in CCSS implementation, to see if there were differences between professional organizations and CCSS organizations. Close

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

83

reading has been a controversial topic in applying the CCSS to classroom instruction (Gewertz, 2012; Hinchman & Moore, 2013; Hodge & Benko, 2014). Close reading in its traditional sense considers texts as artistic objects independent of their historical contexts (Sperling & Dipardo, 2008) and is often associated with the outdated theoretical perspective of New Criticism (Pearson & Cervetti, 2015). Thus, the term was controversial because some worried that the CCSS authors powerfully endorsed a decontextualized, ahistorical approach to reading (Smith et al., 2014). For example, the CCSS lead authors published an influential curriculum guide (Coleman & Pimentel, 2012), in which they used the term “close reading,” over 50 times (Catterson, 2017), introducing new discussion about the term and its components. SNA is a helpful data visualization tool to see patterns in two-mode, qualitative data; in this case, SNA allows visualization of the number of instructional resources from different organizations containing a particular message about the meaning of close reading and, therefore, which messages about the nature of close reading are most prevalent. In this study, we identified the standards and topics included in stateprovided ELA resources from the 10 most central organizations. Then, we examined the messages about instruction the resources provided around the topic of close reading, as well as the extent to which messages were consistent within and across organizations. Filtering our sample to resources sponsored by one or more of the 10 most central organizations left us with 361 resources; however, not all resource types were codable for instructional messages. Links to PD, instructional aids, curriculum guidelines, and articles could be coded for instructional messages; organizational homepages or links to standards or additional collections of resources could not. This further reduced the sample to 177 resources. We coded each resource for the standards and topics it explicitly addressed. We then visualized the sociogram of resource-sponsoring organizations to resource topics to identify the topics on which multiple organizations focused. Then, to identify messages about close reading, we reduced the sample to 31 resources focused on close reading. Because multiple states linked to some of the same resources, the sample of close reading resources was composed of 16 unique resources. We define instructional messages as explicit statements about the purpose and/or major elements of close reading, as well as recommended instructional strategies for close reading. We generated an initial deductive list of close reading’s purpose and

84

E. M. HODGE ET AL.

elements based on controversies in PD guidance, research, and organizational reports. We added several other codes about close reading as we read each resource for messages about what teachers “should” do as part of close reading. We first focused on identifying the standards and topics included in the resources from the most influential organization providing CCSS resources. Interestingly, only 63 of the 177 resources, or 35.6%, addressed one or more specific standards. While all resources were included on webpages labeled as providing standards resources, almost two-thirds were not linked to any particular standards. Of the resources that did refer to specific standards, 93.7% referenced one or more reading standards, about half referenced one or more writing (54%) or speaking and listening (49.2%) standards, and 41.3% addressed language standards. Among the resources addressing reading standards, the two most frequently referenced standards were those most clearly linked to close reading. Over half (57.1%) of resources in this group referred to reading anchor standard one (“Read closely to determine what the text says explicitly and to make logical inferences from it; cite specific textual evidence when writing or speaking to support conclusions drawn from the text [CCSS.ELALITERACY.CCRA.R.1]”), and two-thirds (66.7%) referred to reading anchor standard four (“Interpret words and phrases as they are used in a text, including determining technical, connotative, and figurative meanings, and analyze how specific word choices shape meaning or tone [CCSS.ELA-LITERACY.CCRA.R.4]”). In terms of topics on which the resource set focused, commonly emphasized topics were reading informational text, reading literature, vocabulary, argument writing, and complex text/academic language. Resources did not often focus on the topics of meeting the needs of special student populations, instructional supports, curriculum design/evaluation, or narrative writing. Recall that close reading has been a contested practice in CCSS implementation with varied interpretations. To identify instructional messages about close reading in resources from these influential organizations, we conducted a discourse network analysis (Leifeld, 2013), constructing a network of resources and sponsoring organizations’ messages about the definition of close reading and how teachers should enact it. We generated two sociograms shown in Figs. 2 and 3: the first was composed of organizations to their position on close reading, aggregating individual

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

85

Fig. 2 Sociogram of sponsoring organizations to messages about elements of close reading (Note White circles indicate resources’ sponsoring organizations. Black squares indicate specific instructional messages about close reading. Tie strength notes the number of times a particular organization sponsored a resource in the sample expressing a message about close reading. The node size of close reading messages [black squares] indicates the number of resources expressing that message [larger nodes indicate that more resources expressed a particular message about how teachers should enact close reading])

resource coding to the level of sponsoring organization since some influential organizations provided multiple close reading resources. All but one organization’s resources agreed that close reading should “Use textdependent questions” and “Use complex text.” A less common message was that teachers should incorporate historical or other background knowledge into close reading. However, the second sociogram presented close reading messages in each of the 16 individual resources, demonstrating some internal inconsistencies in messages about close reading at the organizational level. Four of the six lesson plans from the website of Student Achievement Partners (achievethecore.org), an organization founded by the CCSS lead authors, either did not recommend that students draw on background knowledge or specifically said that students should not draw on background knowledge when reading. Two of the lesson plans, however said teachers could incorporate students’ historical knowledge into their reading of a historical text.

86

E. M. HODGE ET AL.

Fig. 3 Sociogram of resources to messages about elements of close reading (Note White circles indicate individual resources with messages about close reading. Black squares indicate specific instructional messages about close reading. Tie strength indicates resources that were duplicated in the sample [e.g., 12 of the 31 resources with messages about close reading were the Publisher’s Criteria]. The node size of close reading messages [black squares] indicates the number of resources expressing that message was present [larger nodes indicate that more resources expressed a particular message about how teachers should enact close reading])

This study provides insight into the positions, standards, and topics that were most and least cited in this resource set. Coding qualitative data and visually representing it as a sociogram provides a user-friendly presentation of major themes. A limitation of our approach is that we used ties simply to demonstrate agreement; two organizations would be connected to a position on close reading if they agreed on that element as part of close reading. However, we did not include disagreement as a type of relationship. Future research might visualize disagreement as a tie attribute for a more accurate perspective on positions in qualitative data.

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

87

A Structural Perspective on State Decision-Making: Understanding Shared Organizational Ties in Curriculum Policy Networks The studies described earlier in this chapter take a connectionist perspective, and illuminated the other states and/or organizations that state educational agencies were linking to as sources of information about new standards, the types of resources they were providing, and some of the messages those resources contained about what teachers “should” do to be teaching “to the CCSS.” In other words, ties were seen as pipes carrying information about new standards from organizations creating resources to teachers. However, what seemed to predict states endorsing resources from particular organizations? Here, we use SNA from a structuralist perspective of ties as bonds or girders, seeking to explain the extent to which states with similar network positions or other attributes made similar choices about organizational linkages (Salloum et al., 2020). In this work, we are seeking to explain similarities in network outcomes in terms of the diffusion of innovations, and we view states making similar choices about organizational ties as a type of convergence across organizations. Based on the idea that there was widespread uncertainty in the years following CCSS adoption, we used an institutional theory lens to examine attributes associated with pairs of states having shared organizational ties (DiMaggio & Powell, 1983). We use multiple regression quadratic assignment procedure (MRQAP), a quantitative model that takes into account the interdependencies of data generated from a network, to understand how various state attributes (e.g., adopting the CCSS, geographic region, and the degree of local control over curriculum, among others) are related to the number of shared organizations that states turn to for information. The majority of states adopted the CCSS during a brief window in 2010; in fact, an event history analysis of factors associated with state adoption used a month-to-month scale of the probability that states would adopt in any given month because of the compressed timeline (LaVenia et al., 2015). In many states, the rapid adoption of the standards was coupled with changes to largescale assessments, as well as teacher evaluation policies relying on new observational rubrics and measures of student performance on state assessments. This rapid, widespread change was then followed by grassroots backlash. Several different coalitions

