Policy Diffusion Dynamics in America

Policy Diffusion Dynamics in America

Policy Diffusion Dynamics in America integrates research from agenda setting and epidemiology to model factors that shape the speed and scope of public policy diffusion. Drawing on a data set of more than 130 policy innovations, the research demonstrates that the “laboratories of democracy” metaphor for incremental policy evaluation and emulation is insufficient to capture the dynamic process of policy diffusion in America. A significant subset of innovations triggers outbreaks – the extremely rapid adoption of innovation across states. The book demonstrates how variation in the characteristics of policies, the political and institutional traits of states, and differences among interestgroup carriers interact to produce distinct patterns of policy diffusion. Graeme Boushey is Robert Wood Johnson Scholar in Health Policy Research at the University of Michigan, on leave from his post as Assistant Professor of Political Science at San Francisco State University. His teaching and research are organized around practical and theoretical questions of state and federal policy making. He recently coauthored a review of individual and organizational decision making for the Handbook of Public Policy, and he also has coauthored an article in the Journal of Comparative Policy Analysis on immigration policy in federations.

Policy Diffusion Dynamics in America

GRAEME BOUSHEY San Francisco State University

cambridge university press Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sao ˜ Paulo, Delhi, Dubai, Tokyo, Mexico City Cambridge University Press 32 Avenue of the Americas, New York, NY 10013-2473, USA www.cambridge.org Information on this title: www.cambridge.org/9780521762816 © Graeme Boushey 2010 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2010 Printed in the United States of America A catalog record for this publication is available from the British Library. Library of Congress Cataloging in Publication data Boushey, Graeme. Policy diffusion dynamics in America / Graeme T. Boushey. p. cm. Includes bibliographical references and index. ISBN 978-0-521-76281-6 (hardback) 1. Policy sciences. 2. Diffusion of innovations – Political aspects – United States. JK468.P64B73 2010 320.60973–dc22 2010005435

I. Title.

ISBN 978-0-521-76281-6 Hardback Cambridge University Press has no responsibility for the persistence or accuracy of urls for external or third-party Internet Web sites referred to in this publication and does not guarantee that any content on such Web sites is, or will remain, accurate or appropriate.

To my parents, Homer and Virginia Boushey

Contents

List of Figures List of Tables Acknowledgments 1

Contagion in the Laboratories of Democracy

2

Incrementalism and Policy Outbreaks in the American States Policy Agents: Innovation Attributes and Diffusion Dynamics Innovation Hosts: State Characteristics and Diffusion Dynamics

3 4 5 6

Policy Vectors: Interest Groups and Diffusion Dynamics Conclusion

page ix xi xiii 1 22 62 92 139 169

Appendix A: List of Innovations Collected Appendix B: Policies Collected by Historical Era

187 193

Appendix C: Innovations Collected by Policy Type and Target Appendix D: State Receptivity to Innovation Ranked by Policy Type References Index

197 201 205 215

vii

Figures

1.1. 1.2. 2.1.

The epidemiologic triad of disease. page 11 The epidemiologic framework of innovation diffusion. 14 Distinct patterns of policy diffusion: The death penalty and state lotteries. 27 2.2. Distinct patterns of policy diffusion: Charter schools and the Amber Alert. 27 2.3. Episodic patterns of policy diffusion: Gubernatorial term limits and English Only legislation. 28 2.4. Newell’s bands of rationality. 35 2.5. S-shaped adoption curve. 39 2.6. S-shaped adoption curves representing three different rates of innovation diffusion. 43 2.7. R-shaped exponential adoption curve. 44 2.8. Simulated theoretical diffusion curves. 48 2.9. Cumulative distribution of adoption times: All policies. 56 2.10. Cumulative distribution of adoption times: 1900–1929. 57 2.11. Cumulative distribution of adoption times: 1930–1959. 58 2.12. Cumulative distribution of adoption times: 1960–2006. 59 3.1. Cumulative distribution of adoption times: All policies. 83 3.2. Cumulative distribution of adoption times: Governance policy. 84 3.3. Cumulative distribution of adoption times: Morality policy. 85 3.4. Cumulative distribution of adoption times: Regulatory policy. 86 3.5. Cumulative distribution of adoption times: Children’s policy. 87 ix

Figures

x

3.6. 4.1. 4.2. 4.3. 4.4.

Cumulative distribution of adoption times: Licensing policy. Map of state receptivity to innovation, 1960–2006. Map of state receptivity to morality policy innovation, 1960–2006. Map of state receptivity to regulatory policy innovation, 1960–2006. Map of state receptivity to governance policy innovation, 1960–2006.

88 102 121 122 123

Tables

2.1. 3.1. 4.1. 4.2. 4.3. 4.4. 4.5.

Statistical Tests for Normality by Historical Era Statistical Tests for Normality by Policy Type State Receptivity to Innovation, 1960–2006 State Receptivity to Innovation by Historical Era State Predictors of Innovation Receptivity, 1960–2006 State Receptivity to Innovation by Policy Type Predictors of State Receptivity to Regulatory, Morality, and Governance Policy, 1960–2006 (Baseline Model) 4.6. Predictors of State Receptivity to Regulatory, Morality, and Governance Policy, 1960–2006 (Full Model) 5.1. Interest Group Variation in Organization and Strategic Behavior

page 56 83 101 104 114 119 131 132 150

xi

Acknowledgments

This work explores the diffusion of public policy innovations in the United States. I became interested in the topic as a graduate student at the University of Washington, where I was introduced to research on policy making in federations. Although I appreciated the purported benefits of decentralized policy making, I could not reconcile ideal models of innovation and diffusion with my experiences growing up in California, where many of the prominent policies that the state adopted in the 1990s did not match the neat, cost-benefit decision-making processes outlined by researchers of an earlier generation. As I began to read research on agenda setting, it became clear that the process of public policy innovation and diffusion is dynamic, one in which incremental decision making is often interrupted by sudden moments of attention-driven policy change. I became interested in identifying what determines the pattern of innovation diffusion, whether by gradual increments or by sudden outbreak. In pursuing this interest, I could not help but notice that a similar pattern emerged in my own work – long periods of gradual improvement interrupted by new insights and sudden productivity. Unlike models of the policy process, the causes of these breakthroughs are easy to explain. They came after meetings and conversations with those friends and colleagues who graciously showed an interest in this project, and who took the time to offer suggestions for improvement. I wish to thank each of them for their attention and support. I am fortunate to have worked with an extraordinary group of people at the University of Washington, and thank each of them for the time and energy they invested in the development of this research. This book xiii

xiv

Acknowledgments

would not have been possible without the mentorship of my committee, Bryan Jones, Mark Smith, and Erik Wibbels. Bryan Jones encouraged my interdisciplinary approach to modeling diffusion dynamics and provided feedback and support for this research from its conception. My interest in policy diffusion grew out of challenging discussions with Mark Smith and Erik Wibbels, whose respective mastery of the literature on interest groups and federalism provided an invaluable resource as I began to explore the questions of why and how innovations spread across states. A number of others provided key advice at various stages of the project. Peter May volunteered feedback as I began to explore how the initiative process shaped the diffusion of innovations. I benefited enormously from the friendship and advice of David Olson, whose knowledge of state and local politics is limitless. Anne Ganley provided insight on how to organize and execute a major research project. John Ahlquist, Christian Breunig, and Josh Sapotichne were valuable critics and good friends. They pushed me to expand my conceptualization of the processes leading to innovation diffusion and often took time from their own research to help me work through various technical challenges that emerged as I worked on this project. I was surrounded by a group of colleagues and friends who made it a joy to study in the Department of Political Science at the University of Washington. Chris Koski, Rose Ernst, Sebastien Lazardeux, Ashley Jochim, Michelle Wolfe, and Samuel Workman each provided encouragement and assistance at important stages of this project. My thinking on public policy diffusion grew better from the exchanges we had over coffee in Gowen and Smith Halls. A number of others provided important comments on various stages of this research. Andy Karch read one of the first drafts of Chapter 2 and later provided feedback when I presented a completed manuscript at the University of Texas–Austin. Frank Baumgartner – a coauthor and former student of Jack Walker – made valuable suggestions that helped clarify how interest groups influence the process of policy diffusion. Frances Berry, John Fulwider, Michael Mintrom, Christopher Mooney, Karen Mossberger, Craig Volden, and Dick Winters offered criticism and comments at various panels over the past few years. I would also like to thank seminar participants at the University of Texas–Austin, who read the manuscript and provided lively feedback during a workshop and panel discussion.

Acknowledgments

xv

I was welcomed to San Francisco by a wonderful community of scholars. Max Neiman, of the Public Policy Institute of California, took an early interest in this project and graciously gave detailed comments on an entire draft of the manuscript. Richard DeLeon commented on a full draft, adding the perspective of a scholar who has made a career of studying the innovative politics of San Francisco. Jesse Cohen provided technical support for ArcGIS and helped produce the maps that appear in Chapter 4. The faculty and students in the political science department at San Francisco State University were supportive audiences as I worked on this research. I especially want to thank the students in my graduate seminar in American politics, whose comments on the manuscript gave me a fresh perspective as I neared completion. I am grateful to the School of Behavioral and Social Science and the Office of Faculty Affairs and Professional Development at San Francisco State University for their support of this research. Lew Bateman of Cambridge University Press has been a supportive editor and a valuable critic. He skillfully kept the project moving, providing important feedback at each stage of the review and revision process. I am in debt to the anonymous reviewers for their generous and useful comments. Each clearly invested a great deal of time and effort reviewing the manuscript, and I am certain that responding to their concerns improved this book. Their comments were careful and comprehensive and not only highlighted issues and ideas in need of improvement, but also provided concrete and useful directions that made revisions much easier. Any errors or omissions that remain are entirely my own. I am blessed to have a family that has provided unwavering support over the years that I have worked on this research. As I collected data for this project, my parents Homer and Virginia Boushey became active students of policy innovation and diffusion. My mother forwarded newspaper clippings about interesting new policies and the problematic legacy of California’s initiative process. My siblings Geoff and Sarah Boushey were close confidants when I became excited by a new idea or frustrated by a setback. My father, a professor of medicine at the University of California San Francisco, was always willing to read drafts of my manuscript, and provided perspective on examples I chose from studies of epidemiology. Finally, I would like to thank my wife, Sara Levine, whose friendship, support, encouragement, and patience have sustained me as I have worked

xvi

Acknowledgments

late evenings and long weekends on this project. I always found it easier to return to writing after the long hikes we took on the coast or in the redwood forests of Northern California, where I gained perspective in the calm that comes during a long walk with a good friend. I cannot imagine completing this book without her extraordinary love and support.

1 Contagion in the Laboratories of Democracy

In July of 1997, Dallas area child protection activists appealed to local police and media broadcasters to launch the nation’s first Amber Alert system, a crime prevention program enabling law enforcement agencies to activate regional emergency broadcast systems to announce missing children alerts.1 From these origins, the Amber Alert system evolved into one of the most successful interstate innovation campaigns in recent history. With strong support from child-protection and victim’s-rights advocates, every state in the union adopted the Amber plan between 1999 and 2005.2 The Amber Alert proved to be such an appealing response to kidnapping that identical versions of the child protection law were soon adopted internationally. Between 2002 and 2004, every Canadian province adopted the Amber program.3 In 2006, the United Kingdom launched its own version of the Amber plan called the Child Rescue Alert.4 1

2 3

4

Demands for the Amber Alert grew out of local outrage following the brutal kidnapping and murder of nine-year-old Amber Hagerman in 1996. Although a neighbor had witnessed the child’s kidnapping and contacted the police with a description of the vehicle, there was no way to broadcast the event to the broader public. For a brief history, see http://www.iowabroadcasters.com/ambrhist.htm; accessed August 2007. Oklahoma became the first state to adopt the Amber Alert in 1999. By 2003, the Amber Alert had been adopted by every state save Alaska and Hawaii. In the United States, Amber legislation stands for America’s Missing Broadcast Emergency Response. The Amber Alert legislation is therefore both a memorial tribute to Amber Hagerman and a description of the program. Interestingly, Canadian provincial Amber plans retained the tribute to Amber Hagerman in its legislation, speaking to the power of the image associated with the policy innovation. A summary of the efforts to internationalize Amber Alert legislation can be found on the website for the Center for Missing and Exploited Children www.missingkids.com; accessed August 2007.

1

2

Policy Diffusion Dynamics in America

Although the Amber Alert was exceptional in the sheer speed and scope of its implementation, such abrupt patterns of policy adoption are far from unique in American politics. The reenactment of the death penalty, prohibition, term limits, tax revolts, state auto lemon laws, English Only language legislation, “three strikes” sentencing guidelines, mandatory child auto-restraint requirements, and sex-offender registries stand as prominent examples of policy innovations that moved rapidly and extensively throughout the nation. Most of these innovations were championed by well-organized interest groups, and appealed broadly to voters across the states. In many cases, the innovation was adopted by more than 30 states in fewer than six years.5 In other cases, innovation spread suddenly over a subset of states before abruptly stopping. The sudden and rapid diffusion of innovations challenges traditional conceptions of policy making in the United States. Students of American government argue that federalism should exert a conservative pressure against rapid policy change.6 The implementation of identical public policies across states should be slowed by the multiple veto points of policy making in a federation, because innovation adoption requires an independent legislative decision by 50 state governments. Yet as the Amber Alert demonstrates, new innovations can and do spark positive feedback cycles leading to the sudden implementation of identical policies across states. Although such rapid standardization of state policies is often stimulated by intervention of the federal government through grants and other inducements,7 there is little evidence to suggest that rapid diffusion depends on the power and resources of the national government. In the case of the Amber Alert, 32 states had adopted the program before the federal government passed enabling legislation providing grants for state Amber Alert programs.8 In the case of the term-limitation movement, during which government reform activists imposed strict legislative term 5

6

7

8

This requirement for the scope and speed standard for unusually rapid diffusion was proposed by Savage (1985a) in his study of the rapid diffusion of public policies whose “time has come” (111). Baumgartner and Jones (1993; 2005) provide a thorough review and critique of models of policy change in federations. For a summary, see Agendas and Instability in American Politics, Chapter 11. National Interaction models of public-policy diffusion explore how federal intervention shapes public-policy diffusion. For a recent study of national interaction effects in policy diffusion, see Andy Karch’s “National Intervention and the Diffusion of Policy Innovations.” American Politics Research 34(4): 403–426 (2006). In 2003, the same year the federal government passed legislation to fund Amber Alert programs across the states, an additional 15 states enacted Amber Alert programs.

Contagion in the Laboratories of Democracy

3

limits on politicians across 20 state legislatures through the first half of the 1990s, interstate policy diffusion occurred absent the involvement of either the federal or state governments.9 Surprisingly, the rapid and sudden adoption of innovations across states is not well explained by extant studies of policy diffusion – the formal study of how ideas move from one jurisdiction to another in federations. Political scientists have generally explained policy diffusion as resulting from a process of incremental political learning by state governments (Walker 1969; Gray 1973; F. Berry and Berry 1990). The diffusion of innovations occurs through the “science of muddeling through” (Lindblom 1959), as government officials identify and emulate those policy innovations that present convenient or popular solutions to existing social or economic problems (Walker 1969; F. Berry and Berry 1999; Volden 2006). In their most common form, theories of public-policy diffusion anticipate that state decision makers identify policy problems and policy goals; engage in a limited solution search by exploring the policy solutions of peer jurisdictions; evaluate competing policy experiments for their efficacy; and, finally, select the “best” available policy solution. Diffusion research therefore gives primacy to the decision making of formal elected and appointed officials in state government, who identify, evaluate, and adopt emerging innovations that meet the challenges presented by interstate economic competition or address pressing social policy problems. Current research in state policy diffusion overlooks the causes of varying rates of innovation diffusion. Whereas the earliest studies in policy diffusion assumed an expressly comparative orientation to the study of policy innovation and adoption,10 modern research has assumed a narrower approach to documenting the processes leading to public-policy

9

10

The diffusion of state legislative term limitations overcame the significant opposition of elected representatives in state governments, who were reluctant to vote themselves out of office. A number of states (MA, WA, OR, ID, UT, WY) later repealed their laws. The two articles responsible for focusing political science research on the diffusion of innovations adopted a comparative approach to the study of policy innovation and adoption. Walker’s groundbreaking article, “The Diffusion of Innovations Among the American States,” (1969) explored general patterns of policy adoption across 88 distinct innovations that diffused across the states. Gray’s (1973) “Innovation in the States: A Diffusion Study” compared temporal and spatial patterns of policies across a range of different issue areas. These two articles sparked an important debate about the validity of generalizations drawn from comparative research on policy diffusion that continues to shape diffusion research today.

4

Policy Diffusion Dynamics in America

diffusion.11 The modern standard in diffusion research focuses on single case studies that are used to document the political and decision-making processes underlying a demonstrative case of innovation diffusion.12 This perspective fails to capture the complexity of policy making in American federalism. Often, a series of states adopt nearly identical policies in a very short time frame, suggesting decision making driven by sudden policy imitation rather than gradual incremental learning. Just as importantly, innovations often spread through the channels of direct democracy, beyond the direct control of state legislatures and without the input of bureaucrats or elected officials. Finally, diffusion research has understated the role of nongovernmental actors in policy diffusion. The diffusion of innovations is driven not simply by sequential emulation across state governments, but rather by carefully orchestrated pressure campaigns of organized interests that strategically work to see policies adopted in as many states as is feasible. The term-limitation movement of the 1990s demonstrates this dynamic. Term-limit activists operated outside of state legislatures and were uninterested in evaluating the impact that term limits would have on the future quality of state policy making. Term-limit activists instead coordinated initiative campaigns to push for governmental reform in as many states in as short a time as possible. The diffusion of state term limits shows little evidence of the incremental learning process familiar to state politics researchers. Perhaps because most research anticipates incremental learning as a centerpiece of the diffusion process, many of the most prominent and compelling cases of policy diffusion in recent U.S. history simply do not conform to the existing theoretical frameworks for how ideas move from one state to another. Although research in state policy diffusion has produced excellent descriptive studies of how economic competition or social-policy learning lead to innovation diffusion across state governments, this approach has been insensitive to the political and 11

12

This standard is partially shaped by the limitations of the event history models currently favored in innovation and diffusion research. These logistic time-series models permit researchers to model how changes in state internal dynamics and interstate interactions increase a state’s probability of adopting a single innovation at a given time. Perhaps the most widely known and cited piece of recent research in policy diffusion is a detailed study of how economic competition and geographic proximity spurred the diffusion of state lottery programs across states from 1960–1987 (F. Berry and Berry 1990). The popularity of the article is in no small part due to its groundbreaking introduction of the event history framework for diffusion research. However, it is telling that this preeminent piece of research on interstate policy diffusion addresses the gradual diffusion of a significant but not highly salient economic policy innovation.

Contagion in the Laboratories of Democracy

5

decision-making processes leading to both incremental policy adjustment and sweeping policy change. Diffusion research currently provides no framework to distinguish between the processes leading to the sudden diffusion of innovations like the Amber Alert, the gradual and steady diffusion of innovations like state lottery programs, and the episodic and periodic diffusion of public policies, such as term limits. To understand why policies follow such remarkably different temporal and spatial patterns of diffusion requires a new and explicitly comparative approach to the study of diffusion – one which moves away from single case studies and instead studies factors leading to variation in patterns of diffusion. If states truly are “laboratories of democracy,” then diffusion research must begin to account for the important causes of variation in the contagion and virulence of innovations that lead to such different patterns of policy diffusion. Overview of Research This book explores the underlying causes of diffusion dynamics – the processes underlying the stable, gradual diffusion of innovations over time and the sudden policy shocks precipitating positive feedback cycles and rapid policy mimicking across states. It advances an epidemiologic framework to understand the factors leading to variations in the rate of innovation diffusion, the relative susceptibility and immunity of states to innovation, and the critical role that interest groups play as carriers or vectors of innovations from one state to another. To understand the causes of diffusion dynamics, this project addresses two primary areas of inquiry. First, it updates the behavioral model of political decision making underlying the distinct patterns of incremental and nonincremental policy diffusion. In so doing, it distinguishes between the decision-making processes leading to gradual policy emulation and the pressures leading to sudden policy imitation. Second, it explores the characteristics that propel certain innovations across some states much more rapidly than others. In addressing these two areas of inquiry, this book provides a framework for the study of the contagion and virulence of innovations, and advances a theory for understanding the causes of policy outbreaks – a process characterized by a positive feedback cycle leading to the extremely rapid adoption of policy innovation across states. The book begins by generating a theoretical and empirical critique of theories of incrementalism in public-policy diffusion. The first section demonstrates that the popular model of incremental decision making

6

Policy Diffusion Dynamics in America

provides only a partial understanding of the behaviors leading to innovation diffusion. To model the joint processes of gradual policy change and sudden positive feedback cycles, this section develops an agenda-setting model of attention-driven political choice to capture the decision-making processes leading to diffusion dynamics. The second section of the book identifies factors leading to both positive feedback cycles and incremental patterns of diffusion. This section borrows from the study of epidemiology to conceptualize the distinct factors leading to the diffusion of innovations across states. This study argues that epidemiology can serve as a useful guide for the study of innovation diffusion. Policy innovations are the specific agents that are being transmitted through the population of states. States are the susceptible hosts that can adopt innovation. Finally, interest groups are the carriers or vectors that transmit policies from one state to another. A model of diffusion dynamics cannot be built around a single causal factor or process, but rather must account for how variation in the agents, carriers, and hosts of innovation shapes the process of diffusion. In three separate chapters, this research develops how systematic variation in the characteristics of policy innovations, the political and institutional traits of states, and differences among interest-group carriers all contribute to nonincremental patterns of policy diffusion. These two stages of the project are closely connected, and taken together the epidemiologic framework yields considerable insight into learning and decision making in federal systems. Diffusion dynamics are shaped by variations in the interactions of individual policies, state sociopolitical institutions, and interest-group organizations to produce different decision-making responses to policy innovations in the federal system. Different policy ideas produce nonincremental patterns of policy diffusion because they affect distinct decision-making processes by state decision makers, by elevating issue salience and arousing a sense of urgency, or by limiting issue salience and encouraging satisficing – a decision-making shortcut in which decision makers adopt the first available solution that is “good enough.” Differences in state receptivity to innovations are shaped by variations in the political and institutional capacities of state governments to process simple or complex political information. States are not uniformly receptive to all forms of innovation. Instead, variation in state political and institutional attributes makes them systematically more or less prone to adopt different forms of innovation. Finally, the interest groups that act as carriers or vectors

Contagion in the Laboratories of Democracy

7

of innovation produce drastically different patterns of diffusion, in part because they adopt different strategies when organizing pressure campaigns for innovation adoption. Variations in both the resources and the rhetorical strategies of interest groups agitating for innovation influence how state decision makers respond to calls for policy change. Political Decision Making and the Diffusion of Innovations The idea that policy change results from two distinct decision-making processes has recently gained traction in public-policy studies. Founders of an important research program documenting policy dynamics in American politics, Bryan Jones and Frank Baumgartner (1993; 2002; 2005) observe that “dramatic policy change occurs regularly in American politics, even if most issues most of the time are characterized by routine developments” (2002, 1). Periods of policy stasis and dramatic policy change are caused by changes in the allocation of government attention (B. Jones and Baumgartner 2005). Incremental policy change occurs through a negative feedback process, as risk-averse decision makers operating under severe time constraints make marginal adjustments to policy regimes in order to maintain the status quo. Sudden and dramatic policy change occurs through positive feedback cycles, as an event focuses mass political attention to a specific issue area, leading to increased demands and support for dramatic policy change.13 A growing body of policy research confirms the dynamics of negative and positive feedback cycles in policy making across an impressive array of American political institutions. Research in presidential decision making (Larsen 2006), congressional attention (Baumgartner and Jones 1993), and state budgeting (Koski and Breunig 2006), has demonstrated that policy making in American political institutions displays both extended periods of policy stasis and sudden moments of policy change. Similar dynamics have been documented in policy areas as diverse as gun laws (True and Utter 2002), crime control legislation (A. Schneider 2006), and environmental and nuclear energy policy (Baumgartner and Jones 1993). Policy change in each of these issue areas has occurred

13

A focusing event can be produced by a number of different factors, ranging from a natural catastrophe, an alarming shift in accepted policy indicators, or increased media attention on a policy problem. It need not be an exogenous shock to the political system.

8

Policy Diffusion Dynamics in America

through long periods of gradual policy adjustment interrupted by dramatic moments of sweeping policy reform.14 Despite anecdotal evidence that the process of policy diffusion is likewise prone to positive and negative feedback cycles, research has yet to connect the process of public-policy diffusion to the decision-making processes documented in the study of policy dynamics. To explain the causes of diffusion dynamics, this book begins by connecting the behavioral model underlying the larger study of policy dynamics to the decisionmaking processes leading to the diffusion of innovations. As with other research linking agenda setting to policy outputs, the processes of policy diffusion cannot be explained through a single static decision-making model, but rather must account for attention-driven pressures leading to both incremental policy adjustments and sudden nonincremental policy outbreaks. Chapter 2 develops an agenda-setting model of public-policy diffusion to account for attention-driven pressures leading to both gradual incremental diffusion and policy outbreaks. Drawing on research in individual and organizational decision making, this analysis demonstrates that differences in diffusion dynamics occur because state decision makers prioritize information differently based on issue salience, perceived importance, and issue complexity. State political institutions disproportionately respond to innovations that activate emotional considerations or address relevant and highly salient issues. They give scant attention to issues with less immediacy. Such disproportionate information processing by decision makers across the federation leads to different diffusion dynamics, and can be used to reconcile models of incrementalism with policy outbreaks. The incremental model of information processing proves accurate for a large subset of innovations that encourage neither elevated issue attention nor a sense of urgency; however, in certain instances innovations channel mass political attention leading to mass policy mimicking. In these cases, diffusion occurs as a positive feedback cycle, absent any form of instrumental program evaluation familiar to incremental models of diffusion. Chapter 2 introduces a stochastic process model to empirically evaluate the degree of incrementalism in the diffusion of innovations. This

14

Similar decision-making dynamics have been identified by a number of researchers studying individual and systemic decision making. For example, most social customs spread gradually through populations; however in social fads, many members of a group imitate a behavior nearly simultaneously.

Contagion in the Laboratories of Democracy

9

chapter introduces a unique data set measuring the state years of adoption for 133 policy innovations covering a wide range of innovations from the nineteenth, twentieth, and twenty-first centuries. This data set is used to evaluate general patterns of policy adoption in state policy making, asking specifically whether patterns of policy diffusion can be characterized as resulting from a process of incremental or nonincremental political decision making.15 The findings presented in Chapter 2 reveal that policy diffusion displays punctuated dynamics that are inconsistent with incremental policy learning and emulation. Instead, policy diffusion has occurred more rapidly than expected in incremental learning models, indicating a process of incrementalism interrupted by sweeping policy outbreaks. Modeling Diffusion Dynamics: Are Public Policies Some Kind of Disease? Taken by itself, analysis of models of decision making can provide only cursory insight into the causes of diffusion dynamics in American federalism. The distinct patterns of policy diffusion instead suggest some compelling questions about the process driving the diffusion of innovation. Why do some policy innovations spread much more rapidly than others? Why are some states receptive to certain forms of innovation when others appear policy-resistant to even the most popular state-level reforms? How does the involvement of nongovernmental actors shape the diffusion of innovations from one state to another? Resolving each of these questions leads to a greater understanding of the dynamics underlying the diffusion of innovations. To model the causes of diffusion dynamics, the second section of this book conceptualizes the diffusion of innovations from the perspective of epidemiology, a discipline expressly dedicated to evaluating how changes in the environment, the virulence of agents, the behavior of vectors, and the attributes of susceptible and resistant hosts interact to shape the distribution and determinants of disease.16 The basic approach of epidemiologic research encourages comparison to the study of the diffusion of 15

16

Chapters 2 and 3 rely on distributional analysis to compare empirical patterns of policy diffusion to simulated patterns associated with incremental diffusion. A more detailed discussion of the approach is described in each of these chapters. Although there are clear limits to comparing the diffusion of innovations to the communication of disease, it is worth noting that diffusion researchers have long drawn inspiration from epidemiology to describe the underlying processes of diffusion. The

10

Policy Diffusion Dynamics in America

innovations. Whereas the study of policy diffusion is focused on understanding the determinants and distribution of policy ideas and innovations across policy-making jurisdictions, epidemiologists have focused on understanding both the distribution and determinants of “health-related states or events in specified human populations” (Last 2001, 62). The epidemiologic framework is especially appropriate for evaluating the incremental and nonincremental diffusion dynamics of central interest in this book. Epidemiologists have explored factors contributing to the speed and scope of outbreaks over time: the comparative virulence of the causes of disease (bacterium, virus, toxin, etc.); the distribution and activity of the carriers of “vectors” transmitting the pathogenic agent; and the susceptibility of the populations exposed (see Text Box 1.1). These areas of inquiry are similar to studies of public-policy diffusion that have explored how the internal dynamics of states make them more or less susceptible to innovation (Walker 1969; Savage 1978; Canon and Baum 1981; Carter and LaPlant 1997); how the distribution, activity, and interactions of interest-group carriers shape the diffusion of innovation (Gray 1973; Mintrom 1997); or how changes in the policy idea itself can lead to the sudden spread of policy innovation (Savage 1985a; Mooney and Lee 1995). Importantly, the epidemiologic framework encourages researchers to move away from descriptive studies documenting individual policy diffusion and toward new questions about comparative diffusion dynamics and the joint processes of incremental and nonincremental policy change. Mapping the Diffusion of Disease in Epidemiology Figure 1.1 shows how public health researchers explore variation in each of four general factors to understand the dynamics of disease in human populations. A researcher interested in modeling the incidence, severity, and rapidity of transmission of disease in a population must account for change in environmental conditions, the characteristics of the carriers or vectors of disease, the genetic or behavioral traits of the host determining susceptibility to infection, and the unique attributes of the agent.

event history models currently favored in diffusion research were pioneered by epidemiologists interested in understanding when and why individuals in a community succumb to illness. In the emerging discipline of memetics, researchers argue that cultural norms and common policy ideas may actually replicate and spread like viruses (Dawkins 1989; Aunger 2002).

Contagion in the Laboratories of Democracy

11

Host Characteristics

Environment

Agent Characteristics

Vector Characteristics

figure 1.1. The epidemiologic triad of disease. Source: Gordis, Leon. 2004. Epidemiology. Philadelphia: Elsevier Saunders, p. 16.

Of course, the study of public health processes is complex. Although health scientists organize their research to isolate and evaluate each of these elements of epidemiology, these factors interact to produce distinct patterns of disease. For example, the introduction of a new strain of the influenza virus in a dense urban setting can lead to a sudden outbreak of flu, as the denser population exposes a greater number of susceptible individuals to the more virulent strain of the disease. Whereas introduction of the same strain into a sparsely settled rural population may result in only a few isolated cases, with the agent most often “dying out” before it can be transmitted. Text Box 1.1 illustrates how an epidemiologist might organize research into each factor of disease dynamics by discussing a hypothetical emergence of a new, dangerous form of the malaria parasite in the South. The Epidemiologic Framework of Interstate Policy Diffusion The epidemiologic triad maps nicely as an organizing framework for the study of policy diffusion. A survey of recent research in the diffusion of policy innovations suggests the epidemiologic framework can be applied to isolate and classify factors contributing to the diffusion of policy innovations across jurisdictions. Researchers interested in how state geographic proximity (F. Berry and Berry 1990; Rincke 2004) or media agenda setting and public opinion (Hays 1996) shape patterns of diffusion have focused on how changes in the political environment spur policy

12

Policy Diffusion Dynamics in America

text box 1.1 Thinking about Diffusion: Tracing an Outbreak of Malaria Consider the potential causes of an outbreak of malaria in the United States. An epidemiologist would recognize that the epidemic could be spurred from a number of factors, including a change toward environmental conditions more favorable to malaria, a change in the carrier of the malaria parasite, a shift in the virulence of the malaria parasite, or a shift in the genetics or behaviors of the human population threatened by disease. The researcher would begin by exploring the many well-known environmental factors that have long been associated with the spread of malaria. In the 1940s and 1950s, the Centers for Disease Control and Prevention (known at that time as the “Communicable Disease Center,”) (CDC) eradicated malaria by eliminating favorable environmental conditions for the breeding of mosquitoes, first draining stagnant water pools where the Anopheles mosquito breeds and then spraying pesticide and petroleum to exterminate mosquito larvae (CDC 2004). As a first step, researchers interested in the resurgence of malaria would begin evaluating changes in the environmental conditions associated with the disease. For example, they could ask if recent flooding increased the number of stagnant pools of water favored by mosquitoes for breeding. Perhaps a shift away from pesticide use or other mosquito-control techniques contributed to a marked increase in the malariacarrying mosquito population. Finally, a change in climate or an increase in temperature could have increased mosquito breeding and forced human populations outside, and thus into contact with the disease carriers more frequently. Each of these lines of inquiry stems from a common observation. Environmental conditions are closely related to the spread of malaria, and a small change in environmental conditions could precipitate a sudden uptick in the incidence of disease. Of course, because malaria has been dormant in the United States for more than 50 years, it is unlikely that these environmental conditions would be the only factor contributing to hasten the spread of the malaria parasite. Students of mosquito vectors have observed considerable genetic and behavioral variation in the different Anopheles species that carry the malaria parasite (CDC 2004). Some Anopheles mosquitoes are more effective at carrying and transmitting the malaria parasite than others. Although there are 430 species of Anopheles mosquitoes, only 30 to 40 are known carriers of the parasites (CDC 2004). Of these, many Anopheles mosquitoes are relatively poor hosts for the malaria parasite, whereas others are especially effective at transmitting the parasite. Even when the ability of the mosquitoes to carry the malaria parasite is equal, some vectors are more likely to transmit the disease than others while

Contagion in the Laboratories of Democracy

13

feeding. Many Anopheles mosquitoes are zoophilic. Some species prey aggressively on humans, whereas others feed primarily on other animals. Finally, some Anopheles vectors have developed a resistance towards pesticide or repellent and thus are more able to overcome existing abatement programs to address the disease. In addition to measuring changes in the environment, public health officials dealing with an outbreak of malaria would begin documenting any changes in the vectors or carriers of disease. The introduction of a new species of mosquito could lead to an influx in mosquito-borne disease, whereas a change in the behavior of a native mosquito could also lead to a reemergence of malaria transmissions. In addition to documenting changes in the carriers of malaria, researchers would also be interested in measuring changes in the contagion of the agent, and begin researching potential changes in the malaria parasite itself. As with the vectors of malaria, there is considerable variation across different forms of the malaria parasite. The P. falciparum malaria parasite is associated with higher incidence of human mortality, while the P. vivax and P. ovale parasites remain dormant in humans for months and years before proliferating, making the best-laid malaria-control programs difficult (CDC 2004). A mutation in any malaria parasite could make transmission from vector to host more likely than before, especially if malaria parasites become immune to treatment methods. An introduction of a new form of the parasite could easily account for a sudden increase in the rate of infection in a region where malaria had previously been eradicated. Finally, a student of public health might become interested in the changes in the hosts of malaria, and begin to explore genetic and behavioral traits of people who became infected with the parasite. Many individuals of African ancestry have sickle cell trait and are resistant to common forms of the malaria parasite. A demographic shift increasing the population without this trait could lead to an upsurge in transmissions of the mosquito-borne disease. Community health officials might also explore the behavioral factors associated with exposure to the malaria parasite. A decline in preventative behaviors such as the use of bed nets or mosquito repellent could also shape the rate of transmission leading to an outbreak of malaria in the region. The example of malaria provides some context to the epidemiological approach to understanding the dynamics of disease. Rather than being limited to a single explanation or cause, the framework encourages researchers to generate theory about a number of factors leading to the outbreak of malaria, including how environmental conditions, the traits of the vector, the characteristics of the agent, and the behavior and characteristics of the host itself all contribute to the transmission of disease.

Policy Diffusion Dynamics in America

14

State Characteristics

Issue Salience National Mood

Innovation Characteristics

Interest Groups & Professional Organizations

figure 1.2. The epidemiologic framework of innovation diffusion.

diffusion. Researchers documenting the role of professional or political networks in the diffusion of innovation have explained the role of carriers in the innovation and diffusion process (Gray 1973; Mintrom 1997). Studies on the internal political, social and institutional characteristics of early and late adopting states have identified how the characteristics of hosts shape state receptivity to innovation (Walker 1969; Savage 1978; Canon and Baum 1981; Carter and LaPlant 1997). Finally, a select few researchers have focused on the virulence of the idea itself, and have offered theory for how specific characteristics of policy agents shape the rate and extent of diffusion (Savage 1985a; Mooney and Lee 1999). As illustrated in Figure 1.2, the movement of policy innovations across jurisdictions is shaped by the changes in the political environment, the carriers of innovation, the distinct traits of each American state or jurisdiction, and the unique characteristics of the innovation agent. This framework has significant advantages for capturing the dynamic processes of policy diffusion. The epidemiologic framework connects research from students who have focused on such varied factors leading to policy innovation, adoption, and diffusion as the strategic behavior of interest-group communities, the characteristics of states, changes in national mood, or the emergence of a new policy idea into a single conceptual framework. Moving towards an epidemiologic framework further encourages researchers to consider intriguing questions about the virulence of policy ideas, the different strategies and behaviors of policy entrepreneurs and interest groups, and the varying resistance and susceptibility of states to innovation. It is an explicitly comparative approach. It attempts to understand the dynamics of diffusion by differentiating

Contagion in the Laboratories of Democracy

15

between the attributes of states, interest groups, and policy innovations that determine the distribution of innovation through the federation. Using the epidemiologic framework as a loose guide for organizing research in diffusion dynamics allows us to draw a number of compelling implications for the causes of policy outbreaks. Text Box 1.1 demonstrates how a political scientist might follow the epidemiologic framework to conceptualize key factors associated with the sudden emergence and spread of anti-crime policies in the United States. text box 1.2 Thinking about Diffusion: Tracing the Spread of Anti-Crime Policies To illustrate how the epidemiologic framework can be transferred to the study of policy diffusion, consider how a political scientist familiar with the basic framework of epidemiology would study the sudden spread of anti-crime policies like the Amber Alert or three-strikes sentencing guidelines in the 1990s. The researcher would recognize that the sudden diffusion of these anti-crime innovations could be examined from a number of different perspectives, ranging from a change in national or state political conditions more favorable to “tough on crime” legislation, an evolution in the type of anti-crime policy favored by state legislatures, the emergence of new interest groups advocating for “tough on crime” legislation, or a change in political, cultural, or institutional makeup of the states adopting new crime policies. In organizing their research around these questions, the political scientist would be following a framework familiar to epidemiologists. This research would explore how the variation in the agents, carriers, and hosts of anti-crime innovation shaped rates and patterns of public-policy diffusion. As a starting point, a crime policy analyst would begin by looking at how changes in the political environment shaped the sudden emergence of state anti-crime policies. For example, analysis of Gallup’s national “most-important problem” (MIP) data reveals that Americans were disproportionately concerned with the crime problem in the mid 1990s (www. policyagendas.org/datasets/index.html#mips).∗ While historically fewer than 5% of Americans have identified crime as one of the most important problems facing the nation, from 1994 through 2001 an average of 25% of respondents (continued) ∗

Trends in Gallup’s national “most-important problem” (MIP) data can be accessed through the Policy Agendas website, at www.policyagendas.org/datatools/toolbox/ analysis.asp. To view trends in national opinion on crime and crime control, navigate to this webpage and select the “Most Important Problem (1947–2007)” and “all subtopics in Law, Crime and Family issues” options. These MIP data are included as part of a larger collection of data made available through the Policy Agendas Project, directed by Bryan Jones, Frank Baumgartner, and John Wilkerson.