88

E. M. HODGE ET AL.

were opposed to the standards, assessments, and/or use of assessments for teacher evaluation (McGuinn & Supovitz, 2016). Some perceived federal government overreach in the RTTT incentive to adopt the standards, and the CCSS became linked to broader conservative concerns about the Obama administration’s role in policy change. For example, some tweets referred to the standards as “ObamaCore” (Supovitz & Reinkordt, 2017). Networks of public-school teachers and parents were also opposed to the use of high-stakes assessments for students and especially for teacher evaluations. These controversies led to a volatile policy environment, in which states held public hearings on the CCSS, revised or repealed the standards they had just adopted, and left assessment consortia they had just entered. DiMaggio and Powell’s (1983) framework of institutional isomorphic change is useful for understanding organizational decision-making in times of uncertainty. Further, Borgatti and Foster (2003) cite DiMaggio and Powell’s framework specifically as a way of explaining similarity in the structuralist tradition—or that similar actions are based on similar structures or holding similar positions within a network. Based on Meyer and Rowan’s (1977) principles of neoinstitutional theory, “institutionalized organizations” stand in contrast to “technical organizations.” While technical organizations have clear, physical inputs and outputs, institutionalized organizations do not produce tangible goods. Institutionalized organizations create structures that may have a practical purpose, but also have symbolic value in signaling their legitimacy to others. Over time, institutionalized organizations often come to resemble each other by adopting structures and practices to signal that they “belong” and “count” as that type of organization. Further, organizational similarities can provide explanations for how particular practices come to dominate an organizational field and how innovations diffuse across settings as organizations imitate each other. DiMaggio and Powell (1983) propose three processes of isomorphism (the process by which institutions come to resemble each other in their structures and practices): coercive, mimetic, and normative. Coercive isomorphism refers to both formal and informal pressures placed upon an organization by other layers of governance. Pressures could be formal, like a mandate or a strong incentive, such as RTTT’s application requirements, or informal, like the broad adoption and momentum of the CCSS signaling to non-adopting states that they should also adopt. Mimetic isomorphism describes when organizations look to others for signals for

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

89

what to do, perhaps states imitating each other across an interstate group or imitating a neighboring state. Normative isomorphic change occurs through participation in professional networks. In this study, we wanted to identify which (if any) mechanisms of isomorphic change—coercive, mimetic, and/or normative—are related to SEAs turning to the same organizations for ELA resources, as well as to identify the state attributes important in predicting shared organizational ties. We conducted a MRQAP to investigate the factors associated with states having shared organizational ties. Shared organizational ties mean that states are making similar choices about where to direct teachers for information about new standards. Our variables include those designed to test each form of isomorphism, as well as state attributes. For coercive isomorphism, we included whether states had adopted the CCSS or not, as well as whether or not states applied to RTTT. For mimetic isomorphism, we included geographic region and test consortium participation as two means through which states may have imitated other states. For normative isomorphism, we included the state ELA coordinator’s membership in a professional network sponsored by the Council of Chief State School Officers the year before data collection. State attributes included student achievement (National Assessment of Educational Progress); RTTT status (winning or losing); state governance structures based on Smith and Gasparian’s (2018) typology; local control over curriculum (a subset of Smith and Gasparian’s [2018] typology called curriculum guidelines); and governor’s political party. We also included two variables related to the network data: the first was the difference between the number of external resources between pairs of states. This was an important control to account for how pairs of states providing many external resources could be more likely to turn to more of the same organizations simply because they had more external connections. The second measure to account for network structure was a categorization of whether states were located in the core or the periphery of the network using the core-periphery analysis feature in UCINET. To run the MRQAP, each variable must be in the form of adjacency matrices, with both rows and columns bearing states’ names. We operationalized the outcome variable as the number of shared organizational ties between pairs of states, so that the number in the cell indicated the number of organizations to which both states turned. Similarly, the independent variables have to be in the form of state-to-state adjacency

90

E. M. HODGE ET AL.

matrices. If variables are categorical, the matrices are binary, based on states having shared status or characteristics. For example, if states share the status of winning RTTT or losing RTTT, a “1” is placed in the cell. A “0” indicates that states do not share the same status. If data are measured on an interval scale, such as NAEP scores, we operationalized that data so that the number in the cell at the intersection of any pair of states was the absolute value of the difference between the NAEP scores for the two states. The MRQAP analysis (2000 permutations) explained more than 12% of the variance in the outcome; the results are shown in Table 3. When controlling for state attributes, the two variables representing coercive isomorphism were both statistically significant. States that both applied (or did not apply) to RTTT in the first round of applications were 0.17SD more likely to turn to the same organizations (p ≤ 05). Shared CCSS status also had a positive and statistically significant relationship with shared organizational ties (β = 0.14, p ≤ 0.1). Several of our control variables had a statistically significant relationship with shared organizational ties including governance structure, degree of control over curriculum, whether states were classified as being in the core or the periphery, and having similar numbers of external ties. Table 3

MRQAP regression model

Variable

B

Constant RTTT Application—Phase 1 CCSS Status RTTT Winner Governor Party Geographic Location Testing Consortia membership SCASS membership NAEP 8th Reading Curriculum Guidelines Government Structure Number of External Resources Core v. Periphery R2

−0.784*** 0.933* 0.829~ 0.025 −0.067 −0.082 0.252 −0.045 0.000 0.554* 0.390* 0.017* 0.621* 0.122

Note ***p ≤ 001, **p ≤ 01, *p ≤ 05, ~ p ≤ 0.1 The MRQAP model was run with 2000 permutations