16

Policy Diffusion Dynamics in America

in the Gallup MIP poll viewed crime as a pressing problem facing the nation, suggesting political conditions were ripe for crime-policy innovation. Other political scientists might explore whether changes in crime rates or unemployment statistics were associated with periods of state anti-crime innovation. Each of these questions are related to the assumption that state policy-making activity is in some way responding and reacting to the surrounding policy environment. Yet it is unlikely that elevated attention to the problem of criminal justice policy would be the only factor driving the rapid spread of anti-crime policies like the Amber Alert or the three-strikes law. Students of the policy process have commented on the important role of policy activists and organized interests as the carriers of innovation in the United States. Political scientists have documented a good deal of variation in the size, strength, and strategies of interest groups operating in the United States. Interest groups specialize according to lobbying strategies, membership size, financial resources, and organization. Some nationally prominent interest groups benefit from a reserve of financial resources or possess the advantages of mass membership and grassroots participation. Other interest groups have limited membership and struggle to capture resources needed to generate political support for their policy goals. These distinctions relate to the ease with which different carriers of innovation could orchestrate a diffusion campaign across the 50 states. The involvement of well-organized and well-funded groups – such as the National Center for Missing and Exploited Children or Mothers Against Drunk Driving – could accelerate the diffusion of an anti-crime initiative by actively carrying it from state to state using existing networks and traditional pressure campaigns. Interest groups that suffer from a relative dearth of financial or human resources – such as the National Organization for the Reform of Marijuana Laws – would struggle to simultaneously pressure for policy change across states. Other political scientists might explore how the characteristics of the innovation agent themselves shape diffusion, exploring how the cost, complexity, or rhetorical framing of an innovation shaped patterns of policy adoption. Child protection policies like the Amber Alert or Megan’s Law directed public attention towards the visceral need to protect young children from violence. The three-strikes law cast mandatory sentencing laws in populist and easily accessible language that appealed broadly to American voters. Although these policies varied according to the ultimate cost, none involved terribly complex or sophisticated expertise to understand how the policy was intended to diminish crime. A similar dimension can be used to explain the slow spread of criminal justice reform policies in recent history. Efforts to address social or public health problems associated with drug use face the challenge of convincing voters that drug use is a problem of public health rather than criminality. In each of these cases, researchers can assess how

Contagion in the Laboratories of Democracy

17

changes in the scope, cost, or presentation of individual policies shaped their rates of diffusion. A final set of questions revolves around how variations in the political, institutional, and cultural traits of states could lead to their resistance or susceptibility to anti-crime policies. These questions implicitly ask how the unique makeup of each of the 50 state hosts of policy innovation shape patterns of policy adoption. For example, 24 states in the union allow citizens to pass either constitutional amendments or legislation through direct democracy (www. iandrinstitute.org). These states thus have an additional institutional avenue for innovation adoption that is especially responsive to interest-group influence and majoritarian politics. Conversely, states without direct democracy might be generally resistant to populist anti-crime movements. Crime policy researchers have operationalized a number of ways for how variations across each of the fifty states (racial demographics, political culture, wealth, professional legislatures, etc.) shape anti-crime policy innovation and adoption. All of these studies explore common questions for how variations in state innovation hosts shape the probability of public-policy adoption. This discussion provides some context for how an epidemiologic framework can be applied to the study of public-policy diffusion. Instead of being limited to a single explanation for the diffusion of innovation, the framework encourages public-policy researchers to think broadly about how changes in the political environment, interest-group carriers, the unique attributes of the innovation, and the political, cultural, and institutional characteristics of states all contribute to public-policy diffusion.

Innovation Characteristics and the Diffusion of Innovations A first major set of implications drawn from the epidemiologic framework of policy diffusion is that the characteristics of innovations or policy agents themselves contribute to diffusion dynamics.17 As with strains of malaria or the flu, not all innovation agents interact the same way with states in the federation. Instead, policy ideas possess attributes that make them more or less likely to produce policy outbreaks. For example, the Amber Alert possessed special qualities – targeting the protection of children through the use of existing emergency broadcast technologies – that made it especially appealing to voters across states and countries. 17

The question of how variation across different innovations shapes rates and patterns of diffusion has been largely overlooked in policy diffusion research. For a summary of research and future directions in this area, see Andrew Karch’s (2007b) “Emerging Issues and Future Directions in State Policy Diffusion Research.” State Politics and Policy Quarterly 7(1): 54–80.

18

Policy Diffusion Dynamics in America

Chapter 3 identifies a number of key innovation attributes that contribute to diffusion dynamics. First and foremost, policy innovations represent issues with different natural levels of issue salience. Because diffusion dynamics are in part a function of attention allocation in political decision making, high-salience issues should be especially prone to positive feedback cycles across states, whereas low-salience issues should encourage incremental decision making, as they fail to focus mass political attention. Many policies – like those related to education, defense, or government fiscal policy – have naturally elevated issue attention, whereas others are traditionally lower priorities for governments. However, issue salience is also shaped by the changing ways governments describe a policy problem, and shifting rhetoric around the way people understand a policy problem can shape mass political attention and elevate issue salience. Beyond salience, a number of other characteristics shape the probability that a given innovation will interact with states to produce a policy outbreak. Innovations vary according to issue complexity and program cost. Innovations that require professional technocratic program analysis or the allocation of significant government resources will tend to diffuse more slowly than cheap innovations that require little in the way of policy expertise. These attributes are not unique to individual innovations and can be assessed to classify policies that are more or less likely to produce policy outbreaks in the federation. Researchers who have studied common attributes of distinct classes or types of public policies have been interested in how “policy determines politics” (Lowi 1972, 300). Recent research connecting policy types to the diffusion of innovation suggests that public policies with common attributes follow similar patterns of policy diffusion (Mooney and Lee 1999). For example, morality policy is characterized by elevated issue attention, low complexity, and high emotional appeal. If elevated issue salience and diminished issue complexity are connected to rates of diffusion, morality policies should be especially prone to policy outbreaks. On the other hand, state regulatory policy – a policy form typified by high technical complexity and low salience – should conform closely to incremental patterns of policy diffusion, as this class of policies rarely engages mass political attention. Finally, governance policy – a policy type where citizens regulate the behavior of elected government officials and political institutions – should be especially prone to nonincremental patterns of diffusion. Governance policy is not only marked by high salience and low complexity, but is also strongly associated with the political institution of the direct statutory citizen initiative.

Contagion in the Laboratories of Democracy

19

Governance policy should therefore be entirely unrelated to incremental patterns of policy diffusion, and especially prone to policy outbreaks, although the outbreak may be limited to states with citizen initiatives. Chapter 3 evaluates how the characteristics of the policy agent encourage diffusion dynamics. Building upon the statistical method advanced in Chapter 2, Chapter 3 demonstrates that different policy types produce dramatically different diffusion dynamics. Some classes of innovation produce patterns of diffusion that closely match incremental decisionmaking processes, while others are much more likely to produce policy outbreaks. State Characteristics and Diffusion Dynamics A second major implication of the epidemiologic framework is that the characteristics of the states should, insofar as they shape susceptibility to policy innovations, lead to considerable variation in patterns of diffusion across states. The diffusion of innovations is not simply determined by whether or not a given state came into contact with innovation at a certain moment. Instead, states are themselves more or less receptive to distinct forms of innovation. The diversity of state political, institutional, and socioeconomic characteristics causes considerable variation in state receptivity to innovation. Chapter 4 assesses state receptivity to classes of morality, regulatory, or governance policy.18 Because these policy forms encourage different processes of political decision making by state governments, state receptivity to morality, governance, and regulatory policy innovation will vary based on each state’s capacity to engage with decision making associated with specific types of innovation. A quick profile of state political and institutional characteristics provides good reason to expect that state receptivity to morality, governance, and regulatory policy will differ depending on differences across key state political and institutional attributes. States with the direct statutory initiative process are most receptive to governance policy reforms, as 18

Chapter 3 elaborates on the distinction between regulatory, morality, and governance policies. State regulatory policy encompasses economic, environmental, and professional regulatory regimes where government uses its coercive authority to shape the behavior of private businesses, organizations, and individuals. Morality policy refers to social regulatory policies where government uses legal authority to dictate social, community, and moral values and practices (Tatalovitch and Daynes 1998). Governance policy addresses those policies which delineates the terms and limits of governmental authority itself.

20

Policy Diffusion Dynamics in America

the initiative process provides an additional venue for direct citizen regulation of government institutions and elected officials. States without the initiative process should be resistant to governance policy, as citizen activists must appeal directly to government officials to regulate their own behavior. State receptivity to regulatory policy (addressing economic, environmental, or professional regulation) is shaped by an entirely different set of characteristics. State receptivity to regulatory policy is directly related to a state’s capacity to engage in complex policy analysis and experiment with technically complicated regulatory design and implementation. Thus, states with more professional legislatures are more receptive to regulatory policy innovation, as legislators are provided with the resources and the staff support to engage in technical program analysis. Likewise, those states with citizen legislatures are resistant to regulatory policy innovation, as they have neither the resources nor the capacity to engage in technocratic regulatory policy analysis. Instead, states with citizen legislatures lag in the adoption of regulatory policy innovation, as they must rely on the analysis and experiences of other states to suggest appropriate state regulatory regimes. Finally state responsiveness to morality policy is shaped by citizen ideology and state levels of electoral competition. Because morality is high salience and encourages mass participation, state receptivity to morality policy is shaped by state legislative responsiveness to citizen demands. Chapter 4 evaluates how variation in the internal political, institutional, ideological, and sociodemographic characteristics of states shapes receptivity to classes of innovation. Chapter 4 builds upon the decisionmaking model to evaluate how state decision-making capacities lead to differences in receptivity to morality, regulatory, and governance policies. This chapter first ranks states according to the speed with which they adopt emerging regulatory, morality, and governance policies. It then generates a statistical model to evaluate those attributes that make states receptive or resistant to different classes of public policy. Policy Vectors: Interest-Group Networks and Diffusion Dynamics A final implication of the policy contagion model is that the carriers of policy innovation produce distinct patterns of diffusion dynamics. Although state governments hold a central place in studies of state policy diffusion, elected officials are not necessarily the primary vectors of innovation in the United States. Instead, policies are communicated across states by interest-group activists, who carry innovation from one state to another

Contagion in the Laboratories of Democracy

21

through organized interest-group networks. Variation in the behavior and organization of policy vectors has important implications for patterns of policy diffusion. Some groups possess the resources and membership to aggressively pursue interstate diffusion campaigns across all states in the federation, whereas other interest groups are limited in their ability to advocate for the diffusion of innovations beyond a small subset of receptive states. Chapter 5 explores how variation in the organization and behavior of interest-group networks and professional organizations shapes diffusion dynamics. This chapter draws upon a series of historical case studies to illustrate how variation in interest-group organization and behavior precipitates both incremental and nonincremental patterns of policy diffusion. This section augments empirical findings in the previous sections of the book by documenting how activists and interest-group organizations are integral to interstate policy diffusion. These case studies consider how political activists or “policy entrepreneurs” strategically exploit the multiple venues of policy making in America, revealing how strategic issue framing and reframing, venue shopping, and the organizational structure of interest-group networks interact to shape diffusion dynamics. Implications This book contributes to the understanding of policy making in federations by exploring how innovations emerge and spread from one jurisdiction to the next. The project advances a model for understanding the considerable variation in the patterns of policy diffusion across states. The book first demonstrates that policy diffusion exhibits punctuated dynamics that are indicative of a mixed model of policy change. This approach confirms prior findings of incremental learning across state legislatures, but then further demonstrates how attention-driven policy prioritization causes positive feedback cycles leading to policy outbreaks. The book then introduces an epidemiologic model of public-policy diffusion in order to provide theoretical and empirical leverage for understanding how variations across states, innovations, and interest-group networks produce positive feedback cycles leading to policy outbreaks across states. This approach moves the focus away from a narrow study of state legislative decision making, and instead reveals the importance of the interactions between interest-group activists, the nature of the policy innovations, and the receptivity of states to the communication of innovations in federal policy making.

2 Incrementalism and Policy Outbreaks in the American States

This chapter builds upon recent advances in the study of policy dynamics to examine the decision-making processes associated with the diffusion of policy innovations. It demonstrates that patterns of policy diffusion display punctuated dynamics inconsistent with a single process of incremental learning, but instead indicate multiple underlying decision-making processes. This is perhaps shown most forcefully by the abrupt and rapid spread of policy innovations across states in American history, a diffusion that cannot be explained through the process of incremental learning but must rather reflect decision making under extreme pressures for change placed on state governments. A more nuanced theoretical understanding of the process of publicpolicy diffusion can be gained by integrating research on innovation diffusion with studies of agenda setting in political decision making. The agenda-setting perspective demonstrates that government attention is unequally allocated in political decision making. State decision makers prioritize and respond differently to competing streams of information based on perceived issue importance, salience, and urgency. The diffusion of public policies often conforms neatly to the process of policy identification, evaluation, and emulation central to most modern research on innovation diffusion. However, a significant subset of policy innovations attracts immediate and widespread attention. These policies encourage immediate political responses, leading to the outbreak of nearly identical policy innovations across states. If diffusion dynamics are truly driven by two or more paths of organizational information processing, then the existing incremental decision-making model is insufficient to explain the processes of diffusion dynamics. 22

Incrementalism and Policy Outbreaks in the American States

23

To evaluate how agenda-setting dynamics shape innovation diffusion, this chapter explores distinct patterns of policy diffusion, drawing upon a data set of state adoption times for 133 different policy innovations. The chapter introduces a method – familiar to students of agenda setting – to compare the distributions of innovation-adoption times in state policy making. It compares historical patterns of policy diffusion in the American states to simulated patterns that replicate diffusion processes if diffusion were purely driven by incremental learning. This comparison is especially appropriate for modeling the diffusion of innovations because observational data of both incremental decision making and the diffusion of innovations are expected to produce an S-shaped normal curve of policy adoptions over time (Gray 1974; Rogers 2003). A simple comparison of real and simulated diffusion curves indicates how well diffusion processes fit incremental learning patterns (Padgett 1980; Rogers 1983; Baumgartner and Jones 1993). This chapter proposes that the degree to which a given empirical pattern deviates from the simulated S-shaped pattern indicates the degree of incrementalism versus positive feedback cycles in policy diffusion.1 This chapter then extends this analysis by exploring diffusion dynamics in three historical eras spanning the twentieth century. A number of researchers have observed that advances in communications technology have shaped the pace of public-policy diffusion in the modern era (Savage 1985a; Mossberger 2000). The chapter extends the analysis by comparing the underlying patterns of adoptions for policies diffusing in the early, middle, and late twentieth century. Disaggregating policies by historical era allows for a simple test of whether nonincrementalism in public-policy diffusion is shaped by changes in communications technology across historical eras. The findings presented in this chapter complement recent research in political decision making (B. Jones and Baumgartner 2005). Diffusion dynamics result from two distinct processes of state institutional decision making – one process representing incremental policy adjustments and another, representing sudden moments of policy imitation. These two 1

This chapter employs distributional analysis to compare the distributions of simulated and real diffusion data. This approach is facilitated because of the properties associated with data collected from incremental diffusion processes. When plotted, the pattern of policy adoptions over time is expected to be not only S-shaped, but also normally distributed. Because policy diffusion is traditionally believed to be normally distributed, a series of statistical tests determining how well historical diffusion data matches a normal distribution allows speculation about models of policy diffusion.

24

Policy Diffusion Dynamics in America

processes result in remarkably different patterns of diffusion over time. Many policies (like state lotteries or the diffusion of living-will legislation) produce S-shaped patterns of adoption consistent with boundedly rational information processing and policy emulation by state legislatures (Rogers 1983; Mooney and Lee 1999). However, as outlined in the previous chapter, many of the most sudden and interesting cases of diffusion – movements such as term limits, “three strikes” laws, and child protection laws such as the Amber Alert – produce punctuated diffusion patterns that indicate decision making driven by elevated attention, emotional reasoning, and policy imitation. These policies depart from the traditional S-shaped curve in a number of interesting ways. Some, like the Amber Alert, produce steep S-shaped curves as policy adoption spreads extremely rapidly across states. Others, like the diffusion of the death penalty, produce R-shaped patterns of adoption, indicating policy diffusion driven by sudden agenda-setting pressures. Still other policies produce step patterns as policy adoptions occur through short bursts of state attention rather than gradual policy adjustment. These findings qualify conventional wisdom for understanding the diffusion of innovations by demonstrating that policy diffusion frequently departs from the process of incremental learning. States may well emulate successful policies, but the careful evaluation of costs, benefits, and outcomes is less common than theory suggests. Incrementalism and Diffusion Models of Policy Adoption It is one of the happy incidents of the federal system that a single courageous state may, if its citizens choose, serve as laboratory; and try a novel social or economic experiment without risk to the rest of the country. (Justice Louis D. Brandeis, 1932).

Central to nearly all studies of policy diffusion2 is an effort to understand the processes of policy evaluation and adoption implied in the “laboratories of democracy” metaphor. Here, an organizing assumption of the study of policy diffusion is that patterns of adoptions are not independent events but rather the product of interstate influence in a federal 2

By convention in the diffusion literature, a policy innovation is defined as any public policy that is “new to the state adopting it” (Walker 1969; Gray 1973, 1174) no matter how much time has passed between the original invention and subsequent adoption. Policy diffusion can be defined as “any pattern of successive adoptions of a policy innovation” (Eyestone 1977, 441).

Incrementalism and Policy Outbreaks in the American States

25

system. The primary research questions underpinning most studies of policy diffusion revolve around why and when states adopt a given policy innovation (F. Berry and Berry 1999). Answers to these questions have been located in two general approaches. Following Walker’s pioneering work (1969), one school compares characteristics of early and late adopters to evaluate the internal dynamics leading states to innovate (Walker 1969; Savage 1978; Canon and Baum 1981; Nice 1994; Carter and LaPlant 1997). Other researchers are interested in modeling the effects of social learning through interstate networks on patterns of adoption (Gray 1973; F. Berry and Berry 1990, 1992; Mintrom and Vergari 1998). The event history framework introduced by F. Berry and Berry (1990) permits researchers to simultaneously model both internal dynamics and interactive processes. This approach has produced stimulating studies of how interstate economic competition (F. Berry and Berry 1990, 1992; Boehmke and Witmer, 2004), social policy learning (Mintrom and Vergari 1998; Boehmke and Witmer, 2004; Rincke 2004), and ideological similarities (Grossback, Nicholson-Crotty, and Peterson, 2004; Volden 2006) encourage the diffusion of innovations across states. An interesting recent addition to the study of diffusion addresses the phenomenon of policy reinvention, exploring how the scope and language of legislation evolves over a diffusion cycle (Glick and Hays 1991; Hays 1996). Although these studies identify different factors for explaining patterns of policy diffusion, they build upon a common behavioral model of political decision making. Following the study of organizational decision making advanced by Lindblom (1959), researchers have accepted that policy diffusion results from a process of incremental learning (Walker 1969; Gray 1973; F. Berry and Berry 1999; Mooney and Lee 1999) and have argued that the diffusion of innovations stems from the satisficing behavior of state decision makers operating under time constraints and considerable uncertainty (Walker 1969; Gray 1973; F. Berry and Berry 1999; Mooney and Lee 1999). Rather than taking a comprehensive solution search for each policy problem, governments borrow heavily from their neighbors or ideological peers (F. Berry and Berry, 1990; HaiderMarkel 2001; Grossback, Nicholson-Crotty, and Peterson, 2004; Rincke 2004; Weyland 2005). For example, a state government faced with problems of increasing identity theft may look to other states for model laws addressing privacy and consumer protection. This is, however, only a refinement of the incremental decision-making model of policy diffusion; it still has state legislatures looking to the experience of other states as

26

Policy Diffusion Dynamics in America

a way of reducing both the information costs and uncertainty regarding the outcomes of a new policy (Lindblom 1959; Glick and Hays 1991; Weyland 2005; Volden 2006). The incremental decision-making model holds for innovation diffusion driven by economic competition or social-policy learning. A state government losing revenue as a result of a neighbor’s lottery program will look to regional programs as a starting point for crafting new legislation. A politician pressured to introduce living-will legislation often will turn to neighboring states with similar demographics for guidance on drafting legislation. Although diffusion scholars have documented the process of policy reinvention, in which policies are modified over time as they are adopted by states, research in innovation diffusion nonetheless suggests that states begin by borrowing from the legislation of their peers. Walker (1969, 881) describes an instance where a policy was copied with so few amendments and revisions by other states that many accidentally reproduced a typo in the original legislation. This incremental learning model confirms the benefits of decentralized policy making in federations, as states are more likely to emulate policy successes than failures (Volden 2006). Diffusion occurs as state decision makers identify, evaluate, and adopt the successful policy experiments of their neighbors. In this formulation, even the adoption of a dramatic new policy innovation that is a stark nonincremental departure from the status quo is viewed as the result of an incremental decisionmaking process. As F. Berry and Berry (1999) note, “By showing how emulation of other states’ innovations can aid in simplifying complex decisions, policy diffusion theorists have demonstrated how the adoption of non-incremental policies can be consistent with the logic underlying incrementalism” (171). Challenges to Incrementalism in Policy Diffusion The movement of policy innovations across states clearly does not always follow the clean trajectory hypothesized by students of diffusion. Often a series of adoptions occur within such a short time frame that the type of policy evaluation and lesson-drawing implied in incremental learning models becomes nearly impossible. As with interest-group-sponsored initiative campaigns, some innovations are enacted with limited involvement of elected governments. Although diffusion studies have done well to fit incrementalism to case studies of diffusion driven by economic competition and social-policy learning, they offer an incomplete understanding of how policy ideas move across states. This dynamic is illustrated in

27

50 40 30

Death Penalty ---

20

--- State Lotteries

10 2005

2000

1995

1990

1985

1980

1975

1970

1965

0 1960

Number of Adopters

Incrementalism and Policy Outbreaks in the American States

Year

figure 2.1. Distinct patterns of policy diffusion: The death penalty and state lotteries.

Figures 2.1 and 2.2, which illustrate the diffusion of death penalty reenactments, state lotteries, charter school legislation, and the Amber Alert. Each case demonstrates not only a different pattern of adoption, but also shows tremendous variation in the rates of innovation diffusion. As Figure 2.1 illustrates, there was a pronounced difference in the rate of adoption of the death penalty and state lotteries in the second half of the twentieth century. A similar dynamic emerges in Figure 2.2. Charter school legislation was adopted by 40 states over the 1990s. The Amber Alert was adopted in all 50 states within six years. Factors beyond rates of speed of adoption challenge the incremental learning assumption underlying diffusion research. Consider the episodic patterns of policy diffusion displayed in the adoption of state “English

50 Number of Adopters

45 40 35 30

Charter Schools ---

25 20

--- Amber Alert

15 10 5 0 1990

1995

2000

2005

Year

figure 2.2. Distinct patterns of policy diffusion: Charter schools and the Amber Alert.

Policy Diffusion Dynamics in America

28 40 Number of Adopters

35 Gubernatorial Term Limits ---

30 25 20 15 10

English Only Laws ---

5 0 1780

1805

1830

1855

1880 1905 Year

1930

1955

1980

2005

figure 2.3. Episodic patterns of policy diffusion: Gubernatorial term limits and English Only legislation.

Only” legislation or gubernatorial term limits3 (Figure 2.3). Rather than resembling the S-shaped distribution expected of a process of gradual but consistent issue uptake and policy adoption, the diffusion of these two public policies has been marked by temporal bursts of adoption activity from subsets of states interspersed with long periods of state policy inactivity. Louisiana passed an English Only law in 1811, whereas the next state to follow suit was Nebraska in 1920. The remaining 28 states to adopt English Only legislation did so between 1980 and 2000. New Hampshire passed gubernatorial term limits in 1787, followed by a select few states in the mid-nineteenth century, and a large group of states in the early 1990s. These innovations indicate remarkable nonincrementalism in policy diffusion, as these policies are characterized by extraordinary periods of inactivity interrupted by sudden periods of policy change.4 Agenda Setting and Diffusion Dynamics Research in agenda setting suggests that studies of policy diffusion have overemphasized the role of incremental decision making in state 3

4

Both state English Only laws and gubernatorial term limits were excluded from the statistical analysis in this research because of the abnormal length of time that elapsed between the first and second adoptions of these innovations. Interestingly, a similar dynamic emerges in the historical diffusion of state lottery programs and the death penalty. Although researchers generally treat these policies as new innovations, they have in fact been the object of significant policy-making activity throughout U.S. history. In various forms, state lotteries and the death penalty have been banished and reinstituted by states throughout U.S. history, suggesting that even the diffusion of these programs are driven by shifts in governmental attention.

Incrementalism and Policy Outbreaks in the American States

29

legislatures while excluding other important factors shaping the process of policy diffusion. First, studies across agenda-setting perspectives demonstrate that changes in the policy environment can shape patterns of policy making. Elevated issue attention caused by a focusing or a mobilizing event that galvanizes mass public and political attention can open a “window of opportunity,” creating conditions for rapid and nonincremental policy change (Kingdon 1984; Baumgartner and Jones 1993; Glick and Hays 1997). Even absent a focusing event, research on framing effects (Zaller 1992; Kahneman and Tversky 2000), issue definition and redefinition (Baumgartner and Jones 1993; B. Jones and Baumgartner 2005), and policy targeting (A. Schneider and Ingram 1993) demonstrates that drawing public attention to certain dimensions of a policy innovation can lead to swings in public opinion and support for new policy ideas. Like other policy-making processes, patterns of interstate policy diffusion are sensitive to shifts in mass political attention, especially when new political attention short-circuits an incremental learning process leading to sudden public-policy diffusion. Perhaps more importantly, research in policy-process theory suggests that the idea that states are laboratories of democracy is a useful but limited framework for understanding the processes of innovation and diffusion in America. These “laboratories of democracy” are linked venues and are poorly insulated from those interest groups that sponsor innovations and pressure policy adoption. Although elected officials clearly play a central role in evaluating and implementing public policies, they are not the only sources of change in the political system. Federalism encourages venue shopping, a process in which activists and interest groups strategically exploit the multiple venues of government to secure support for their legislative programs (Baumgartner and Jones 1993; Holyoke 2003; Pralle 2003). This process increases the number of sites where new ideas can enter the political systems and can create conditions in which “new ideas or policy images may spread rapidly across linked venues, thus setting in motion a positive feedback process” (Baumgartner and Jones 1993, 240). The notion that states act as independent policy laboratories therefore presents a somewhat unrealistic understanding of policy learning and diffusion, as the pressures generated by activists and interest groups can lead to innovations being adopted across states before decision makers have had the opportunity to evaluate the costs, benefits, and implications of a policy experiment. Several recent and controversial innovations seem to stand as prominent examples of this phenomenon. In the 1990s, “three strikes” sentencing guidelines were adopted by many states in response

30

Policy Diffusion Dynamics in America

to heightened public concern over chronic violent felons. Three-strikes policies – which require repeat violent offenders to serve mandatory life sentences after a third offense – were enacted well before states could assess the impact such policies would have on prison crowding, crime rates, or the legal system. Likewise, term limits were adopted across 20 states before the benefits and costs of term limitations could be evaluated. Only recently have the first state legislatures felt the impact of the retirement of a generation of politicians, and only recently have states been positioned to evaluate the political consequences of high turnover and short tenures in government. This is not to say that states necessarily erred when passing these laws – only that they passed each piece of legislation with an incomplete understanding of efficacy or costs. Such patterns emerge partly because of the influence of interest groups in the political process. The legislative objectives of interest-group activists and policy entrepreneurs who invent and advocate for the passage of new innovations are often distinct from the goals of elected state representatives. Whereas policy makers may have a sincere interest in evaluating the costs and benefits of a particular policy innovation (e.g., evaluating the effectiveness and costs of three-strikes sentencing laws or statewide smoking bans), interest-group activists have issue-specific agendas and will look to capitalize on a window of opportunity to galvanize public support for policy change. These nongovernmental actors are not constrained by the same considerations of policy feasibility, complexity, and program cost as are elected officials, nor are they bound by the pressures of reelection. Instead, interest groups are driven to pass legislation that best meets the interests and preferences of their members and are sometimes capable of mounting sophisticated campaigns to win legislative support for new innovations (Balla 2001). When activists encounter political opposition to innovation, they adopt a number of strategies to encourage policy adoption. Wealthy and well-organized trade, professional, or peak associations can apply insider pressure across state legislatures, providing expert testimony, sample legislation, and campaign contributions for politicians. Mass-member citizen advocacy organizations, such as Mothers Against Drunk Driving, can organize pressure campaigns, mobilizing voters to demand policy adoption. When working with the state legislature fails, interest groups may seek policy changes at other venues of government, pursuing reforms in municipal governments or through the courts. In nearly half of the states, policy entrepreneurs can bypass statehouse governments entirely, using direct citizen initiatives to change public policy.

Incrementalism and Policy Outbreaks in the American States

31

Research that folds agenda-setting variables into studies of policy diffusion have provided some promising support for understanding how political attention and issue characteristics shape the rates of diffusion. In a study linking political salience and mass public opinion to policy diffusion, Scott Hays (1997) considers the possibility that innovation under heightened salience will result in a positive feedback cycle, arguing that “if political conditions within a given state political system favor adoption at the same time that an issue gains attention and is placed on the agenda, rapid – if not immediate – policy adoption is imminent” (497). His research on the temporal diffusion of living-will legislation revealed a moderate interaction between elevated national salience and policy diffusion. Studies of the electoral connection in the diffusion of innovations have expanded on the link between mass preferences, electoral agenda setting, and policy diffusion. Andrew Karch (2007a) looks beyond the traditional focus of state legislative decision making to understand how broader stages of problem identification, policy enactment, agenda setting, information generation, and customization shape the process of policy diffusion. Karch hypothesizes that time-constrained officials will be drawn to innovations that are visible and politically salient (Karch 2007a). Politicians may advocate innovation adoption not simply as a result of problem identification and program emulation, but also because sponsoring nationally popular initiatives wins electoral support from voters. As public support for innovation increases, so too does the pressure and expected benefit for innovation adoption. Research on the diffusion of morality policy supports the notion that the broad emotional appeal of a policy innovation can accelerate patterns of diffusion. Christopher Mooney and Mei-Hsien Lee (1999) discover that the reenactment of the death penalty followed a much more rapid pattern of adoption than commonly expected in diffusion research. Although standard diffusion patterns are expected to follow an S-shaped cumulative adoption curve, the diffusion of the death penalty exhibited a dramatic, sudden issue uptake across a large subset of states (see Figure 2.1). Mooney and Lee suggest that the high salience and technically unsophisticated characteristics of the policy result in faster adoptions across receptive states. Theories of policy learning in public administration and public policy likewise propose that the processes leading to the diffusion of innovation expand beyond a process of incremental problem identification and policy emulation. Peter May (1992) argues that the conceptualization of

32

Policy Diffusion Dynamics in America

goal-oriented instrumental policy learning central to the most stylized studies of policy diffusion represents only one narrow form of analysis and decision making in the policy process. Instead, May (1992) identifies three distinct forms of learning in the policy process: instrumental policy learning, social policy learning, and political learning. He explains: Instrumental policy learning entails lessons about the viability of policy instruments or implementation designs. Social policy learning entails lessons about the social construction of the policy problem, the scope of the policy, or policy goals. Political learning entails lessons about policy processes and prospects (332).

These three forms of policy learning are not mutually exclusive; instead they indicate that what diffusion researchers often conceive of as learning – the evaluation of program success or the expected political gains of program implementation – involves different dimensions of lessondrawing. The considerations leading to a political choice are not necessarily linked to questions of policy design, program complexity, or cost. Instead, political choices are informed by considerations as distinct as a desire to achieve previously identified policy goals, a strategic choice to ensure electoral victory, or a new way of framing a policy problem to advance a political agenda. Interestingly, there is a degree of inconsistency in political learning even in the domain of instrumental policy learning, where the goal-oriented, rational, and analytic process of program evaluation and imitation is believed to be central to the selection and implementation of policy instruments. May (1992) differentiates between true instrumental learning indicated by “policy elites’ increased understanding of policy instruments and designs,” (337) and superstitious instrumental policy learning, in which “beliefs about effectiveness of particular actions or individuals dominate any understanding of evaluation of performance” (336–337). This distinction has important implications for understanding the process of learning underlying the diffusion of public-policy innovations. As May (1992) points out: As with trial and error learning, learning can simply entail judgments about whether a given course of action or a given policy tool is still preferred to the alternatives currently being promoted. Copying or mimicking entails adoption of policy ideas without such understanding. For example, competition among states for certain kinds of industry may spur economic development fads (e.g., tax credits, enterprise zones) that are inappropriate for the situation some states face (333).

Incrementalism and Policy Outbreaks in the American States

33

May’s research on learning challenges the incremental diffusion model’s simplification of policy learning. The factors compelling states to adopt innovation frequently do not match the process of gradual incrementalism identified in recent studies of diffusion research. Instead, state decision makers superstitiously imitate new policies either because they are popular, or because they are persuaded by expected future benefits. Studies of public-policy diffusion have identified cases where policy learning approximated both superstitious and true instrumental policy learning. In his study of the diffusion of the Children’s Health Insurance Program (CHIP), Volden (2006) discovered that state governments emulated those programs that proved most successful in raising the proportion of poor children covered. Interestingly, the decision-making process leading to program adoption was a mixture of instrumental and political policy learning, as the incentive for state decision makers to identify the most successful program was driven by the election-seeking behavior of politicians. Mossberger’s (2000) excellent evaluation of political decision making in the diffusion of free enterprise zones finds less compelling evidence for true instrumental policy learning. In interviews with policy experts involved in the design of state free-enterprise zones, she discovered that “although learning occurred, it can be characterized as limited. The knowledge held by most participants involved only one or two generalizations about state zones, and the participants conducted no active search for information” (Mossberger 1999, 49). Mossberger instead speculates that the technical information diffusing was limited and that it was insufficient for true rational policy learning to have occurred. Instead, the information diffusing resembled first-order “policy labels” rather than a broader set of detailed, program-specific policy information (2000). Thus decision makers gained knowledge of a program’s innovation and objectives, but obtained less policy-relevant information about policy instruments, program design, and actual performance. Taken together, research in agenda setting and policy learning offers an interesting possibility for modeling the diffusion of innovations. Rather than following a single, uniform learning model, policy diffusion is better described by several distinct, underlying decision-making processes. A large subset of policies moves across states through a process of incremental learning – in which state decision makers encounter and emulate a successful policy innovation (Rogers 1983; Glick and Hays 1991; Volden 2006). However, theory and research suggests a second set of policies moves not through any formal learning process per

34

Policy Diffusion Dynamics in America

se, but rather through elevated issue attention, emotional reasoning, and imitation. Decision-Making Models in Policy Diffusion To clarify how policy diffusion can be conceived of as resulting from distinct decision-making processes, it is useful to briefly review how researchers have explained the decision-making process leading to publicpolicy diffusion. The process of incremental learning leading to policy diffusion mirrors the process of trial-and-error learning in individuals (Rogers 1983; F. Berry and Berry 1999).5 Everett Rogers (2003) argues that the decision to adopt an innovation occurs through five stages of knowledge, persuasion, decision, implementation, and confirmation6 : 1) Knowledge occurs when an individual (or other decision-making unit) is exposed to an innovation’s existence and gains an understanding of how it functions. 2) Persuasion occurs when an individual (or other decision-making unit) forms a favorable or unfavorable attitude towards the innovation. 3) Decision takes place when an individual (or other decision-making unit) engages in activities that lead to a choice to accept or reject the innovation. 4) Implementation occurs when an individual (or other decisionmaking unit) puts a new idea into use. 5) Confirmation takes place when an individual seeks reinforcement of an innovation-decision already made, but he or she may reverse this previous decision if exposed to conflicting messages about the innovation. Rogers’s model of decision making in the diffusion of innovations is a more general form of the incremental learning model favored by students of public-policy diffusion. Policy diffusion occurs through a process of innovation and emulation as decision makers encounter a novel policy 5

6

This conceptualization is entirely consistent with the broader research tradition of bounded rationality in individual and organizational decision making. An explicit aim of bounded rationality is to connect the process of organizational decision making to the preferences and decisions of individuals who populate an institution or an organization (B. Jones, Boushey, and Workman 2006). Rogers’s conception of the innovation diffusion process is similar to the heuristic stages of the policy cycle in public-policy research, in which policy makers identify a policy problem, form competing solutions, arrive at a political decision, implement the public policy, and evaluate the efficacy of the policy solution.

Incrementalism and Policy Outbreaks in the American States Intendedly

35

Rational Band

Information Impression Emotion Attention Problem Representation Solution Search Choice

Cognitive Band Information Impression Impression Choice

figure 2.4. Newell’s bands of rationality. Source: Jones, Bryan D. 2001. Politics and the Architecture of Choice, Chicago: University of Chicago Press, p. 101.

solution, form preferences through evaluation, make a formal policy choice, implement that choice, and reevaluate the intended and unintended consequences of the innovation. It is a process identical to the process of problem identification and adoption taken by individuals in trial-and-error problem solving. Yet as elaborated in the previous section, this decision-making process represents only one path of preference formation leading to policy adoption and diffusion. Although studies of diffusion frequently identify cases that conform to the process of preference formation outlined in Rogers’s innovation and diffusion model, research in both individual and organizational choice in the policy process challenges this simplification of the decision-making process. Researchers across disciplines have noted how cognitive heuristics (Kahneman and Tversky 2000), emotional reasoning (B. Jones 1994), or the influence of prejudice and bias short-circuit the linear process of problem definition, evaluation, solution, and implementation in both individual and organizational decision making (B. Jones 1994; B. Jones, Boushey, and Workman 2006). Instead, there are multiple paths to preference formation and choice for both individuals and social systems. Such a mixed decision-making model of policy diffusion is consistent with models of information processing advanced by cognitive psychologists. Newell (1990) identifies two bands of human information processing: a cognitive band and an intendedly rational band. As Figure 2.4 demonstrates, the intendedly rational band closely resembles Rogers’s diffusion-of-innovations model, and represents a process of evaluation and learning as individuals generate alternatives and compare expected

36

Policy Diffusion Dynamics in America

outcomes (B. Jones 1994). However, the cognitive band involves relatively little formal reasoning, as it produces “a system that engages in a knowledge search but it does not engage in a problem search” (Newell 1990, 139). In the cognitive band, a decision is arrived at with little formal evaluation of expected costs or outcomes. To connect these two cognitive processes to systems-level decision making in policy diffusion, the intendedly rational band can be viewed as a process of incremental learning and emulation, whereas rapid, fad-driven policies are similar to the impressionistic, emotional reasoning typified by reasoning at the cognitive level. Such parallel dynamics of information processing in social systems are well documented in other diffusion study analogs. For example, epidemiologists have well-developed procedures for identifying and distinguishing between “point source” outbreaks, which occur when a subpopulation is simultaneously exposed to a common contaminant, and propagated person-to-person outbreaks, which appear when a contagious disease is communicated via person-to-person interactions (CDC 2009). In a point-source outbreak, a subgroup within a population shows symptoms of disease almost simultaneously as individuals respond to a common exposure. In propagated outbreaks, disease transmission occurs at varying rates, as the individuals first infected transmit the disease to others, who in turn transmit it more broadly within the population. Epidemiologists have developed a number of applications for modeling and interpreting the comparative speed of outbreaks from the first known observed case. In the social sciences, communications scholars and marketing analysts have also researched how mass-media and advertising campaigns shape the awareness of a media event or a new product across a population, and the extent to which consumer behavior is driven by person-to-person interactions (Bass 1969; Valente 1993). In diffusion driven by mass-media effects, awareness of innovation can occur nearly immediately across a population (Valente 1993). In diffusion driven by word of mouth, learning occurs more gradually (Valente 1993). In both the social sciences and epidemiology, researchers have developed distinct models of information processing to capture the dynamic process of diffusion. Punctuated Equilibrium Theory and Policy Diffusion The notion that the dynamics of individual and organizational decision making represent two distinct processes of preference formation and information processing is not foreign to students of the policy process. Students of punctuated equilibrium observe that policy making in

Incrementalism and Policy Outbreaks in the American States

37

American politics is characterized by periods of policy stasis interrupted by pronounced policy punctuations (B. Jones, Sulkin, and Larsen 2003; Baumgartner and Jones 2005; Koski and Breunig 2006). These two distinct processes are explained by different forms of decision making. Policy stasis occurs through a process of incremental policy adjustment in a policy subsystem monopolized by a group of actors. Policy punctuations result when a focusing event elevates issue attention, changes the dominant attributes of a policy image, and focuses broader political attention on the politics of a particular subsystem. Elevated issue attention coupled with a new understanding of a policy problem leads to increased demands for policy change and results in a positive feedback cycle as a series of new policies are passed representing new preferences for a policy domain (True, Baumgartner, and Jones 1999, 102). Such dramatic patterns of policy change have been documented in studies of congressional attention (Baumgartner and Jones 1993; B. Jones and Baumgartner 2005), presidential decision making (Larsen 2006), federal budgeting (Jones, Baumgartner, and True 1998), and state budgeting (Koski and Breunig 2006). There is good reason to expect the same dynamic process to appear in the diffusion of policy innovations. In developing their theory of punctuated equilibrium, Baumgartner and Jones (1993) argue that policy diffusion is characterized by the same underlying dynamics of incremental adjustment and sudden positive feedback cycles captured in punctuated equilibrium theory. They write that “policy diffusion, with its S-Shaped curve, is remarkably like a punctuated equilibrium model in which the system shifts rapidly from one stable point to another” (Baumgartner and Jones 1993, 17). Empirical Models of Innovation Diffusion These theoretical models of information processing in social systems can be matched to distinct patterns of innovation adoptions. One nearly axiomatic finding across studies of diffusion-of-innovations theory is that patterns of adoptions follow a bell-shaped curve when plotted over time, and that the cumulative number of adopters in the life of innovation diffusion generally follows an S-shaped curve. In his classic Diffusion of Innovations, Rogers (2003) reviews over 5,000 research publications on diffusion and concludes that the S-shaped curve is so common that it is a general characteristic of innovation diffusion. The S-shaped curve of diffusion has been identified in a striking number of diffusion studies ranging from diffusion of farmers’ adoption of hybrid corn (Ryan and

Policy Diffusion Dynamics in America

38

Gross 1943), hate-crimes legislation (Grattet, Jenness, and Curry 1998; Rogers 2003), new-product adoptions (Rogers 1976), and across a range of education, welfare, and civil rights policies (Gray 1973). This generalization emerges from the processes that describe the cumulative adoptions over time as a product, behavior, or policy spreads across actors in a social system. The number of innovation adopters is initially limited, as only a few early pioneering individuals have identified and embraced an emerging innovation. As time progresses, the number of adopters accelerates as more and more individuals in a social system encounter and adopt a new innovation. This acceleration continues until a tipping point, when more than half of the actors in a social system have adopted an innovation. The adoption curve then continues to grow at a slower rate as fewer and fewer individuals in a system adopt a new behavior. Mathematically, this common diffusion process can be described by an internal influence diffusion model, where a new product or behavior is first introduced into a social system by a few pioneering individuals, and is then adopted throughout a population as more actors within a social system encounter and adopt that product or behavior. This can be described by the internal influence diffusion model, given by Mahajan and Peterson (1985) as dN(t) = bN(t)[N − N(t)] dt

(1)

where dN(t) is the rate of diffusion at time t, N(t) is the cumulative number dt of adopters at time t, N is the total number of potential adopters at time t, and b is the constant of imitation or internal influence. The cumulative adopters distribution function of the internal diffusion model produces an S-shaped curve, and can be represented by the following equation, given by Mahajan and Peterson (1985) as N(t) =

N (N − N0 ) 1+ exp[−bN(t − t0 )] N0

(2)

The internal diffusion model converges on an S-shaped curve because the model represents how innovation diffusion occurs through interpersonal communication. The initial period of adoption occurs gradually, as relatively few units in the social system can spread an innovation to others. As more actors in a social system encounter and adopt the innovation, the rate of diffusion increases rapidly, as there is a rapid growth in the

Incrementalism and Policy Outbreaks in the American States

39

0.0

0.2

Percentage of Adopters 0.4 0.6 0.8

1.0

S-Shaped Curve of Diffusion

0

5

10

15 Time

20

25

30

figure 2.5. S-shaped adoption curve.

number of units that can spread the new behavior. Finally, as the number of prior adopters approaches the number of possible total adopters, the rate of diffusion invariably decreases, as fewer individuals remain that have failed to adopt the innovation (Mahajan and Peterson 1985). Figure 2.5 simulates the standard life cycle of innovation diffusion. The S-shaped curve is the cumulative frequency of innovation adoption over time. In Figure 2.5, innovation diffusion begins gradually, accelerates and decelerates proportionally as more actors in a social system adopt the innovation, and then tails off as fewer and fewer laggards adopt the new innovation. This S-shaped curve is called the diffusion curve because it provides a general pattern of innovation adoption over time. It is important to note that the rate of innovation adoption may vary according to the rate of internal influence, b, changing the slope and asymptote of individual adoption curves (Mahajan and Peterson 1985). However, regardless of the rate of internal influence, the diffusion-of-innovations curve will almost always follow the process of limited initial adoption, described by a gradual introduction, an increasingly rapid period of adoption, and a decreasing period of late adoption. Because innovation diffusion is routinely characterized by these Sshaped cumulative distribution curves, diffusion theorists have argued that innovation diffusion converges on a normal distribution (Rogers

40

Policy Diffusion Dynamics in America

2003). Rogers advanced a generalization that “adopter distributions follow a bell-shaped curve over time and approach normality” (2003, 275). According to Rogers, this generalization is supported by both theory and empirical data collected across an impressive range of innovation and diffusion research. Comparing the innovativeness of individuals and organizations to other human traits, such as height, intelligence, or the learning of new information, Rogers writes: We expect a normal adopter distribution for an innovation because of the cumulatively increasing influences upon an individual to adopt or reject an innovation, resulting from the activation of peer networks about the innovation in a system. This influence results from the increasing rate of knowledge and adoption (or rejection) of the innovation in the system. We know that the adoption of a new idea results from information exchange through interpersonal networks. If the first adopter of an innovation discusses it with two other members of the system, each of these two adopters passes the new idea along to two peers, and so forth, the resulting distribution follows a binomial expansion, a mathematical function that follows a normal shape when plotted over a series of successive generations” (Rogers 2003, 272).

The assumption that adopter distributions approach normality has been used to support a general theory of the characteristics of adopters in a social system. Rogers (2003) sorts innovation adopters into five distinct categories according to how rapidly they embrace a new innovation. Innovators represent the small subset of actors that first introduce a new innovation into a social system and are classified as the first 2.5% of adopters in the population. Early Adopters are opinion leaders or trendsetters that first imitate the policy after it is introduced and make up the next 13.5% to take up an innovation. The Early Majority are those that embrace a new innovation before the general population and represent 34% of all adopters. The Late Majority are skeptical adopters that resist a new behavior or innovation and make up the next 34% of the population, and Laggards are the remaining few traditionalists who are the last adopters of innovation and represent the final 16% of the population (282–285). Anomalies Models in the Diffusion of Innovations If the S-shaped curve of innovation adoption can be applied as a tool to explore the common traits of innovation diffusion, it is also useful as a starting point for locating and describing divergence from traditional patterns of policy diffusion. Although Rogers argues that the tendency

Incrementalism and Policy Outbreaks in the American States

41

towards normality is a general characteristic of diffusion data, he also observes that innovation processes do occasionally depart from the standard S-shaped curve. Mathematical modelers have likewise employed a number of alternate distributions for fitting S-shaped diffusion curves. These diffusion theorists have used comparisons to the standard normal S-shaped curve to draw a number of inferences about the processes driving innovation diffusion. Researchers have examined the non-normality of innovation data to understand factors leading to unexpectedly rapid diffusion resulting from both external and internal pressures on social systems and to draw inferences about how crises, agenda-setting pressures, or contagion shapes innovation diffusion. Ryan and Gross (1943) compared the diffusion of hybrid corn to an expected normal distribution and discovered that the diffusion of corn seed occurred more rapidly than anticipated by a normal model. Valente (1993) demonstrated how the diffusion of three distinct innovations – hybrid corn, awareness of Eisenhower’s stroke, and physician prescriptions of new drugs – conformed to or deviated from an S-shaped distribution, demonstrating that diffusion of awareness from exposure to external events occurred more rapidly than diffusion through internal processes. Mooney and Lee (1999) demonstrate how the diffusion of the death penalty followed an R-shaped curve rather than the traditional S-shaped curve, speculating that the characteristics of morality policy led to more rapid innovation diffusion. More generally, Baumgartner and Jones (1993) speculate that innovation diffusion can spark positive feedback cycles in the United States, producing a more rapid and steeper-than-expected S-shaped logistic diffusion pattern. Findings from this body of research on diffusion “anomalies” have interesting implications for modeling public-policy diffusion. Many of these studies begin with an assumption that the standard model of innovation diffusion conforms closely to an S-shaped normal distribution. Researchers then compare empirical patterns of innovation in order to measure how strongly a given pattern matches an expected normal distribution of diffusion. These comparisons are used to draw inferences about innovation processes. Importantly, S-shaped distributions do not immediately indicate normal distributions. Virtually any unimodal distribution can produce an S-shaped curve, and researchers have employed logistic, log-normal, and Gompertz functions to model innovation diffusion (Mahajan and Peterson 1985). An extremely rapid pattern of innovation diffusion may

42

Policy Diffusion Dynamics in America

produce an S-shaped curve, but one which deviates strongly from a normal curve. One primary source of non-normality in adopter distributions emerges from the dynamics of internal influence in social systems, which can produce steep and non-normal S-shaped distributions. Ryan and Gross (1943) discovered that the diffusion of hybrid corn seed in Iowa deviated sharply from a normal distribution after an initial period of innovation. They speculate that normal frequency “does not appear to be a concept closely adapted to this condition where pressures, or reasons, for adoption become increasingly acute with passing time” (1943, 23). The diffusion of hybrid corn increased in speed as pressures to adopt the innovation increased dramatically on farmers over time. Baumgartner and Jones speculate that the rapid diffusion of policies across states would converge on a logistic distribution as demands for immediate policy adoption increase through a bandwagon effect or positive feedback cycle, as occurred with the term-limitation movement or the diffusion of the Amber Alert. The initial period of diffusion occurs slowly as new ideas are gradually evaluated and implemented; but then diffusion occurs extremely rapidly through a positive feedback cycle, as states rush to adopt emerging innovation. Such positive feedback cycles occur when innovation spreads rapidly through internal dynamics. Both Ryan and Gross and Baumgartner and Jones have implicitly connected variation in patterns of public-policy diffusion with processes familiar to epidemiologists and marketing researchers. Not all innovations produce like patterns in social systems. As the sheer contagion of an innovation increases, the likelihood also increases that it will lead to diffusion that is more rapid than traditionally expected. This dynamic can be captured in the internal diffusion model cited earlier. As the rate of innovation by imitation increases, policies will take off more rapidly than expected. These dynamics can also be modeled with the internal influence model. Figure 2.6 illustrates the S-shaped curve of diffusion of policies spreading at three different rates. As the contagion of internal influence increases, the speed of policy adoption and the slope of the S-shaped curve increase dramatically. While S-shaped distributions are by far the most common distribution of adoptions, diffusion researchers have observed alternate patterns of diffusion that are neither S-shaped nor normally distributed, which also merit discussion. As observed in Mooney and Lee’s (1999) research on the death penalty and Henrich’s (2001) exploration of the cultural

43

0.0

0.2

Percentage of Adopters 0.4 0.6 0.8

1.0

Incrementalism and Policy Outbreaks in the American States

0

5

10

15 Time

20

25

30

figure 2.6. S-shaped adoption curves representing three different rates of innovation diffusion.

transmission of innovation diffusion, adoptions can be spurred not by gradual person-to-person interaction, but rather by exposure to an external common stimulus, such as a focusing event. Such phenomenon is known to researchers in epidemiology or communications who study how exposure to a common external parameter shapes innovation adoption. In these rare cases, adoption will occur more suddenly than expected, as units in a social system immediately encounter and respond to innovation. When plotted over time, the distribution of innovations in these instances will resemble an exponential curve, as innovation adoption across a large number of actors in the system is nearly immediate. Mathematical modelers of innovation diffusion have identified this process as an external influence model of innovation diffusion. Here, the innovation adoption is driven by immediate awareness of adoption, given by Mahajan and Peterson (1985) as dN(t) = a[N − N(t)] dt

(3)

where dN(t) describes the rate of innovation adoption, N(t) is the cumudt lative number of adopters at time t, and a is a coefficient or constant of external influence describing the influence of all factors other than interpersonal communication.

Policy Diffusion Dynamics in America

0.0

0.2

Percentage of Adopters 0.4 0.6 0.8

1.0

44

0

2

4 Time

6

8

figure 2.7. R-shaped exponential adoption curve.

As the constant a increases, the cumulative distribution function of the external influence model produces an R-shaped exponential distribution over time. This represents innovation adoption driven by responses to an external stimulus rather than internal dynamics. The cumulative distribution function is given by Mahajan and Peterson (1985) as N(t) = N[1 − exp(−at)]

(4)

When diffusion processes are shaped by external influences, such as an exogenous event or a common source influence on diffusion processes, innovation adoption can happen nearly immediately. When plotted over time, the external influence diffusion model produces an Rshaped curve, deviating strongly from an S-shaped innovation diffusion curve. Figure 2.7 illustrates the exponential adoption curve of diffusion driven by external influence. Of course, students of innovation diffusion have realized that both external and internal influences likely shape the rate of diffusion. Because of this fact, many researchers have preferred to represent the diffusion of innovations through a mixed-influence model developed by Bass (1969). The Bass diffusion model has been applied to estimate the rates of innovation diffusion driven by both external and internal influences (Bass 1969; Mahajan and Peterson 1985; Srinivasan and Mason 1986, Valente 1993). The Bass model returns estimates of two rate parameters of a diffusion

Incrementalism and Policy Outbreaks in the American States

45

process: the constant a, representing the rate of innovation adoption due to innovation or external influence, and the constant b, measuring adoption driven by imitation. The Bass Model therefore provides a relatively simple representation of the processes driving the diffusion of innovations. The constant a describes the rate of initial innovation adoption, and the constant b represents the rate of new adopters of innovation at a given time as a fraction of those that have already adopted the policy. It is a preferred model of innovation diffusion because it provides a more realistic representation of the real-world processes of innovation diffusion than the internal or external influence model alone. The mixed-influence equation models diffusion as a response to both external and internal pressures for adoption. It is given here by Mahajan and Peterson (1985) as dN(t) = (a + bN(t))[N − N(t)] dt

(5)

with a cumulative distribution function of a(N − N0 ) exp[−(a + bN)(t − t0 )] (a + bN0 ) N(t) = b(N − N0 exp[−(a + bN)(t − t0 )] 1+ (a + bN0 ) N−

(6)

The mixed-influence model is useful for diffusion scholars for several reasons. Perhaps most importantly, it is a more theoretically satisfying model of innovation diffusion, as it incorporates how both external and internal pressures shape diffusion patterns. Diffusion can be an immediate response to an exogenous shock, followed by a period of internal learning, or diffusion can occur through internal dynamics at vastly different rates. The mixed-influence model provides a tool for distinguishing between external and internal dynamics driving innovation diffusion, and a method for capturing both dynamics simultaneously.7 This review of mathematical models provides two important takeaways for the subsequent analysis in the book. First, like the larger theory of incrementalism, the a priori expectation is that policy diffusion data will follow an S-shaped distribution when plotted over time. The distribution of policy adoptions will converge on a normal distribution. This 7

The Bass Model returns three constants, a, b, and M. An increasing constant, a, suggests more rapid diffusion from external influence. A larger constant, b, indicates more rapid innovation diffusion from internal influence. M is a measure of the expected (or total) number of adopters.

46

Policy Diffusion Dynamics in America

expectation is supported by both theory and evidence developed in the study of innovation diffusion writ large, and the study of public-policy innovation more specifically. Data will generally conform to an S-shaped normal distribution because most diffusion processes capture the process of trial-and-error learning in social systems. However, this discussion suggests that the diffusion of innovations can deviate from normality in two principal ways. First, exogenous agendasetting pressures can make innovation adoption more immediate, resulting in an R-shaped distribution resembling an exponential distribution. Second, the dynamics of internal innovation diffusion can shape the pattern of innovation diffusion. As the transmission of innovation increases through internal influence, the diffusion of innovation can deviate from a normal distribution. This occurs when there is a sudden acceleration in the number of adopters following an initially gradual introduction of innovation into the social system. Measuring Decision-Making Models in Policy Diffusion This review of theoretical and mathematical models of innovation diffusion presents an interesting question for research in policy diffusion. To what extent do patterns of diffusion match behavioral incrementalism? Are policy outbreaks uncommon aberrations, or does the rapid diffusion of innovations represent an important and distinct process of policy diffusion? Finally, is it possible to explain the rapid and sudden diffusion of innovations within the context of existing incremental diffusion theory? Prior challenges to incremental learning models outside of diffusion research have employed stochastic processes models to evaluate political learning and policy change (Padgett 1980; Baumgartner and Jones 1993; B. Jones, Sulkin, and Larsen 2003). This method is justified by an empirical nuance of incremental decision making. The observational data of an incremental learning process are normally distributed, providing researchers with a relatively simple method for verifying incrementalism (Padgett 1980; B. Jones and Baumgartner 2005). As Jones and Baumgartner (2005) explain, this method greatly simplifies hypothesis testing, as “any normal distribution of policy changes must have been generated by an incremental process,” whereas “any time we observe a non-normal distribution of policy change, we must conclude that incrementalism cannot have caused it” (123). This approach is especially appropriate for the study of policy diffusion. First, as the previous section described, there is a strong assumption

Incrementalism and Policy Outbreaks in the American States

47

that the cumulative pattern of adoptions over time follows an S-shaped curve, with an initial period of early adopters pioneering an innovation, a middle period of more rapid uptake around an inflection point, and a final phase of uptake by late adopters (Rogers 1983; Mooney and Lee 1999). Perhaps more importantly, the patterns of adoptions are assumed to be normally distributed (Rogers 1983). Thus, a plot of the probability density function of a typical diffusion process should generate a distribution resembling a Gaussian curve, and derivation of the cumulative distribution function (CDF) should yield a plot following the CDF of a normal distribution (Rogers 1983). That observational data collected from processes of incrementalism and policy diffusion share the same underlying distribution is not coincidental. In Diffusion of Innovations, Rogers (1983) explains that the distribution of times to adoption in diffusion is normal exactly because it represents an incremental learning process. He writes: Psychological research indicates that individuals learn a new skill, or bit of knowledge, or set of facts, through a learning process that, when plotted over time, follows a normal curve. When an individual is confronted with a new situation in the psychologist’s laboratory, the subject initially makes many errors. After a series of trials, the errors decrease until a learning capacity has been reached. . . . If a social system is substituted for the individual in the learning curve, it seems reasonable to expect that the experience with the innovation is gained as each successive member in the social system adopts it. Each adoption in the social system is in a sense equivalent to a learning trial by an individual. (Rogers 1983, 244)

Thus it is not an abstraction or a matter of mathematical convenience to suggest that the underlying probability distributions of incrementalism and policy diffusion are the same. They are the same because they both measure the behavioral process of incremental learning. The diffusion of innovations model is an incremental learning model.8 Nonincremental Patterns of Policy Diffusion As described in the two prior sections, theory and empirical evidence suggest two distinct patterns of learning associated with the process of public-policy diffusion. Figure 2.8 simulates and plots the CDF of data 8

Interestingly, Rogers recognized that the diffusion of innovations did occasionally deviate from this expected cumulative frequency distribution. Likewise, Gray (1973) observed that the diffusion of public policy also occasionally departed from the S-shaped cumulative normal curve. However, neither author seemed to ask exactly what such departures indicated for the process of innovation and diffusion.

Policy Diffusion Dynamics in America

48

0.6

0.8

Policy outbreaks

0.4

Incrementalism

Hyperincrementalism

0.0

0.2

Percentage of Adopters

1.0

Theoretical Diffusion Curves

0

5

10

15 Time

20

25

30

figure 2.8. Simulated theoretical diffusion curves.

from three hypothetical diffusion processes: hyperincrementalism, incrementalism, and policy outbreaks. The two S-shaped curves show the cumulative incremental adoption curve at different rates, indicating variation in communication across units in a social system. The hyperincrementalism curve is represented by the gradually sloped thin line at the bottom of Figure 2.8. The incremental learning curve is the S-shaped distribution curve in the middle. The policy outbreaks curve is represented by the thick exponential distribution at the top of the figure. Both the hyperincrementalism curve and the incrementalism curves are simulated normal curves with different means and standard deviations indicating differences in time elapsed between adoptions.9 If the process of diffusion occurs through a process of incremental problem identification, policy evaluation, and emulation, then the observational data associated with public-policy diffusion should follow the S-shaped normal curve, although the steepness of the curve may vary.10 9

10

Both the hyperincrementalism curve and the incrementalism curves are simulated normal curves with different means and standard deviations indicating changes in time elapsed between adoptions. It is important to note that both rapid and slow policy diffusion can be normally distributed. The speed of diffusion is at least partially related to how frequently units in a

Incrementalism and Policy Outbreaks in the American States

49

An alternate to the incremental curve is captured in the policy outbreaks curve, when an accelerated diffusion process results in serial adoptions resembling a policy outbreak. As with the pattern of death penalty readoptions displayed in Figure 2.1, this curve exhibits extremely rapid issue uptake as imitation leads to a sharp and immediate rise in innovation adoption. The policy outbreaks curve is represented here by a steep simulated exponential curve.11 Policies matching this distribution exhibit patterns of adoption consisting with policy mimicking rather than incremental learning. Figure 2.8 simulates and plots data from three hypothetical diffusion processes – hyperincrementalism, incrementalism, and policy outbreaks. The hyperincrementalism curve is represented by the gradually sloped thin line at the bottom of Figure 2.8. The incremental learning curve is the S-shaped distribution curve in the middle. The policy outbreaks curve is represented by the thick exponential distribution at the top of the figure. Expectations The extent to which a diffusion distribution matches the normal distribution can be used as an indicator of incrementalism in a diffusion process. Here, the assumption that the probability distributions of diffusion processes are normally distributed allows us to assess whether incrementalism is the appropriate decision-making model for policy diffusion. If incrementalism is the dominant decision-making process in diffusion, then the data from diffusion processes should conform closely to the S-shaped cumulative normal curve. If diffusion processes display punctuated dynamics, then we expect a distribution converging towards the exponential curve in Figure 2.8. Incremental Diffusion and Historical Eras In order to evaluate the consistency of nonincrementalism in diffusion research, it is important to control for temporal variation in diffusion patterns. Modern advances in communication technology have made it

11

social system come into contact with and encounter new policy innovations. As mentioned in the prior section, researchers have used Gompertz or logistic curves when S-shaped diffusion patterns have not matched the assumptions of normality. A number of alternate distributions could be used to match the rapid issue uptake of data collected from policy outbreaks. Although the exponential distribution is proposed here, other researchers might model such diffusion data using a Paretian distribution.

50

Policy Diffusion Dynamics in America

easier for decision makers to acquire and evaluate information about new policy innovations (Savage 1985a; Mossberger 2000). The proliferation of radio, television, Internet, and mass-communications technology permits the rapid flow of information, making positive feedback cycles and the rapid diffusion of innovations more common in the modern era. The importance of communications variation across historical eras suggests a second research proposition regarding the diffusion of innovations. Newer policies will display especially pronounced punctuated dynamics, whereas older policies will tend to display more gradual incremental learning patterns resembling an incremental, normal distribution. Data To evaluate patterns of policy diffusion in the American states, this investigation gathered information on state years of adoption for 133 different innovations, following a purposive sampling procedure designed to ensure a balanced representation of state public policies by historical era, policy type, and speed of diffusion. To identify innovations, this study followed Walker’s definition of an innovation as a public policy that is “new to the state adopting it” (1969, 881). The research therefore includes only innovations that were formally enacted by state governmental institutions, excluding informal policies, legal strategies, and other unofficial policy positions. Furthermore, because the study of diffusion is concerned with the timing of political decisions, coders entered information for the year of legislative enactment rather than the year of implementation. Despite these limitations, the data cover a wide range of policy types, representing economic, social, and procedural policies throughout American history. To ensure a balanced and representative sample of innovation diffusions, this research followed the lead of both Walker (1969) and Savage (1978), collecting policies representing a wide range of policy innovations. Walker (1969) collected his sample of innovations from a list of issue areas suggested by the Council of Governments’ Book of the States (Walker 1969, 882). He gathered information on state years of adoption for legislation from different issue areas: welfare, health, education; conservation; planning, administrative organization; highways; civil rights; corrections and police; labor; taxes and professional regulation (1969, 882). In constructing an even more expansive set of policies, Savage (1978) built a data set of 181 policy innovations drawn from a similar list of issue areas including “agriculture, business

Incrementalism and Policy Outbreaks in the American States

51

regulation, conservation, crime and corrections, education, electoral regulation, governmental structure and operation, local government, health, professional licensing, race relations, taxation, transportation, and welfare” (214). This research followed the sampling protocol developed in each of these two earlier studies as a guide to construct a balanced sampling universe of state innovations. As explained below, it first replicated Walker’s (1969) data. When gathering new innovations to build the larger sample of 133 innovations, it followed the protocol outlined by Walker and Savage and identified policies across a similar universe of state issue areas, including civil rights, crime and policy, public health, highway and transportation safety, education, economic regulation, local government, and governance reform. The sampling procedure used to identify and collect innovations in this study is consistent with other sampling protocols used to construct large samples of state legislation in prior research on policy diffusion. Because the goal here is to evaluate how well incrementalism explains patterns of diffusion, a significant portion of the cases included in this research are drawn directly from innovations that have been featured prominently in prior research in public-policy diffusion. Of these, Walker’s replication data set provides a large portion of the data analyzed for this study. These data alone supply information on the years of adoption for 86 policies in the 48 contiguous states. The research also includes policies and years of adoption from more recent studies of policy diffusion. To identify policies for inclusion, this study identified policies through keyword searches for policy diffusion research in Expanded Academic Index and JSTOR. The data set includes research in state lotteries (F. Berry and Berry 1990); child abuse reporting and crime victims’ compensation (Hays 1996); smoking regulations (Volden and Shipan 2006); same-sex marriage bans (Eskridge 1999); death penalty reenactments (Lee and Mooney 1999); charter schools (Rincke 2004); living-will laws (Glick and Hays 1991); state medical savings accounts (Bowen 2005); no-fault divorce laws (Nakonezny, Shull, and Rogers 1995); statutory rape laws (Cocca 2002); the repeal of state sodomy laws (Disarro 2005); and hate-crimes legislation (Disarro 2005). When appropriate, it extended and updated this information through searches of state legislative statutes. Information for the remaining public policies was gathered through information provided by issue organizations, government agencies, the National Conference of State Legislatures, and keyword searches of

Policy Diffusion Dynamics in America

52

individual state legislative statutes. These policies include issues such as term limits, three-strikes laws, medical marijuana legislation, and bottle deposit legislation. A full list of policies can be found in Appendix A. To facilitate comparison of timings of adoption across decades, the data have been organized in a duration format, indicating the time each state took to adopt a given public policy. Here, the first adopters are assigned a zero (indicating immediate adoption), and subsequent adopters are indicated by the number of years elapsed between the first adoption and their own adoption of the innovation. A censoring variable is included for states that have not yet adopted a policy. To classify policies by historical era, this researcher followed the procedure suggested by Savage (1978), placing a policy in the period when the first 10 states adopted the innovation.12 Following Savage’s research on the temporal consistency of innovation scores in modern U.S. history (1978), policies were then grouped into one of four historical eras spanning the late nineteenth, twentieth, and early twenty-first centuries. This process identified 17 policies that diffused prior to 1900, 28 policies in the 1900–1929 period, 35 policies in the 1930–1959 era, and 52 policies in the 1960–2006 period. A list of policies in each of these categories can be seen in Appendix B. Method These data were employed to model the underlying distribution of times to adoption, and to evaluate whether policy diffusion can be characterized as resulting from an incremental decision-making process. This approach follows prior stochastic process studies of public policy making, which evaluate probability distributions to understand the underlying decision-making process in public budgeting (Padgett 1980) and governmental attention (Baumgartner and Jones 1993; B. Jones and Baumgartner 2005; Koski and Breunig 2006). These studies suggest that incremental processes “invariably lead to a normal outcome change distribution,” whereas nonincremental decision-making processes leading to sudden change should deviate strongly from a normal curve, following an exponential or Paretian distribution (B. Jones and Baumgartner 2005, 123;

12

To check the validity of this sorting protocol, a secondary classification sorted policies by historical era by first taking the mean year of adoption for all states,and then subtracting one standard deviation from the mean to find the year of early diffusion. This process provided an almost identical classification system to the process proposed by Walker.

Incrementalism and Policy Outbreaks in the American States

53

Padgett 1980). Transferring this method to the study of decision making in policy diffusion is appropriate given the strong prior assumption that temporal patterns of policy diffusion resemble a normal S-shaped curve (Gray 1973; Rogers 1983). To measure how well diffusion data matched an incremental learning curve, this research compared the empirical cumulative distribution function (ECDF) for the aggregated pooled data of policy diffusions to a simulated Gaussian cumulative distribution of the same mean and variance as the empirical data. Here, the degree to which a plotted empirical cumulative distribution function conforms to the expected incremental or Gaussian curve can be evaluated through a visual plot. The plotted CDF of the hypothetical normal curve provides the familiar S-shaped distribution of an incremental learning process. If the empirical duration data collected in this research follows a similar distribution, then it should closely follow the simulated normal cumulative distribution function. This analysis of distributions is then extended to identify distinct patterns of diffusion across historical eras. An identical procedure is used to compare the empirical cumulative distribution functions across three groups of policies: those diffusing from 1900 to 1929, 1930 to 1959, and 1960 to 2006. Each of these distributions is then plotted alongside their expected Gaussian distribution. Again, the research hypotheses can be tested through a visual plot. Distributional Tests of Normality and Kurtosis In addition to the visual plots, several goodness-of-fit tests were employed to assess the normality of diffusion data. To examine the normality of the distributions of adoption times, this research relied on two standard, onesample tests of normality. First, this research assessed the normality of the diffusion data with the Shapiro-Wilk (S-W) test, which provides a powerful evaluation of normality and is appropriate for small-to-medium sample sizes (Conover 1999; Koski and Breunig 2006). The S-W test allows researchers to measure normality by returning both a p-value and W statistic, with smaller values for W indicating non-normality in the data. To confirm this test, the research also evaluated the distribution of diffusion data using the Anderson-Darling (A-D) test of normality. The A-D test is a powerful variation of the nonparametric Kolmogorov-Smirnov test, but gives more weight to the tails of a distribution in evaluating normality. The A-D test returns both an A statistic and a p-value for confirming whether or not a given distribution is consistent with a normal

54

Policy Diffusion Dynamics in America

distribution. Together, these goodness-of-fit tests provide a straightforward method for assessing the normality of diffusion data. To measure the general shape of diffusion distributions, this research also evaluates diffusion data for positive and negative kurtosis. Kurtosis (K) measures the “peakedness” of a given distribution, and with diffusion data can be used to indicate a positive feedback cycle from a sudden uptick in the number of individuals adopting an innovation as it gains momentum. A normal distribution is described as mesokurtic and is typically smooth and symmetrical around the mean. Deviations from a normal distribution can be represented in two common shapes. A platykurtic distribution has thin tails and is relatively flat in the center, whereas a leptokurtic distribution is indicated by a sharp central peak and fat tails. As a first measure of kurtosis, this research first employs a measure of Lkurtosis (LK), which represents the fourth moment of a probability distribution around the mean (Hosking 1990; Breunig 2006). Recent research applying distributional analysis to study the shape of distributions from the policy process has preferred LK to the more traditional measure of kurtosis, as it is less influenced by outliers and results in a more stable estimate of shape (Hosking 1990; Breunig 2006; Baumgartner, Breunig, Green-Pedersen, Jones, Mortensen, Neytemans, and Walgrave 2009). LK provides a relatively simple method for interpreting kurtosis, returning a value in intervals between 0 and 1, with values of 0.1226 representing the LK of a normal distribution. Marginally higher values are indicative of departures from a normal distribution, with an exponential distribution having an expected LK score of 0.1667 (Hosking 1990). Lower values likewise suggest non-normality, and are indicative of abnormally flat or bimodal distributions common in binomial or uniform distributions. In keeping with prior research in distributional analysis, this research also presents the standard measure of kurtosis, measuring the shape of a distribution (DeCarlo 1997). Normal distributions have kurtosis values of 3, with larger numbers indicating leptokurtic distributions and smaller, negative values indicating platykurtic, or flatter bimodal or unimodal distributions. As a final measure of the shape of diffusion data, each distribution of adoption times was measured for L-skewness. This measure of skew represents the third moment of a probability distribution around the mean and provides a final method for evaluating the shape of diffusion data. Here, normal distributions typically have L-skewness of 0. Exponential distributions produce L-skew values of 0.333 (Hosking 1990). Higher values of skew indicate a greater departure from a normal distribution.

Incrementalism and Policy Outbreaks in the American States

55

Together these measures of distribution shape provide cursory measures for understanding the normality and shape of diffusion data. As a rule, distributions with higher K and LK are indicative of punctuations in the middle of a diffusion process. A higher measure of L-skewness may be indicative of more immediate policy uptake, followed by a series of later adoptions, indicative of an exponential distribution. To illustrate how values of K, LK and L-skewness change depending on the distribution of diffusion data, it is useful to refer back to Figures 2.5 and 2.7 of this chapter. The curve in Figure 2.5 represents a cumulative normal curve. The distribution underlying the curve has a kurtosis of 2.73, LK of 0.103, and L-skewness of 0.030. Figure 2.7 represents a simulated exponential distribution, with kurtosis of 6.65, LK of 0.151, and L-skewness of 0.333.13 Results Figure 2.9 plots an empirical cumulative distribution function of all policy adoptions against a simulated normal cumulative distribution curve of the same mean and variance. Here, the x-axis represents years to policy adoption, and the y-axis represents the probability of adoption. As indicated in Figure 2.9, the empirical cumulative distribution of the pooled diffusion duration data deviates sharply from the expected normal distribution. Policy adoptions occur more rapidly than expected, resembling an exponential distribution rather than the simulated normal distribution. The separation between these two curves is indicative of a nonincremental learning process and lends preliminary support for the hypotheses that policy diffusion deviates from an incremental learning process. Across cases and historical periods, policy diffusion is marked by nonincremental decision making. The non-normality of the aggregate diffusion data is confirmed by the goodness-of-fit statistics presented below in the first row of Table 2.1. Both the Anderson-Darling and Shapiro-Wilk tests permit us to reject the assumption that diffusion data is normally distributed. The two measures of kurtosis (LK = 0.120, K = 4.08) do not indicate unusually high levels of positive kurtosis; however, the measure of L-skewness suggests that the skew around the mean of 0.32 is close to the expected skew in an exponential distribution. 13

These measures of kurtosis and skew shift slightly as the mean, standard deviation, or rate of the normal and exponential curve changes.

Policy Diffusion Dynamics in America

56

table 2.1. Statistical Tests for Normality by Historical Era Historical Era

A-D Test (A) P<

S-W Test (W) P<

All Years 216.871 0.000 0.847 1900–1929 32.003 0.000 0.909 1930–1959 97.318 0.000 0.764 1960–2006 89.088 0.000 0.758

0.000 0.000 0.000 0.000

Kurtosis L-Kurtosis L-Skewness 4.080 3.767 5.713 5.635

0.120 0.092 0.199 0.184

0.322 0.231 0.405 0.427

Figures 2.10, 2.11, and 2.12 show the distribution of adoption times for three historical eras spanning the twentieth century. As hypothesized, these three visual plots identify considerable variation in patterns of diffusion across the three historical periods, with policies from the early twentieth century following a much more gradual diffusion trajectory than policies from later periods. As Table 2.1 indicates, policies diffusing

0.8 0.6 0.4 0.0

0.2

Probability of Adoption

1.0

All Policies Combined

0

20

40 60 Years to Adoption

80

100

figure 2.9. Cumulative distribution of adoption times: All policies. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

57

0.0

0.2

Probability of Adoption 0.4 0.6 0.8

1.0

Incrementalism and Policy Outbreaks in the American States

0

20

40

60 Years to Adoption

80

100

figure 2.10. Cumulative distribution of adoption times: 1900–1929. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

in the early twentieth century are characterized by decidedly lower levels of L-kurtosis, kurtosis, and L-skewness than policy diffusion in later generations. Figure 2.10 represents a plot of the empirical CDF of policies diffusing from 1900 through 1929 against a simulated normal distribution of the same mean and variance. The results of this plot are interesting. The two curves conform closely at the outset; however, the simulated normal distribution predicts faster policy adoption as time progresses. This curve approaches the theoretical hyperincremental diffusion curve of a normal distribution with a broader variance around the mean identified earlier in this chapter. In keeping with the graphical findings, distributional tests indicate that policies from 1900 to 1929 approach a normal distribution. Whereas the Anderson-Darling and Shapiro-Wilk tests lead us to reject the hypothesis that the diffusion data are normally distributed, there is evidence that observational data from this period more closely approaches its simulated normal distribution. The higher value of the W statistic,

Policy Diffusion Dynamics in America

0.0

0.2

Probability of Adoption 0.4 0.6 0.8

1.0

58

0

20

40

60

80

100

Years to Adoption

figure 2.11. Cumulative distribution of adoption times: 1930–1959. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

the kurtosis score of 3.67, and the LK score of 0.092 indicate a smaller deviation from the normal than identified in the other areas. Furthermore, this distribution exhibits much lower values of skewness than publicpolicy innovations that diffused in later generations. Figures 2.11 and 2.12 present the cumulative distribution of adoption times for policies diffusing from 1930 through 1959 as well as the distribution for policies diffusing from 1960 to 2006. As indicated in Table 2.1, the observational data for these time periods deviate from the expected normal distribution. Likewise, the kurtosis scores for these time periods are higher than those for the early twentieth century. Policies in the middle and late twentieth century are therefore characterized by a larger degree of policy punctuation. The visual plots support this finding. Both distributions deviate sharply from the simulated normal distribution, with a sharp curve extending upwards at the earliest time periods. This effect appears especially pronounced in the modern era, in which the gross majority of policies occur relatively early in the diffusion process.