Std. Errors

β

p

0.000 0.427 0.546 0.200 0.105 0.163 0.205 0.315 0.000 0.179 0.214 0.008 0.350

0.000 0.167 0.136 0.005 −0.013 −0.014 0.048 −0.008 0.000 0.101 0.051 0.245 0.118

0.000 0.023 0.076 0.386 0.250 0.333 0.112 0.487 1.000 0.011 0.048 0.010 0.038

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

91

That variables representing coercive forces had a statistically significant relationship to shared organizational ties provides empirical evidence that simply applying for RTTT shaped states’ behavior. A prior study of states’ RTTT aspirations (LaVenia et al., 2015) demonstrated that RTTT application was associated with CCSS adoption, but here, we demonstrate that applying for RTTT also influenced CCSS implementation in terms of the organizations to which states directed ELA teachers for information about how to enact new standards. States’ shared CCSS status was marginally significant. Widespread adoption of the CCSS may have created cultural pressures for remaining states to adopt the standards. We believe that RTTT and the CCSS likely reinforced each other as relatively more and less formal coercive forces. These findings also indicate that shared CCSS status leads to similar numbers of shared organizational ties. This means that states not adopting the CCSS had similar, smaller numbers of shared organizational ties, and states adopting the CCSS had similar, but larger numbers of shared organizational ties, providing further evidence of how standards-adoption influenced states’ behavior around implementation. We found no evidence of geographic proximity—one measure of mimetic isomorphism—as a significant predictor of shared organizational ties. This is important because proximity is often a key explanation for the diffusion of innovations that the CCSS may be interrupting. States do not seem to be turning to the same sources as their regional neighbors. This analysis indicates that the CCSS theory of action is at least partially coming to fruition, as states with shared CCSS status are directing teachers towards the same organizations for instructional support. Further, this analysis demonstrates that RTTT was a powerful policy tool in inducing states not only to adopt the CCSS, but in shaping states’ decisions about the organizations they should direct teachers towards to learn about ELA instruction under new standards. This study took a structuralist network perspective, explaining how similar network positions or other attributes related to states making similar choices about organizational linkages. We provide quantitative evidence that CCSS adoption and RTTT application, as well as states’ network position in the core or periphery, had a meaningful relationship with states’ organizational ties.

92

E. M. HODGE ET AL.

New Directions for Social Network Analysis in Education Our work using SNA in education has taken both connectionist and structuralist perspectives: we have viewed ties as pipes to understand the connectivity between states, organizations, and ideas contained in instructional resources, and we have viewed ties as bonds to understand how states’ network positions influenced decision making about instructional resources. The first study viewed instructional resources as ties connecting resource-sponsoring organizations and SEAs that would carry information about CCSS instruction to teachers. The second study also used a connectionist perspective, but closely examined the information about instruction flowing through the pipes that the ties represent. The third study viewed ties as structural bonds, partially determining state action based on structural similarities and shared characteristics. Educational research is rife with studies of advice-seeking using a connectionist perspective. Advice networks provide insight into how information flows through a network, but seldom provide detailed information about the information itself. Sometimes, social network surveys also ask about the content of conversations, but generally in a topical way (not only whom do you talk to, but what do you talk about?). A deeper, qualitative analysis of the content of what is being transmitted between actors, whether via a recorded conversation or a document that is written and then read, provides greater insight into human behavior and decisionmaking. Less common in educational research are three approaches to SNA that we believe could be fruitful in examining discourse around educational issues, understanding policy change, and identifying forces relevant to policy implementation and diffusion. One promising line of research takes a connectionist approach, but rather than a whole-network study of a school or district, takes on the analysis of large-scale data sets from social media, using web scraping or other tools to collect hundreds of thousands of Tweets or Facebook posts about educational issues including opt-outs (Green Saraisky & PizmonyLevy, 2020; Paquin Morel, 2019), the CCSS (Supovitz et al., 2017; Wang & Fikis, 2019), and the Every Student Succeeds Act (Curran & Kellogg, 2017). These studies identify how information is flowing through a social network, who the influential actors are, and in some cases, the content of interactions. For example, Paquin Morel (2019) examines the frames used

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

93

by those advocating for students to opt-out of state assessments. Similarly, Supovitz et al. (2017) identify prominent frames used by interest groups tweeting about the CCSS; this group also used a psychological, lexical analysis tool similar to sentiment analysis to identify the mood and motivations (among other dimensions) of the Tweets. One particularly underdeveloped area of SNA work in education is focused on the diffusion of innovations—work like the third study described in this chapter. One team of education researchers used event history analysis to model the probability of states or other governance units making similar choices based on internal and external characteristics (e.g., LaVenia et al., 2015); however, to our knowledge, other scholars taking up a similar approach are from political science rather than education. Although few in education are combining political science theory with SNA methods, there are a few notable exceptions on the vanguard of this line of inquiry. There is a robust body of work using SNA to identify the recipients of foundation funding (Reckhow, 2012; Reckhow & Snyder, 2014) and the funding networks of alternative teacher certification and charter schools (Au & Ferrare, 2014; Kretchmar et al., 2014). Building on this work, others have used the Advocacy Coalition Framework as a theoretical perspective to understand policy change using a network approach. Two studies in education have used discourse network analysis (Leifeld, 2013), an approach combining qualitative analysis of text with SNA. Galey-Horn et al. (2020) studied congressional hearings to understand how teacher evaluation and student accountability policies rose through the advocacy of “idea brokers,” or those who sponsor particular ideas, connect others to them, and make those ideas more commonplace. Wang (2020) studied a smaller set of Congressional hearings focused on the implementation of the Every Student Succeeds Act, using discourse network analysis to identify sets of actors and shared positions. GaleyHorn and Ferrare (2020) propose that Advocacy Coalition Framework, combined with discourse network analysis and policy narratives, represent an ideal combination for studying policy change in education. The discourse network analysis approach comes from a political science tradition, using qualitative coding that can be exported as a matrix, but focused on policy change. Other approaches to combining qualitative analysis with SNA are less tethered to a particular disciplinary tradition. In a 2019 book chapter, González Canché describes creating an edgelist

94

E. M. HODGE ET AL.

with the dimensions of actors, positions, and time, which could then be used to create a dynamic animation of how a focus group conversation changes over time, as different actors participate and alter the direction of the conversation. For those using SNA with qualitative data, it is important to note that once qualitative data has been reduced to a matrix, SNA can be used to not only visualize these matrices, but compare them. For example, a researcher could conduct two sets of interviews on gun control with parents, National Rifle Association members, and policymakers—one set of interviews before a school shooting, and the other set after a school shooting. After the interviews are coded for participant, role, and position on gun control, the two sociograms at time one and time two could be compared with a QAP correlation to discern whether the networks are significantly different after the event of a school shooting. In other words, does the event create a statistically meaningful change in discourse? Alternately, examining newspaper articles before and after a similar event could measure discourse changes on a larger scale, perhaps detecting softening towards gun control legislation. In conclusion, we have outlined three applications of SNA to the implementation of a large-scale educational policy: the Common Core State Standards. Two of the studies view ties as pipes carrying information about instruction, and they seek to understand who is transmitting that information and the instructional messages contained in resources SEAs are sharing for teachers. The final study views ties as bonds, as states’ network position was related to their shared organizational ties. We conclude with the directions that we hope the field of educational policy will take up. SNA is a flexible analytic tool for both description and prediction of a wide set of relationships between many different sorts of entities, especially when considering two-mode networks. SNA can be used to study the diffusion of innovations across states and localities, and it is an ideal tool for visualizing and analyzing relationships in qualitative data, such as coalitions of actors with shared points of view or understanding the prevalence of themes in a qualitative dataset.