59

0.8 0.6 0.4 0.0

0.2

Probability of Adoption

1.0

Incrementalism and Policy Outbreaks in the American States

0

20

40

60 Years to Adoption

80

100

figure 2.12. Cumulative distribution of adoption times: 1960–2006. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

Neither curve matches the simulated normal cumulative distribution, and both are characterized by higher than expected values of kurtosis and skew, indicating a sharp and potentially immediate punctuation in the diffusion process. Discussion This chapter evaluated how well the process of policy diffusion fits with theories of incremental decision making. The idea that diffusion is driven by an incremental process of “muddling through” is central to current research in the diffusion of innovations. However, there is ample theoretical reason to believe that the process of incremental decision making is inadequate to capture the dynamic process of policy diffusion. Rather, theory suggests that the diffusion of innovations is better conceptualized as resulting from a mixture of distinct decision-making processes. The dominant process remains policy incrementalism, as the preponderance of

60

Policy Diffusion Dynamics in America

legislative activity by state governments occurs in policy subsystems and fails to galvanize mass attention. However, as anecdotal and empirical evidence suggest, policy diffusion produces patterns that are inconsistent with the incremental learning model. In certain cases, a single innovation triggers a positive feedback cycle and the diffusion of innovations across states sparks a process of policy mimicking consistent with a policy outbreak, as shown in the examples of the Amber Alert, term limits, or the readoption of the death penalty. Evidence of this phenomenon emerges in the aggregate analysis of diffusion, as patterns of diffusion deviate sharply from the expected S-shaped curve of incremental learning, indicating that these periods of punctuated diffusion occur more frequently than existing research appreciates. In each of the diffusion distributions tested the empirical distribution of adoption times deviates from the hypothetical normal distribution expected in incremental learning processes. These findings offer strong support for the agenda-setting model of decision making in policy diffusion introduced at the outset of this chapter. There is reason to believe that policy diffusion represents at times both incremental learning and sudden policy outbreaks. More importantly, the shape of the empirical cumulative distribution curves provides preliminary insight into the decisionmaking processes underlying diffusion dynamics. With the exception of the group of policies spreading in the early twentieth century, diffusion distributions exhibit patterns consistent with an exponential distribution. This indicates two dominant processes of decision making underlying policy diffusion – a rapid decision-making process typified by attentiondriven choice, and the gradual process of policy incrementalism. The incremental evaluation of policies such as state lotteries, charter schools, or identity-theft prevention programs represents a different process than the sudden imitation of programs like the Amber Alert. Measuring the influence of modernization on the diffusion of innovation likewise provides fascinating results. Beginning with Walker (1969), students of policy diffusion have consistently remarked on the potential influence of advances in communications technology on the speed of innovation and diffusion (Savage 1985b; Mossberger 2000). The diffusion of innovations is expected to increase over time as decision-making units in the social system are able to more rapidly acquire and disseminate relevant information about innovation. A comparison of the distribution of adoption times across three periods in the twentieth century demonstrates this expected dynamic. In each successive era, patterns of policy adoptions exhibited increasingly rapid

Incrementalism and Policy Outbreaks in the American States

61

issue uptake and greater deviation from the normal distribution. This effect was especially pronounced when comparing policies diffusing over the first 30 years of the twentieth century to policies from the later periods. Earlier diffusion patterns display a gradual diffusion curve. Later policies have a sharper curve and depart significantly from the simulated normal distribution. This finding is entirely consistent with research on the importance of communication in the diffusion process. Absent the same advantages of modern communications systems, diffusion in earlier eras occurred more gradually, as state exposure to the agents of innovation occurred less rapidly. These findings are suggestive, but provide little insight into the actual processes driving both incremental and nonincremental patterns of policy diffusion in the American states. Although this chapter challenges the incremental learning model underpinning much of diffusion research, it does little to explain why or when diffusion will resemble an incremental learning curve or will produce pronounced policy punctuations and rapid policy diffusion. To understand these dynamics, the book argues that the study of policy diffusion should move beyond studies of incremental learning and policy emulation in state legislature, toward a more expansive study of agenda-setting dynamics in policy diffusion. In the remaining chapters, this research evaluates the causes of diffusion dynamics in America. Borrowing directly from research in epidemiology, these chapters assess how variation in three key components of policy diffusion – the nature of the policy idea, the characteristics of states, and the capacity of interest-group vectors – all contribute to distinct patterns of innovation and diffusion. This research not only confirms findings of incrementalism in policy diffusion but provides additional leverage for understanding the processes leading to positive feedback cycles and policy outbreaks across states.

3 Policy Agents Innovation Attributes and Diffusion Dynamics

In recent years, members of the scientific community have become increasingly alarmed by the threat of a global pandemic caused by a strain of H1N1 influenza popularly known as “bird flu” or “swine flu.” The cause of concern revolves around the especially virulent attributes of the strain of influenza responsible for an estimated 30 to 50 million deaths between 1918 and 1919.1 The emergence of that strain is thought to have resulted from recombination of portions of the influenza genome of human and avian strains, creating a virus that could be transmitted from human to human, yet so different from previous strains that prior infection conferred no immune protection (CDC 2007).2 The principle, well established in study of the epidemiology of infectious disease, is that some agents possess attributes that elevate the risk of an epidemic. Interestingly, although studies of policy diffusion and epidemiology ostensibly explore similar diffusion dynamics, research in interstate policy diffusion has largely overlooked how differences in the properties of policy innovations shape patterns of adoption. Diffusion researchers have focused on understanding how state internal dynamics and interstate interactions produce diffusion patterns, but the influence of the characteristics of policy innovations themselves receives little attention in diffusion 1

2

The great influenza pandemic of 1918 and 1919 is believed to have led to the death of 675,000 within the United States and between 30 and 50 million worldwide. For more information on the great flu, see the U.S. Department of Health and Human Services website on “The Great Pandemic.” http://1918.pandemicflu.gov/; accessed 10/5/08. For specific information about current government concern over the avian flu, refer to the Centers for Disease Control and Prevention’s Avian Flu webpage at http://www.cdc .gov/flu/avian/; accessed 10/5/08.

62

Policy Agents

63

research3 (Karch 2007b). This is surprising because the study of policy diffusion is in essence the study of how new policy ideas interact with states to produce patterns of policy change in federalism. There is good reason to expect that variations in the attributes of public-policy innovations will lead to differences in the patterns of policy diffusion. Policy innovations vary across a number of different dimensions, including target, issue complexity, program cost, issue salience, and issue fragility – the degree of perceived political opposition to innovation. Each of these attributes shapes the way state governments process policy information when deciding whether or not to adopt an innovation. The policy target identifies the group receiving benefits or burdens from government and influences how mass publics and policy makers respond to an emerging policy. Issue salience shapes the way political systems prioritize information. A highly salient innovation engenders a different form of political participation than a low-salience policy reform. Likewise, issue complexity determines how decision makers arrive at political choice. Highly technical, complicated innovations require technocratic program evaluation, best performed by policy experts and bureaucratic specialists. Less complex innovations can be resolved by elected representative and mass participation in the policy process. Innovation cost dictates the ease of policy implementation. Lower-cost policies are generally easier for state governments to implement than policies requiring the allocation of significant new resources. Issue fragility also determines the ease of policy adoption. Issues with low fragility engender little formal opposition to policy adoptions. Highly controversial issues encourage the mobilization of strong policy opposition. Such opposition introduces friction in the political system and can lead to periods of inertia followed by abrupt movements of policy change. Diffusion dynamics are shaped by the way state governments prioritize and react to innovations based on distinct innovation attributes. Decision makers prioritize issues that invoke a sense of urgency and elevate mass 3

In his review of the state policy diffusion literature, Karch (2007b) argues that diffusion research has overlooked intriguing questions of how changes in innovation characteristics contribute to policy diffusion. He argues that important breakthroughs in our understanding of the diffusion process will come from studies that explore how changes in policy innovations shape the process of diffusion. There are a number of ways researchers can explore how innovation characteristics shape diffusion dynamics. First, studies of policy reinvention demonstrate that policies evolve over a diffusion cycle. The scope and extent of policies sharing the same name is therefore quite different across states. In a more basic form, the characteristics of policy innovations themselves also shape the speed and extent of policy diffusion.

64

Policy Diffusion Dynamics in America

political attention. They underattend to lower salience issues that fail to engender mass public support or that require technocratic rather than intuitive policy analysis. The processes leading to nonincremental publicpolicy diffusion are significantly influenced by the way the attributes of innovation encourage sudden policy imitation or gradual program evaluation and implementation. Distinct diffusion patterns emerge because the characteristics of innovations invite fundamentally different political decision-making processes across state governments. This chapter explores how the attributes of innovations shape diffusion dynamics. It draws upon research in policy typologies to explore how dimensions shared by common types of public policies lead to differences in patterns of diffusion. The chapter traces patterns of diffusion across three well-documented types of public policies: state regulatory policies (addressing state economic, environmental, and professional regulatory regimes), morality (social regulatory) policies, and governance policies.4 It theorizes that variation in the complexity, cost, fragility, and salience across types of public policies systematically produces distinct patterns of diffusion. Governance and morality policy are commonly characterized by elevated issue salience, high levels of mass participation, and low technical complexity. These attributes make the diffusion of these two policy regimes especially prone to dramatic moments of policy change resembling policy outbreaks. On the other hand, state regulatory policies are typically characterized by higher technical complexity and lower issue salience. The diffusion of these policies will conform closely to patterns of diffusion driven by incremental decision making. As virologists are working to identify the particular features of certain strains of influenza virus that account for their high rates of transmission, political scientists may find it useful to identify the features of policy innovation especially likely to produce outbreaks across states in the federation. To evaluate patterns of diffusion in regulatory, morality, and governance policy, this chapter extends the distributional analysis introduced in Chapter 2. To understand the degree of nonincrementalism in the diffusion of each policy regime, it first sorts innovations by policy type. It then uses these data to compare the historical diffusion patterns of adoptions of morality, regulatory, and governance policies. This comparison 4

It is important to note that state regulatory policy, morality policy, and governance policy all ultimately seek to codify or alter behavior through regulatory regimes. By convention, however, political scientists have distinguished between what is classically understood to be state regulatory policy (the regulation of economic, professional, or environmental behavior through governmental policy) and state morality and governance policies, which address unique regulatory issues.

Policy Agents

65

demonstrates how certain policy types trigger increasingly rapid diffusion cycles across states. As in the prior chapter, the degree to which the empirical distribution of diffusion data matches a simulated normal distribution of the same mean and variance indicates the degree of incrementalism or positive feedback cycles in the diffusion of innovations. The chapter then extends this analysis to model how differences in the target of policy innovations shape patterns of diffusion across states. The findings in this chapter offer insight into how the common attributes of each of three distinct classes of innovations shape diffusion dynamics. These data show that changes in issue salience, fragility, cost, complexity, and policy targets systematically lead to different patterns of interstate policy diffusion. The chapter begins by explaining how innovation attributes facilitate or hinder the rapid diffusion of innovations across states. It then turns to recent research on morality, governance, and state regulatory policies, demonstrating that the individual traits of innovation are not unique to specific innovations, but rather can be used to model the diffusion of common classes of public policy. Finally, this chapter advances a stochastic process model to identify diffusion dynamics in each of these three policy regimes. In so doing, it provides a theory for understanding how the contagion of innovations shapes patterns of policy making in America. Innovation Attributes and Diffusion Dynamics Research on state politics has identified a number of policy traits that shape patterns of policy making in federations. For example, researchers have commonly observed that innovation complexity and cost (Clark 1985; Gormley 1986), innovation salience (Gormley 1986; Hays 1996; Karch 2007a), and issue fragility (Savage 1985a) all determine the ease of policy evaluation, imitation, and adoption across state governments in the federation. These characteristics therefore present straightforward but important components for modeling variation in the contagion of policy innovation.5 Perhaps the most common observation regarding differences in innovations is that public policies vary considerably according to issue complexity and cost (Clark 1985; Gormley 1986; Eshbaugh-Soha 2006). 5

In addition to this partial list of policy characteristics, researchers have proposed a number of other attributes that shape the ease of policy adoption. For example, Mossberger (2000) suggests that innovation flexibility – the ease with which decision makers can modify innovations – is another important determinant of the adoption and diffusion process.

66

Policy Diffusion Dynamics in America

Because many innovations involve relatively simple policy proscriptions, they require little expert training to engage in political decision making, policy design, and policy implementation. Other policy innovations require the creation or evaluation of complicated new policy instruments and require highly technical expert analysis. Policies with greater issue complexity are more difficult to evaluate and implement, and encourage the involvement of policy experts (Gormley 1986; B. Gerber and Teske 2000). As with issue complexity, innovation cost can present a significant hurdle to innovation adoption, as many states lack the resources to experiment and adopt emerging innovations (Walker 1969; Rogers 1983). In both cases, increasing program complexity and cost should slow patterns of diffusion, as these innovations require greater government program evaluation and emulation before policy adoption. Complexity and cost are therefore directly related to the speed of diffusion. To illustrate how cost and complexity shape the speed of innovation diffusion, consider the difficulty state governments face in evaluating and adopting a relatively simple policy such as the Amber Alert, as opposed to a more complex policy such as state clean air regulations limiting greenhouse gas emissions. The Amber Alert represents a nearly perfect policy for rapid innovation diffusion. Not only does the kidnapping prevention policy appeal broadly to voters; it is also simple to understand and relatively cheap to implement. The Amber Alert permits state lawenforcement agencies to use existing state emergency broadcast alert systems to circulate the details of a kidnapping across jurisdictions. State Amber programs require little new infrastructure and offer easily understood and tangible benefits to policy makers and publics alike. State greenhouse gas regulations demonstrate the difficulty of legislating when policies are complex and costly and when outcomes are uncertain. Greenhouse gas monitoring policies mandate the creation of new regulatory and monitoring systems; require expert analysis regarding the impact of anti-pollution controls on the environment, public health, and the economy; and beg questions of cost and efficacy. The diffusion of clean air innovation is thus predictably slowed by the high cost, complexity, and uncertainty surrounding the innovations. In addition to issue complexity and cost, researchers in state policy making have also identified issue salience as a general innovation trait (Rogers 1983; Gormley 1986; Hays 1996; Karch 2007b; Mooney and Schuldt 2008). Issues vary by general levels of public attention and interest. For example, issues such as education, crime, and government tax

Policy Agents

67

policy are generally salient to attending publics. Elected officials should prove especially responsive to demands for policy change across these highly salient issues. Eshbaugh-Soha (2006) explains, “A salient policy affects a sizeable portion of the general population (Gormley 1986, p. 598) and salience demands that elected representatives respond to the public or face electoral consequences or a decline in public support” (227.) In simple terms for understanding policy diffusion, lower salience issues should be characterized by slower diffusion cycles than higher salience issues. The diffusion of higher salience issues is accelerated by the interaction of mass public opinion and responsive policy making, while lower salience policy innovations encourage technocratic decision making and policy satisficing. Issue salience does not in and of itself guarantee rapid diffusion. Issues with higher salience may precipitate sudden change; however, they may also invite mass political conflict and disagreement, resulting in institutional gridlock and policy inertia. As students of punctuated equilibrium theory have noted, such friction does not necessarily imply incrementalism but rather can create conditions for nonincremental patterns of policy change (B. Jones, Sulkin, and Larsen 2003). As pressures for reform build, conflict in high-salience areas can result in rapid policy reform across states. Finally, a number of policy researchers have observed that the targets of innovations themselves shape patterns of state decision making (A. Schneider and Ingram 1993; Donovan 2001). Students of social constructivism have noted that different levels of political conflict emerge from the way that society defines and understands target populations as “groups of people delimited by some set of shared characteristics who are identified through legislative language as the recipient of a benefit, a burden, of special treatment under [sic] law” (Donovan 2001, 4). Policies addressing different target populations engender dramatically different responses from actors in the political system. For example, state governments are generally supportive of policies that impose burdens on negatively constructed groups like drug users or criminals (A. Schneider and Ingram 1993). Conversely, officials are usually eager to allocate resources and policy support to positively constructed groups like children and senior citizens (A. Schneider and Ingram 1993).6 6

Similarly, governments are generally reluctant to provide policy benefits to marginalized deviant populations and are reluctant to impose costs on advantaged positively-viewed groups. For a summary of this research, see A. Schneider and Ingram (1993).

68

Policy Diffusion Dynamics in America

Innovations systematically produce different levels of opposition depending on the target group they address. Savage (1985a) echoed research on the social construction of target populations to model the sudden diffusion of a “policy whose time has come” (111). Savage also argued that policy fragility – the degree of perceived organized resistance to policy adoption – is central to understanding rapid patterns of innovation diffusion. Comparing the rapid diffusion of state auto lemon laws and child auto-restraint policies, Savage argued that rapid diffusion occurs when the targets of innovation diminish the organized resistance to innovation adoption. Savage applied the theory of issue fragility to anticipate the future rapid diffusion of child protection policies, reasoning that the elevated salience and limited formal opposition to policies protecting children would lead to their extremely rapid diffusion. His analysis proved prescient. From the child-abuse reporting requirements of the 1960s, the missing children’s clearinghouse programs established across states through the 1980s, to the recent diffusion of children’s memorial protection policies such as the Amber Alert, Megan’s Law, and Jessica’s Law, the diffusion of children’s policies occurred in dramatic and rapid succession across states. The impact of the social construction of target populations on diffusion patterns can be further illustrated by the diffusion of crime control and prevention programs in the United States. Mandatory sentencing laws such the three-strikes guidelines of the 1990s, or mandatory sex-offender registries like those mandated by Megan’s Law target criminals by imposing specific burdens on them. These policies are in line with the general public’s belief that crime can be controlled by targeting and assigning burdens to criminals, and so they receive little mass opposition. Likewise, policies such as mandatory child-abuse reporting requirements or the Amber Alert protect children from criminals. These policies extend benefits (in the form of protection) to positively constructed groups. Because these policies match conventional understandings of the appropriate designation of government burdens and benefits on the targeted population, they diffuse rapidly, benefiting from mass public support and limited opposition. On the other hand, a select few recent crime control policies have challenged the received wisdom for burdens and benefits and have attempted to legislate crime control by conferring public benefits to “deviant” criminal groups. Recent efforts to adopt needle exchange programs and state-sponsored drug treatment programs stand as two recent examples.

Policy Agents

69

These policies have countered common approaches for drug-crime control by increasing public spending to provide support for and benefits to a group engaged in criminal activity. Although these policies may have larger goals of limiting prison overcrowding or preventing public health crises, their diffusion is slowed by the significant public opposition to the reforms. All of these policy attributes shape patterns of diffusion in part because they shape the way that mass publics are involved in the policy process. Innovations marked by high salience invite mass participation in the policy process. Lower-salience issues are largely ignored by the public. The relative complexity of innovation shapes the ease with which a lay public can understand and advocate for policy reform. Finally, issue fragility and the policy target reflect how organized public support for or opposition to innovation will emerge. In all cases, elevated mass political involvement should lead to nonincremental policy change, whereas policies that deter mass involvement should tend towards incrementalism and gradual diffusion. This discussion allows us to draw clear expectations for how variation in the attributes of individual policies can shape diffusion dynamics. Variations in issue salience, program complexity, innovation cost, and the fragility of targeted populations all shape the way state decision makers respond to emerging innovations. Technically complicated policies will spread less rapidly than unsophisticated policy solutions, as higher issue complexity requires significant policy analysis by policy experts. Costly policies will diffuse more slowly than policies that require few new resources, as expensive policies require states to dedicate scarce resources to a new innovation. Likewise, issue salience shapes state information processing leading to diffusion by either elevating or diminishing the urgency and priority decision makers place on immediate policy adoption. Finally, targeted populations with different levels of issue fragility encourage distinct diffusion patterns, as policies addressing specific groups encourage diminished political opposition to policy implementation. Researchers in policy diffusion have usually invoked ideas of issue complexity, salience, and fragility to explain the unusual pattern of diffusion for independent policy innovations. However, there is no reason why these characteristics cannot be organized to explain patterns of diffusion across common classes of policy innovations. The next section considers innovation attributes as general traits of classes of policy. As with the case of child protection policies, this section expands the discussion of

70

Policy Diffusion Dynamics in America

diffusion to argue that specific types of policies with common characteristics will produce distinct diffusion patterns. While some policy forms encourage incremental decision making and gradual policy diffusion, the characteristics of an important subset of morality and governance policies encourage episodic diffusion patterns marked by positive feedback cycles. Approaches to the Policy Classification The idea that the common characteristics of policy innovations can be sorted according to a policy typology has been captured in a number of frameworks. For example, students of public policy argue that a useful distinction could be drawn between substantive policies, representing a tangible action taken by government to achieve a policy goal, and procedural policies – which establish the rules and procedures government follows in pursuing a policy action (Anderson 1997). In a similar distinction between tangible and intangible policies, Edelman (1985) distinguished between material and symbolic policies, differentiating between rhetorical statements of values and tangible government actions. Both of these approaches observe important differences between general classes of public policy; however, they suffer insufficient rigor to be translated into a meaningful classification system. As is the case with many general policy typologies, these dichotomous sorting systems suffer from some conceptual slippage under empirical scrutiny. Procedural policies often require significant allocation of government resources to establish a protocol for regulating a target population’s behavior. Material policies are laden with symbolic content. Lowi (1964; 1972) advanced a more expansive theory for differentiating between policy types in a series of essays organized around a system of classification of policies by dominant characteristics. Lowi developed a policy taxonomy centered on the coercive power of government, observing that “different ways of coercing provide a set of parameters, a context, within which politics takes place” (1972, 299). Lowi sorted policies by two dimensions of government coercion: the channels of government application of coercion, and the likelihood of the application of coercion by government. This system suggested four theoretical types of public policy: distributive policy (coercion targeting individuals, likelihood of coercion remote,); regulatory policy (coercion applied to individuals, likelihood of coercion immediate); redistributive policy (coercion applied through environment, likelihood of coercion immediate); and

Policy Agents

71

constituent policy (coercion applied through environment, likelihood of coercion remote) (1972, 300). Lowi believed the policy typology permitted researchers to evaluate how “policy determines politics,” clarifying the different decision-making patterns that emerged when crafting distinct forms of policy (299). Lowi’s insights held immediate appeal for students of public-policy diffusion, who were interested in the idea that policies with common characteristics encourage distinct patterns of uptake of the innovations. Gray (1974) argued that studies of policy diffusion should shift focus away from the internal characteristics of states and instead focus on the dynamics produced by variation across innovations themselves. Although her seminal (1973) study on the effects of interstate communication on diffusion focused on a small set of select cases, Gray later suggested that diffusion research could be extended to understand general patterns of diffusion associated with policy types. She wrote: “Theories which specify linkages between types of policies and types of politics ought to be more useful at the state level than theories simply linking ‘progressive’ or ‘liberal’ policy to certain gross political characteristics” (Gray 1974, 699).7 According to Gray, studying how policies interact with state decision-making institutions would lead to a more compelling understanding of the diffusion of innovations than simply focusing on traditional ideological relationships between governments and public policy. Yet studies on the diffusion of innovations have been slow to incorporate theories relating policy types to the temporal and spatial patterns of diffusion. This is in no small part because of the difficulties faced in operationalizing the conceptual frameworks proposed by students of the policy typologies. Although studies of policy types have provided a useful framework for organizing thought about how government policy will produce different levels of political conflict, they have been less successful in providing a meaningful system for classifying public policies. In a short review of policy typologies, K. Smith (2002) notes that efforts to classify policies following Lowi’s or other alternate typologies confirmed that “it was virtually impossible to objectively classify policy: Scholars classified the same policies differently; some policies overlapped categories, and others seemed to shift categories over time or in response to changes in the broader political environment” (380). 7

Gray’s interest in how policy types shaped diffusion dynamics is discussed in some detail in Savage’s review of policy diffusion in the federation (1985b).

72

Policy Diffusion Dynamics in America

A Narrower Classification of Policy Types: Regulatory Policy, Morality Policy, and Governance Policies in American State Politics The study of regulatory policies has emerged as an important exception to the problem of classifying and studying patterns of decision making associated with Lowi’s policy typology. Research on both national- and state-level regulatory policies has proliferated across disciplines in recent years, ranging from studies of state economic regulatory enforcement to environmental protection and morality (social regulatory) policy restrictions. Although regulatory policies cover a diverse range of issue areas, they share a common set of policy instruments and goals. Regulatory policy involves the use of government power to control the behavior of groups or individuals (Lowi 1998). Regulatory policy is expressed in mainstream politics through a variety of instruments as government attempts to control or minimize the consequences of the behavior of individuals or groups – establishing standards of conduct, designing licensing regimes to sanction and enforce standards of conduct, regulating behavior through taxes and tariffs, or regulating the economic or public sector to encourage competition and ensure the fair provision of a service or public good (Lowi 1998). Regulatory policy scholars have distinguished between characteristics of both social and economic regulatory policy to understand different patterns of political conflict (Tatalovich and Daynes 1998; Mooney 1999, 2001; Mooney and Schuldt 2008). Gormley (1986) proposed a classification system for modeling politics of federal regulatory policy making based on characteristics of issue complexity and issue salience. Gormley initiated his study of state regulatory policy by first sorting policies according to the attributes of salience and complexity. Social regulatory policy on matters such as gun control, abortion, and the sale of pornography could be characterized as having high public salience and low issue complexity. Issues like regulatory licensing policy, banking regulation, and insurance reforms had opposing characteristics of being low in salience and extremely high in complexity. Gormley argued that the dimensions of these regulatory policies could be used to anticipate political involvement and conflicts between actors in the public-policy process. Eshbaugh-Soha (2006) explains: The salience and complexity dimensions of public policy present different incentives for political actors to participate in the policy making process. Because different policies comport differently with these dimensions, different policies will present a different set of opportunities for involvement in the policy process, influencing who will play a prominent role in its adoption or implementation and who will not (225).

Policy Agents

73

Gormley anticipated that changes in the dimensions of regulatory policy would shape patterns of state policy making. Gormley argued that the direct involvement of elected officials and citizen activists in political decision making would be highest when issue salience was high and complexity was low. When decision making involved more complex and less salient regulatory policy, political decision making would rely upon bureaucratic and professional expert policy analysis. This theory matches the decision-making model articulated in Chapter 2. The characteristics of policies invite different decision-making responses by state governments. Gormley’s research has immediate implications for understanding how dimensions of state regulatory policy shape the diffusion of innovations. It reinforces the idea that issue salience and issue complexity are major determinants. The very attributes that encourage state bureaucratic decision making or widespread public pressure and representative involvement in policy making will lead to important differences in diffusion dynamics. Morality/Social Regulatory Policy In recent years researchers have begun to offer a series of interesting refinements to the classification of types of regulatory policy. Most prominent is the emerging concept of state morality policy, a form of social regulatory policy where the government practices “the exercise of legal authority to affirm, modify, or replace community values, moral practices, and norms of interpersonal conflict” (Tatalovitch and Daynes 1998, xxx). Researchers in political science have produced a number of compelling case studies of the regulation of morality, including same-sex marriage bans (Haider-Markel 2001), hate crimes (Disarro 2005), abortion laws (Mooney and Lee 1995), pornography (Daynes 1998), and statutory rape legislation (Cocca 2002). These case studies suggest that morality policies produce systematically different patterns of political behavior than other policy types. Conflicts in morality policy reflect disputes over first principles and core moral values (Mooney 1999; Cocca 2002; K. Smith 2002; Mooney and Schuldt 2006). Compared to other policy types, morality policies are characterized by technical simplicity and high salience (Tatalovich and Daynes 1998; Mooney 1999; 2001; Cocca 2002; K. Smith 2002; Mooney and Schuldt 2006). Social regulatory policy involves broad citizen participation and tends to be dominated by noneconomic interest groups (Mooney and Schuldt 2006). These traits reduce the importance of technically sophisticated decision making by trained policy experts and imply

74

Policy Diffusion Dynamics in America

a strong link between public opinion on political decision making in morality policy legislation (Mooney 1999; Cocca 2002).8 In recent years, diffusion studies have made prominent advances in understanding how morality policies are adopted across the states (Lee and Mooney 1999; Haider-Markel 2001; Cocca 2002; Soule 2004; Disarro 2005). Most relevant for this analysis is the potential extension of Mooney and Lee’s (1999) research on the temporal diffusion of death penalty reenactments. Mooney and Lee specifically argued that the characteristics of morality policy could lead to extremely rapid diffusion across states. In their study of temporal diffusion of death penalty reenactments, the authors found that the pattern of diffusion departed from the typical S-shaped pattern expected in diffusion research, indicating much more rapid issue uptake than expected. If policy attributes such as low issue complexity and elevated issue salience precipitate rapid patterns of diffusion, then nonincremental diffusion should emerge as a general pattern of morality policy adoption. This could be complicated, however, by political friction from the strong opposition sometimes generated by morality policies. This could result in policy stagnation, a ceiling effect in which the policy is taken in only a receptive subset of states (see Chapter 4) or in episodic “punctuated” outbreaks of policy uptake. Regardless of the precise pattern of uptake, morality policies encourage the positive feedback cycles of elevated issue attention and representative responsiveness, encourage policy mimicking, and result in patterns of diffusion that deviate sharply from an S-shaped normal curve. Governance Policy In a recent and important attempt to clarify the role of direct democracy in state constituent government policies, Tolbert (2002) introduced a new category of governance policies, those innovations that change the rules of the political system and “regulate how the state should proceed to govern” (80). Tolbert reviewed a series of state level political reforms that delineated the behaviors of government and elected officials – including term limitations, tax limitations, campaign finance reform, and open 8

Although morality policy is a relatively new introduction into the policy typology literature, it appears to be have high construct validity. In a study of a taxonomical approach to classifying public policies, K. Smith (2002) demonstrated that coders proved remarkably reliable in differentiating between morality policies, standard economic and social regulatory policies, and policies with no common dimension (386–390).

Policy Agents

75

primaries. She argued that each of these policies was unified by an unusual regulatory goal, as “governance policies modify the behavior of the public sector and government officials” (Tolbert 2002, 80). Thus, government policy is a type of regulatory policy in which citizens attempt to control the rules and behaviors of elected officials and government representatives. As with morality policies, governance policies share attributes that distinguish them from other regulatory policy forms. First, governance policies reverse the direction of regulation and political conflict, as citizens regulate the behavior of government. As with social regulatory policies, governance policies are also characterized by arguments over first principles and values. Whereas social regulatory policy aims to regulate the social behavior of citizens, governance policy defines the appropriate relationship between governments and the people. Second, governance policies are often described by high levels of salience and generally encourage mass political support. Tolbert (2002) explains, “Governance policies tend to be supported by large majorities, and thus do not lead to sustained political conflict, but they are opposed by elected officials” (79). Finally, governance policy is strongly associated with the tradition of direct democracy, as this policy-making institution presents one of the few ways that citizen activists can implement governance policy regulations. This is fundamentally important for distinguishing the politics of governance reforms from other regulatory policy types. Governance policy is frequently drafted by citizen advocacy groups and implemented directly through a state popular vote. Although many other policies are drafted and implemented through direct democracy, the initiative process is especially suitable as a vehicle for governance policy reforms, as politicians are often reluctant to themselves introduce legislation that fundamentally alters the structure, organization, and power of government.9 It is interesting to observe the expected relationship between governance policy and processes of political decision making leading to policy diffusion. Governance policy may involve technically complex issues; however, it shares a relationship to morality policy in that debates over government regulation are generally reduced to first principles – in this case, principles dealing with fairness, corruption, and democratic norms. A second significant distinguishing characteristic of governance policy 9

This close relationship to the initiative process also suggests that many states will be extremely susceptible or resistant to governance policy reforms. As explored in Chapter 4, states with direct democracy are far more likely to adopt governance policy reforms than the remaining states without this institution. The diffusion of governance policies may thus be restricted by a ceiling effect, where many states simply resist adopting reforms.

76

Policy Diffusion Dynamics in America

extends from its unique relationship to the political institution of direct democracy. Reforms on the powers and behaviors of government and elected officials have been associated with the initiative process since populists and progressives passed and used the initiative to rid state politics of corrupt and entrenched government at the turn of the century (Tolbert 2002). Because governance policy often spreads through a mechanism entirely outside of the formal decision-making processes in state legislatures, it is likely that political decision making associated with governance policy adoption will be distinct from decision making commonly associated with representative deliberation. Importantly, many governance policies that are adopted across states are necessarily high-salience policies, as they have been the focus of an initiative-gathering campaign and a public referendum. Given its high salience, high levels of mass participation, focus on first principles, limited complexity, and association with the tradition of direct democracy, governance policy should exhibit pronounced departure from the incremental learning curve, spreading much more rapidly than traditionally expected in innovation and diffusion research. Nonincrementalism should emerge in governance policy diffusion for two reasons. First, like morality policy, governance policy invites rapid policy change that occurs when mass political attention is focused on a single issue across states. Second, governance policy is frequently developed and implemented by citizens’ interest groups, and can diffuse as quickly as activists can qualify a governance reform on the ballot. Unlike morality policy, governance policy innovations are unlikely to generate friction from widespread political opposition in the electorate. So at least in the states with direct citizen initiatives, these policy innovations can be expected to be taken up with great speed. State Regulatory Policy: Economic, Environmental, and Professional Regulatory Regimes in American Politics Finally, researchers have identified a broad class of economic, environmental, and professional regulatory policies that are conventionally referred to as state regulatory policy that exhibit higher levels of issue complexity and lower levels of political salience. At the state level, students of government regulatory policy have explored issue areas including energy regulation (Gormley 1983), air pollution standards (Ringquist 1993), agricultural waste (Koski 2007), health care market reforms (Carter and LaPlant 1997; Stream 1999), and car dealerships (Savage

Policy Agents

77

1985b). Policy development in each of these diverse issue areas is united by a common regulatory relationship between state government and affected individuals. In each instance, government attempts to shape the behavior of private industry or citizens by using its coercive authority to achieve policy goals. Importantly, these forms of regulatory policy are characterized by high degrees of technical complexity, high cost, low political salience, and high issue fragility. B. Gerber and Teske (2000) explain: While regulatory policy is unique partly for its constraints on private behavior, it is often highly technical, requiring significant bureaucratic expertise, yielding a concomitant delegation of substantial policy-making authority to bureaucrats (852).

The type of decision making encouraged by this class of economic, environmental, and professional regulatory policy does not engender mass political attention and responsive policy making. Instead, decision making in this policy domain is shaped by technocratic policy analysis. Just as importantly, because state regulatory policy defers the cost of regulation on to a regulated group, this policy form should also be marked by higher levels of issue fragility as organized interest groups oppose interstate reforms. Unlike morality and governance policy, state economic, environmental, and professional regulatory policy development is generally dominated by economic interests groups. If morality and governance policies are likely to produce policy outbreaks because of their high salience and low technical complexity, then regulatory policies with opposing characteristics should produce much different patterns in the policy process. Technocratic regulatory policies are especially prone to incremental diffusion and are unlikely to produce policy outbreaks. State regulatory policy making is generally attended to by policy experts and state bureaucrats rather than mass publics and elected officials. Decision making in economic, environmental, and professional regulatory policy therefore encourages incremental policy adjustment rather than sweeping policy reform. Patterns of diffusion in the domain of state economic and social regulatory policy making should conform closely to a normal curve, indicating an incremental learning process of program identification, evaluation, and implementation. Policy Types and Policy Dynamics These expectations can be restated to permit comparison of diffusion dynamics across the three policy types. Of the three policy forms,

78

Policy Diffusion Dynamics in America

governance policy – characterized by high salience, low complexity, and direct citizen policy making through institution of direct democracy will deviate most strongly from the incremental learning curve, and will exhibit the greatest tendency toward positive feedback cycles and policy outbreaks, although these outbreaks may be limited to states with citizen initiatives. Governance policy often diffuses through channels entirely outside of statehouse governments, so the traditional pressures of legislative incrementalism will not constrain the diffusion of governance policy innovations. Morality policies will also deviate sharply from the incremental learning curve, suggesting that diffusion of morality policy cannot be explained through a single incremental diffusion model. While the degree of nonincrementalism in the diffusion of morality policy will be less pronounced than for governance policy, the elevated issue salience and diminished issue complexity of morality policies indicate this policy form will depart significantly from incremental patterns of public-policy diffusion. Instead, morality policy will exhibit pronounced punctuations as citizen interest groups and mass publics pressure state governments for rapid policy adoption. Finally, state regulatory policies will conform most closely to incremental patterns of diffusion. The pressures leading to policy adoption in this class of policy will be strongly associated with the behaviors leading to incremental policy change – time constraints, high uncertainty, and technical complexity in the domain of regulatory policy making encourage the satisfying behavior and lead to an S-shaped pattern of adoption over time. Diffusion Dynamics and Policy Targets, and Issue Fragility Of course, the salience and perceived complexity of a policy innovation change with the way a problem is defined by politicians and publics. Studies in agenda setting have revealed how problem definition and issue framing alter political responses to policy problems. Shifting arguments surrounding policy reform can invite renewed interest in a policy problem, expanding the scope of conflict to new actors and eventually leading to policy change (Schattschneider 1975; Baumgartner and Jones 1993). For example, the movement to regulate nuclear energy in the 1970s gained political traction when nuclear energy was redefined as a potential public health and environmental threat rather than as a cheap and renewable source of energy (Baumgartner and Jones 1993).

Policy Agents

79

Although this research explores interest-group involvement in issue framing and reframing of innovations in Chapter 5, it is possible to get a sense of how the framing of a particular policy innovation shapes diffusion patterns here by exploring the diffusion dynamics of policies targeting very different groups. The rate of policy diffusion is in no small part shaped by the degree of organized opposition or “fragility” for policies proscribed for a particular social group. Distinct patterns of policy making are shaped by how politicians prioritize and act upon a specific innovation. Some policy targets spur rapid political action, whereas others demand less urgent responses. As an example, consider how policy makers respond to public policies addressing the welfare of children. Research on the development of child protection policies has documented that politicians and publics respond quickly once they become aware of threats to child welfare. In her study of the politicization of child abuse, Nelson (1984) argues that the rapid diffusion of child-abuse reporting requirements in the late 1960s was specifically shaped by the salience of the issue, the cost of the policy, and the incentives for political actors to become involved in legislating child protection. Legislators “considered the physical abuse of children disgusting” and seized the opportunity to benefit from “no-cost rectitude,” taking a position consistent with mass opinion (1984, 76). Child-abuse reporting laws spread so rapidly that Nelson observed that patterns of child-abuse reporting could not be explained using traditional diffusion models. She wrote: Clearly there is little utility to trying to fit the standard S-curve to this diffusion pattern. The problem with fitting the adoption pattern to the S-curve is, of course, largely technical, created by the fact that all fifty states adopted the laws within five years. (Nelson 1984, 80)

The characteristics that shaped policy development in the diffusion of child-abuse reporting requirements appear generalizable across a class of policies that aim to protect children. As Savage (1985a) notes, there is extremely limited opposition for implementing policies addressing children. Furthermore, child protection policies are generally of elevated salience and are centrally important to voters. If this body of research is correct, then policies targeting children should generally encourage atypically rapid diffusion patterns. Although the target of some policies will generally elevate issue attention, other policy targets will engender less mass political involvement and present fewer opportunities for high-profile position-taking by politicians.

Policy Diffusion Dynamics in America

80

For example, consider the levels of public interest and involvement in the development of professional licensing policies, which set standards and regulate entry into professions such as hairdressing, accounting, nursing, and law. Unlike child welfare policies, professional licensing policies are characterized by low issue salience and high complexity. Efforts to impose regulatory burdens such as licensing requirements should engender considerable opposition from the targeted group, but are not likely to invoke large public response. Although historically, publics have mobilized to protect the well-being of children, it is difficult to imagine similar moments of widespread public outcry demanding the sudden development of new barber’s licensing requirements. In addition to the relatively low salience of this issue, there is generally a higher degree of opposition to professional licensing regulation, as regulated groups organize to oppose government intervention. The diffusion of professional licensing policies should be far less prone to positive feedback cycles than child welfare legislation. To illustrate how policy targets shape diffusion patterns, this chapter explores the diffusion patterns of policies addressing children and those licensing professional groups. Children’s policies should encourage rapid diffusion patterns and deviate sharply from incremental decisionmaking.10 The diffusion of government licensing policy should closely follow the normal S-shaped curve of incremental diffusion. Data The data collected to evaluate patterns of diffusion across policy types are identical to the data collected and described earlier in Chapter 2. A simple coding procedure was used to classify each of the 133 policies collected by policy type. Three graduate student coders at a large West Coast research university, who were familiar with the policy typologies literature, identified characteristics of each policy as predominantly regulatory, morality (social regulatory), governance, or none of these.11 Because policies 10

11

A. Schneider and Ingram (1993) identify children as dependents and therefore expect that policy burdens will be oversubscribed and policy benefits will be undersubscribed. However, research in innovation diffusion suggests that children are suitably positively constructed as to benefit from the same levels of policy attention as advantaged groups. The coding procedure was manifest. Students were provided with a brief description of each policy and asked to classify it by a primary characteristic. When a policy could not be coded as primarily morality, governance, or regulatory, coders were instructed to classify it as “other.”

Policy Agents

81

are often overlapping (involving both regulatory and morality components), coders were instructed to classify only the predominant element of the public policy. Only policies with perfect intercoder agreement were included in the aggregated measure. This method helped identify 40 regulatory policies, 22 morality policies, and 10 governance policies. A survey of recent literature was used to confirm the reliability of the classification of policies in each grouping. Policies included in the morality category are consistent with cases presented in the two most prominent recent anthologies on morality and social regulatory policy (Tatalovich and Daynes 1998; Mooney 2001) Coders generally agreed with experts in morality policy and identified policies addressing abortion, gay rights and homosexuality, euthanasia, alcohol restrictions, drugs, and women’s rights as being primarily morality policy. However, in this study coders departed from recent literature by including a class of crime victims’ rights policies as morality policy. Coders classified policies like childabuse reporting requirements, the reenactment of the death penalty, and crime victim’s rights amendments as morality policies. Policies classified as governance policy were compared to cases selected in Tolbert’s (2002) study on state governance policy. Coders followed Tolbert in classifying term limits, supermajority requirements for tax increases, and voting rules for primary elections as governance policy. The remaining policies identified by coders – laws restricting the use of eminent domain, municipal home rule, and budgeting standards – are consistent with the definitions offered by Tolbert at the outset of her study. Finally, state regulatory policies were matched against B. Gerber and Teske’s (2000) review of the theories and research on state regulatory policy making. Gerber and Teske identify a broad range of state economic, environmental, and professional regulatory policies governing energy, telecommunications, insurance, banking, trucking occupational licensing, environmental regulation, consumer protection, worker safety, and health. Almost all of the regulatory innovations identified by coders in this study can be sorted into this grouping, indicating a high level of validity on the regulatory policy classification used in this research. In addition to classifying policies by policy type, data were also sorted according to the policy targets, specifically identifying policies targeting children and a set of policies addressing professional licensing This procedure identified 7 children’s policies and 11 licensing policies. A list of policies in each of these categories can be seen in Appendix C.

82

Policy Diffusion Dynamics in America

Method The methods employed in the distributional analysis of this chapter are identical to the procedures followed in Chapter 2. To identify distinct patterns of diffusion across policy regimes, this chapter compared the empirical cumulative distribution functions for innovations aggregated by policy type and policy target. As a reference, it also reproduced the distributional analysis of all policies together from Chapter 2. This chapter compares the empirical distribution for each group of policies to their expected simulated Gaussian distribution of an identical mean and variance. The degree to which a plotted empirical cumulative distribution function conforms to the expected incremental or Gaussian curve can be tested through a visual plot. The plotted CDF of the hypothetical normal curve provides the familiar S-shaped distribution. If the duration data follow a similar distribution, data should closely follow the simulated normal cumulative distribution function. To confirm the findings in each visual plot, researchers conducted several standard statistical tests of the normality and shape of the data. As in Chapter 2, the analysis included the Shapiro-Wilk and Anderson-Darling tests to evaluate the normality of diffusion data. This investigation also measured the data for positive kurtosis and skew to evaluate the shape of the empirical distributions. Results Figure 3.1 reproduces the empirical cumulative distribution function of all policy adoptions against a simulated normal cumulative distribution curve of the same mean and variance. Here, the x-axis represents times to policy adoption, and the y-axis represents the probability of adoption. As indicated in Figure 3.1, the empirical cumulative distribution of the pooled diffusion data deviates sharply from the expected normal distribution. A greater number of policy adoptions occur more rapidly than expected, indicating a nonincremental and non-normal distribution. The separation between these two curves is indicative of a nonincremental learning process. Across cases, policy diffusion is marked by nonincremental decision making. Figure 3.2 captures the diffusion of governance policies in American politics. As expected, the diffusion of governance policies produces punctuated dynamics, deviating sharply from the simulated normal cumulative distribution curve of the same mean and variance. The non-normality of governance policy adoption times is confirmed by all distributional tests found in Table 3.1. Governance policy has a high LK score of 0.279 and

Policy Agents

83

table 3.1. Statistical Tests for Normality by Policy Type Historical Era

A-D Test (A) P<

S-W Test (W) P<

All Years 216.871 0.000 0.847 Governance 37.234 0.000 0.699 Policy Morality 34.590 0.000 0.801 Policy Regulatory 51.421 0.000 0.877 Policy Licensing 7.419 0.000 0.937 Policy Children’s 8.461 0.000 0.883 Policy

Kurtosis L-Kurtosis L-Skewness

0.000 4.080 0.000 6.729

0.120 0.279

0.322 0.463

0.000 3.352

0.096

0.366

0.000 4.260

0.134

0.278

0.000 4.146

0.154

0.175

0.000 2.597

0.029

0.239

0.0

0.2

Probability of Adoption 0.4 0.6 0.8

1.0

All Policies

0

20

40 60 Years to Adoption

80

100

figure 3.1. Cumulative distribution of adoption times: All policies. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

Policy Diffusion Dynamics in America

84

0.0

0.2

Probability of Adoption 0.4 0.6 0.8

1.0

Governance Policies

0

20

40

60

80

Years to Adoption

figure 3.2. Cumulative distribution of adoption times: Governance policy. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

a kurtosis score of 6.729, both significantly above the values expected in a normal distribution. Furthermore, the skewness of governance policy is far higher than the expected skew in a normal distribution, and the S-W statistic indicates that the observational data from the adoption of governance policies are remarkably different from a normal distribution expected in the incremental learning model. More than any other policy type, governance policies are characterized by sudden policy change, exhibited by the extremely rapid issue uptake indicated in the visual plot. Figure 3.3 represents a plot of the empirical CDF of morality policy against a simulated normal CDF of the same mean and variance. As hypothesized, the empirical CDF of morality policies deviates sharply from the simulated normal CDF. As with governance policy, morality policies appear to be characterized by a rapid diffusion pattern, with a faster initial takeoff than anticipated by a simulated normal curve. These findings are confirmed by the two measures of normality of distributions

Policy Agents

85

0.0

0.2

Probability of Adoption 0.4 0.6 0.8

1.0

Morality Policy

0

20

40 Years to Adoption

60

80

figure 3.3. Cumulative distribution of adoption times: Morality policy. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

in Table 3.1. Neither the A-D nor the S-W tests of normality confirm the data came from an “expected” normal distribution. The measures of the shape of morality policy diffusion are also interesting. Surprisingly, the kurtosis scores for morality policy do not indicate a much higher level of excess kurtosis than expected in a normal distribution; they show a smaller deviation than do either governance or regulatory policy forms. However, like all other forms of diffusion data, morality policy does show somewhat higher skewness around the mean. Figure 3.4 represents a visual plot of the empirical and simulated CDFs of state regulatory policy. Here we see an unexpectedly large departure from the normal distribution. These regulatory policies are characterized by a more gradual takeoff than either morality or governance policy; however, the distribution of regulatory policy adoptions still deviates considerably from an expected incremental learning distribution of an identical mean and variance. The non-normality of regulatory policy is confirmed by the goodness-of-fit statistics presented in Table 3.1.

Policy Diffusion Dynamics in America

86

0.8 0.6 0.4 0.0

0.2

Probability of Adoption

1.0

Regulatory Policy

0

20

40

60

80

Years to Adoption

figure 3.4. Cumulative distribution of adoption times: Regulatory policy. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

Figures 3.5 and 3.6 present an interesting comparison of children’s and licensing policies. As expected, each of these policies displays remarkably different patterns of distributions. Children’s policy is characterized by an extremely rapid uptake across policies. Interestingly, both the simulated and empirical distribution for children’s policies indicate that as a policy form, the diffusion of children’s policy – through either legislative emulation or positive feedback cycles – will occur more rapidly than most other policy types. Although distributional statistics indicate that children’s policy data are non-normally distributed, it is interesting to note that children’s policies have the smallest kurtosis scores of all classes of policies in the sample. Indeed, the L-kurtosis (LK) score of 0.029 falls well below the typical kurtosis expected in a normal distribution. Such low values of LK are often indicative of a bimodal distribution, suggesting that children’s policy occurs through a sudden and short period of policy adoptions in time. Figure 3.6 presents the empirical distribution of adoption times across licensing policies – an issue area characterized by especially high technical

Policy Agents

87

0.0

0.2

Probability of Adoption 0.4 0.6 0.8

1.0

Policies Targeting Children

0

20

40 Years to Adoption

60

80

figure 3.5. Cumulative distribution of adoption times: Children’s policy. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

complexity and low salience. In the case of licensing policy, the empirical cumulative distribution function and the simulated normal distribution function produce almost identical curves. Furthermore, the Shapiro-Wilk (S-W) test of normality is high for licensing policy, indicating that these data approach a normal distribution. This suggests that licensing policy adoption times are closer to an expected normal curve. Although this finding is suggestive, both the visual plot and the S-W score should be read with some caution. The Anderson-Darling (A-D) test and the slightly high kurtosis scores indicate that the data deviate from an expected normal diffusion curve. Discussion In each of these graphs, the empirical distribution of adoption times deviates from the hypothetical normal distribution expected in an incremental learning process. These findings present strong evidence that the diffusion of innovations changes with the characteristics of an innovation. Understanding these processes permits important insight into the causes

Policy Diffusion Dynamics in America

88

0.0

0.2

Probability of Adoption 0.8 0.4 0.6

1.0

Licensing Policies

0

20

40 Years to Adoption

60

80

figure 3.6. Cumulative distribution of adoption times: Licensing policy. In this figure, the sawtooth line represents the empirical cumulative distribution function for all adoption times gathered. The lighter line represents a simulated normal cumulative distribution function of the same mean and variance as the empirical data.

of diffusion dynamics. As with the prior chapter, this distributional analysis presents evidence that the diffusion of innovations represents at times both a process of incremental learning and sudden policy feedback cycles. Governance policy – the policy type considered most likely to inspire positive feedback cycles – deviates most sharply from the expected incremental normal distribution. Morality policy – characterized by high emotional appeal and arguments over first principles – is marked by a tendency toward rapid diffusion. Surprisingly, even state regulatory policy reforms deviate from the expected normal distribution, although the more gradual takeoff in the empirical cumulative distribution curve does suggest that the higher technical complexity and lower salience of these policies encourage a more gradual process of program evaluation and emulation. The possibility that policy diffusion is characterized by a mixture of decision-making processes is reinforced by the stark comparison of licensing and children’s policy. Licensing policies, marked by low salience, high complexity, and high fragility conform closely to the incremental diffusion curve. This finding is not surprising, as it is difficult to imagine an

Policy Agents

89

event leading to an interstate public outcry demanding state regulations of beauticians, dentists, or real-estate brokers. It is somewhat easier to imagine mass public support for the adoption of a range of policies addressing the welfare of children, which are characterized by a favorable policy target, consistently high salience, and broad emotional appeal. This dynamic is confirmed in Figure 3.5, which demonstrates that diffusion of children’s policies occurs extremely rapidly across states. Implications This analysis provides some important insights into the cause of diffusion dynamics. As suggested by the epidemiologic framework, the characteristics of policy agents contribute to the unique patterns of public-policy diffusion. Policies with appealing policy images, low technical complexity, and high issue salience are far more likely to produce policy outbreaks than less salient and more complex innovations. Interestingly, the policy type that deviates most sharply from the expected incremental learning model is state governance policy. There are some intriguing implications that can be taken from the unique pattern of governance policies. Many governance policies are implemented through direct democracy rather than through state legislative decision making. This additional venue creates the opportunity for sudden policy imitation and extremely rapid diffusion, as was the case with term limits in the early 1990s. The punctuated dynamics of governance policies are entirely consistent with an agenda-setting model of policy diffusion. Governance policies such as term limits benefit from strong organized interest-group involvement, high salience, and technical simplicity.12 As Chapter 5 describes in detail, term-limit activists were able to exploit national outrage at governmental corruption to place term-limitation measures on the ballots of 21 states between 1990 and 1994. By relying almost exclusively on state ballot initiatives as a vehicle to pass term limitations, activists were able to pass legislation rapidly, and without the normal process of legislative compromise that occurs in statehouse governments. These findings also suggest that the ways policies interact with state political institutions and interest-group actors shape patterns of diffusion. 12

Although it is certainly possible that morality policies could be imitated through state ballot initiatives – as was the case with initiatives to prohibit gay marriage in the first decade of the twenty-first century – this policy form more commonly spreads through responsive state policy making.

90

Policy Diffusion Dynamics in America

Although both morality policy and governance policy share common characteristics of high emotional appeal, mass involvement in policy making, elevated issue salience, and low complexity, they display somewhat different patterns of policy diffusion. This chapter speculated that the unique patterns result from the differences in political institutions used across states to implement governance or morality policy reforms. Governance policy often diffuses through the channels of direct democracy, entirely outside of statehouse governments. Conversely, theory suggests that the electoral connection will be strongest in the diffusion of state morality policies, as state representatives engage in mimicking highsalience, low-complexity policies to satisfy the demands of mobilized publics. Although not directly observed in this chapter, it is important to note that both governance and morality policy legislation may invite more limited patterns of diffusion than other policy forms. States without the additional policy-making institution of direct democracy may be far more resistant to governance policy reforms than states with the initiative system. In select cases, this may impose a ceiling effect on the extent of public-policy diffusion, even when an emerging governance policy innovation triggers widespread attention and interest across states. Likewise, morality policy diffusion often emerges when heightened political attention triggers a reform movement. However, because morality policy invites political conflict over core political values and principles, there will often be states that resist innovations that are too restrictive or permissive of social behaviors. For example, in the 1970s, death penalty reenactments spread quickly across states that embraced capital punishment as a suitable deterrent; however, a large subset of states appear to have no interest in implementing the reform. Because of these characteristics, the diffusion of both morality and governance policy may be limited by a ceiling on the extent of diffusion across states. Although economic competition over regulatory policies may eventually encourage every state to embrace an innovation, the debates over core values and first principles may lead to policy that is unsuitable for implementation across all 50 states. Morality and governance policies may invite abrupt policy outbreaks across a subset of states, rather than a gradual diffusion driven by cost-benefit analysis. A third dimension of state policy making emerges from the study of economic and professional regulatory policy making. Because the analysis and adoption of these policies require significant technological analysis and complex program design, theorists have argued that state

Policy Agents

91

regulatory policy making addressing economic, environmental, or professional behavior will be determined by bureaucrats and policy professionals. Because of the low salience of the issues, neither state representatives nor citizen activists are likely to attend to incremental reforms in state regulatory regimes. This presents an interesting question for understanding diffusion dynamics in federalism. Although the analysis of diffusion patterns across different innovations has provided an understanding of why certain innovations produce positive feedback cycles and policy outbreaks, it has only provided a cursory assessment of the state decision-making processes facilitating the adoption of these diverse public policies. The next chapter explores how state political, ideological, and socioeconomic characteristics shape state responses to governance, morality, and regulatory policy. As this chapter suggests, each of these policies should be associated with a different attribute of state policy making. Chapter 4 evaluates how differences in key state sociopolitical attributes make certain states receptive or resistant to specific classes of public policy.

4 Innovation Hosts State Characteristics and Diffusion Dynamics

Diffusion dynamics result not only from differences in the attributes of the innovation, but also from variations in the internal characteristics of those states encountering, responding to, and adopting new public policies. That state regulatory, morality, and governance policies each produce different patterns of policy diffusion revives a series of interesting questions regarding the diffusion of policy innovations in American states. What explains state receptivity to innovation? Why are some states receptive to innovation though others appear policy-resistant?1 Are states equally responsive across types of public policy, or does state receptivity to innovation shift depending on the characteristics of the agent of innovation? The diffusion dynamics identified in Chapter 3 suggest the interesting possibility that state receptivity to innovation changes in response to the characteristics of policy innovation. Just as epidemiologists have discovered that populations with certain genetic or behavioral traits suffer from elevated risks for contracting a particular disease, states with particular political, economic, or ideological attributes are especially receptive to different classes of innovation. This expectation matches common explanations for variation in state policy making. For example, states with the 1

This chapter departs from prior research in public-policy diffusion by referring to rapid adopters of innovation as receptive or susceptible rather than innovative. The idea that the earliest adopting states are innovative imposes a highly stylized way of thinking about the diffusion of innovations by calling attention to the agency of state decision makers in adopting positive innovations. Because a number of factors make states more susceptible to arguments and innovations, this chapter instead refers to early adopters as receptive or responsive to innovation.

92

Innovation Hosts

93

initiative process are responsive to innovations that regulate the behavior of elected government officials, as they have the institutional means available to pursue governance reforms (Tolbert 2002). Religious and ideologically conservative states are more responsive to morality policies that regulate behaviors deemed “socially deviant” (Haider-Markel 2001). This chapter explores state receptivity to public-policy innovation in the modern era by building a profile of state characteristics related to policy adoption, and then examining how these factors elevate or diminish the likelihood that a state will be an early adopter of innovation. The chapter begins by revisiting a long-standing question in the diffusion-ofinnovations literature by evaluating whether certain states are systematically more likely to be early adopters of all forms of public policy. However, in keeping with the analysis presented in prior chapters, it then distinguishes between state receptivity to regulatory, morality, and governance policy. To understand why certain states are especially receptive or resistant to emerging innovations, the chapter evaluates how key institutional, ideological, economic, and demographic state characteristics shape responsiveness to innovation. This research identifies considerable variation in the predictors of state receptivity to state regulatory policy (the regulation of economic, environmental, and professional standards), morality policy (government regulation of social behaviors and norms), and governance policy (regulations and limits on the powers of government itself). State receptivity to regulatory policy is strongly correlated with levels of state legislative professionalism, as a state’s capacity to engage in complex policy analysis shapes the speed with which it can process technically complex information. Receptivity to morality policy is shaped by state political competition and citizen interest-group involvement in the policy process, as increased partisan competition makes state policy makers more responsive to citizen demands for morality policy change. State receptivity to governance policy is strongly influenced by the citizen initiative process, as this particular political innovation encourages direct citizen participation in the policy process and permits activists to directly legislate the behavior of government. These findings have important implications for understanding the broader processes of diffusion dynamics. The decision-making processes leading to incremental diffusion or positive feedback cycles are not simply caused by changes in the issue attention cycle and political environment. Instead, distinct patterns of policy diffusion emerge across policy types in part because innovations interact differently with the decision-making

94

Policy Diffusion Dynamics in America

processes of state governments. There is a considerable variation in political institutions across state governments. These variations make certain states systematically more receptive or resistant to innovation. Modeling State Responsiveness to Innovation in Diffusion Research Research into the attributes of early and late adopters of innovation is one of the oldest and most robust areas of inquiry in the diffusionof-innovations research. Disciplines as varied as marketing, sociology, anthropology, epidemiology, and public policy have all generated models for understanding the timing and extent of innovation adoption (Rogers 2003). Although these studies have conceptualized the innovation and diffusion process differently – for example, linking a population’s behavioral characteristics to susceptibility to disease, or looking to media consumption to understand the innovative behavior of trendsetters – the basic form of inquiry remains constant across disciplines. To understand the factors shaping susceptibility or resistance to innovation, researchers have developed a number of approaches to identify those common attributes of early, late, and nonadopters of an agent, innovation, or behavior. The idea that states can be classified as policy leaders and policy laggards was introduced to the study of public policy in Jack Walker’s (1969) study of innovation and diffusion in the American states. Although Walker’s research spawned a groundswell of later research on the role of interstate interactions and influence in the process of innovation diffusion, his study focused in part on identifying and understanding the internal attributes of innovation leaders and laggards in the American states. For Walker, modeling the mechanisms of diffusion was part of a larger research agenda to assess what he termed the “policy innovativeness” of American states, a concept introduced to capture the general speed with which states adopted new policies. His model of the correlates of state innovativeness has become the benchmark for modern diffusion research, as his findings on the importance of state wealth, population size, urban density, and educational attainment in innovation adoption remain among the most replicated and robust findings in policy diffusion research. Research on the correlates of state “policy innovativeness” has since been imitated by a number of researchers interested in policy leaders and laggards in American federalism. Walker’s research design has been modified to rank and model the adoptions of state public policies (Savage 1978;

Innovation Hosts

95

Nice 1994), tort laws (Canon and Baum 1981), health care innovations (Carter and LaPlant 1997), governance policies (Tolbert 2002), and uniform state laws (Quaile-Hill and Hurley 1988). Each of these studies has evaluated how state characteristics such as ideology, culture, political institutions or socio-demographic characteristics shape patterns of public-policy adoptions. In spite of its appeal, the conceptual framework used to model “state innovativeness” has been roundly criticized on both theoretical and empirical grounds (Gray 1973; Eyestone 1977; F. Berry and Berry 1999). Critics contend that models of policy innovativeness suffer from specification problems, as the measure of policy innovativeness (the dependent variable) is constructed with data covering a broad time span, whereas variation in the correlates of innovation (the independent variables) are estimated using cross-sectional data from the median year of adoption (Carter and LaPlant 1997; F. Berry and Berry 1999). Findings from these models are therefore insensitive to temporal changes in state characteristics such as wealth, ideology, or political competition. Researchers have also offered compelling theoretical challenges to the concept of policy innovativeness as a general trait of American states. In her now classic rejoinder to Walker’s essay, Virginia Gray (1973) argued that the processes leading to innovation, diffusion, and policy adoption were so distinctive across specific policies that the idea of policy innovativeness as a trait was inherently flawed. Gray discovered no systematic patterns of innovation and diffusion in her analysis of the diffusion in state welfare, civil rights, and education policy. Instead, she argued that there was such impressive variation across state leaders and laggards from one innovation to the next that the notion of state policy innovativeness was virtually meaningless. This critique is echoed by F. Berry and Berry (1999), who argued that the theoretical and empirical challenges of diffusion research require that researchers forgo broad conceptualizations of the innovativeness of states and instead focus narrowly on processes associated with the diffusion of a single interesting case. They explain: When it is unfeasible to measure important variables for as many units as pooled state data analysis requires, the only reasonable alternative is to sacrifice the benefits available from large-sample quantitative research for the gains secured by intensive analysis of a small number of cases, via case studies or small-sample comparative design. The theories need not change, only the approach to empirical testing (194).

96

Policy Diffusion Dynamics in America

F. Berry and Berry (1999) demonstrated that diffusion researchers could correct for both the empirical and theoretical limitations of past diffusion models using event history methods. In their groundbreaking study of the causes and timing of state lottery diffusion (1990), they demonstrated how changes in both internal state characteristics and interstate interactions over time shaped the likelihood of policy adoption. Following Berry and Berry’s introduction of the event history method, studies on the causes and timing of state policy adoption have proliferated as researchers have attempted to understand how both interstate interactions and internal state characteristics promote policy innovation and diffusion (F. Berry and Berry 1990; F. Berry and Berry 1999). Theorists have tested a broad range of factors increasing a state’s susceptibility and resistance to public-policy adoption. The now classic findings on the diffusion of policy innovation demonstrate the role of geography in state policy adoption, as a given state’s choice to adopt a specific policy is strongly influenced by the policy choices of neighboring states (F. Berry and Berry 1990; 1999; Boehmke and Witmer 2004; Rincke 2004; W. Berry and Baybeck 2005).2 Although geography is an important predictor of innovation and diffusion, researchers have identified a number of other mechanisms explaining the timing of state policy adoption. Research on the influence of political ideology has qualified the determinism of geography by demonstrating that states imitate policy decisions of states with similar ideological and political preferences regardless of geographic proximity (Grossback, Nicholson-Crotty, and Peterson 2004). State receptivity to innovation is

2

Event history models have highlighted the causal mechanisms underlying the “neighborhood effect” in innovation diffusion. Research into both economic competition and social policy learning has provided insights into the contagious regional influence of policy adoption in federalism (Boehmke and Witmer 2004). In keeping with the theory of fiscal federalism, research on economic competition has repeatedly documented how state revenue loss to a pioneering neighbor leads to program emulation (Boehmke and Witmer 2004; W. Berry and Baybeck 2005). This dynamic is famously articulated in a series of articles on state lottery diffusion by William Berry et al., which demonstrate that lottery adoption can be explained by the proximity of a state population to a lottery state (F. Berry and Berry 1990; F. Berry and Berry 1999) and even more accurately by the proximity of major population centers to states with lottery programs (W. Berry and Baybeck 2005). Geography also plays a role in the diffusion of innovation through social policy learning. Researchers have argued that voters and decision makers disproportionately attend to the policy choices of their neighbors. Diffusion occurs as decision makers emulate those successful or politically popular programs of their neighbors to win electoral support from their own voters (Boehmke and Witmer 2004; Volden 2006).

Innovation Hosts

97

in part determined by the policy-making activity of ideologically and culturally similar states in the union, even when such states are not geographically proximate. (Grossback, Nicholson-Crotty, and Peterson 2004). Researchers have also documented the important role that interest groups and professional associations play in the diffusion of innovations. Interstate policy networks (Mintrom 1997; Mintrom and Vergari 1998) and professional associations (Balla 2001) have been shown to be a crucial source of influence and information for activists and policy makers across states.3 Beyond these external influences, event history studies have explored how the internal characteristics of states shape the speed of program adoption, looking at the common attributes of innovation leaders and laggards in the diffusion of innovations. Multiple studies have confirmed Walker’s initial findings regarding the demographic and economic correlates of policy innovativeness. State wealth, population size, urban density, and educational attainment all contribute to state responsiveness to innovation (Nakonezny, Shull, and Rogers 1995; Tolbert 2002; Volden 2006). Larger, richer, denser states have the infrastructure and resources to generate policy ideas and experiment with innovation (Walker 1969; F. Berry and Berry 1999). Researchers have also identified a number of state political and ideological attributes that are central for evaluating the internal decision-making dynamics in policy adoption. Students have evaluated the link between policy innovation and state legislative professionalism (Walker 1969; Carter and LaPlant 1997; Tolbert 2002; Volden 2006); political competition (F. Berry and Berry 1990; F. Berry and Berry 1992; Karch 2006); ideology (Grossback, Nicholson-Crotty, and Peterson 2004; Nakonezny, Shull, and Rogers 1995); and state political culture (Tolbert 2002). These findings suggest that politically competitive, progressive, and professional states are leaders in policy innovation.4 Finally, case studies of public-policy innovation and adoption have modeled how state responses to the problem environment lead to innovation and adoption in a specific issue area. (Eyestone 1977; Nice 1994). Carter and LaPlant (1997) demonstrated how problem severity in health 3 4

Chapter 5 reviews how interest-group activists and policy entrepreneurs shape the movement of innovations from one state to another. Students have also connected state policy innovation to state public opinion and ideology. For example, studies have connected the diffusion of no-fault divorce laws (Nakonezny, Shull, and Rogers 1995) and abortion innovation (Mooney and Lee 1995) to state ideology and religiosity.

98

Policy Diffusion Dynamics in America

care precipitated early adoption of state level health care reforms. Volden (2006) linked state policy innovation in the Children’s Health Insurance Program to levels of uninsured poor children. These findings provide some reassuring evidence that state policy innovation responds to the real-world challenges facing a state.5 Limitations of Event History and Internal Dynamics Models Although single case studies provide important insights into how state characteristics shape the diffusion of innovations, they afford somewhat limited insight into state receptivity to innovation. The popular event history studies permit a thorough evaluation of the predictors of state policy adoption for only a single policy innovation at a time. Any generalizations about characteristics of state innovativeness must therefore be extended from individual case studies or a careful comparison across a select few cases of innovation diffusion.6 Furthermore, few researchers have connected research into state policy types to the process of policy innovation and diffusion. This is surprising, as the range of research on the individual determinants of policy adoption suggests that classes of public policies may provoke demonstrably different patterns of policy diffusion. The factors leading to the early adoption of state lottery programs should be remarkably different from those causing state adoptions of anti-crime legislation. The notion of “state innovativeness” as a generalizable trait remains conceptually appealing; especially if states with common ideological, political, and sociodemographic characteristics respond in systematically similar ways to different classes or types of public-policy innovations. To evaluate how state characteristics shape the diffusion of innovations, this chapter modifies Walker’s initial research on state policy innovativeness to model state receptivity to innovations. To model receptivity, the chapter begins by updating Walker’s measure of state innovativeness, ranking states according to their average speed of policy adoption from 1960 through 2006. It then develops and tests a series of hypotheses 5

6

Studies of state problem environments extend beyond research in health policy innovation. For a review of this research tradition, see Nice’s (1994) Policy Innovation in State Government. There is evidence that students of policy diffusion are growing increasingly interested in the comparative dynamics of innovation diffusion. In Democratic Laboratories: Policy Diffusion Among American States (2007), Karch compares five distinct health and welfare policy innovations to understand agenda-setting pressures leading to the diffusion of innovations.

Innovation Hosts

99

for how variations in state political, institutional, and sociodemographic attributes shape the general receptivity of states to innovation in the modern era. The chapter then repeats this process to model state receptivity to morality, governance, and regulatory policies in the modern era. This two-step process permits us to evaluate whether state receptivity to innovation is constant across policy types, or if certain policy types are especially virulent to groups of states with common political and institutional attributes. The American states display tremendous variation in political, institutional, cultural, and demographic characteristics. This study asks whether certain common characteristics make states more or less receptive to innovation. In keeping with the focus on decision-making dynamics developed in the preceding sections, this chapter connects the correlates of state receptivity to innovation to the processes of political decision making. Diffusion dynamics are shaped by the distinct ways that states interact and evaluate different forms of public policy. Variations in state receptivity to innovation emerge because states are differently equipped to engage in regulatory, morality, and governance policy making. Calculating State Receptivity to Innovation: The Innovation Index To identify those states that are the most responsive to adopting new policy innovations, this study constructed a composite innovation score for each state following a research design common in innovation and diffusion research (Walker 1969; Canon and Baum 1981; Nice 1994; Carter and LaPlant 1997). State receptivity scores were calculated using the same policy innovations data set gathered for the distributional analysis in the preceding chapters. For every policy innovation, a state was assigned a score ranging between 0.000 and 1.000 corresponding to the proportion of the adoption period that remained when the state adopted the reform (Canon and Baum 1981; Carter and LaPlant 1997).7 The first state to adopt a policy innovation received a score of 1.000, whereas those states 7

The innovation score is a ratio that measures the total time between the state’s adoption of a specific program and the last state’s adoption of that innovation, against the total time elapsed between the first and last state’s adoption of the innovation. It is derived using the following equation: IS = [(LY + 1) − SY]/[(LY + 1) − FY] where IS is the innovation score for the state, LY + 1 represents the year after the last state adopted the policy innovation, SY represents the year the state adopted the innovation, and FY represents the first year a state adopted that policy innovation.

100

Policy Diffusion Dynamics in America

that failed to adopt a given innovation within the observation period received a score of 0.000. The last state to adopt an innovation received a score slightly above 0.000, differentiating it from nonadopting states by assigning it a small positive value as a late adopter of innovation.8 For example, to calculate Washington’s innovation score for adopting the Amber Alert in 2002, coders used the date when the Amber Alert was first adopted in Oklahoma (1999), and the year Hawaii became the final adopter (2005). Researchers then subtracted the time elapsed between one year after the last adoption of the Amber Alert and Washington’s adoption of the Amber alert (2006–2002), and divided it by the time elapsed between the first observed adoption of the Amber Alert and a year after the last observed adoption (2006–1999). Converting this proportion into a percent gives Washington an innovation score of 0.571 for the program, whereas Oklahoma’s score for the Amber Alert is a 1.000, and Hawaii receives a low score of 0.143. This procedure was followed for each state and each innovation. The composite innovation score for each historical era and policy type was calculated by taking the mean of the innovation scores for all policies in each grouping for each state. This measure provides a rudimentary indicator of the speed with which a state adopts policy innovation in a historical period. States that are quicker to adopt innovation – those states that are the most receptive to policy innovation – have innovation index scores closer to 1.000, indicating that they generally adopt policies sooner than other states. Policy resistant or laggard states have scores closer to 0.000, indicating that they are slow to adopt innovations, and are less receptive to emerging policies. Ranking the States: Innovation Responsiveness 1960–2006 Which states have been the most responsive to emerging innovations in the modern era? Table 4.1 and the corresponding map (Figure 4.1) present state receptivity scores for all 50 states for the period from 1960 to 2006. There is a good deal of variation in state receptivity scores, 8

Walker’s initial calculation of state innovation scores did not differentiate between the last state to adopt a policy innovation and a state that had not adopted the innovation at all. Thus, both last adopters and nonadopters received innovation scores of 0.00. To differentiate between late and nonadopting states, I follow the recommendation of both Canon and Baum (1981) and Carter and LaPlant (1997), assigning the last state to adopt an innovation a score slightly higher than zero by adding a year to the last date of adoption, as explained in the formula above.

Innovation Hosts

101

table 4.1. State Receptivity to Innovation, 1960–2006 State

Index Rank

State

Index Rank

California Oregon Connecticut Florida New Jersey New York Washington Texas Nevada Delaware Minnesota Rhode Island Colorado North Carolina Massachusetts Michigan Louisiana

0.6228 0.5523 0.5398 0.5241 0.4719 0.4702 0.4540 0.4507 0.4372 0.4283 0.4283 0.4268 0.4205 0.4176 0.4168 0.4147 0.4116

Illinois Utah Ohio Oklahoma Arizona Kansas Maine Idaho New Mexico Alaska Hawaii Missouri Tennessee Virginia Montana Wisconsin Maryland

0.4116 0.4102 0.4082 0.4079 0.4022 0.4017 0.4013 0.3921 0.3914 0.3913 0.3905 0.3896 0.3895 0.3852 0.3837 0.3755 0.3716

State

Index Rank

Iowa New Hampshire Pennsylvania Nebraska Vermont North Dakota Indiana South Carolina Georgia Kentucky Alabama West Virginia Wyoming Arkansas Mississippi South Dakota

0.3592 0.3538 0.3496 0.3305 0.3219 0.3189 0.3180 0.3099 0.3096 0.3013 0.2945 0.2882 0.2877 0.2723 0.2716 0.2391

ranging from a minimum of 0.239 for South Dakota to a maximum of 0.623 for California. The average state innovation score is 0.390, with a standard deviation of 0.075. Larger states appear to be more responsive to innovation. Following California in the top ten are several states that have featured prominently in national politics in recent decades, including New York, Florida, and Texas. The grouping of low scores for the bottom 10 states is also interesting. Arkansas, Mississippi, and South Dakota all lag in the adoption of innovations, with receptivity scores under 0.275. As expected from research on geographical determinants of policy diffusion, there is a regional clustering in state receptivity rankings for the modern era. Eight of the bottom 10 states are from the South, whereas the majority of states in the top 10 are clustered in the Northeastern and Western United States. Minnesota leads the Great Lakes region as the tenth most receptive state; however, the remaining states in the upper Midwest are slow to adopt policy innovations. With its position as a regional economic leader, it is surprising to see Illinois ranked out of the top 15 in speed of policy adoption. It is likewise interesting to note that both Wisconsin and Pennsylvania appear sluggish in adopting policy

102

figure 4.1. Map of state receptivity to innovation, 1960–2006.