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

95

References Au, W., & Ferrare, J. J. (2014). Sponsors of policy: A network analysis of wealthy elites, their affiliated philanthropies, and charter school reform in Washington State. Teachers College Record, 116(11), 1–24. http://www.tcrecord.org/Con tent.asp?ContentID=17387. Benko, S., Hodge, E., & Salloum, S. (2020). Policy into practice: Understanding state writing resources. Journal of Literacy Research, 52(2), 136–157. https:// doi.org/10.1177/1086296X20915538. Borgatti, S. P. (2002). NetDraw software for network visualization. Analytic Technologies. https://sites.google.com/site/netdrawsoftware/home. Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002). UCINET for Windows: Software for social network analysis. https://sites.google.com/site/ucinetsof tware/. Borgatti, S. P., & Foster, P. C. (2003). The network paradigm in organizational research: A review and typology. Journal of Management, 29(6), 991–1013. https://doi.org/10.1016/S0149-2063(03)00087-4. Cappella, E., Kim, H. Y., Neal, J. W., & Jackson, D. R. (2013). Classroom peer relationships and behavioral engagement in elementary school: The role of social network equity. American Journal of Community Psychology, 52(3–4), 367–379. https://doi.org/10.1007/s10464-013-9603-5. Catterson, A. K. (2017). Close reading in secondary classrooms: A 21st-century update for a 20th-century practice (Publication No. 10281978). Doctoral dissertation, University of California, Berkeley. ProQuest Dissertations and Theses Global. Coburn, C. E. (2016). What’s policy got to do with it? How the structure-agency debate can illuminate policy implementation. American Journal of Education, 122(3), 465–475. https://doi.org/10.1086/685847. Coburn, C. E., & Russell, J. L. (2008). District policy and teachers’ social networks. Educational Evaluation and Policy Analysis, 30(3), 203–235. https://doi.org/10.3102/0162373708321829. Coleman, D., & Pimentel, S. (2012). Revised publishers’ criteria for the Common Core State Standards in English language arts and literacy, grades 3– 12. Washington, DC: Council of Chief State School Officers, Achieve, Council of the Great City Schools, National Association of State Boards of Education. Curran, F. C., & Kellogg, A. T. (2017). Sense-making of federal education policy: Social network analysis of social media discourse around the Every Student Succeeds Act. Journal of School Leadership, 27 (5), 622–651. https:// doi.org/10.1177/105268461702700502. Daly, A. J., & Finnigan, K. S. (2010). A bridge between worlds: Understanding network structure to understand change strategy. Journal of Educational Change, 11(2), 111–138. https://doi.org/10.1007/s10833-009-9102-5.

96

E. M. HODGE ET AL.

Daly, A. J., & Finnigan, K. S. (2011). The ebb and flow of social network ties between district leaders under high-stakes accountability. American Educational Research Journal, 48(1), 39–79. https://doi.org/10.3102/000283 1210368990. Davis, E. A., & Krajcik, J. S. (2005). Designing educative curriculum materials to promote teacher learning. Educational Researcher, 34(3), 3–14. https:// doi.org/10.3102/0013189X034003003. Drake, C., Land, T. J., & Tyminski, A. M. (2014). Using educative curriculum materials to support the development of prospective teachers’ knowledge. Educational Researcher, 43(3), 154–162. https://doi.org/10.3102/001318 9X14528039. DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48(2), 147–160. https://doi.org/10.2307/2095101. Finnigan, K. S., & Daly, A. J. (2013). Systemwide reform in districts under pressure: An overview. Journal of Educational Administration, 51(4), 476– 497. https://doi.org/10.1108/09578231311325668. Freeman, L. C. (1979). Centrality in social networks: Conceptual clarification. Social Networks, 1(3), 215–239. https://doi.org/10.1016/0378-873 3(78)90021-7. Galey-Horn, S., & Ferrare, J. J. (2020). Using policy network analysis to understand ideological convergence and change in educational subsystems. Education Policy Analysis Archives, 28, 1–28. https://doi.org/10.14507/epaa.28.4508. Galey-Horn, S., Reckhow, S., Ferrare, J. J., & Jasny, L. (2020). Building consensus: Idea brokerage in teacher policy networks. American Educational Research Journal, 57 (2), 872–905. https://doi.org/10.3102/000283121 9872738. Gewertz, C. (2012, April). Common standards ignite debate over prereading. Education Week. http://www.edweek.org/ew/articles/2012/04/25/29prer eading_ep.h31.html. González Canché, M. S. (2019). Geographical, statistical, and qualitative network analysis: A multifaceted method-bridging tool to reveal and model meaningful structures in education research. In M. B. Paulsen & L. Perna (Eds.), Higher education: Handbook of theory and research. Springer. https:// doi.org/10.1007/978-3-030-03457-3_12. Green Saraisky, N., & Pizmony-Levy, O. (2020). From policy networks to policy preferences: Organizational networks in the opt-out movement. Education Policy Analysis Archives, 28. https://doi.org/10.14507/epaa.28.4835.

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

97

Hargreaves, A., Lieberman, A., Fullan, M., & Hopkins, D. (Eds.). (2010). Educational change in Finland. In Second international handbook of educational change (Vol. 23, pp. 323–348). Springer Science & Business Media. https://doi.org/10.1007/978-90-481-2660-6. Hinchman, K. A., & Moore, D. W. (2013). Close reading: A cautionary interpretation. Journal of Adolescent & Adult Literacy, 56(6), 441–450. https://doi.org/10.1002/JAAL.00033. Hodge, E. M. (2015). Rigor for all? The enduring tension between standardization and differentiation in the era of the Common Core State Standards (Publication No. 10609580). Doctoral dissertation, The Pennsylvania State University. ProQuest Dissertations and Theses Global. Hodge, E. M., & Benko, S. L. (2014). A “common” vision of instruction? An analysis of English/Language Arts professional development materials related to the Common Core State Standards. English Teaching: Practice and Critique, 13(1), 169–196. Hodge, E., Benko, S., & Salloum, S. (2020). Tracing states’ messages about Common Core instruction: An analysis of English/language arts and close reading resources. Teachers College Record, 122(3). https://www.tcrecord. org/Content.asp?ContentId=23019. Hodge, E., Salloum, S., & Benko, S. (2019). The changing ecology of the curriculum marketplace in the era of the Common Core. Journal of Educational Change, 20(4), 425–466. https://doi.org/10.1007/s10833-019-093 47-1. Kornhaber, M. L., Griffith, K. M., & Tyler, A. (2014). It’s not education by zip code anymore—but what is it? Conceptions of equity under the Common Core. Education Policy Analysis Archives, 22(4), 1–30. https://doi.org/10. 14507/epaa.v22n4.2014. Kretchmar, K., Sondel, B., & Ferrare, J. J. (2014). Mapping the terrain: Teach for America, charter school reform, and corporate sponsorship. Journal of Education Policy, 29(6), 742–759. https://doi.org/10.1080/02680939. 2014.880812. LaVenia, M., Cohen-Vogel, L., & Lang, L. B. (2015). The Common Core State Standards initiative: An event history analysis of state adoption. American Journal of Education, 121(2), 145–182. https://doi.org/10.1086/679389. Leifeld, P. (2013). Reconceptualizing major policy change in the Advocacy Coalition Framework: A discourse network analysis of German pension politics. The Policy Studies Journal, 41(1), 169–198. https://doi.org/10.1111/psj.12007. Massell, D. (1998). State strategies for building capacity in education: Progress and continuing challenges (CPRE Research Report Series RR-41). Philadelphia, PA: Consortium for Policy Research in Education. https://doi.org/10. 1037/e384422004-001.