Innovation Hosts

103

innovations and are ranked in the bottom half for receptivity to innovation in the modern era. On the other hand, North Carolina and Louisiana stand out as having higher regional innovation scores, standing well away from their Southern peers in the upper third of states. Ranking the States: Innovation Responsiveness by Historical Era It is interesting to observe the stability of state receptivity to innovation through the twentieth century. Prior research on the stability of state innovativeness in public policy (Savage 1978) and tort laws (Canon and Baum 1981) identified considerable variation in policy receptivity to innovation across historical eras. Table 4.2 presents a comparison of state receptivity rankings for three periods: the early twentieth century (1900–1929), the middle twentieth century (1930–1959), and the late twentieth and early twenty-first century (1960–2006). This comparison not only permits us to evaluate the consistency of the receptivity index over time, but also allows us to evaluate changes in individual state receptivity to innovation across the twentieth century. Looking across state rankings presented in Table 4.2 confirms some variation in state receptivity to innovation across historical eras. In keeping with prior findings in state innovativeness across eras, the data presented in Table 4.2 reveal moderate correlation across historical periods – with a Pearson correlation coefficient of .600 for innovation scores in the early and mid-twentieth century, and a Pearson correlation coefficient of .471 for receptivity scores in the mid- and late-twentieth century.9 As Table 4.2 shows, a state’s receptivity to innovation is not necessarily fixed over time. A state with a high or low innovation score in the early twentieth century is not certain to display identical susceptibility to innovation in the future. Table 4.2 reveals a degree of stability at the top and bottom of the index rankings over time. States in the Northeast and West are consistently rapid adopters of policy innovation, whereas Southern states have historically lagged in adopting emerging innovations. Louisiana has remained a more responsive state to innovation than its Southern neighbors and is consistently ranked in the upper half of state receptivity. Perhaps more intriguing are those states that have experienced marked surges or declines in innovation responsiveness. Both Florida and Texas emerged as rapid responders to policy innovations in the last 9

These correlation coefficients are both significant at p < .001.

Policy Diffusion Dynamics in America

104

table 4.2. State Receptivity to Innovation by Historical Era

Early Twentieth Century (1900–1929)

Mid-Twentieth Century (1930–1959)

Late Twentieth to Early Twenty-First Century (1960–2006)

State

Index Score

Rank

Index Score

Rank

Index Score

Rank

Alabama Alaska∗ Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Hawaii∗ Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania

0.398 NA 0.542 0.383 0.682 0.520 0.520 0.389 0.382 0.443 NA 0.463 0.539 0.404 0.482 0.456 0.433 0.464 0.442 0.466 0.594 0.547 0.540 0.331 0.428 0.487 0.476 0.398 0.416 0.534 0.336 0.551 0.427 0.493 0.560 0.476 0.607 0.502

T40 NR 9 43 1 T13 T13 42 T44 T26 NR 24 11 39 19 25 T30 23 28 22 3 8 10 48 32 18 T20 T40 35 12 47 7 33 17 6 T20 2 16

0.355 NA 0.296 0.386 0.512 0.489 0.520 0.343 0.373 0.282 NA 0.371 0.450 0.434 0.237 0.313 0.338 0.431 0.362 0.417 0.578 0.524 0.441 0.265 0.241 0.335 0.307 0.238 0.388 0.592 0.413 0.669 0.379 0.379 0.423 0.314 0.454 0.540

29 NR 42 23 7 T8 6 32 26 43 NR 27 13 15 48 37 33 16 28 18 3 5 14 45 46 34 40 47 22 2 T19 1 T24 T24 17 36 12 4

0.295 0.391 0.402 0.272 0.623 0.421 0.540 0.428 0.524 0.310 0.391 0.392 0.412 0.318 0.359 0.402 0.301 0.412 0.401 0.372 0.417 0.415 0.428 0.272 0.390 0.384 0.331 0.437 0.354 0.472 0.391 0.470 0.418 0.319 0.408 0.408 0.552 0.350

45 T26 T22 T48 1 13 3 T10 4 T42 T26 25 T17 41 35 T22 44 T17 24 34 15 16 T10 T48 T29 32 38 9 36 5 T26 6 14 40 T20 T20 2 37

Innovation Hosts

105

table 4.2 (continued)

Early Twentieth Century (1900–1929)

Mid-Twentieth Century (1930–1959)

Late Twentieth to Early Twenty-First Century (1960–2006)

State

Index Score

Rank

Index Score

Rank

Index Score

Rank

Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming

0.418 0.376 0.414 0.414 0.408 0.514 0.443 0.433 0.576 0.435 0.565 0.382

34 46 T36 T36 38 15 T26 T30 4 29 5 T44

0.485 0.308 0.311 0.353 0.280 0.398 0.346 0.413 0.455 0.334 0.489 0.306

10 39 38 30 44 21 31 T19 11 35 T8 41

0.427 0.310 0.239 0.390 0.451 0.410 0.322 0.385 0.454 0.288 0.376 0.288

12 T42 50 T29 8 19 39 31 7 T46 33 T46

Innovation scores were not calculated for Alaska and Hawaii for these periods because of missing data for policies prior to statehood.

half of the twentieth century, while Pennsylvania and Wisconsin dropped dramatically. State Characteristics as Predictors of State Responsiveness to Innovation The variation in state responsiveness to innovation displayed in the receptivity indexes in Tables 4.1 and 4.2 presents an interesting question for understanding patterns of innovation and diffusion in American federalism. Why are some states much more responsive or receptive to policy innovation than others? The following section develops key political, institutional, ideological, demographic, and economic characteristics that shape state policy receptivity to innovation. The majority of these concepts have been suggested by prior research in state innovation and diffusion; including them in the analysis here allows us to measure state innovation responsiveness in the modern era against research on state innovativeness in previous eras. Furthermore, political indicators of institutional and ideological characteristics allow us to measure how variation across state decision-making capacities shapes the speed of public-policy adoption.

106

Policy Diffusion Dynamics in America

Modeling Determinants of State Responsiveness to Innovation The role of state political institutions in the innovation and diffusion process is of primary interest for students of public-policy diffusion. Policy adoptions are the result of formal political, institutional decision making, and the policy responsiveness of states should be highly correlated with the capacity of states to identify and make formal decisions regarding a policy problem. The speed of innovation should be enhanced both by the capacity of state legislatures to engage in problem identification and policy evaluation and by the responsiveness of elected officials to their constituents. Although formal institutions clearly play a central role in the diffusion of innovations, research in the broader policy process identifies a number of other factors shaping state receptivity to innovation. The speed of policy adoptions is also shaped by the involvement of policy entrepreneurs galvanizing public opinion, by campaign promises made during a heated election, or by technocratic analysis and recommendation from legislative committees and state bureaucracies. The majority of policy adoptions result from careful decision making by statehouse governments, but others are enacted by direct citizen initiative. Putting the decision-making process leading to policy adoption in the context of the larger policy process allows us to draw additional expectations for the role played by political, institutional, and ideological correlates of policy diffusion. This analysis identifies a broad range of social, political, and demographic factors that increase a state’s hazard for policy responsiveness. These factors measure variations in formal political institutions, nongovernmental actors, and differences in state culture and ideology. Political Institutional Characteristics and State Policy Receptivity One of the primary considerations when modeling state legislative outputs extends from the marked differences in full-time and part-time citizen legislatures across the American states. Although most state legislatures underwent significant modernization from 1960 through 1990 (Mooney 1995; Hamm and Moncreif 2004), students of state politics continue to observe considerable variation in the professionalism of state legislatures (Squire 1992; Mooney 1995; King 2000; Hamm and Moncreif 2004). Some states have lengthy annual legislative sessions; others meet biennially or have relatively short annual legislative sessions. States with more frequent and longer legislative sessions should prove more responsive

Innovation Hosts

107

to innovations, as they are more frequently poised to make legislative decisions. In addition to the frequency and duration of legislative sessions, the level of support provided to state representatives varies considerably across legislatures. Though the most professionalized states provide representatives with a full-time salary and a well-paid legislative support staff, states with citizen legislatures provide a small stipend to part-time representatives and provide little in the way of research or administrative support (Squire 1992; Mooney 1996; King 2000; Hamm and Moncreif 2004). Differences in legislative professionalism have clear implications for state receptivity to innovation. States with more professional legislatures should prove more likely to identify, evaluate, and rapidly implement emerging innovations than their less professional peers (Mossberger 2000). This advantage should be especially pronounced when addressing complex policy issues. As Carter and LaPlant (1997) note, professional state legislatures should prove “better prepared to handle increasingly complex policy issues” (22) whereas less professionalized states must rely on the program analysis and innovation of their peers. A state like California with a full-time legislature supported by a large professional staff should be able to identify and craft legislation more rapidly than a state like Montana, which has a part-time legislature and limited support staff. The concept of legislative professionalism is operationalized through an index developed by King (2000), who tracks changes in the professionalism of U.S. state legislatures from 1964 through 1994. Building upon Squire’s (1992) conceptualization of state legislative professionalism, the King index compares the professionalism of U.S. state legislatures – measured by levels of financial compensation, days in session, and expenditures for services and operations per legislature – against identical measures of professionalism in the U.S. Congress (Squire 1992; King 2000). States legislatures closely approximating the levels of professionalism in the U.S. Congress have index scores approaching 1.00 (indicating perfect imitation of Congressional professionalism), whereas citizen legislatures have scores approaching 0.00. If professionalism facilitates the early adoption of emerging innovations, we should expect a strong positive correlation between legislative professionalism and state receptivity to innovation. In addition to the professionalism of state legislatures, researchers in state policy diffusion have explored the link between political competition and state innovativeness (Walker 1969; Glick and Hays 1997; Karch 2006; Volden 2006). Researchers have advanced two central hypotheses

108

Policy Diffusion Dynamics in America

regarding partisan competition and state innovativeness. First, researchers have posited that a greater level of state partisan competition for elected office is expected to result in higher state receptivity to innovation, as states with higher levels of turnover between parties are likely to be more accountable and responsive to citizen concerns (Glick and Hays 1997; F. Berry and Berry 1999). A related argument offers the theory that frequent party turnover in statehouses and governors’ mansions results in higher innovativeness, as newly elected majorities look to pass an agenda ignored by their predecessors (Walker 1969).10 Political competition is operationalized using the folded Ranney index, a common indicator of state party competition created by converting the standard Ranney measure of party control into an indicator of party competition (Ranney 1976; Bibby and Holbrook 2004). The Ranney index calculates the degree of Republican or Democrat party control of state legislatures with an index between 0.00 and 1.00, indicating the proportion, duration, and frequency of party domination or control. Lower values indicate Republican domination of elections; moderate scores indicate a two-party system; whereas scores approaching 1.00 indicate strong Democratic control of state elections (Ranney 1976; Bibby and Holbrook 2004). The folded Ranney index modifies this measure of party control by indicating competition regardless of party control.11 Lower scores of 0.50 indicate absolute partisan domination, whereas higher scores approaching 1.00 indicate increasing party competition. If state party competition is linked to innovativeness, then there should be a positive correlation between the states with higher levels of party competition and state innovation scores. Nongovernmental Actors and State Policy Receptivity A number of state political institutions beyond state legislatures shape state receptivity to innovation. One important institutional variable often 10

11

There is some reason to expect that in select cases, policy will be negatively correlated with state political competition. As F. Berry and Berry have observed, politicians in closely contested elections may shy away from supporting policies that are “widely unpopular or at least sufficiently unpopular with some segment of the population to be deemed controversial” (1999, 182). The folded Ranney index modifies this measure of party control by indicating competition regardless of party control. Lower scores of 0.50 indicate absolute partisan domination, while higher scores approaching 1.00 indicate increasing party competition. The formula for the folded Ranney competition index is 1 − |0.5 − Ranney Index Score| (See Bibby and Holbrook 2004, 89).

Innovation Hosts

109

overlooked in the state policy diffusion literature is found in the tradition of the statutory initiative process, an additional policy-making institution that permits citizens in 24 states to directly implement legislation by approving qualified ballot initiatives through a popular vote (Cronin 1989; E. Gerber 1999; Tolbert 2002). Here the tradition of direct democracy is measured with a simple indicator variable for the state statutory initiative.12 Because the citizen initiative provides an additional legislative venue that is directly responsive to citizen concerns, we expect a positive relationship between state initiatives and state policy receptivity. A final measure of state political context includes a measure of the population of nongovernmental actors attempting to influence state policy outputs through organized interest-group politics. The prominent role that nongovernmental actors play in the official legislative process is well documented by researchers of American interest-group politics (Gray and Lowery 1996; Baumgartner and Leech 1998; Balla 2001). Interest groups generate policy ideas, provide research support and expert testimony, draft legislation, and mobilize support or pressure campaigns for innovation. Research in American social movements suggests that citizen advocacy groups should play an especially important role in the innovation and diffusion of public policies. Whereas for-profit economic interest groups often oppose reforms that disrupt the status quo, citizen advocacy groups are instrumental in galvanizing public support for populist innovation. Unfortunately, state interest-group influence has proved a difficult concept to measure for state politics researchers. Most research linking interest-group influence to the diffusion of innovations has focused on single case studies modeling how issue networks or professional association influences state innovation adoption (Mintrom 1997; Balla 2001). There has been some important research on the ecology of interest-group populations across the American states (Gray and Lowery 1996); however, these data provide only a general sense of the proportion and percentage of registered interest groups in the American states. Although measures of state-level interest-group populations exist, they are somewhat more limited than other measures of state political phenomenon – such

12

There is a good deal of variation in state requirements to qualify a statute for the popular vote. For example, state rules governing signature requirements impose different standards for the percentage and geographic distribution of signatures needed to qualify an initiative for the ballot (Cronin 1989; Boehmke 2005). Such variations clearly have implications for state receptivity to policy innovations in future research.

110

Policy Diffusion Dynamics in America

as legislative professionalism or political competition – which provide a more nuanced picture of political dynamics across states. To evaluate how differences in the ecology of state interest-group populations shape policy innovativeness, this research includes two measures of the size and makeup of state interest-group populations. As a first rough measure, the research includes a simple count variable measuring the number of registered interest groups by states. This variable evaluates whether states with larger interest-group populations are more receptive to innovation. To understand how interest-group dynamics shape state receptivity to innovation, this research also includes Gray and Lowery’s (1996) measure of the proportion of not-for-profit interest groups across states. This measure assesses whether increased presence of citizen advocacy groups as a proportion of a state’s interest-group population increases receptivity to innovation. State Political Ideology and Political Culture on State Policy Innovation Of course, the political characteristics of states extend well beyond the institutional capacity of states to evaluate and enact legislation. As anyone who has encountered the popular red state/blue state media analysis of U.S. presidential elections can attest, states vary considerably with regards to political culture and ideology. In the classic conception of the relationship between diffusion and ideology, liberal and progressive states are believed to be policy responsive, whereas conservative, traditional states are expected to oppose changes to the status quo. A number of researchers have documented and studied variations in state political ideology. Using a sample of public opinion data from 1977 to 1988, Erikson, Wright, and McIver (1993) constructed a static measure of the ideological liberalism and conservatism of state legislatures. They found a strong correlation between state political ideology and state policy liberalism – an index score calculated in part using Walker’s measure of policy innovativeness (Erikson, Wright, and McIver 1993). More recently, William Berry and his coauthors generated an alternative measure of both state institutional and citizen ideology to track changes in the conservatism and liberalism of the electorate and state elected officials from 1960 through 2004 (W. Berry, Ringquist, Fording, and Hanson 2007). Here they departed from traditional samples of state public opinion, instead evaluating ideological positions of states, institutions, and citizens by measuring interest-group ideology ratings for state elected officials and for differences between elected officials and incumbents in

Innovation Hosts

111

state political contests (W. Berry, Ringquist, Fording, and Hanson 1998). This approach has the advantages of capturing variation for state political ideology for both citizen and institutions from 1960 through 2004 (W. Berry, Ringquist, Fording, and Hanson 2007). The relationship between both institutional and citizen ideology on state receptivity to innovation is operationalized using the measures developed by W. Berry et al. (1998).13 Following the traditional conceptions of progressivism and conservatism, we expect more liberal states in both institutional and citizen ideology will be more receptive to innovations. Political Culture Many students of policy diffusion have included measures of political culture to evaluate how broader state political traditions shape policy outputs (Walker 1967; Tolbert 2002). For the purposes of understanding innovation and adoption, the relationship of state political culture to program innovation and initiative is most relevant. The most common conception of state political culture stems from the seminal work of Daniel Elazar (1984), who argued that political culture expresses “the particular pattern of orientation to political action in which each political system is embedded” (109). Elazar (1984) identified three predominant categories for political cultures across the states, moralist, individualist, and traditionalist. As Ira Sharkansky (1969) explains: Moralist cultures welcome new programs for the good of the community; in the Individualist culture, new programs would be initiated only if they could be described as political favors that would elicit favors in return from those who initiated the program; and the Traditionalist states would accept new programs only if they were necessary for the maintenance of the status quo (69).

These attributes of state political culture result in clear expectations for state policy responsiveness. Moralist states should be responsive to new policy innovations, whereas traditionalist states should be resistant to new innovation. Individualist states will fall between these two extremes. To assess how political culture shapes state receptivity to innovation, this research includes a dummy variable for traditionalist political culture. 13

Both measures have been used to measure ideology in innovation and diffusion research. Berry’s measure is employed here because it includes observations for Hawaii, Alaska, and Nevada, three states excluded from Erikson, Wright, and McIver’s earlier research (1993).

112

Policy Diffusion Dynamics in America

Traditionalist states should be less responsive to innovation than either moralistic or individualistic states. State Demographic and Economic Characteristics Students of policy diffusion have long acknowledged that state innovativeness is not simply shaped by political characteristics, but is also a function of state population characteristics and economic productivity. Some of the most robust findings in state innovation and diffusion literature explore the link between state wealth and policy innovativeness (Walker 1969). Wealthier states are expected to pioneer policy innovations, as they have the capacity to invest in policy analysis and can afford policy experimentation and policy failure. Following the common operationalization of state wealth in the diffusion of innovation literature, state wealth is measured here using per capita personal income by state (Walker 1969; Tolbert 2002; Volden 2006). Prior research in policy diffusion has also suggested a number of demographic characteristics that shape state innovativeness. State population density and the percent of the population living in urban centers have both been linked to higher innovation scores (Walker 1969; Carter and LaPlant 1997). A measure of the percentage of the population living in urban centers is included to evaluate this relationship. In keeping with previous research and theory, we expect a positive relationship between urban density and state innovativeness. Finally, following the well-documented connection between education and economic and social innovation (Walker 1969; Florida 2003), a simple measure (the percentage of the state population that have college diplomas) is included to assess whether more educated states are more receptive to emerging policy innovations. Analysis of State Responsiveness Scores The dependent variables for the analysis are the state receptivity scores for the period from 1960 through 2006 presented in Table 4.1. The independent variables are operationalized as the average values for each independent variable for available data from 1960 through 2000.14 Because the

14

Averaging independent variables across a time period is a departure from past research in state innovativeness. Prior research models selected variables from the median year of adoptions in the sample. Here, scores were averaged to produce a more accurate measure of independent variables for the time period. For example, taking the median

Innovation Hosts

113

dependent variable is represented with a continuous variable bounded between 0 and 1, ordinary least squares (OLS) regression was used to model the relationship between state innovation responsiveness and key institutional, ideological, economic, and demographic variables. As previous diffusion researchers have observed, estimating state innovation scores with cross-sectional data for such a broad time period introduces some measurement errors, as the responsiveness for early adopting states is modeled using data averaged for a later time period, even though the responsiveness of late adopting states is estimated using data from a much earlier time (Carter and LaPlant 1997; F. Berry and Berry 1999). To evaluate how such variability in the data shaped the statistical model, additional OLS regression models were estimated holding constant the dependent variables and using independent variable data gathered for each decade from 1960 through 1990. These models were then compared to the model estimated with the average values across all periods. Two general trends emerged from this reliability analysis. First, with a few exceptions, the coefficients and significance levels of the variables remained stable. Second, a comparison of the Adjusted R2 and F-Statistic of the five models suggested that the model’s fit became more robust with inclusion of data from later decades. Table 4.3 represents the results of the OLS regression analysis with robust regression coefficients.15 Given the large number of independent and control variables and the relatively small sample size, this study estimated both a base model, including only major variables suggested from prior research in policy diffusion, as well as a full model, including measures of educational attainment, traditional political culture, and not-for-profit interest-group populations. As Table 4.3 demonstrates, the baseline and full models produce similar findings for the influence of state characteristics on general state receptivity to innovation. The first set of robust regression coefficients in Table 4.3 present the link between state institutional political variables and state public-policy

15

year approach of political competition would have resulted in estimating the effects of political competition on the entire model by including a measure of partisan competition for the 1981–1988 period. A better estimation of state competitiveness for the entire time period can be gained by a broader measure of legislative turnover, generated by averaging all competition scores for the modern era. Standard regression diagnostics identified several observations exerting higher degrees of leverage and influence on the model. This research estimated several models, excluding each influential observation with a censoring variable. Rather than exclude a number of states from the statistical model, this research employed a robust regression analysis to control for influential data.

114

Policy Diffusion Dynamics in America

table 4.3. State Predictors of Innovation Receptivity, 1960–2006 Independent Variable Political Institutional Variables State Legislative Professionalism Political Competition Direct Statutory Initiative

Partial Model −0.020 (0.063) 0.345∗∗∗ (0.113) −0.003 (0.014)

% Not-For-Profit Interest Groups Ideology and Culture Citizen Ideology Institutional Ideology

−0.001 (0.001) 0.001 (0.001)

Traditional Political Culture Economic Demographic Variables Per Capita Personal Income % Urban Population

0.000 (0.000) 0.003∗∗∗ (0.001)

% College Graduates Constant

−0.117 (0.087)

Full Model 0.051 (0.066) 0.258∗∗ (0.114) 0.013 (0.015) 0.053 (0.124) −0.001 (0.001) 0.002∗ (0.001) −0.070∗∗∗ (0.019) 0.000 (0.000) 0.004∗∗∗ (0.001) −0.001 (0.004) −0.203 (0.087)

N = 50 ∗ p < .10. ∗∗ p < .05. ∗∗∗ p < .01.

responsiveness. As Table 4.3 indicates, there is a strong positive association between the average levels of state partisan competition between 1960 and 2006 and state receptivity to innovation. States with higher degrees of party competition are quicker to adopt emerging policy innovations. More importantly, partisan competition has a significant impact on state policy responsiveness, as an increase in one unit on the political competition index results in an increase of 0.345 in the state innovation score. The effect of moving from the lowest to the highest levels of competition nearly doubles state responsiveness to innovation. Other measures of state institutional political characteristics do not have a statistically significant impact on state innovation receptivity. Most interestingly, the empirical model cannot confirm the expected

Innovation Hosts

115

positive relationship between state legislative professionalism and state responsiveness to innovation, suggesting that the importance of legislative professionalism in identifying, evaluating, and implementing new policy innovations may be overstated in the innovation and diffusion literature. Likewise, the expected positive association between both the direct statutory initiative process and the percentage of registered interest groups that are not-for-profit cannot be confirmed at standard levels of statistical significance.16 In terms of political institutional variables, electoral connection emerges as the single strongest predictor of state receptivity to innovation. Measuring the influence of citizen ideology and state political culture on state policy receptivity permits us to confirm weak significant relationships between ideology, culture, and state responsiveness to policy innovations. The full model allows us to confirm that state political culture – measured here by including a dummy variable for state traditional political culture – has a weak negative relationship with policy innovativeness. This makes intuitive sense, as a prominent trait of states that have traditional political cultures is a resistance to innovation that alters the status quo.17 Although this relationship is statistically significant, it is important to note that the effect is extremely small, as the impact of moving from the control group of states with moralistic and individualist political cultures to the reference group of states with traditionalist political cultures results in only a −0.07 decline in state innovation receptivity scores. Controlling for state political culture in the full model reveals a second weak correlation between state institutional political ideology and state innovation receptivity. More liberal state governments are more likely to be more responsive to new policy innovations, and a 10-point increase in state institutional liberalism results in a commensurate 0.02 increase in state responsiveness to innovation. In absolute terms, the effect of moving from a state with the most conservative institutional political

16

17

Interest-group influence was also measured using a count variable representing the number of registered interest groups for each state. This measure did not change the model and was dropped, as it provided a less nuanced picture of the different makeup of state interest-group populations. This research also estimated the model alternatively, using indicator variables for state moralist and individualist political cultures. Including the indicator variable for moralist political cultures produced a statistically significant and strong finding, as moralist states had higher innovation adoption scores. As expected, the model could not confirm a relationship between traditionalist political culture and state policy receptivity.

116

Policy Diffusion Dynamics in America

ideology to the most liberal state results in a modest 0.12 increase in a state’s responsiveness to innovation. The final grouping of control variables allows us to measure the effects of state economic and demographic characteristics on modern receptivity to innovation. Here the model confirms the well-documented positive association between urban density and state innovativeness (Walker 1969). However, the model cannot confirm the expected positive relationship between either state wealth, as measured by per capita personal income, or state population education, as measured by the percentage of college graduates inhabiting the state. Discussion and Implications: Policy Receptivity as a General Trait Why are some states much more responsive to public policy than others? Drawing upon prior research in policy innovation and diffusion, this analysis profiled those state characteristics expected to shape state responsiveness to innovation from 1960 through 2006. The research tested several competing hypotheses regarding the influence of political, ideological, cultural, economic, and demographic predictors of state policy responsiveness. The model confirmed a relatively weak relationship between ideology and a strong relationship to state political competition.18 In keeping with prior research on policy diffusion, liberal urban states with higher levels of political competition are more responsive to innovation than their traditionalist conservative peers. Remarkably, the model failed to confirm research hypotheses regarding the influence of the remaining institutional or ideological predictors of state innovativeness. Neither of the measures of nongovernmental activity in state politics proved to be a significant predictor of state responsiveness. The initiative process may permit citizens to implement innovations without involving state legislatures; however, this additional institutional mechanism does not lead to earlier adoptions of policies. Although diffusion researchers have provided strong empirical and theoretical reasons 18

That this approach failed to replicate known correlates of innovation in a multipleregression model using modern data is not entirely surprising. Early researchers in state policy innovativeness estimated the correlations between state political innovativeness and state characteristics with zero-order correlation coefficients (Walker 1969; Nice 1994). The failure to replicate here is not necessarily an indicator of the instability of independent variables or the modern measure of innovativeness, but more likely a result of the covariance on independent variables included in the multiple regression model.

Innovation Hosts

117

to expect that interest groups play a strong role in the diffusion of innovations, this model found no link between the proportion of not-forprofit interest groups and state receptivity to innovation.19 The model also failed to confirm the expected positive relationship between state legislative professionalism and public-policy diffusion. Professional state legislatures should hold a significant advantage in the early identification and adoption of public policies. No such association emerged. An interesting theoretical explanation for the failure to confirm hypotheses regarding the link between institutional political variables and state innovativeness can be accounted for by revisiting the discussion of the decision-making processes underlying policy diffusion highlighted in earlier chapters of this book. As suggested in the distributional analysis of the diffusion of policies by policy type, evidence suggests that patterns of policy diffusion are inconsistent with any single model of incremental policy change. Instead, the policy dynamics displayed in Chapters 2 and 3 are the product of a mixed model of political decision making, one driven at times by careful technocratic program evaluation and emulation by state legislatures, and at other times by responsive policy mimicking propelled by elevated issue salience and mass public involvement in the policy process. Including policies driven by each of these processes in the aggregate measure of state receptivity masks the links between state receptivity and legislative professionalism, public opinion, or the initiative process. Each of these political institutional variables may be highly correlated with policies resulting from one set of governmental decision making and entirely uncorrelated with another. For example, state legislative professionalism should facilitate the early adoption of technocratic, complex, and lowsalience regulatory policies; however, professionalism is less likely to play an important role in the adoption of technically unsophisticated populist morality policies. Citizen ideology may prove to be a strong predictor of high-salience morality policies, but a poor predictor of licensing policies requiring the regulation of a poorly-organized target population. Likewise, theory suggests that the proportion of citizen-advocacy groups in a state may be strongly correlated to morality policy receptivity, as morality policy is largely determined by mass public participation in the policy

19

This research also estimated a model using a simple count of the number of registered interest groups to measure interest-group environment. This specification of the model also failed to establish a relationship between the size of state interest-group population and innovation diffusion.

118

Policy Diffusion Dynamics in America

process. Not-for-profit interest groups are less likely to be involved in economic regulatory policy development, which tends to be shaped by professional interest groups employing traditional lobbying strategies. Finally, the direct citizen initiative process should be a strong predictor of citizen efforts to limit the reach of government, but less strongly correlated with other policy types. As with the analysis of the distributional properties of all policies taken together, the measure of state general policy responsiveness provided here is insensitive to important differences of political processes that emerge across policy regimes. The next section measures state receptivity to morality, governance, and regulatory policy. Following the approach taken in the preceding analysis, it ranks state responsiveness to each of these three types of public policy. It then generates a statistical model to evaluate general predictors of state resistance and receptivity to state regulatory, morality, and governance policies, holding constant the independent variables across the three models. This section compares these models to assess how specific institutional and ideological measures associated with political decision making facilitate the process of problem identification, evaluation, and adoption implied in innovation and diffusion research. If policy diffusion represents a mixed model of political decision making and policy change, the correlates of state responsiveness should differ considerably across these three policy types, supporting the findings in the preceding chapters that policy innovation and diffusion display considerable dynamics that cannot be explained through a single model of policy identification, evaluation, and emulation. More importantly, this analysis tests the expectation that each policy type is correlated with a distinct form of political decision making. Regulatory policy – marked by low levels of salience and high issue complexity – should be strongly correlated with state legislative professionalism. Morality policy – characterized by high emotional appeal and low issue complexity – is shaped by political competition, citizen interest-group involvement, and state ideology. Governance policy – a policy form that explicitly regulates the behavior of government – is almost exclusively related to state initiatives. Ranking the States: Innovation Responsiveness by Policy Type, 1960–2006 Table 4.4 and the corresponding maps (Figures 4.2, 4.3, and 4.4) present composite receptivity scores for morality, state regulatory, and governance policies from 1960 to 2006. Rankings of state receptivity to each

Innovation Hosts

119

table 4.4. State Receptivity to Innovation by Policy Type Morality Policy 1960–2006

Regulatory Policy 1960–2006

Governance Policy 1960–2006

State

Index Score

Rank

Index Score

Rank

Index Score

Alabama Alaska Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island

0.342 0.400 0.425 0.324 0.676 0.478 0.493 0.451 0.531 0.347 0.340 0.382 0.433 0.399 0.362 0.405 0.391 0.468 0.406 0.455 0.297 0.316 0.444 0.259 0.368 0.366 0.349 0.514 0.305 0.444 0.466 0.362 0.433 0.318 0.345 0.416 0.563 0.347 0.412

38 24 17 41 1 6 5 11 3 T35 39 28 T15 25 T32 22 26 7 21 10 46 43 T13 49 29 T30 34 4 45 T13 8 T32 T15 42 37 19 2 T35 20

0.200 0.433 0.356 0.270 0.548 0.397 0.731 0.530 0.457 0.254 0.448 0.399 0.396 0.289 0.498 0.490 0.283 0.395 0.458 0.400 0.567 0.418 0.558 0.299 0.383 0.334 0.298 0.264 0.386 0.609 0.418 0.711 0.575 0.378 0.473 0.413 0.631 0.330 0.485

49 21 36 46 8 28 1 9 T17 48 19 27 29 43 11 12 44 30 16 26 6 T23 7 41 33 39 42 47 31 4 T23 2 5 T34 15 25 3 40 14

0.277 0.544 0.391 0.246 0.474 0.634 0.246 0.263 0.485 0.177 0.250 0.613 0.146 0.205 0.246 0.238 0.165 0.468 0.330 0.077 0.470 0.500 0.115 0.279 0.541 0.388 0.134 0.504 0.192 0.176 0.200 0.092 0.276 0.100 0.364 0.485 0.398 0.238 0.254

Rank 25 3 16 T30 12 1 T30 27 T9 40 29 2 44 37 T30 T33 42 14 22 50 13 T6 T47 24 4 17 45 5 39 41 38 49 26 49 18 T9 15 T33 28 (continued)

Policy Diffusion Dynamics in America

120 table 4.4 (continued) Morality Policy 1960–2006

Regulatory Policy 1960–2006

Governance Policy 1960–2006

State

Index Score

Rank

Index Score

Rank

Index Score

Rank

South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming

0.234 0.270 0.383 0.423 0.402 0.312 0.445 0.458 0.328 0.366 0.279

50 48 27 18 23 44 12 9 40 T30 47

0.457 0.189 0.438 0.525 0.276 0.385 0.429 0.378 0.339 0.487 0.350

T17 50 20 10 45 32 22 T34 38 13 37

0.496 0.332 0.500 0.477 0.356 0.115 0.154 0.327 0.208 0.336 0.225

8 21 T6 11 19 T47 43 23 36 20 35

policy type are organized in Appendix D. There is considerable variability across the state responsiveness to each policy type, indicated by the extremely weak correlation between the index rankings for regulatory, morality, and governance policy. A Pearson’s product correlation of .383 indicates a weak correlation between morality and regulatory policies in the modern era,20 whereas the anemic correlation between governance policies and either morality policy (Pearson’s correlation coefficient of .192) or regulatory policies (Pearson’s correlation coefficients of .008) cannot be confirmed at standard levels of significance. A review of the leaders and laggards for each policy type establishes that there is a good deal of movement in the rankings for state receptivity to regulatory, morality, and governance policy. No state is consistently ranked in the top or bottom 10 for state receptivity for all three policy types, although California and Oregon remain consistently responsive to all policies, and Georgia, Nebraska, North Dakota, and Vermont are perennial laggards. There are some remarkable changes in state receptivity to innovation. Nevada, a state which is highly receptive to morality and governance policies, ranks in the bottom five states for responsiveness to regulatory policy. Massachusetts is surprisingly resistant to morality policies but 20

The correlation between morality and regulatory policies is significant at the p < .05 level.

121

figure 4.2. Map of state receptivity to morality policy innovation, 1960–2006.

122

figure 4.3. Map of state receptivity to regulatory policy innovation, 1960–2006.

123

figure 4.4. Map of state receptivity to governance policy innovation, 1960–2006.

124

Policy Diffusion Dynamics in America

appears receptive to regulatory and governance policy. Finally, New York – traditionally considered an innovation leader – is ranked in the bottom tier of states for receptivity to governance and morality policies.21 Determinants of Morality, Procedural, and Governance Policy Responsiveness What causes the considerable variation in state responsiveness to morality, state regulatory, and governance policies identified in Table 4.4? With a few important exceptions (Mooney and Lee 1995; Mooney and Lee 1999; Tolbert 2000), research in the diffusion of innovations has done little to connect research in policy typologies to the diffusion of innovation. Instead, research on state innovativeness as a general trait of American states would lead us to hypothesize that the same characteristics that predict general state responsiveness to innovation should also prove to be stable predictors of state receptivity to each of the three key subsets of policies. If current theory is correct, then wealthy urban states with professional legislatures and high levels of political competition should prove systematically more receptive to morality, regulatory, and governance policies. Furthermore, these states should be consistently more receptive to innovation than poorer agricultural states with little political competition and citizen legislatures. There is ample reason to question the assumption that the predictors of state responsiveness are stable across policy types. As discussed in Chapter 3, there are clear, generalizable differences between classes of morality, governance, and regulatory policy. If policy causes politics (Lowi 1972), it is reasonable to expect that these different forms of policies will interact differently with political and ideological characteristics of states. The traits that make a state responsive to technically sophisticated, low-salience regulatory policies are fundamentally different from 21

It is also interesting to observe the instability of geography as a predictor of innovation responsiveness. There is some geographic clustering in each of the index rankings; however, it is not as pronounced as in the general measure of policy responsiveness. The Northeast makes up a healthy leadership in state regulatory policies, but lags in the adoption of governance policies and morality policies. The West Coast remains innovative in morality policies, but is less pioneering than expected in either regulatory or governance policies. The Mountain West and Midwest produce leaders of governance policies, but lag in the adoption of regulatory or morality policies. In spite of these vague trends, the strength of geographic clustering or of national prominence to explain variation in innovation and adoption is less pronounced across historical eras.

Innovation Hosts

125

those characteristics that make a state susceptible to high-salience morality policy. The following section revisits the institutional, ideological, demographic, and economic predictors of state policy responsiveness to regulatory, morality, and governance policy. It departs from traditional diffusion theory by arguing that the correlates of state innovativeness will differ across each of the three policy types. It theorizes that state characteristics associated with different channels of information processing and state decision making are associated with differences in state receptivity to regulatory, governance, and morality policies. Political Institutional Correlates of Policy Responsiveness by Policy Types The influence of state legislative professionalism is of central interest to evaluating the correlates of state responsiveness to morality, regulatory, and governance policy. As a number of researchers have pointed out, state legislative professionalism is an important indicator of a state’s capacity to identify, evaluate, and implement new policy ideas (Walker 1969; Carter and LaPlant 1997; Mossberger 2000; Volden 2006). Because professional legislatures provide representatives with much policy research and analysis support, professionalized state legislatures will hold a relative advantage in the analysis of complex or technically sophisticated public policies. In this sense, legislative professionalism is strongly related to the process of program identification, evaluation, and emulation central to most theories of public-policy diffusion. There is good reason to expect that state responsiveness to regulatory policy will be strongly influenced by legislative professionalism. As discussed in Chapter 3, state policies addressing economic, environmental, and professional regulations are characterized by low levels of public salience and higher levels of issue complexity. States with more professional legislatures benefit from larger staff and professional representatives who are able to engage in complex problem analysis and policy design. If such a relationship holds, then we expect that state legislative professionalism will be a strong and positive predictor of state regulatory policy responsiveness. There is less reason to expect a connection between state legislative professionalism and morality policy. The significant advantages of professional legislatures – the ability to identify and perform relevant policy

126

Policy Diffusion Dynamics in America

analysis on complex political issues – are relatively unimportant for evaluating policy arguments surrounding first principles over right and wrong. Indeed, it is possible that states with professional legislatures will be less responsive to morality policy, as they are likely to identify broader difficulties of constitutionality, program cost, or other secondary considerations when evaluating morality policies. Because the advantages of state legislative professionalism are unrelated to the attributes of morality policy innovation and adoption, we expect a weak negative relationship between state legislative professionalism and state responsiveness to morality policy. There is likewise little reason to anticipate a positive relationship between governance policy and state legislative professionalism. Governance policy represents an effort by citizens to regulate the behavior of elected state representatives and delineate the procedures of government institutions. Even the least cynical students of American state politics have observed that legislatures are reluctant to limit the scope of their own power. Though governance policy shares attributes of issue complexity and moral arguments over first principles, its uneasy relationship with state governments suggests there will be little connection between governance policy and state legislative professionalism. A second key political institutional variable is the level of political competition across states. As the previous analysis of all innovations demonstrated, political competition proved to be the best predictor of state policy responsiveness from 1960 through 2006, and there is good reason to expect that political competition will be a strong predictor across policy types. Party turnover introduces new ideas into statehouse governments and disrupts previously stable relationships between interest groups and state legislative committees. This leads to policy innovation as newly elected majorities pass restrictive regulatory policy, change government procedures, or pass partisan morality policy. There is some reason to expect that political competition will be a stronger predictor of morality policy than either regulatory or governance policy. In recent years, a number of pundits have observed the importance that morality policy plays in political campaigns, as candidates look to mobilize their voting base by sponsoring symbolic high-salience morality issues. As the recent state battles over abortion, gay marriage, and anti-drug legislation suggest, contested political elections may lead to considerable innovation in the domain of morality policy. In keeping with this argument, we expect that state responsiveness to morality

Innovation Hosts

127

policy is positively and strongly correlated to levels of state political competition. Nongovernmental Actors in the Policy Process Evaluating the link between the statutory initiative process and governance, morality, and regulatory policy permits us to refine and revise expectations for how the institution of direct democracy shapes state receptivity to public policies. First, we expect a strong and statistically significant relationship between the early adoption of governance policy and the initiative process. Absent constitutional amendments, the state initiative process is often the only mechanism available for citizens to regulate the power of elected representatives or alter the rules of their political institutions. The link between the initiative process and regulatory and morality policy innovation should be much less pronounced. There are instances in both cases where the initiative process has been used to enact economic or social regulatory reform; however, these are relatively rare events. Most state innovation in regulatory or morality policy is expected to result from traditional bill making in state legislatures. We therefore expect no relationship to emerge between the initiative and state responsiveness to governance and morality policy. Exploring state responsiveness to these three types of public policy further allows us to refine our understanding of interest-group involvement in state policy innovation and adoption. No comprehensive measure of the involvement of nongovernmental actors in the adoption of policy innovations has been developed; the measure used here is an indicator of what percentage of a state’s interest-group populations are not-for-profit (Gray and Lowery 1996), for these interest groups are likely to reflect citizen involvement in the policy process.22 General diffusion theory suggests that interest groups are the key carriers of policy reform. However, this hypothesis fails to distinguish between the very different interest groups advocating for economic, morality, or governance policy reform. A careful reading of literature on the politics of regulatory policy development allows us to draw some clear expectations 22

This variable was preferred over a simple count measure of the number of registered interest groups, as it allows a cursory insight into how differences in the makeup of a state’s interest-group population shapes receptivity to innovation.

128

Policy Diffusion Dynamics in America

for how interest-group activity should influence the policy process. The size of a state’s not-for-profit interest-group population should be especially correlated to state morality policy innovation, as theory and research suggests that citizen interest-group involvement in the policy process is most pronounced in battles over social regulatory policy. This is not to say that interest groups are not influential in regulatory or governance policy reform, but rather that the specific way that interest-group populations have been operationalized in this research will provide a measure that is closely associated with only one of the three forms of regulatory policy. The percent of not-for-profit interest groups is expected to be strongly associated with morality policy responsiveness, as research on social movements and interest has repeatedly documented the role of grass roots interest-group politics in policy reform. Political Ideology and Culture Finally, it is interesting to observe how state political ideology shapes responsiveness to each of these three policy types. The traditional conception of political ideology suggests that more liberal progressive states will be more receptive to policy innovation, and conservative traditionalist states will remain entrenched and resistant to policy change. However, the diffusion of these different forms of policy permits us to refine these expectations. After all, the expectation that liberal states are universally more innovative poorly matches recent policy experiments in a number of conservative policy innovations. For state regulatory policy, it is reasonable to expect that the traditional conceptualization of the relationship between state political ideology and policy innovation should hold. Conservative and traditionalist states should prove reluctant to impose government regulation on business, preferring instead market-based solutions or simply the status quo. Conversely, liberal states should be highly responsive to regulatory policy innovation, as they are more receptive to government intervention in private enterprise. It is more difficult to determine a clear direction of influence for state ideology on patterns of adoption of morality policy. Both theory and the expected link between state partisan competitions indicate that public ideology should be critically linked to state responsiveness to morality policy, but it is theoretically unclear whether liberal or conservative states are less likely to respond rapidly to morality policy. For example, the range of morality policies moving across states in recent years – from policies

Innovation Hosts

129

dealing with stem cell research, to prohibitions of gay marriage and the legalization of gay civil unions, to the use of medical marijuana and support for the death penalty – represents quite different ideological and cultural preferences. Absent a clear expectation for influence on morality policy, state ideology is included as a control variable with no clear expectations for influence. The relationship between state governance policy and citizen ideology is also a complicated one. The tradition of direct democracy emerged from the populist and progressive movements at the turn of the century, and it is likely that more progressive states are far more likely to regulate the procedures of government. Yet in recent years, the diffusion of governance policy has taken a decidedly conservative turn, as citizens have passed initiatives limiting the ability of government to raise taxes or engage in regulatory takings of private land. Because the measure of governance policy is here operationalized as policies adopted from 1960 through 2006, we expect that conservative states will be more responsive to governance policy than their liberal counterparts. Finally, to remain consistent across models, this research also includes measures of state wealth, the percentage of the population that is urban, and education as control variables. Here the expectations are unchanged from the prior discussion. Educated, wealthy urban states should be more responsive to policy innovation. Analysis of State Responsiveness to Policies by Policy Type To estimate state responsiveness to regulatory, morality, and governance policy in the modern era, this section followed a procedure identical to the one used to model all policies presented in Table 4.3. With the exception of the measure of not-for-profit interest groups, values for the independent variables represent the average value for each variable from 1960 through 2006. The dependent variable in each model is the state index score for regulatory, morality, and governance policy. It is important to note that sorting by both policy type and historical era limited the number of innovations in each innovation index. The measure of morality policy was calculated with 20 policies in the modern era, whereas regulatory policy was measured by averaging 15 policies, and the measure for governance policy was calculated with only 5 policies.23 23

To identify whether the limited sample used to calculate innovation scores for governance and regulatory policy had a significant impact on the coefficients in the model, I also

130

Policy Diffusion Dynamics in America

State receptivity to regulatory, morality, and governance policy was estimated using ordinary least squares (OLS) regression analysis. Because of the limited number of observations and the large number of independent variables of interest, I estimated both a baseline and full model for each policy type. To control for the influential effects of outliers, the models for regulatory and morality policy were estimated each using OLS regression with robust regression coefficients. The values reported for the governance policy models are unstandardized regression coefficients.24 Tables 4.5 (baseline) and 4.6 (expanded) report the regression coefficients for state political, institutional, ideological, cultural, economic, and demographic predictors of state responsiveness to regulatory, morality, and governance policy.25 With the important exception of regulatory policy, coefficient values remain stable between two models. As indicated in the first column, legislative professionalism is positively correlated with state responsiveness to regulatory policy, although this effect cannot be confirmed in the larger model. This provides some support for the argument that states with professionalized state legislatures are more likely to adopt regulatory policies, whereas citizen legislatures tend to lag in regulatory policy adoption. The model also reveals an unexpected negative relationship between morality policy and direct democracy, suggesting that states with citizen initiatives are marginally less responsive to regulatory policy innovations. None of the control variables appear to be strong predictors of state regulatory policy responsiveness. Although there was reason to expect a negative relationship between state ideology and regulatory policy responsiveness; this effect could not be confirmed. As expected, the third column indicates a strong positive correlation between governance policy responsiveness and the initiative process. The

24

25

estimated an expanded model that included measures of state innovativeness constructed from a broader set of policies spanning the twentieth century, and using data taken from 1960 to capture the independent variables in the baseline model. The coefficient values were similar although significance levels improved for key observational variables. In short, improving the measure by gathering more data would improve the model, but not substantively change the direction of relationships of the coefficients. Regression diagnostics found no problematic outliers in the governance policy model. There was therefore no need to control for the leverage of outliers through robust regression. The robust regression for state morality policy receptivity drops California from the analysis in the baseline model. When compared to standard OLS regression with all cases included, the significance levels for all variables are the same save for state legislative professionalism, which is no longer statistically significant.

Innovation Hosts

131

table 4.5. Predictors of State Receptivity to Regulatory, Morality, and Governance Policy,+ 1960–2006 (Baseline Model) Independent Variable Political Institutional Variables State Legislative Professionalism Political Competition Direct Statutory Initiative Ideology and Culture Citizen Ideology Institutional Ideology Economic Demographic Variables Per Capita Personal Income % Urban Population Constant

∗

Regulatory Policy

Morality Policy

Governance Policy

0.242∗ (0.143) 0.293 (0.254) −0.075∗∗ (0.031)

−0.266∗∗∗ (0.084) 0.464∗∗∗ (0.125) 0.002 (0.015)

0.288 (0.185) (0.037) 0.328 0.121∗∗∗ (0.040)

0.001 (0.002) 0.000 (0.002)

−0.004∗∗∗ (0.001) 0.003∗∗∗ (0.001)

−0.004 (0.003) 0.002 (0.002)

0.000 (0.000) 0.001 (0.002) −0.125 (0.195)

0.000 (0.000) 0.003∗∗∗ (0.001) −0.147 (0.096)

0.000 (0.000) 0.002 (0.002) 0.299 (0.252)

F = 4.77∗∗∗ N = 50

F = 7.38∗∗∗ N = 49

Adj R2 = 0.2355 N = 50

p < .10. ∗∗ p < .05. ∗∗∗ p < .01. Policy values are estimated without robust regression coefficients.

+ Governance

effect of moving from the control group to states with direct democracy results in a full increase of 0.121 in state receptivity to governance policy. The model could not confirm any relationship between interest-group activity and governance policy, suggesting that the strength of not-forprofit, interest-group representation in a state is unrelated to governance policy receptivity. Likewise, neither conservative political ideology nor any of the other measures included in the model have a significant impact on morality policy adoption. The coefficients for the predictors of state morality policy in the second column reveal some intriguing differences between the correlates of morality policy and both state regulatory and governance policy responsiveness. First, both the baseline and full model indicate a statistically significant and negative relationship between morality policy innovation and state legislative professionalism. The very institutional attribute that

132

Policy Diffusion Dynamics in America

table 4.6. Predictors of State Receptivity to Regulatory, Morality, and Governance Policy,+ 1960–2006 (Full Model) Independent Variable

Regulatory Policy

Political Institutional Variables State Legislative Professionalism

0.209 (0.151) Political Competition 0.423 (0.268) Direct Statutory Initiative −0.051 (0.035) % Not-For-Profit Interest Groups −0.158 (0.295)

Ideology and Culture Citizen Ideology Institutional Ideology Traditional Political Culture Economic Demographic Variables Per Capita Personal Income % Urban Population % College Graduates Constant

Morality Policy −0.292∗∗∗ (−0.057) 0.523∗∗∗ (0.101) 0.005 (0.013) 0.322∗∗∗ (0.111)

Governance Policy 0.296 (0.201) 0.035 (0.358) 0.118∗∗ (0.047) 0.095 (0.394)

0.003 (0.003) −0.001 (0.002) 0.074 (0.058)

−0.002∗ (0.001) 0.002∗∗∗ (0.001) 0.069∗∗∗ (0.022)

−0.004 (0.004) 0.002 (0.003) 0.010 (0.078)

0.000 (0.000) 0.000 (0.002) 0.008 (0.008) −0.364 (0.264)

0.000 (0.000) 0.003∗∗∗ (0.001) 0.001 (0.003) −0.405∗∗∗ (0.100)

0.000 (0.000) 0.002 (0.002) 0.001 (0.011) 0.264 (0.352)

F = 3.64.∗∗∗ F = 11.52.∗∗∗ Adj R2 = 0.1787 N = 50 N = 50 ∗ +

p < .10. ∗∗ p < .05. ∗∗∗ p < .01. N = 50 Governance Policy values are estimated without robust regression coefficients.

makes states more responsive to regulatory policy appears to insulate them from morality policy innovation. As expected, the relationship between morality policy and state partisan competition is both pronounced and highly significant. State responsiveness to morality policy innovation increases sharply as state elections become more contested, indicating that state political competition results in increasing activity in the innovation and early sponsorship of morality policy. The model also confirms the importance of nongovernmental

Innovation Hosts

133

actors in state receptivity to morality policy. States with greater representation of not-for-profit interest groups are decidedly more responsive to morality policy innovation. The model allows us to clarify the relationship between state ideology and morality policy responsiveness. State citizen ideology is negatively correlated with morality policy innovation, indicating that more conservative states are early adopters of state morality policies. This finding is consistent with the connection between state traditional political culture and morality policy. More puzzling is the relationship between state institutional ideology and morality policy. Although liberal citizen populations are negatively correlated with responsiveness to morality policy, more liberal state political government institutions are innovative. Both effects are small; however, the findings suggest that elected decision makers depart from the preferences of populations in morality policy innovation. It is worth observing that the model accounts better for morality policy receptivity than the other two regulatory policy forms. The strength of the morality policy model confirms a series of hypotheses about state policy-making behavior for high-salience policies that engender mass public support. The measures included for state political competition, citizen ideology, not-for-profit interest groups, and political culture were all expected to engage state decision-making processes that are most sensitive to highly publicized arguments over first principles. These variables are stronger predictors of morality policy responsiveness because they represent the places where state morality policy decision making occurs. These same variables are less strongly associated with other policy forms. Discussion and Implications: State Responsiveness to Regulatory, Morality, and Governance Policy The predictors of state responsiveness to state regulatory, morality, and governance policy permit further evaluation of patterns of state institutional decision making and the processes of policy innovation and adoption. Each model was estimated with a set of political, institutional, and ideological correlates of state decision making. State legislative professionalism captured variation in the support and resources given to state representatives in crafting and evaluating public policy. State political competition measured how electoral competition and partisan turnover shape state innovation adoption. Variables for the initiative process and state interest-group representation provided rough measures of the

134

Policy Diffusion Dynamics in America

involvement of nongovernmental actors in the policy process. Measures of state ideology and political culture evaluated how mass social preferences and ideology shape innovation receptivity. A comparison of the predictors of state policy innovation across models provides some important insights into the factors leading to state receptivity to innovation. As expected, the analysis allows us to reject the hypothesis that predictors of state policy innovation are stable across policy types. Instead, there is considerable variation across state responsiveness to morality, regulatory, and governance policy. Key predictors of state innovation responsiveness were masked by including policies correlated with fundamentally different political and social processes into a single mode of state policy innovativeness. The relationships that emerged between state institutional political characteristics and policy responsiveness are compelling. In Chapter 3, we observed that of the three policy types, regulatory policy converged most closely to the expected incremental learning curve. This observation led to a hypothesis in this chapter that state regulatory policy responsiveness to regulatory policy would be strongly influenced by legislative professionalism. This expectation was formed because the predominant characteristics of regulatory policy – high issue complexity, uncertainty in policy design, potential unintended consequences, low issue salience– indicate that policy innovation and responsiveness will be greater in states with the capacity to identify, evaluate, and implement regulatory policy reform. Discovering that states with citizen legislatures are resistant to regulatory policy innovation is also consistent with an incremental model of policy diffusion. Because these states lack the capacity to engage in complex policy analysis, design, and implementation, they must rely on the experiences of their expert peers for experimentation and analysis before adopting innovation. This finding fits nicely with traditional research on leaders and laggards in diffusion research. States like California, with a strong advantage in legislative professionalism, can dedicate more time and resources to new policy innovation and implementation than states with smaller part-time legislatures. On the other hand, Montana, which has a part-time legislature that meets every other year, has neither the time nor the resources to engage in large-scale environmental policy legislation, and instead must learn from the experiences of larger neighbors. Further insight into diffusion dynamics was gained from the surprising negative relationship between legislative professionalism and morality policy responsiveness. Technical legislative expertise is not simply

Innovation Hosts

135

unrelated to the adoption of morality policy. Instead, professional state legislatures delay state adoption of morality policy, perhaps because they are more likely to identify secondary problems in the implementation of morality policy innovations. A state with a citizen legislature may defer to a larger, more professional state for the development of technocratic regulatory policy; however, they are less constrained when evaluating and adopting morality policies that revolve around first principles of right and wrong, such as a same-sex marriage ban or parental-consent laws for abortion. Conversely, more professional legislatures may identify hidden costs in morality policy development, for example, anticipating constitutional or other legal challenges to consent laws for abortion. The very research expertise that makes a professional legislature more likely to innovate in state regulatory policy may make them reluctant to adopt morality policy legislation. State receptivity to morality policy instead reflects a connection between political agenda setting, citizen preferences, and issue attention. Morality policy is strongly shaped by state political competition, indicating a strong electoral connection between state political competitions and morality policy innovation. Morality policy may be championed in political contests less because of program analysis or an identification of a policy problem and a policy solution, but rather because morality policy innovation mobilizes voters. To take this a step further, morality policy innovation is correlated with state political competition because voters respond to morality policy innovation in elections. This finding also matches well with the nonincremental patterns of diffusion observed in patterns of morality policy adoption identified in Chapter 3. The recent and rapid diffusion of same-sex marriage bans in the United States corresponds with the electoral use of morality policies. Battles over morality policy innovation prompted conservative activists to raise samesex marriage bans exactly because they mobilized and engaged the mass public. Finally, it is interesting to observe the importance of direct democracy in influencing state responsiveness to governance policy. In Chapter 3, we speculated that the deviation of governance policy from the incremental learning curve suggested the critical role of the initiative process in the diffusion of government regulatory policy. In this context, the separation of governance policy from the incremental learning curve made sense, as governance policy was often implemented through channels entirely independent of state legislative decision making. The strong association that emerged between direct democracy and state governance policy adoption

136

Policy Diffusion Dynamics in America

confirmed this relationship. The link between direct democracy and governance policy reform also supports a broader theory for understanding why some states are resistant and others susceptible to governance policy innovation. Because entrenched politicians of either party are often reluctant to limit the scope and extent of their power, direct democracy often stands as one of the few opportunities to implement governance policy reforms. With the exception of Louisiana, legislative term limits were implemented through direct citizen initiative. For certain policies, a ceiling may exist on the number of states that are truly at risk of adopting an innovation. States in the union possess institutional traits that make them significantly more or less susceptible to emerging innovations. Conclusion How does this analysis of state receptivity to regulatory, morality, and governance policy innovation fit with the distributional analysis of political decision making in public-policy diffusion presented in Chapter 3? As stated at the outset of the book, research into the diffusion of policy innovations has relied on models of incremental decision making to explain the diffusion of innovations from one state to another. According to the dominant decision-making model in diffusion theory, policy diffusion occurs as state legislative decision makers operating under considerable time and resource constraints identify, evaluate, and adopt the innovations of their neighbors. Although there is good reason to expect that the process of policy identification, evaluation, and emulation holds for a large subset of policies, findings from distributional analysis of policy diffusion in Chapters 2 and 3 suggest that incrementalism alone cannot explain the considerable policy dynamics identified in the diffusion of innovations. Importantly, different types of policies encouraged different patterns of policy diffusion. Governance and morality policies deviated sharply from incremental learning curves, whereas regulatory policy approached the expected simulated learning curve. Evaluating the correlates of state policy receptivity has provided insight into the causes of diffusion dynamics. Diffusion dynamics are not simply the product of policies moving at different rates across the same set of innovative states. State receptivity to distinct forms of innovation is shaped by variations in state political and institutional characteristics. The adoption of regulatory policy – the policy that conformed most closely to the expected incremental learning curve – is closely correlated with

Innovation Hosts

137

state legislative professionalism, a concept which at least in part captures the capacity of states to perform complex policy analysis. Morality and governance policy are related to entirely different predictors of policy responsiveness. It is not surprising that state receptivity to morality policy – which deviated sharply from the incremental learning curve – is influenced by electoral competition and interest-group involvement in the policy process rather than legislative professionalism. Governance policy, which deviated most sharply from an incremental learning curve in Chapter 3, is predicted by direct democracy and appears to be unaffected by either legislative professionalism or state political competition. Of course, the processes leading to state policy adoption are often complex. Many of the variables used to capture complex concepts like political competition, state political culture and ideology, or interestgroup environment stand as rough proxy measures for what are dynamic and complex processes. Some measures of state politics used in this chapter stand as better approximations of political and institutional traits than others. Although certain variables – such as legislative professionalism – provide a robust measure of the variation that exists in the length of legislative session, staff size, and salary provided to state politicians, others, such as the measures of state political culture, have sorted the broad political values of diverse states into three overlapping categories. All measures applied in this chapter are accepted in the state political community, but select measures are better than others. It has been especially difficult to measure the impact of interest groups and professional organizations on the policy process. Because interestgroup activists are responsible for much policy invention and advocacy in American politics, theory suggests that interest-group populations will be influential for all forms of policy across all models. However, because interest-group population was measured here with an imperfect proxy of the proportion of not-for-profit interest groups, the model identified interest groups as having a clear and important role only in state innovation and receptivity to morality policy. A similar relationship could not be confirmed in the models for regulatory or governance policy. This is surprising, as a cursory reading of American social movements indicates the strong role of citizen interest groups in the innovation and diffusion of regulatory and governance policy. The next chapter analyzes the role of interest groups as vectors in the diffusion of innovations in American politics. This chapter connects the process of policy diffusion to the larger policy process in America. Because development of a comprehensive measure of interest-group activity has

138

Policy Diffusion Dynamics in America

so far proved elusive, this chapter departs from the statistical analysis of patterns of policy diffusion taken in the previous chapters, and instead develops illustrative cases that demonstrate the role of interest groups in selecting test venues, experimenting with policy frames for preferred innovations, and mounting diffusion campaigns to enact preferred public policies across states. In keeping with the analysis of the prior chapters, Chapter 5 analyzes how the involvement of interest-group populations shapes diffusion dynamics. The model follows research in agenda setting in the policy process. It suggests that variation in the speed and extent of innovation diffusion results from the capacity of interest groups to shape public opinion regarding an innovation, select political venues, and mount a diffusion campaign across states.

5 Policy Vectors Interest Groups and Diffusion Dynamics

Interest groups play a central role in the diffusion of innovations in America. As advocates for policy change, interest groups interact with government decision makers and political institutions to achieve legislative goals. Activists introduce novel policies for legislative consideration and work diligently to pressure government to enact them. Groups that are well represented in mainstream politics may work closely with elected government to enact legislation, whereas marginalized outsider groups attempt to influence legislation by framing and reframing policy proposals to shape mass perceptions and support for legislation, or by agitating for policy innovation at the state or municipal government level. In this sense, interest groups and individual policy advocates are important carriers of innovation in the United States. The spread of policy innovation is often driven by the dedicated work of policy entrepreneurs and interestgroup activists who appeal to local, state, and national governments to secure legislative change. Although research in public-policy diffusion has recognized the central role that interest groups and policy entrepreneurs play as the primary carriers of innovation, studies of interest groups have most often narrowly focused on single case studies documenting the existence and influence of interest-group networks and professional associations. This descriptive orientation overlooks how variations in the organization and behavior of policy vectors shape patterns of policy diffusion. Interest groups pursue a number of distinct organizational and rhetorical strategies to achieve policy change. The ways in which interest groups organize and pressure for innovation lead to very different outcomes for innovation diffusion. 139

140

Policy Diffusion Dynamics in America

Interest groups often face considerable political and institutional opposition to legislative change, as from competing organized interests with different if not contradictory legislative goals. Their greatest hurdle, however, may be the status quo bias characteristic of American political institutions. In Congress and state legislatures, the committees charged with developing policy in specific issue areas tend toward a conservative, risk-averse, and incrementalist approach to drafting and adopting new policies. Despite this strong tendency toward the status quo, interest groups do not simply acquiesce to the incremental bias in American politics. This chapter explores the role of interest-group networks as vectors of innovation. The chapter is organized around a series of observations regarding the unique role of policy entrepreneurs and interest-group organizations in the diffusion of policy innovations in America. Policy entrepreneurs actively experiment with carefully targeted policy frames to expand support for legislative agendas (Baumgartner and Jones 1993; A. Schneider and Ingram 1993). They strategically select venues to exert political pressure, moving across jurisdictions to secure piecemeal legislative change (Baumgartner and Jones 1993; Pralle 2003). Actors throughout the interest community draw lessons from politically successful campaigns (Mintrom and Vergari 1998), modify marginal campaigns, and abandon politically unsuccessful tactics. Rapid policy diffusion occurs in those rare moments when a “successful” innovation is communicated throughout an interest-group network and becomes the focus of mimicking campaigns across venues. Policy outbreaks occur when interest-group activity shapes mass political attention and facilitates the sudden implementation of innovation across states. The organization and strategic choices of interest groups are therefore important components for understanding factors leading to both incremental and nonincremental patterns of public policy diffusion. Interestgroup activists experiment with the selection of venues and the framing of issues, employing trial-and-error strategies to secure legislative progress in different jurisdictions. In the overwhelming majority of cases, these trial-and-error experiments fail. However, when interest groups are able to capitalize on changes in public attention, policy image, or an increase in receptive venues, diffusion can occur through a positive feedback cycle, as linked members of an interest-group network mimic successful strategies to secure policy change. This chapter begins by conceptualizing how interest groups shape the diffusion of innovation across states. It argues that variations in

Policy Vectors

141

interest-group organization and resources have a significant impact on the ability of interest groups to mount a diffusion campaign. Larger, wealthier, and centrally-organized interest organizations are better equipped to meet the resource challenges to pressure for political change simultaneously across states. Smaller, decentralized interest groups lack the capacity to mount sweeping diffusion campaigns and must employ alternative strategies to influence the policy diffusion. Interest groups engage in both strategic issue framing and venue shopping to legitimize policy proposals and shape mass political responsiveness to innovation. The framing and reframing of issues not only changes mass public opinion but also helps build alliances with other groups to pressure for policy change. Venue shopping leads to policy legitimization as previously controversial policy solutions become accepted by recognized political entities. A large, well-organized, and well-funded interest group with chapters across the country can exploit a change in mass public opinion and legitimization of a policy innovation to trigger a positive feedback, so that support for innovation in one state can lead to sudden adoption across multiple states. To illustrate how organizational structure, venue shopping, and issue framing shape diffusion campaigns, this chapter focuses on interest-group involvement in the diffusion of four critical innovation areas: alcohol prohibition, legislative term limits, child-protection policies, and state medical marijuana reforms. Analysis of these cases illustrates how variations in interest-group organization, strategic framing, and venue contribute to diffusion dynamics. Patterns of sudden and gradual diffusion and policy change are shaped by the organization and behavior of interest-group vectors as they communicate innovation from one state to another. Interest Communities and Policy Entrepreneurs Students of public-policy diffusion have long recognized the crucial role that interstate networks of interest-group activists and professional associations play in policy innovation and diffusion (Gray 1973; Mintrom 1997, 2002; Balla 2001). Policy entrepreneurs serve as the catalyst for policy invention and the advocates for policy reforms across states (M. Schneider, Teske, and Mintrom 1995; J. Campbell 2002; John 2003). Interest-group networks provide the crucial pathway linking policy entrepreneurs across venues, as networks facilitate communication between activists linked in a policy network, “a group of actors who share an interest in some policy area, and who are linked by their direct and

142

Policy Diffusion Dynamics in America

indirect contacts with one another” (Mintrom and Vergari 1998, 128). These policy networks encourage policy learning as individuals unified around common causes communicate political problems and successful strategies (Howlett 2002). Surprisingly, research in the diffusion of innovations has been narrowly occupied with documenting when and how interest-group activists communicate policy ideas across venues, rather than how changes in the types of organizations and the strategies they employ shape diffusion patterns.1 Diffusion scholars use case studies to document how interstate state professional organizations and interest-group networks contribute to innovation across a range of issue areas; however, very little research has explored how differences across these communications networks shape patterns of diffusion.2 This exception is curious, as the process of diffusion is shaped by the organizational choices of activists pursuing policy change across the states. Interest groups adopt a number of strategies when mobilizing for policy change across states; these strategies shape political outcomes leading to the diffusion of innovation. The following sections outline how variations in interest-group organizations, issue framing, and venue shopping shape diffusion dynamics. The Organization and Resources of Interest-Group Networks Studies of interest-group behavior distinguish between two general strategies that organizations pursue to exert influence in American politics – insider and outsider lobbying. Whereas financial and corporate interests rely upon costly direct insider lobbying strategies to influence the policy process, grassroots mass-membership interest groups employ outsider lobbying strategies to pressure legislators. Insider lobbying relies on close personal contacts with government, through which lobbyists meet with legislatures and assist in providing and interpreting policy-relevant information, drafting legislation, and arranging expert testimony. Outsider 1

2

Studies of policy diffusion have documented how the strength of ties between actors in a specific issue network or the timing of contact between actors in a network shapes policy adoption (Mintrom 1997). In most cases, research on interest-group networks in the diffusion of innovations has been descriptive, and has documented how the presence or absence of linked interest-group actors shapes diffusion. This research overlooks interesting questions about how variations in the strategies pursued by policy entrepreneurs and interest groups produce different patterns of policy diffusion. One exception is the literature on the influence of federal intervention on state diffusion. National interaction models demonstrate that the standardization of public policies occurs more rapidly when campaigns are orchestrated by the federal government (Karch 2006).

Policy Vectors

143

lobbying works through applying electoral pressure on decision makers. When a piece of important legislation comes before Congress or a state legislature, interest organizations engage in public opinion campaigns, hold press conferences, or encourage members to voice their concerns through letter-writing campaigns, protests, or outright boycotts. This strategy works because of the threat of electoral support – congressmen seeking votes are typically eager to pass legislation that is monitored and preferred by large, organized blocks of their constituents. Although the instruments for strategic pressure campaigns are available to all interest groups, variations in size, resources, issue type, and organizational goals suggest that pressure campaigns will be more effective for some groups than others. Large, wealthy, mass member organizations like the National Rifle Association or the American Association of Retired Persons are able to put greater and more meaningful pressure on national and state representatives than highly specialized issue niche organizations (Baumgartner and Leech 2001). Such organizations can hire full-time professional staff and policy experts; they can employ professional lobbyists and policy analysts who can measure public opinion and carefully craft legislation. (Baumgartner and Leech 2001). Smaller groups may not even be able to support a full-time professional lobbyist and must operate on a shoestring budget with volunteers and few full-time staff. Such differences in financial resources and membership support have important implications for interest-group involvement in diffusion campaigns. The speed and extent of policy diffusion is partially shaped by the considerable transaction costs associated with mobilizing to achieve policy change across states. For example, in state initiative campaigns, wellfunded interest-group organizations have the advantage of being able to simultaneously hold signature drives and sponsor initiative campaigns across multiple states. Smaller interest organizations may face difficult choices between funding an initiative campaign in a single state, supporting signature drives in several states, or supporting public awareness campaigns to raise the profile of a policy problem for future legislative activity.3 A campaign by a weak interest network should therefore spread more gradually than a campaign funded by a strong interest network, as a lack of resources constrains the ability of interest groups to 3

The same advantages hold for more traditional state pressure campaigns. Smaller interest groups lack the resources to maintain a professional staff in each state, and thus must be more selective in where to exert political pressure.

144

Policy Diffusion Dynamics in America

simultaneously coordinate and pursue campaigns across multiple venues. The patterns of innovation diffusion are therefore directly related to the human and financial resources of interest-group communities. Recent studies in political science have begun to correct the assumption that interest groups are uniformly organized to accomplish legislative goals (E. Gerber 1999). Some interest groups are highly centralized hierarchical organizations that coordinate political campaigns at both federal and local levels with the help of expert and professional staff. Others are highly decentralized, with semi-autonomous activists orchestrating legislative campaigns at state or local jurisdictions. Highly centralized, hierarchical interest groups are best equipped to orchestrate rapid campaigns across venues when they encounter an appealing policy innovation. Decentralized interest groups benefit from increased policy experimentation, but are far slower to mobilize a broader campaign around an appealing policy innovation. The capacity of policy vectors to spread innovation across venues is shaped by differences in member size, resources, legislative goals, and organizational structure. Wealthy mass membership organizations with strong national and state organization can produce simultaneous pressure campaigns across states, leading to extremely rapid policy diffusion. Conversely, marginalized or decentralized interest organizations may initially lack the resources to agitate for policy change simultaneously beyond a select few venues. Interest-Group Behavior The resources and membership strengths of issue organizations are central to their ability to mount diffusion campaigns; however, a number of other strategic choices can precipitate distinct patterns of diffusion. Interest groups may adopt a number of strategies to overcome the influence of opposing interests and break the status quo bias in American politics. One strategy is to frame and reframe legislation to elevate issue attention and build new coalitions of support for innovation. Another strategy is to exploit the multiple policy-making jurisdictions in federalism to legitimize innovation. Both the strategies are directed at broadening support for innovation diffusion by expanding the scope of conflict, elevating issue attention, building a coalition of support for innovation, and legitimizing policy-making activity in an issue area. These strategies are a good deal more haphazard and experimental than generally appreciated. Interest groups are constantly looking to control the rhetoric surrounding a public

Policy Vectors

145

debate, but there is a good deal of uncertainty about the efficacy of framing or venue-shopping strategies. For almost every issue, an opposing group of interests are providing counter frames and pressuring policy makers to oppose legislative reforms. Thus, most efforts fail to disrupt the status quo. Dramatic successes are relatively rare, but can and do occur. When an interest group succeeds, it can trigger sudden policy outbreaks across states. Venue Shopping A first common strategy of outsider interest groups occurs through a process known as venue shopping, as interest groups operate to take advantage of the multiple points of access in the federation to secure policy change at different levels of government (Baumgartner and Jones 1993; Holyoke 2003; Pralle 2003). Rather than limiting themselves to exerting influence at one level or branch of government, interest groups – especially those excluded from the mainstream policy process – select government venues to pursue piecemeal legislative change (Pralle 2003). Such venue shopping occurs at all levels of government; interest groups may file lawsuits, appeal to executive bureaucracies, introduce legislation to subnational or municipal governments, or pursue legislation through direct citizen initiatives. The logic of venue shopping is straightforward. Where interest groups’ efforts to influence the policy process are frustrated, they can secure legislative victories by shifting to other venues. There are some important limitations on how freely interest groups are able to engage in venue shopping. First and foremost, venue shopping can be costly, and interest groups must be selective in how and where they apply political pressure. Perhaps because of the high costs of pursuing many different campaigns across distinct venues, organizations often employ highly specialized venue-shopping strategies, for example, hiring a team of lawyers to push for reforms in courthouses across the states, or using the ballot initiative to push for new laws across receptive states. There remain significant variations in how narrowly or expansively interest groups operate in the federation. Interest groups make strategic choices for the selection of venues; however, these choices can be constrained by the organizational preferences or makeup of the interest group. Groups that have pursued litigation in federal and state courts, for example, find it difficult to shift to new venues to build support for legislative reforms. Similarly, a group committed to the initiative process will find it difficult

146

Policy Diffusion Dynamics in America

to pursue insider lobbying strategies in Congress or state legislatures. Venue shopping influences diffusion dynamics for two key reasons. First, by implementing policy in select venues, previously controversial innovations may become legitimized and accepted as reform. This is especially the case with controversial or costly policy innovations, as the act of innovation adoption in one state legitimizes a program for neighboring jurisdictions. Second, the act of securing policy adoption in one state elevates salience of innovation and direct mass attention to support future innovation adoption.4 The ability of interest groups to secure change in select venues can precipitate incremental policy change through a process of trial-and-error venue shopping, as interest groups that have adopted a state or municipal level policy strategy push reforms through receptive venues. In rare instances, venue shopping can lead to sudden feedbacks as a policy becomes legitimized and the broader political system adopts innovation. Framing and Policy Targeting Organized interests not only experiment with the selection of venues but also actively engage in the framing and reframing of legislative ideas in the hopes of expanding public support for policy change (M. Smith 2007; Baumgartner, De Boef, and Boydstun 2008). Following the dictum that those who define political problems control the debate (Schattschneider 1975), activists seeking to shape public responsiveness to their policy programs experiment with framing of legislation in response to changes in the political environment, looking to attribute new meaning or justification for their preferred policy proposals (Baumgartner and Jones 1993). Interest groups adopt one of two strategies when attempting to frame a policy problem. A first strategy is to simply attach a new justification for an existing policy proposal. Interest groups frame legislation by “directing attention to one attribute [of a policy problem] in a complex problem space” (B. Jones 1994, 104). For example, a victims’ rights group wanting 4

It is important to note that policy legitimization and increasing salience are the goals of this behavior. Venue-shopping strategies often fail to secure widespread attention or policy legitimization. For example, the Oregon Health Care plan failed to legitimize statewide health care innovations. Likewise, California’s Proposition 187, which deprived undocumented immigrants of access to health care or public education, was widely publicized, but failed to trigger a series of imitations. Chapters 3 and 4 provide general insights into why policies may fail to diffuse rapidly.

Policy Vectors

147

to force sex offenders to register may shift the justification for the proposal to protecting children from sex crimes rather than promoting it as a protection for the general population. Often, interest groups justify policy reforms in the context of the major salient “issues of the day,” for example, pushing for tax reform legislation as a solution to a major financial crisis, or advocating for a ban on factory farming in the wake of food safety concerns. Such framing and coupling strategies are commonly described in research on interest-group activity. A second form of strategic interest-group framing happens with policy targeting. In this case, interest groups narrow or expand the conflict surrounding a proposal by narrowing or expanding the scope and application of the policy proposal itself (Schattschneider 1975). Anti-smoking activists favoring outright smoking bans begin with efforts to ban smoking in all places frequented by children. Proponents of gun control target certain types of weapons rather than all guns in general. Targeting policies in this way is different from classic framing, as it does not simply offer new justification for the policy proposal; it alters the scope of the policy innovation. The selection of frames and strategic rhetoric is often a highly experimental process, as interest groups modify how they describe policy problems in the hopes of winning public support for legislation. When a policy frame fails to spark much support, activists will try a new way of discussing or framing a policy problem. Rhetoric is relatively cheap, whereas the payoff from a successful framing can diminish opposition to change and open the door for sudden policy reform. As with venue shopping, issue framing shapes policy dynamics when, through a process of issue framing and reframing, a new way of describing an innovation elevates issue salience, encourages positive consideration of previously controversial innovations, and reduces opposition to innovation across venues. Of course, efforts to redefine or reframe a policy debate generally fail to dramatically engender support for new policy reform. Even when a new way of describing a policy problem succeeds, opposing interest groups often succeed in diffusing the impact of an emerging frame with a counterargument. However, in rare moments, issue redefinition can lead to dramatic turnabouts in support for a policy, leading to sudden reforms across states. Reframing policy ideas successfully unites broad and distinct social interests around a targeted cause. When this happens, diffusion can occur through a positive feedback cycle, as a wider group of supporters call for sudden policy change.