98

E. M. HODGE ET AL.

McGuinn, P., & Supovitz, J. (2016). Parallel play in the education sandbox: The Common Core and the politics of transpartisan coalitions. Consortium for Policy Research in Education. https://doi.org/10.12698/cpre.2016.Parall elPlay. Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83(2), 340–363. https://doi.org/10.1086/226550. Moolenaar, N. M., Daly, A. J., & Sleegers, P. J. (2010). Occupying the principal position: Examining relationships between transformational leadership, social network position, and schools’ innovative climate. Educational Administration Quarterly, 46(5), 623–670. https://doi.org/10.1177/0013161X1037 8689. Paquin Morel, R. A. (2019). Test questions: Organizing, motivating, and mobilizing opposition to accountability testing (Publication No. 22587435). Doctoral dissertation, Northwestern University. ProQuest Dissertations and Theses Global. Pearson, P. D., & Cervetti, G. N. (2015). Fifty years of reading comprehension theory and practice. In P. D. Pearson & E. H. Hiebert (Eds.), Research-based practices for teaching Common Core literacy (pp. 1–40). Teachers College Press. Reckhow, S. (2012). Follow the money: How foundation dollars change public school politics. Oxford University Press. https://doi.org/10.1093/acprof: oso/9780199937738.001.0001. Reckhow, S., & Snyder, J. W. (2014). The expanding role of philanthropy in education politics. Educational Researcher, 43(4), 186–195. https://doi.org/ 10.3102/0013189X14536607. Remillard, J. T. (2005). Examining key concepts in research on teachers’ use of mathematics curricula. Review of Educational Research, 75(2), 211–246. https://doi.org/10.3102/00346543075002211. Rothman, R. (2011). Something in common: The Common Core Standards and the next chapter in American education. Harvard Education Press. Rowan, B., & White, M. (2021). The Common Core State Standards initiative as an innovation network. American Educational Research Journal. https:// doi.org/10.3102/00028312211006689. Russell, J. L., Meredith, J., Childs, J., Stein, M. K., & Prine, D. W. (2014). Designing inter-organizational networks to implement education reform: An analysis of state race to the top applications. Educational Evaluation and Policy Analysis, 37 (1), 92–112. https://doi.org/10.3102/0162373714527341. Salloum, S., Hodge, E., & Benko, S. (2020). State educational agencies in an uncertain environment: Understanding state-provided networks of English language arts curricular resources. Educational Policy Analysis Archive, 28. https://doi.org/10.14507/epaa.28.4494.

INTEGRATING CONNECTIONIST AND STRUCTURALIST …

99

Smith, J., & Gasparian, H. (2018). Development of a 50-state typology of education governance. International Journal of Educational Reform, 27 (2), 127–155. https://doi.org/10.1177/105678791802700202. Smith, M. W., Appleman, D., & Wilhelm, J. D. (2014). Uncommon Core: Where the authors of the standards go wrong about instruction—and how you can get it right. Corwin. Sperling, M., & Dipardo, A. (2008). English education research and classroom practice: New directions for new times. Review of Research in Education, 32(1), 62–108. https://doi.org/10.3102/0091732X07309336. Supovitz, J., Daly, A. J., del Fresno, M., & Kolouch, C. (2017). #commoncore Project. http://www.hashtagcommoncore.com. Supovitz, J., & Reinkordt, E. (2017). Keep your eye on the metaphor: The framing of the Common Core on Twitter. Education Policy Analysis Archives, 25(31). https://doi.org/10.14507/epaa.25.2285. Thornton, P. H., Ocasio, W., & Lounsbury, M. (2012). The institutional logics perspective: A new approach to culture, structure and process. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199601936.001.0001. Volante, L. (Ed.). (2012). School leadership in the context of standards-based reform: International perspectives (Vol. 16). Springer Science & Business Media. https://doi.org/10.1007/978-94-007-4095-2. Wang, Y. (2020). Understanding congressional coalitions: A discourse network analysis of congressional hearings for the Every Student Succeeds Act. Education Policy Analysis Archives, 28. https://doi.org/10.14507/epaa.28.4451. Wang, Y., & Fikis, D. J. (2019). Common core state standards on Twitter: Public sentiment and opinion leaders. Educational Policy, 33(4), 650–683.

Measuring Issue Preferences, Idea Brokerage, and Research-Use in Policy Networks: A Case Study of the Policy Innovators in Education Network Joseph J. Ferrare, Sarah Galey-Horn, Lorien Jasny, and Laura Carter-Stone

A central criticism of democratic governance in education is that bureaucratic structures stifle policy innovation (Chubb & Moe, 1990; McShane & Hess, 2015). The latter critique has gained substantial traction in recent years, as groups from across the political spectrum have grown impatient with persistent inequities that pervade the education system

J. J. Ferrare (B) University of Washington Bothell, Bothell, WA, USA e-mail: [email protected] S. Galey-Horn University of Edinburgh, Edinburgh, UK L. Jasny University of Exeter, Exeter, UK L. Carter-Stone Vanderbilt University, Nashville, TN, USA © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 M. S. Weber and I. Yanovitzky (eds.), Networks, Knowledge Brokers, and the Public Policymaking Process, https://doi.org/10.1007/978-3-030-78755-4_5

101

102

J. J. FERRARE ET AL.

(DeBray-Pelot & McGuinn, 2009). In response, a variety of private organizations have formed networks through which new policy ideas can be generated, spread, and implemented (Debray et al., 2014; Reckhow, 2010). These policy networks are comprised of a variety of actors, ranging from wealthy individuals, private foundations, businesses, and non-profit organizations seeking to support and challenge the influence of traditional institutions in the education policy arena (Au & Ferrare, 2015; Mehta & Teles, 2012; Reckhow, 2013; Scott & Jabbar, 2014). Education advocacy organizations have become central actors within these policy networks working to shape contemporary education reform. These organizations have arisen, in part, as a response to state-level mandates and shifts in governance that have been implemented as a result of federal policies such as No Child Left Behind, Every Student Succeeds Act, and Race to the Top. Despite the network context of these organizations, much of the advocacy work has proceeded in a relatively decentralized fashion. As a result, there is a growing recognition among advocates and researchers that these organizations must begin to coordinate more strategically across states to maximize impact on the policymaking process (McGuinn, 2012). Prior research suggests that building advocacy coalitions involves engaging in discursive debates over core policy beliefs that shape how the public and policymakers think about issues (Sabatier & Weible, 2007). Increasingly, policymakers and the public more generally are voicing expectations that these debates center on rigorous evidence (Brighouse et al., 2018). However, emerging research suggests that, at least in the education policy arena, the sources of evidence tend to come from a wide range of providers, such as universities, think tanks, and other private organizations, each of which may have divergent expectations and conceptions of rigor (Reckhow & Galey-Horn, 2017). Thus, understanding how and where advocacy organizations engage and retrieve research to support their core policy beliefs has become a central concern among researchers and policymakers (Debray et al., 2014; Scott & Jabbar, 2013). When examining the exchange of ideas and information through policy networks, researchers can emphasize the structure of information flow, the contribution of individual actors to the structure (e.g., central actors or “brokers”), or the relationship between the structure and substance of the information being exchanged within the network. Social network analysis as a whole has tended to focus on the structural properties of networks and actors, especially given the theoretical

MEASURING ISSUE PREFERENCES, IDEA BROKERAGE, AND …

103

assumptions guiding these methods (Wasserman & Faust, 1994). This often leads researchers to use network ties as proxies for information flow, while paying little attention to the substance or meaning of what is flowing. Increasingly, however, social network analysts are adapting techniques to highlight the content that moves through these structures, and in some cases connecting the information to the structural properties of the networks and/or individual actors (e.g., Galey-Horn et al., 2020). In this chapter, we use a collection of methodological tools that together allow for a multifaceted exploration of these relationships in policy networks constructed as two-mode matrices—that is, networks consisting of a set of actors and a set of events. In particular, we make use of discourse network analysis (Leifeld, 2013), exponential random graph models (ERGMs) applied to two-mode networks (Jasny, 2012; Snijders et al., 2006), and correspondence analysis (Greenacre & Blasius, 2006). This chapter proceeds by first establishing the context of our approach to examining policy preferences and research use in education advocacy policy networks. Next, we describe a specific instance of one of these networks—the Policy Innovators in Education (PIE) Network—that will serve as the case study for our illustrative analyses. We then work through the methodological tools mentioned above. For each method, we elucidate the possible contributions to this body of work and the limitations researchers may encounter when examining issue preferences and research use in policy networks. In the process, we reveal insights about the structure and substance of the PIE network.

Background: Policy Networks and Education Advocacy Organizations The presence of advocacy organizations in the education policy arena has a long history in the United States, as organizations aligned across the ideological spectrum (e.g., NAACP, American Enterprise Institute) have been working to shape policy outcomes for decades. Historically, these groups have had limited influence in relation to the so-called “education establishment” consisting of major teachers unions (e.g., NEA, AFT) and other groups that represent state and local jurisdictional actors (Moe, 2011). The latter groups have generally resisted attempts to expand school choice, teacher-based accountability systems, and other marketoriented reforms. However, in the past decade these policies have spread

104

J. J. FERRARE ET AL.

widely across states as a new set of advocacy groups have mobilized bipartisan support (DeBray-Pelot & McGuinn, 2009; McGuinn, 2012) and the financial backing of wealthy elites and foundations (Au & Ferrare, 2014; Ferrare & Reynolds, 2016; Ferrare & Setari, 2018; Kretchmar et al., 2014; Reckhow, 2010; Reckhow & Snyder, 2014; Scott, 2009, 2015). Many contemporary education advocacy organizations arose in the wake of the No Child Left Behind Act of 2001, which included a sweeping set of mandates that required states to establish, among other things, accountability and choice-oriented reforms (Galey, 2015). Some of the advocacy organizations that emerged in the process of pushing these reforms stand alone as independent groups, while many others are affiliated with national umbrella organizations. For example, Democrats for Education Reform (DFER) operates a national office and individual chapters in seven states and the District of Columbia, whereas the League of Education Voters focuses exclusively on issues germane to Washington State. While these organizations vary in structure and scope, they are similar in that they function with limited membership dues and instead rely on private donations (Manna & Moffitt, 2014). Education advocacy organizations operate within policy networks with a variety of other private and public organizations that exchange financial, human, and informational resources in an effort to influence policy outcomes (Rhodes, 2006). Policy scholars in education have begun to draw upon the advocacy coalition framework (ACF) as a way to analyze policy formation processes within these networks (e.g., Debray et al., 2014; Reckhow et al., 2016). Traditional ACF theory identifies three levels of policy-related beliefs conceptualized as a nested hierarchy (Sabatier & Weible, 2007): deep core beliefs (e.g., “open markets drive innovation”), policy core beliefs (e.g., “parents should have educational choices”), and policy preferences (e.g., “states should remove caps on charter schools”). Thus, the ACF prioritizes the role of beliefs and preferences in understanding policy formation and change. This means that scholarship focused on policy networks often must operate with a twomode understanding of networks, since ACF assigns conceptual primacy to the intersection of policy actors and ideas. Discourse network theory (Leifeld, 2013) is an attempt to push the ACF tradition to more explicitly address the two-mode network context in policy network studies. This approach argues that advocacy coalitions are comprised of multiple discourse networks that organize around a

MEASURING ISSUE PREFERENCES, IDEA BROKERAGE, AND …

105

policy belief system which drives interdependencies across institutional and political boundaries. Generally, discourse coalitions correspond to meso-level policy core beliefs and are of particular significance because they set the terms of debate in policymaking domains. Put differently, competing discourse coalitions vie to draw the cognitive map that determines how the public and policymakers think about policy issues. Policymakers cannot discuss all possible issues in one domain simultaneously, thus, a few issues tend to dominate the policy discourse at different points in time. For instance, recent work has found that school-based accountability discourse networks were prominent among federal policymakers during the George W. Bush Administration, but this network was transformed by a teacher-based accountability discourse network during the Obama years (Galey-Horn et al., 2020). Importantly, emergent issue area networks can anticipate major policy change, such as the rise of the charter issue network in the 1990s and early 2000s. Factions and sub-divisions within coalitions, or the emergence of a third coalition, are often explained by disagreements over issue areas that challenge dominant belief systems (Sabatier & Weible, 2007). A key objective of discourse network analysis is to understand the ideational dynamics shaping these coalitions. Idea brokers play an important role in driving change within these discourse networks. Traditionally, brokers are conceptualized as actors who occupy positions of power in networks via their ties to disconnected actors (Burt, 1995; Gould & Fernandez, 1989; Marsden, 1982; Obstfeld et al., 2014). The concept of an idea broker is similar, but takes on a unique orientation in discourse networks given the two-mode structure. Rather than occupying the space between disconnected actors, idea brokers in discourse networks are positioned between otherwise disconnected policy preferences. For example, an actor who has a preference for teacher-based accountability and the expansion of charter schools is in a position of idea brokerage if another actor in the discourse network prefers teacher-based accountability but not charter school expansion (we refer to this as an “open brokerage chain” below). In the study by Galey-Horn et al. (2020), being affiliated with idea brokers through Congressional hearings in this way during the Bush Administration was predictive of a shift toward teacher policy consensus during the Obama Administration. In fact, the predictive power of being affiliated with a broker was stronger than party affiliation—a striking finding in today’s highly polarized political climate.

106

J. J. FERRARE ET AL.

In what follows, we illustrate two-mode techniques for measuring and analyzing policy preferences, idea brokerage, and research use behaviors in a prominent education policy network: the Policy Innovators in Education (PIE) network. The PIE network is an instructive context to examine these concepts, as its primary mission is to connect advocacy organizations in an effort to facilitate research and information exchange in order to bolster the influence of their work. In this sense, our analysis of the PIE network connects the literature on the issue preferences of education policy networks to the growing body of literature that seeks to understand the forms and roles of research use in education policy decision-making (Coburn & Talbert, 2006; Coburn et al., 2009; Finnigan & Daly, 2012; Finnigan et al., 2013; Scott & Jabbar, 2013).

Policy Innovators in Education (PIE) Network The Policy Innovators in Education (PIE) network was established in 2007 and has the stated objective to “connect state-level education advocacy organizations with colleagues across the country to amplify their voices and maximize their impact” (PIE, 2019). The PIE network was founded on the assumption that education reform efforts have to extend beyond the federal level to include work that is attentive to state and local complexities (McGuinn, 2012). The formal network context of PIE is intended to help coordinate these efforts through information and resource exchange, which ostensibly enables collaboration, innovation, and strategic response (PIE, 2019). To assist in these processes, the PIE network links state organizations to a variety of pro-reform think tanks (e.g., The Education Trust, National Alliance for Public Charter Schools, Thomas B. Fordham Institute) and foundations (e.g., Gates, Broad, Walton). Thus, unlike many policy networks, which are nebulous and difficult to identify (Ball & Junemann, 2012), the PIE network has clear boundaries that distinguish members from non-members. At present (June 2020), there are 88 members of the PIE network and 24 national partners spread across 32 states and the District of Columbia. The primary members include stand-alone organizations (e.g., Kids Ohio!, Oklahoma Business and Education Coalition) and those that affiliate with a national organization (e.g., Stand for Children Colorado, Democrats for Education Reform Washington). The national partners are made up of research organizations and think tanks that are generally sympathetic to the modern education reform movement that emphasizes

MEASURING ISSUE PREFERENCES, IDEA BROKERAGE, AND …

107

school choice and teacher-based accountability (e.g., Fordham Institute, Center on Reinventing Public Education). As a condition of membership (there are no monetary dues), the lead executive from each member organization must participate in at least one major PIE meeting per year, and member organizations must complete two annual surveys that collect information related to policy work and upcoming priorities. Participation is reviewed annually and failure to meet these conditions can lead to loss of membership.

Data The data used to illustrate two-mode methods of policy network analysis come from publicly available sources. Data collection began by downloading a wide range of content from n = 89 PIE members’ websites, such as mission or advocacy statements, summaries, and press releases. The latter content allowed us to build a broad picture of each organization’s policy preferences. In addition, we collected publicly available policy briefs and reports retrieved from PIE members’ websites. To ensure broad coverage of evidence use within the network, we collected the three most recently published reports from the n = 60 members who had publicly available reports published online during the data collection phase (summer 2018). Thus, a total of 180 briefs/reports were collected from PIE members’ websites. Finally, since the National Partners are meant to serve as a direct source of information and research for members, another source of data consisted of reports, policy briefs, and press releases gathered from the National Partners’ websites. In addition to collecting information related to each organization’s policy beliefs and preferences, we also gathered attributes corresponding to the organizations. The two binary attributes used in our analysis include whether or not a PIE member is part of a national parent organization (e.g., Democrats for Education Reform) and whether or not the organization joined during or after 2016. The cutoff of 2016 was chosen since this was the year that marked the start of a rapid expansion of the network from approximately 35 members to 89 members as of the writing of this chapter. Our illustrative analysis of these data proceeds in the following. First, we demonstrate our qualitative coding procedures that serve as the tools to construct two-mode discourse networks of PIE actors-by-events. In this case, “events” are PIE members’ policy preferences and research use

108

J. J. FERRARE ET AL.

behaviors. Next, we use two-mode applications of ERGMs and brokerage analysis to analyze the structure of policy preferences within the PIE network. Finally, we illustrate how correspondence analysis can be used to examine the structure of research use behaviors within the PIE network.

Discourse Network Analysis of PIE Members’ Policy Preferences and Research-Use Behaviors Our approach to analyzing the discursive aspects of policy networks is rooted in qualitative coding of publicly available documents. This approach affords researchers maximum autonomy over the coding process, enabling the use of theoretically grounded coding schemes and adaptive processes such as the constant comparative method (Corbin & Strauss, 2008). In what follows, we detail our approach to qualitative coding of PIE members’ policy preferences and research use behaviors. Coding Policy Preferences Qualitative coding strategies informed by grounded theory served as the foundation for systematically identifying PIE members’ policy preferences and research use behaviors (Corbin & Strauss, 2008; Glaser & Strauss, 1967; Saldana, 2013). A single coder conducted a line-by-line analysis of 35 randomly selected PIE network members’ and National Partners’ policy proposals while noting recurrent policy priorities. A policy advanced by a minimum of three organizations prompted the writing of an associated code. This process concluded with a total of 79 initial policy categories. These codes were then clustered and sub-categorized according to the primary focus of the proposed reform: schools, district leadership, principals, teachers, state governing bodies, or students. After formulating the preliminary coding scheme for shared policy priorities, further analysis revealed that while PIE network members advocate for a multiplicity of reforms, many members’ advocacy agendas share thematic or ideological principles (henceforth “policy preferences”). The identification of these policy preferences allowed for a more cogent system of organization; the preliminary 79 policy-type codes were thus further organized according to the conceptual threads underlying them. These policy preferences provided seven selective coding categories: (1) Autonomy and Deregulation, (2) Choice, (3) Early Childhood Opportunities, (4) Evaluation, Incentive, and Accountability, (5) Standards and

MEASURING ISSUE PREFERENCES, IDEA BROKERAGE, AND …

109

Quality, (6) Equity, and (7) Support. To provide a common example of the application of this coding taxonomy, a PIE network members’ proposal to tie teacher evaluations to student assessment scores was coded under the selective category “Evaluation, Incentive, and Accountability D.” The “D” subsection houses accountability-inspired proposals focused on teachers, as opposed to those concerned with governing bodies, schools, administration, or student data. This proposal was subcategorized under the policy code “Link teacher evaluation systems to student performance.”1 Following codebook construction, each PIE network member’s policy summaries were imported into Discourse Network Analyzer (DNA), a qualitative, category-based coding software which facilitates the generation of network data structures (Leifeld, 2013). Each PIE network member organization was tagged as an “actor,” and their chief policy proposals were coded as “statements” so that the strength of the ideological and advocacy priorities shared among them could be mapped and quantified. Throughout the coding process, several new policy categories were added through the constant comparative process to capture types of proposals not suggested by the initial sample (Glaser & Strauss, 1967). The number of organizations calling for charter authorizer reform, literacy achievement, and improved systems of principal evaluation justified the creation of novel policy categories, among others, resulting in a final count of 88 policy codes (available upon request). We broadened several of the earlier categories, most often to capture vague or unspecific policy solutions. For instance, the code for “Link teacher tenure to performance” was modified to capture general support for teacher tenure reform, as organizations did not always specify the type of tenure reform they favored. In rare cases, initial codes were combined. “Promote college readiness” was merged with “Promote career readiness,” for example, as these priorities appeared together with overwhelming consistency. For codes at the second level of abstraction, or those reflecting shared policy preferences, it was necessary to split the set of policy codes concerning Early Childhood Opportunities from those within the more

1 The codebook constructed by Galey and Ferrare (2016) and Reckhow et al. (2016) suggested several similar codes adaptable to our schema, particularly pertaining to teacher accountability. The present codebook is modeled after their structural framework, similarly featuring categories of policy proposals arranged within a hierarchy of overarching belief systems.

110

J. J. FERRARE ET AL.

generic Support category, and to divide proposals backing Autonomy and Deregulation from the set of Choice codes. A notable number of PIE network organizations professed a desire for reforms reflecting these values in se, meriting the creation of more precise, self-contained groupings. Approximately 30 anomalous proposals, defined as proposals advanced by fewer than three organizations total, remained outliers to the coding scheme, e.g. the Foundation for Florida’s Future’s support of virtual schooling and Mississippi First’s advocacy for comprehension sex education. These outliers excluded, most of the categories gleaned from the initial sample captured the remaining organizations’ substantive policy solutions and the shared ideologies underlying them. Several alterations and additions aside, the early codes did not require profound structural revision. Encountering few cases of ambiguity, most key policy proposals fell within the bounds of one or several of the initial or refined codes. Once coded, a simple descriptive look at the network offers an important summary of the policy preferences among network members. Figure 1 illustrates the actor-by-policy preference affiliation network. It

Fig. 1 Affiliation network of PIE members and their policy preferences (member node labels suppressed)

MEASURING ISSUE PREFERENCES, IDEA BROKERAGE, AND …

111

is clear from the graph that, at the time of data collection, issues of accountability and standards were the most central policy preferences structuring the PIE network. In particular, teacher standards and teacherbased accountability policies were advocated by more than half of all PIE members (59.3% and 54.3%, respectively). The preferences along the periphery of the network—such as school choice or equity-based funding—make up the advocacy agendas among subsets of PIE members, many of whom remain tied to the central issues of the network (i.e., standards and accountability). This basic graph serves as a foundation for additional analysis presented later in the chapter. Coding Research-Use Behaviors The coding of PIE members’ briefs/reports for the research use analysis followed a similar process as the more general coding of policy preferences. That is, a random sub-set of reports were thematically coded and refined and then the entire set of reports were coded following the constant comparative method as described above (Corbin & Strauss, 2008; Glaser & Strauss, 1967; Saldana, 2013). During the coding process, two distinct sets of codes were created. First, the specific policy topics of each report were coded. In total, there were 49 distinct policy topics covered in the sample of 180 reports. The policy topics covered a wide range, such as rural education, school climate, and teacher quality. In addition to coding the topics of each report, the sources of evidence cited within those reports were also coded. In particular, we coded each instance in which a PIE member used a citation to support a claim related to a policy topic. A total of 14 types of evidence were coded across 4,628 citations. As with the distribution of policy preferences described above, a simple look at the types of sources cited in PIE members’ advocacy content offers insight into the ways members use research to bolster their work. The range of evidence types used in this data set varied widely, such as academic articles, government agencies, National PIE Partners, and other PIE members. Table 1 illustrates the frequency distribution of the 14 sources of evidence that were categorized through the coding process. It can be seen that reports from government agencies represented the most common source of evidence used to substantiate claims, making up over one-third (36.9%) of all coded citations. The reports were sourced from a wide variety of agencies, from federal-level organizations (e.g., U.S. Department of Education) to state and local educational agencies.

112

J. J. FERRARE ET AL.

Table 1 Distribution of types of evidence cited in PIE members’ publicly available policy briefs/reports

Evidence type Government agency Academic Non-profit organization (other) Unknown News source PIE member PIE National Partner Think tank Assessment organization Law/Bill Foundation Consultant Education week Charter Authorizer Total

# of Citations

% of Citations

1709 558 466

36.9 12.1 10.1

387 311 303 210 148 147 146 96 79 50 18 4628

8.4 6.7 6.5 4.5 3.2 3.2 3.2 2.1 1.7 1.1 0.4 100.0

Academic sources—those that appeared in peer-reviewed journals or academic books—were the next most common form of evidence cited in the reports (12.1%). However, reports and briefs published by a variety of intermediary organizations—including foundations (2.1%), think tanks (3.2%), PIE Members (6.5%), PIE National Partners (4.5%), and other non-profit organizations (10.1%)—made up one-quarter (26%) of all evidentiary sources supporting the advocacy work of PIE Members. Thus, the vast majority of evidentiary sources used to substantiate the work of PIE members come from either governmental agencies or intermediary organizations. These descriptive accounts of policy preferences and research use behaviors are informative on their own. However, social network researchers often want to learn something about the structure of these “events” in two-mode networks. In the next two sections, we illustrate different techniques for exploring this structure within the context of PIE members’ policy preferences and research use behaviors.

MEASURING ISSUE PREFERENCES, IDEA BROKERAGE, AND …

113

ERGM Analysis and Identifying Idea Brokers In this section, we employ exponential random graph models (ERGMs) to determine which policy preferences are shaping the structure of the PIE policy network. We also examine patterns of idea brokerage to understand how influential policy actors in the PIE network promote their policy preferences. To begin, we use a two-mode version of ERGMs to estimate the probability of a tie between a PIE actor and a policy preference (for an overview of the model, see Cranmer & Desmarais, 2011). Similar to logistic regression, the variable of interest in the ERG model is binary—a “1” or a “0”—indicating the presence or absence of a tie. However, ERGM analysis accounts for interdependencies between observations, distinguishing it from regression analysis which assumes that only exogenous variables influence the outcome. In other words, in addition to exogenous factors, ERGMs also model endogenous configurations as a way of explaining network structure. This feature of ERGMs is especially relevant for constructivist theories of policy change, which conceptualize policymaking as a dynamic process between actors and ideas. ERGM analysis usually includes a term for edges, which functions like the intercept, as well as endogenous network terms and exogenous terms for node-level attributes.

Network (Endogenous) Terms Actor Activity Actor activity captures the tendency of actors to have multiple policy preferences, using geometrically weighted degree counts for the first mode (actors) in the network (Hunter, 2007). This term counts how many actor nodes have one connection to a policy preference, two connections, etc., and places a lower weight on larger numbers of connections using a geometric decay parameter. The closer decay is to zero, the more lower degree nodes are considered relative to higher degree nodes. The assumption is that larger numbers of connections are less prevalent than fewer connections. Preference Popularity The preference popularity term is known as a “star” term for the second (policy preference) mode because it counts how many actors refer to the

114

J. J. FERRARE ET AL.

same preference as a specific actor. Preference popularity accounts for a policy network’s dominant policy preferences and the structural tendency of policy network actors to cluster around those preferences.

Node-Level (Exogenous) Terms Since our analysis employed a two-mode ERGM, we incorporated nodelevel exogenous terms for actors and events. For the actor mode, we used the two binary attributes described above: whether or not a PIE actor joined in 2016 or later, and whether or not the actor belongs to a national umbrella organization. For the event mode, we included issue area attributes that indicate the relative popularity of particular issues among members of the PIE network. The five issue areas include: accountability, choice, standards, support, and equity. Equity is used as the reference category in our model because it had the lowest number of ties. In addition, we included a “same issue area” binary attribute indicating whether or not actors tended to support policy preferences in the same issue areas. Table 2 presents the coefficients for the ERGM. The results indicate that time of entry into the PIE network was a significant factor Table 2

Results of ERGM analysis of PIE network members’ policy preferences

Probability of a tie between actors and preferences in PIE network

Model 1

Model 2

Model 3

Edges Node Attributes (Mode 1) Joined 2016 or Later Same national network Node Attributes (Mode 2) Accountability Choice Standards Support Same issue area Endogenous Terms Actor activity (mode 1, α = 2) Preference popularity (mode 2, α = 2) AIC BIC

0.15

−1.978***

−2.304*

+ p