148

Policy Diffusion Dynamics in America

Case Selection To understand how interest-group organization, venue shopping, and issue framing have shaped diffusion dynamics in the United States, this study focused on interest-group involvement in four critical cases in American politics: the movement to prohibit alcohol, the term-limit movement; the movement to toughen child-victim crime policy, and the efforts to legalize medical marijuana. This chapter specifically explores how interest groups organize to mobilize activists to participate in American policy making. To this end, the cases selected illustrate the behavior of a subset of interest organizations that rely on activist support and participation to pursue their legislative agenda. This excludes private economic interest groups such as individual businesses and trade associations that overwhelmingly rely on professional lobbyists (Baumgartner and Leech 2001). Despite this limitation, the cases selected here represent a broad range of organizational types in the American political system, ranging from public interest groups (like Common Cause), ideological interest groups (like Americans for Democratic Action), and single issue groups (like the National Organization for the Reform of Marijuana Laws, or the National Rifle Association). Interest-group involvement in prohibition, child-protection, term limitations, and medical marijuana reform were selected to illustrate how interest-group organization, strategic policy framing, and issue venue shopping shape diffusion dynamics. Each case was included to maximize variation across a broad range of interest strategies and organizational forms of interest group in American politics. These groups represent strong and weak interest organizations. They have pursued different venue-shopping and issue-framing strategies. These case studies illustrate processes underlying interest-group involvement in innovation and diffusion, and allow for some insight into the consequences of variation across these three dimensions for future studies. Because each of the cases selected is associated with a successful diffusion campaign, they are clearly unrepresentative of the bulk of interest-group behavior that precipitates neither rapid diffusion, nor diffusion of any form. These cases should therefore be read as illustrations of what should be common processes uniting successful interest groups, rather than as specific tests of hypotheses. With this limitation in mind, these case studies are intended to illustrate why interest groups engage in venue-shopping and issue-framing strategies, how these activities can create conditions for sudden policy

Policy Vectors

149

feedback cycles, and how the organization, professionalism, and membership and resource advantages of interest groups have facilitated or hindered rapid diffusion campaigns across states. Table 5.1 provides a brief summary of the organizational and behavioral attributes that characterize interest-group involvement in the movements to prohibit alcohol, pass child-protection policies, limit legislative terms, and legalize medical marijuana.5 The remaining sections detail how these groups developed issue frames, selected venues to exert political pressure, and organized to facilitate the spread of successful innovations. Organizing for Innovation Diffusion: The Evolution of the Prohibition Movement The movement to abolish alcohol in the United States presents a fascinating example of interest-group organization for an innovation diffusion campaign. The prohibition movement evolved dramatically over the nineteenth and early twentieth century, organizing at first with the grassroots ministries of the American Society of Temperance, reviving after the Civil War with the activism of the Women’s Christian Temperance Union and the Prohibition Party, and finally solidifying itself as the Anti-Saloon League, one of the earliest modern single-issue pressure organizations in American politics (Odegard 1928). Through U.S. history, prohibition activists pursued multiple rhetorical and legislative strategies to galvanize public sentiment for alcohol regulation at local, state, and national levels of government. These strategies included pressuring elected officials to support restrictions on alcohol, placing referenda on ballots across states, and pursuing legal challenges in the courts. This section presents a brief overview of the American prohibition movement, focusing specifically on how prohibition activists organized to regulate and repeal alcohol across states. It then considers the competing rhetorical and venue-shopping 5

Table 5.1 lists one major interest group involved for each issue area. Other groups clearly worked to influence legislation in each issue campaign. The Anti-Saloon League was preceded by the Prohibition Party, the Women’s Christian Temperance Union, and the American Temperance Society. The National Center for Missing and Exploited Children worked with the Klaas Kids Foundation, the Polly Klaas Foundation, and the Megan Nicole Kanka Foundation, among others. U.S. Term Limits coordinated with Americans to Limit Congressional Terms, Citizens for Congressional Reform, and Americans Back in Charge. Other groups involved in the medical marijuana reform movement included the Marijuana Policy Project, the Drug Policy Institute, and Act Up!. The case studies detailing interest-group involvement for each of these issue areas expands on the involvement of these groups.

150

table 5.1. Interest-Group Variation in Organization and Strategic Behavior Organization and Resources

Issue Area

Major Interest Groups Involved

Fragility (Political Opposition)

Resources (Financial and Human)

Organization (Coordination Between National and State Chapters)

Alcohol Prohibition

Anti-Saloon League

High

Strong

Moderate

Trial–and-Error Framing

State Legislatures, Local Governments, Initiative States, Congress

Child Protection

National Center for Missing and Exploited Children

Low

Moderate

Strong

Single Dominant Issue Frame and Policy Target

State Legislatures, Congress

Term Limitations

U.S. Term Limits

Moderate

Moderate

Strong

Initiative States

Marijuana Reform

NORML

High

Low

Weak

Single Dominant Issue Frame and Policy Target Trial-and-Error Framing

Represented Interests and Opposition

Interest-Group Behavior Issue Framing and Policy Targeting

Venue Selection Strategy

Initiative States

Policy Vectors

151

strategies activists pursued in mounting political support for prohibition. Although this section provides only a condensed overview of the Prohibition movement, it is instructive for understanding how interest-group behavior shapes the diffusion of policy ideas across states. Many of the strategies of modern interest groups were pioneered by American prohibition activists. The Evolution of Organization The temperance movement in the United States emerged as religious activists mobilized in opposition to the dangers of alcohol abuse. In the 1820s, clergy formed the religious American Society for Temperance, a network of ministers who advocated for abstinence from alcohol as a precursor to salvation (Aaron and Musto 1981; Billings-Yun, Brittan, and Donahue 1983). The popularity of the movement led to a dramatic proliferation of temperance chapters across the country, and by 1835, one-tenth of the population had joined local temperance organizations and pledged to abstain from alcohol (Aaron and Musto 1981; BillingsYun, Brittan, and Donahue 1983). This early social movement translated into political action as community activists demanded that government develop a political solution to the problem of alcohol (Billings-Yun, Brittan, and Donahue 1983). Responding to the electoral pressure placed by the American Society of Temperance, a majority of states enacted regulatory control on the sale of liquor between 1838 and 1855, and 13 states enacted total prohibition of hard alcohol between 1851 and 1855 (Aaron and Musto 1981). In spite of these gains, the early prohibition movement was subdued by a series of events. The Civil War ended mass interest in the prohibition movement, and liquor laws were dismantled under pressure from liquor industry representatives, who successfully challenged the constitutionality of state alcohol restrictions by appealing to federal primacy in the regulation of interstate commerce. Perhaps more importantly, existing prohibition laws proved difficult to enforce, and were widely circumvented by saloon operators and the liquor industry. Of the 13 states that experimented with the prohibition of hard liquor from 1851 through 1855, only 5 had laws remaining after the Civil War, and these were virtually unenforceable (Aaron and Musto 1981; Kerr 1985). These setbacks led activists in the prohibition movement to conclude that more explicit political organization was needed to carry forward the prohibitionist agenda in American politics. Many prohibitionists believed

152

Policy Diffusion Dynamics in America

that the only way to guarantee the long-term success of the prohibitionist agenda was to take direct control of government. The solution was the formation of the Prohibition Party, a national political party bridging disparate interests in the prohibition movement (Kerr 1985). The Prohibition Party had an immediate impact as a spoiler party in American elections; however, the early political gains of the Prohibition Party failed to translate into meaningful election victories or legislation. The demands of generating a larger political platform created strong divisions between broad-gauge prohibitionists, who believed alcoholism was caused by poverty and insisted on a broad program of economic reforms, and narrow-gauge prohibitionists, who believed that “moral bankruptcy” caused alcohol abuse, crime, and poverty (Billings-Yun, Brittan, and Donahue, 1983). These disagreements became magnified as the party tried to build a larger political platform to address economic and social policy reform. Following a series of embarrassing electoral setbacks in presidential contests, the Prohibition Party dissolved from the factional challenges to organizing and operating as a viable independent party (Kerr 1985). The decline of the Prohibition Party led directly to the emergence of the Anti-Saloon League, a single-issue organization that rejected partisan politics and instead organized around a strategy to support major party candidates in state and national elections based “solely on their willingness to support anti-liquor legislation” (Aaron and Musto 1981, 155). This new organizational vision prevented infighting among members about larger policy problems, and made it easier for the Anti-Saloon League to recruit new activists interested in the alcohol problem (Kerr 1985). The unique organizational structure of the Anti-Saloon League marked an important departure from past efforts to mobilize for collective action and policy change. The Anti-Saloon League had learned not only from the failures of past prohibition organizations, but also from the managerial revolution of the 1800s, when organizations transformed to meet the demands of producing and distributing goods in an increasingly complex network of cities and states (Kerr 1985, 4). The Anti-Saloon League adopted a centralized hierarchical organization to facilitate simultaneous national and state pressure campaigns for alcohol abolition (Aaron and Musto 1981; Kerr 1985). Rather than being controlled by competing party factions, the League was governed by a single National Executive Committee that set political strategies, defined policy goals, and controlled financing for legislative pressure campaigns (Donahue 1983, 7; Kerr 1985). State and local Anti-Saloon chapters were directed by

Policy Vectors

153

professional members of the national organization, directly accountable to the mandates of the national movement. While the League controlled political campaigns across states, it continued to benefit from mass popular support and “very large budgets from the contributions of hundreds of thousands, if not millions, of church people” (Kerr 1985). The Anti-Saloon League was designed with the single purpose of enabling the diffusion of alcohol restriction and prohibition policy across the states. The national chapter operated in Washington, D.C., to combat legal barriers to state-level restrictions on alcohol sales and consumption (Kerr 1985). The organization then focused early legislative activity on local campaigns to regulate, restrict, and ultimately illegalize alcohol state by state (Kerr 1985, 122). This organizational hierarchy corrected for the deficiencies that had prevented the coordination of alcohol reforms in prior movements. The earliest prohibition movements had failed because they were loosely connected grassroots state campaigns. Although they had limited success in banning alcohol before the Civil War, they suffered from both an inability to coordinate interstate campaigns and to combat national judicial and Congressional challenges to state liquor laws. The emergence of a nationally organized and federated Anti-Saloon League permitted state and federal coordination of a carefully controlled campaign to enact alcohol restrictions. In this sense, “the League revolutionized American Politics, as it had shown a new way for ‘minor associations’ to organize and achieve their objectives” through orchestrated pressure campaigns (Odegard 1928; Kerr 1985, 4). Importantly, this organization directed political campaigns and provided a context for coordinating strategies and mobilizing a diffusion campaign. The evolution in organizational form was matched by dramatic changes in the strategic rhetoric and framing designed to persuade publics to support prohibition. Before the Civil War, prohibition activists cycled between issues of morality, religious probity, economics, and public health. The preacher activists of the American Society of Temperance argued that alcohol subverted social structure and endangered families, workers, and social stability (Aaron and Musto 1981). These activists were joined by early public health advocates who had the daunting challenge of persuading Americans that alcohol was a good deal more deadly than they realized (Donahue 1983). Later prohibition advocates altered the justification of alcohol prohibition beyond the initial religious and health considerations in order to galvanize public support for alcohol restrictions. Throughout the 1800s, activists argued that alcohol was the root of criminal behavior (Aaron and

154

Policy Diffusion Dynamics in America

Musto 1981). The Women’s Christian Temperance Union placed alcohol prohibition as a centerpiece for a broader set of social programs to improve the quality of life for women and families. They argued that alcohol precipitated domestic violence and increased divorce rates (Donahue 1983). Populists in the prohibition movement argued that alcohol was linked to political corruption, as saloon operators and liquor corporations financed and bribed officials to oppose popular alcohol regulations (Donahue 1983). The Anti-Saloon League added to this laundry list of considerations by asserting that alcohol corrupted the productivity of American’s labor force and increased “promiscuity, municipal corruption, political radicalism, and a number of other “vices spread by immigrants” (Billings-Yun, Brittan, and Donahue 1983, 13). In the buildup to World War I and the passage of the Eighteenth Amendment, prohibition activists persuaded the national government that alcohol damaged the war effort, as it corrupted soldiers and drained the country of valuable resources that could be better used elsewhere (Aaron and Musto 1981). Although the consequences of alcohol were constantly reframed to accommodate the issues of the day, the solution remained constant – a ban on alcohol. These shifting justifications for prohibition activated new considerations for those who were undecided about alcohol prohibition. Shifting the focus of prohibition policy not only helped direct public attention to new problems presented by alcohol consumption, but also helped activists build crucial political alliances for reform campaigns. The link between alcohol and corruption helped win support from the populists who opposed the influence of the industry on local governments. The economic arguments drew significant political support from industrialists such as John Rockefeller, who was persuaded that alcohol consumption decreased worker productivity, and who contributed generously to the prohibition cause at the turn of the century (Billings-Yun, Brittan, and Donahue 1983, 15). In addition to strategically framing the debate about alcohol regulation, prohibition activists also experimented with limiting the scope of alcohol policy in order to legitimize prohibition. Where the earliest activists pushed for the full abolition of alcohol sales in American states, the Anti-Saloon League adopted a more pragmatic strategy of first pursuing small reforms to alcohol regulatory policy before pushing for total prohibition.6 By securing piecemeal reforms, activists hoped that forms 6

This strategy remains viable today. Anti-abortion activists have pursued incremental legislation to limit access to abortion before pushing for a total ban.

Policy Vectors

155

of alcohol regulation would become legitimized, making broader policy change more politically palatable to skeptical politicians and voters. This strategy proved effective in laying the groundwork for prohibition. As Billings-Yun, Brittan, and Donahue (1983) explain: The organization began with the seemingly modest aim of pressuring lawmakers in states where saloon policy was normally set by the legislature, to pass local option laws allowing individual cities or counties to outlaw saloons if they so chose. Because local option laws were framed as steps towards local self government and direct democracy, they appealed even to dedicated drinkers. In 1907, once those laws had been established, the league started lobbying for state laws banning both the manufacture and the sale of liquor. Not until 1913 did League officials declare their intent for national prohibition (6).

This strategy proved remarkably effective in implementing alcohol regulations across states. “By 1906 most state legislatures had passed local option laws, and 40 percent of the nation’s population, primarily in rural areas, was living in saloon-less territory” (Billings-Yun, Brittan, and Donahue 1983, 9). The Anti-Saloon League’s organization facilitated the coordination of alcohol policy regulation at multiple venues of government. Anti-saloon activists initially focused their energy and resources where they believed they were most likely to win legislative reforms. The League targeted rural states, where support for prohibition was strongest. It then strategically targeted states with weak party majorities, where they could mobilize large groups of voters to compel vulnerable legislators to support prohibition laws. When this strategy failed, they bypassed state legislatures and employed the initiative process to secure political change. Many of the most significant prohibitionist victories came about through referenda rather than legislative amendments. This strategy gradually shaped the regulation of alcohol across states. “In 1906, only three states had prohibition; by 1913, there were 9, with campaigns underway in all the others. By 1916 there were 23 dry states, and in 17 of those states the measure was approved by direct vote of the people” (Aaron and Musto 1981, 157). Although prohibition activists were happy to work to diffuse local liquor restrictions through local option laws or anti-saloon laws, they had developed the organizational capacity to mobilize around a mass national campaign quickly when the opportunity presented itself. After Anti-Saloon League pressure compelled Congress to approve and send the Eighteenth Amendment to the states, prohibition activists were immediately able to pressure rural state legislatures to approve the amendment. States “approved the amendment in record time: the required thirty six

156

Policy Diffusion Dynamics in America

states ratified the 18th amendment in less than fourteen months” (BillingsYun, Brittan, and Donahue 1981, 11). The well-documented history of the prohibition movement allows insight into the strategies interest organizations pursue when excluded and when empowered in national politics. Under the guidance of the Anti-Saloon League, the prohibition movement constantly experimented with shifting justifications for prohibition. When an argument appealed to state publics – as was the case with the local option laws for local control of state liquor laws, the League took advantage of its political empowerment to push for change across receptive venues, and organized a careful venue-shopping strategy to maximize the reach of alcohol restrictions in America. When activists were unable to influence national legislation, prohibitionists selected venues where they could successfully pressure for changes in the liquor laws. However, when national organization and issue attention shifted in favor of national prohibition, Anti-Saloon League activists had developed the organization needed to act quickly to capitalize on prohibitionist sentiment. The length of the prohibition movement suggests that the diffusion of alcohol regulation followed a lengthy pattern of reforms culminating with a sudden diffusion of prohibition policy across states. Protecting Victims and Criminal Justice Policy, 1990–2005 Over the last three decades, child advocacy groups pressuring state and national government have succeeded in securing the rapid and widespread diffusion of a number of well-known child-protection laws. Child welfare activists orchestrated the diffusion of state missing children’s clearinghouses in the 1980s. In the 1990s, child crime prevention activists pushed for the passage of state Megan’s Laws, which require law enforcement officials to make sex-offender registries open to the public; they also pushed for the Amber Alert – an alert system that uses existing state emergency broadcast systems to provide rapid information to the public about kidnappings.7 In each instance, organizations advocating for child-protection policy were able to quickly capitalize on elevated national attention to crime policy to mobilize a national diffusion campaign. The strategies employed by the coalition of interests pressuring 7

Child advocacy groups were also instrumental in the passage of state three-strikes laws and have recently been involved in a campaign to implement Jessica’s Law, a policy that requires convicted sex offenders to wear GPS tracking systems.

Policy Vectors

157

for child-protection policy are informative for understanding how popular pressure organizations are able to exploit single episodes that rivet national attention, in order to dramatically influence state and national politics and produce rapid policy change. This section focuses on strategies used by interest groups in the recent campaigns to pass the Amber Alert and Megan’s Law in each state. As with prohibition movements, the interest-group network appealing for state adoption of the Amber Alert and Megan’s Law benefited from considerable coordination and national organization. Groups like the Klaas Kids Foundation, the Megan Nicole Kanka Foundation, the Polly Klaas Foundation, and Parents for Megan’s Law worked at both the state and national level to promote the passage of child-protection policies. These groups were guided by the prominent National Center for Missing and Exploited Children, a national interest group that had been central in encouraging state adoption of the missing children’s clearinghouses of the 1980s. Child-protection interest groups moved extremely quickly in response to the elevated issue attention to crime policy – engendered by highly publicized discrete events – to pressure state governments to adopt child protection legislation. The 1990s marked a period when Americans were uncommonly alarmed by crime. This unease was exacerbated by a series of high profile tragedies involving the victimization of children. The highly publicized kidnapping, rape, and murder of 12-year-old Polly Klaas by a repeat, violent sex offender in California directly led activists to call for statewide adoption of three-strikes sentencing laws. Publicity surrounding the kidnapping, sexual abuse, and murders of Megan Kanka and Amber Hagerman elevated public attention to the problems in detecting local threats and mobilizing rapid responses to kidnapping. Organizations like the Klaas Kids Foundation and the National Center for Missing and Exploited Children coordinated campaigns to capitalize on elevated public attention to crime in order to mobilize broad public support for both the Amber Alert and Megan’s Law. These interest groups mounted communication campaigns, distributing literature and sample legislation to community and state activists in order to raise awareness of beneficial anti-crime legislation. They directed state campaigns to pressure legislators into innovation adoption. Finally, each of these interest groups publicly tracked the status of Megan’s Law and Amber Plan legislation and announced an ambitious goal of having a sex-offender registry and emergency kidnapping response system in place in every state in the union. Although the bulk of these campaigns were directed at state

158

Policy Diffusion Dynamics in America

legislatures, activists within the issue community appealed to the federal government to provide grant assistance to facilitate standardization of implementation of state child-protection policies. This organizational strategy proved remarkably effective. Over the 1990s, every state enacted Megan’s Law legislation allowing community access to state sex-offender registries. The Amber Alert passed in every state in just six short years. Interestingly, in both cases the federal government lagged behind the states in innovation. The rapid diffusion of these policies was therefore not singularly attributed to federal intervention, but was rather the result of a well-organized interest group pursuing popular policy reforms. It is interesting to observe how child-protection advocates framed legislation in order to build public support for child-protection innovation. As Chapter 3 indicated, child welfare policy holds a natural advantage of having low issue fragility, and there is little formal organized opposition to innovation for child protection. Yet despite this natural advantage, child advocates controlled the framing of policy innovations to maximize mass appeal and strategically diminish political concerns. Proponents of the Amber Alert and Megan’s Law emphasized emotional and moral considerations and directed attention away from program costs or potential ethical problems of policy application (Wood 2005). Although both policies address a range of criminal behavior extending beyond the protection, public discussions focused almost exclusively on rare and narrow examples of egregious violence against white female children (Wood 2005). The very policy label attached to these innovations shaped public opinion by calling attention to a very specific story about crime against vulnerable children. The Amber Alert legislation officially stands for America’s Missing Broadcast Emergency Response; however, the policy name reminds the public of the well-known kidnapping and murder of Amber Hagerman. Megan’s Law has no such acronym, but the policy image draws public attention to a horrific instance of sexual violence against a child. Interestingly, in certain states, activists maximized the appeal of the policy image by adding names of locally salient crime victims. Georgia’s Amber Alert plan is referred to as Levi’s Call (http://alerts.gbi.georgia.gov). In Arkansas, the program is referred to as the Morgan Nick Amber Alert (https://www.ark.org/asp/alerts/mnaa/index.php). The framing of the crime policy innovation focuses attention on the policy aberrations and minimizes public opposition to innovation. This framing technique is especially important for understanding the diffusion of the Megan’s Law sex-offender registry. In some states, Megan’s Law requires nonviolent offenders who have engaged in consensual but

Policy Vectors

159

criminalized sexual behavior to register as sex offenders. Crafting the policy to address the problems of violent child molestation focused public attention on the defense of children from stranger violence rather than on the taboo discussion of family sexual violence or the civil rights of nonviolent offenders. The strong interest-group organization behind the Amber Alert and Megan’s Law made it possible for proponents to press for extremely rapid innovation diffusion, first by carefully mobilizing across states and then by shifting the policy image to capitalize on mass concern for crime policy throughout the 1990s. Organizing for Ballot Initiatives: The Term-Limit Movement, 1990–1998 The diffusion of term-limit initiatives in the early 1990s proved to be one of the most popular and sweeping innovation movements of the twentieth century. Driven by extremely broad public support for the policy idea,8 term-limit measures appeared on the ballot in over 21 states between 1990 and 1994 (Karp 1995). These movements were so sweeping that they were characterized by some pundits as a populist revolt against “an entrenched political class whose principal interest was its own self perpetuation” (Wallison 2000, 39). The rapid diffusion of term-limit initiatives stands as a strong example of how an idea with high issue salience, an appealing frame, and strong interest-group organization can move rapidly across political venues. Term-limit activists framed the issue of legislative limits in truly populist language, promising to restore citizen democracy and limit the opportunities for political corruption associated with career politicians. Although term limitations were by no means a new concept in American politics,9 the public mood opposing career politicians was particularly high at the beginning of the 1990s, as a cluster of scandals in Congress created the sense that corruption plagued state and national governments10 (Wallison 2000). These themes united groups across ideological perspectives by 8 9

10

Most polls showed over 70% of voters favored legislative term limits (Jost 1994, 3). New Hampshire included term limits for governors in their state constitution. Other states adopted governors’ term limits through the eighteenth, nineteenth, and twentieth centuries. State legislative term limits were passed in the 1990s; however, this idea had been discussed since the founding of the Republic (Benjamin and Malbin 1992). Wallison writes, “Voters were inundated with information about corruption in the House of Representatives’ post office and the credit union, about secret pay raises, and even million dollar slush funds that may or may not have existed” (2000, 40).

160

Policy Diffusion Dynamics in America

playing to traditional American distrust of government and elected representatives (Grant 1995). Term-limit initiatives were carefully packaged to channel national outrage. The translation of this raw anger against career politicians into concrete public policies was facilitated by the organizational capabilities of the expanding term-limit network. Organizations such as Americans to Limit Congressional Terms, Citizens for Congressional Reform, Americans Back in Charge, and later the umbrella organization U.S. Term Limits were heavily involved in funding ballot drives and educating citizen activists about the steps needed to place term legislation on a ballot. These organizations lessened the transaction costs associated with placing ballots on the initiatives by providing clear channels of communications across the interest-group network, and just as importantly, by providing vast sums of money to initiative campaigns across states. Stuart Rothenberg explains: National term-limit organizations played an important role in stimulating the public’s interest in term limitations, in motivating key state leaders, and in supporting state limitation efforts. They have sped up the process by providing the resources and skills to allow dozens of states to have term-limit measures ready for the ballot . . . by the end of 1992 (1992, 112).

Rothenberg’s point goes to the heart of how the policy frame, issue salience, and the strength of the interest-group network influence the speed of an idea’s diffusion. The appeal of the populist frame captured by the term-limit movements united a receptive public for policy reform across the states. The strength of the issue network streamlined coordination and diffusion of term-limit policies across the country. These factors contributed to the striking speed of the term-limit movement. Term limits had both the broad issue appeal and the strong network support needed to capitalize on national dissatisfaction with government. From 1990 to 1995, 21 states passed legislative term limits.11 Interestingly, term-limit proponents used the only venue available to enact limits on professional legislatures. Because term limits relied on the state ballot initiative, the movement was largely limited to those states with direct democracy. In 1995, Louisiana became the only state to implement legislative term limits by a vote in the state legislature. 11

Nebraska passed term-limit legislation in 2000. Of the 22 states that passed legislative term limits, 6 have had the laws overturned by legal challenges.

Policy Vectors

161

Piggybacking on Successes: The Informed Voter Ballot Notations of 1996–1998 The diffusion of “informed voter” or “scarlet letter” ballot notations provides an interesting comparison to the diffusion of the original term-limit movement. Drawing upon the organizational advantages of the interestgroup network that emerged during the first term-limit movement, proponents of informed ballot initiatives sponsored measures that demanded candidates and incumbents to pledge their support for a constitutional convention on federal term limits. Those politicians who refused to make such a pledge would have a ballot notation declaring “disregarded voter’s instructions on term limits” placed next to their name on ballots during an election (Fisher 1996, 12A). This caused many pundits to characterize informed voter ballot notations as “scarlet letter” initiatives (Chiang 1999, A18). Informed voter ballot initiatives clearly benefited from the resources of the strong term-limit interest community. U.S. Term Limits, the largest national interest group dedicated to reforms of term-limit laws, liberally spent money in order to place informed voter ballot notations on the ballots across the states (Fisher 1996, 12A). The coordination of these efforts was facilitated by the presence of a state network of policy entrepreneurs who had gathered support for term-limit legislation from 1990 to 1996. The scarlet letter initiatives benefited by addressing a fairly salient policy problem. Although the attention directed to term limits had subsided from a high in 1994, voters continued to support reforms in term-limit initiatives across the country. The major difference between the term-limit initiatives of 1990–1994 and the informed voter initiatives of 1996–1998 can be located in the nature of the policy target. Whereas the term-limit movement surged because of appealing policy image, informed voter laws suffered from diminished salience and a less-appealing issue frame. The efforts to force politicians to call a constitutional convention discouraged libertarians and constitutional purists who had previously supported term limits as a means of restoring democracy to the citizens. These groups believed that a constitutional convention would be costly, chaotic, and potentially destructive. Many more simply opposed any amendment to the constitution (Fisher 1996, 12A; The Phyllis Schlafly Report Online 1996). The language of the informed ballot initiatives also isolated Republican politicians who had famously used popular sentiment for term limits to great success in 1994. Informed voter notations not only mandated that

162

Policy Diffusion Dynamics in America

politicians declare their support for a constitutional convention and term limits, but also threatened to tag them with the same negative “disregard voter’s instruction tag” if they did not support the exact policy proscriptions favored by U.S. Term Limits. Scarlet letter initiatives finally turned off those who objected to the unconstitutional and potentially dangerous precedent such a law would invoke. Editorial writers voiced fears that informed voter notations would lead to other “notations” cluttering ballots, such as information about stances on abortion (Fisher 1996, 12A). The emergence of this strong counter-frame against informed ballot notations acted to dampen public support for the initiative. The combination of a strong network and a limiting frame help explain the strange pattern in the diffusion of scarlet letter initiatives across states. In 1996, with the help of funding from U.S. Term Limits, policy entrepreneurs placed informed voter notations on the ballots of 14 states, with 9 states eventually approving the policy innovation. In 1998, only 3 states considered informed voter ballot notations – California, Idaho, and Nevada. By 2000, this strategy to force politicians to support term limits had faded entirely. The resources of the strong issue network behind the informed voter ballot notations allowed for a coordinated effort to place the policy idea on the ballot in several states simultaneously. However, the failure to frame the policy in a way that would capture broad public support and bring divergent interests under the umbrella of term-limit reform led to slowing rather than increasing diffusion of policy reforms through the informed voter initiatives. Because of the organizational strength and the venue-shopping strategy, scarlet letter laws were implemented in a number of states well before policy evaluation revealed problems in the policies’ constitutionality, leading to the abrupt end of diffusion. The diffusion of the informed ballot notations thus presents a clear instance where states mimicked a policy failure. Activists succeeded in passing ballot notation initiatives across states, only to have each innovation overturned by the courts. Reframing and Expanding the Coalition of Support: The Case of Medical Marijuana The medical marijuana movement of the mid-1990s represents perhaps the most interesting strategic manipulation of policy frames and venue shopping in recent history. Unlike the term-limit and scarlet letter

Policy Vectors

163

initiatives, the interest groups calling for marijuana reform have historically been resource weak. Furthermore, efforts to legalize marijuana have traditionally received tepid support in prior campaigns, as legalization of drug policy is met with skepticism from voters and opposition from elected officials and government. However, through reframing of marijuana policy reform away from the goal of legalizing or decriminalizing marijuana, toward legalizing marijuana as medicine, policy entrepreneurs were able to expand their coalition’s base and achieve significant, albeit comparably limited policy reforms in several states. Medical marijuana laws spread across 10 states from 1996 to 2006. The success of medical marijuana initiatives was driven by an important shift in framing of the policy reform away from legalization or decriminalization for public consumption, toward the “compassionate use” frame that justified allowing doctors to prescribe the drug as a treatment for a variety of illnesses including AIDS, glaucoma, cancer, or even migraines. The reframing of the policy as a question of patients’ rights had the effect of strongly expanding the coalition of supporters of marijuana policy reform. Public opinion polls tracking the legalization of marijuana for medical use show generally strong support for the policy idea, frequently with more than 70% of voters approving of allowing sick patients access to the drug (Koch 1999). Almost the same percentage of people opposed the legalization of marijuana for personal consumption (Koch 1999). Framing of medical marijuana as a therapeutic drug for the chronically ill did more than shift public opinion. It broadened the coalition of supporters by forming an important alliance with patients’ rights activists. The medical marijuana movement grew as an offshoot of the AIDS epidemic. Although medical marijuana legislation was introduced by drug legalization activists, the innovation drew the attention of AIDS activists who were fighting for access to experimental medications and increased research into HIV prevention and treatments. After the government discontinued extremely limited marijuana research and distribution programs, AIDS and cancer activists demanded access to the drug, which purportedly has therapeutic benefits for AIDS wasting and for countering nausea caused by chemotherapy. In response to growing interest in the medical use of marijuana, advocates for drug policy reform began testing medical marijuana policies in limited venues at the municipal level in Northern California. These small successes in California encouraged activists from the National

164

Policy Diffusion Dynamics in America

Organization for the Reform of Marijuana Laws (NORML), the Marijuana Policy Project, and the Drug Policy Institute to organize support for an initiative legalizing marijuana in California and Arizona. These projects were helped immensely by the funding of George Soros and other drug reform activists, who poured money into the medical marijuana initiatives first in California and Arizona in 1996 (Koch 1999), and later in Oregon, Washington, Maine, and Nevada. This funding helped professionalize the initiative campaigns, allowing activists to coordinate resources and mobilize support around medical marijuana initiatives in California and elsewhere (Lacayo 1996). The case of medical marijuana demonstrates how strategic venue shopping and issue framing by marginalized interest groups can lead to moments of policy activity driven by trial-and-error framing and venue shopping. Where prior initiatives to legalize or decriminalize marijuana sponsored by the drug policy reform organizations had failed to garner much support, the notion of medical marijuana provided a bridge to bring new groups into the issue network. Supporters for “compassionate use” initiatives included not only those who favored the legalization of marijuana, but also a number of new groups concerned with patients’ rights. AIDS activists from groups such as ACT UP (Aids Coalition to Unleash Power) and the San Francisco Aids Foundation joined with doctors, libertarians, and drug policy activists to campaign on behalf of the medical marijuana initiatives. The speed of the initiative increased as steadily as the size and resources of the network expanded. However, compared even to the plodding progress of the prohibition movement, medical marijuana has thus far received relatively limited diffusion. Reframing has permitted some small success, but organizational disadvantages and low issue attention has limited the speed and extent of medical marijuana diffusion. Importantly, the success of legal challenges to medical marijuana initiatives has dampened broader support for the movement. Discussion These cases provide some intriguing insights into how interest-group behavior shapes the diffusion of innovations. A first theme to emerge is the significant influence of interest-group organization and resources on the speed and extent of public-policy diffusion. The size, wealth, and public perception of an issue organization provide a significant advantage to interest groups pushing for innovation diffusion across states.

Policy Vectors

165

The rapid and extensive diffusion of child-protection policies like the Amber Alert was facilitated by the vigilant campaign of interstate childprotection organizations. These interest groups were well established to capitalize on the elevated national attention directed toward crime policy after media coverage of the kidnapping and murder of Polly Klaas, Amber Hagerman, and Megan Kanka. These organizations directed state pressure campaigns to encourage innovation adoption, provided sample legislation, and agitated for the federal government to provide block grants to support for child-protection programs. Of course, the success of child-protection policy diffusion is in no small part influenced by the natural appeal of child-protection programs, as there is little organized, mobilized opposition to these groups. Strong interest-group organization also helped both extensive and rapid diffusion of the term-limit movements in the 1990s, and facilitated the final push for prohibition in the early twentieth century. In both these cases, the ability of activists to capitalize on a window of opportunity and press for innovation adoption across a large set of states was facilitated by a coordinated professional staff that organized local innovation campaigns. U.S. Term Limits provided sample ballot initiative legislation, strategic support, and financial backing for term-limit initiatives. The Anti-Saloon League’s hierarchical organization not only permitted it to carefully control the campaign for alcohol restrictions prior to the Eighteenth Amendment, but also allowed prohibitionists to mobilize extremely quickly around the amendment when it was referred to the states for ratification. In fact, the importance of organizational strength suggests an additional advantage of interest-group institutional memory in diffusion campaigns. The diffusion of the Amber Alert, the scarlet letter initiatives, and Prohibition were facilitated by the experience of interest groups involved in prior diffusion campaigns. The diffusion of each of these innovations occurred through the mobilization of an interest-group network that had prior experience mounting diffusion campaigns. The organizational structure emerged and became stronger and more centralized over the life of the trial-and-error diffusion campaigns of the medical marijuana movement. Innovations sponsored by newer interest-group coalitions may be slowed by a need to establish interstate infrastructure for national diffusion campaigns. This importance of interest-group organization and resource advantages should be generalizable beyond the selected cases included in this

166

Policy Diffusion Dynamics in America

chapter. Diffusion campaigns sponsored by Mothers Against Drunk Driving, the National Rifle Association, and the American Cancer Society share a common strength of centralized hierarchical organization. Wealthy mass membership organizations have distinct advantages when compared to smaller-issue niche organizations. A second theme to emerge from the case study analysis is the role of interest-group organizations in shaping conditions for diffusion. As Chapter 3 demonstrated, certain policy types lead to distinct diffusion patterns, in part because state decision makers and publics interact differently with policies having high salience and low issue complexity. The cases in this chapter reveal that interest-group organizations invest considerable time in experimenting with the rhetorical frame and policy target to elevate issue salience, encourage coalition building, or direct public attention away from negative or controversial components of innovation. Issue framing and reframing is most easily observed in smaller outsider interest groups that either champion controversial causes or lack strong central organization. Drug legalization advocates capitalized on the AIDS crisis to reframe calls for marijuana decriminalization. Retargeting decriminalization for a specific target population not only shifted mass public opinion in favor of medicinal marijuana; it also encouraged AIDS activists, cancer activists, and physicians’ organizations to support innovation adoption across states. Whereas strategic framing helps small groups with controversial issues secure limited success across highly receptive states, stronger interest groups have capitalized on issue definition and redefinition to more dramatically shape public opinion and drive extremely rapid innovation diffusion. Prohibitionist organizations strategically attached alcohol regulation to virtually every salient issue of the day. Victims’ rights organizations associated crime policy reforms with the highly publicized victimization of a child. Controlling the debate quells opposition to innovation and encourages decision makers to prioritize policy on an emotional rather than a technical level. Framing and reframing strategies are well known beyond these selected cases. Modern anti-abortion advocates not only carefully control the language of the abortion debate with terms like “pro-life” and “partial birth abortion,” they also strategically limit conflict around abortion reforms by narrowing the scope of abortion regulation – for example, restricting the access to abortion by children, or demanding 24-hour waiting periods before procedures. Tobacco control advocates shift the justification of smoking bans away from the risks to the smoker and instead focus

Policy Vectors

167

on the dangers of secondhand smoke for children, employees, and other patrons and workers. In many cases, these redefinitions are persuasive across jurisdictions, changing public opinion and leading to state receptivity to the diffusion of innovations. Both issue framing and venue shopping are strategically employed to elevate issue salience and legitimize innovation. Prohibitionists and medical marijuana advocates who were unable to secure national support for legislation worked at the municipal and state level to enact gradual legislative change. The term-limit movement was forced to specialize in ballot initiatives because of the significant challenges it faced in persuading career politicians to vote themselves out of a job. Even crime victim advocates selected venues where crime policy was most immediately salient. The act of adopting innovation not only legitimizes the policy solution in American politics; it can also elevate issue salience and lead to interstate demands for policy adoption. The strategies interest groups pursue – strategic framing and reframing and venue shopping – can lead to trial-and-error diffusion as activists attempt to implement innovation one venue at a time. However in rare cases, activists can exploit elevated issue attention and implement rapidly across multiple venues. Interest groups play an underappreciated role in this familiar process. Their strategic choices do more than capitalize on elevated issue attention: Sometimes simply introducing a new idea for policy innovation to a receptive venue can itself lead to elevated issue attention and a policy outbreak. The role of issue organizations in rapid policy diffusion has some important implications for understanding decision-making processes underlying diffusion of innovations. Although interest-group networks are believed to facilitate policy and political learning across linked activists in different jurisdictions, this learning does not always yield the familiar pattern of incremental policy identification, evaluation, and implementation. In the case of child-protection policies like Megan’s Law, the three-strikes law sentencing guidelines, and the scarlet letter initiatives, diffusion occurred with minimal consideration of outside arguments. In the case of the scarlet letter initiatives, nine states simultaneously adopted the innovation before courts ruled the ballot notations unconstitutional. Megan’s Law was framed and targeted as a child-protection policy – and in many states, initially compelled nonviolent, consensual sex offenders to register and report as sex-crime perpetrators. In states

168

Policy Diffusion Dynamics in America

like California, the three-strikes law was pushed through without evolving to include provisions for prison overcrowding. The sudden popularity of these and other innovations short-circuited the decision-making process of state legislatures, leading states to adopt policies without broader considerations of program cost or the possibility of unintended consequences.

6 Conclusion

In many ways, the approach described in this book is a throwback to an older approach to the study of public-policy innovation and diffusion. It follows the pioneering work of Walker (1969), Gray (1973), and Savage (1978) by going back to a large-N comparative analysis of policy innovations spanning more than a century. This stands in stark contrast to the recent trend toward intensive analysis of single cases intended to illustrate the processes contributing to the diffusion of innovations. These single case studies serve well for highlighting specific decision-making processes or for commenting on an anomalous case in which an innovation’s diffusion has deviated from the expected pattern. It is when such deviations occur that the limitations of the case-study approach become apparent, for this approach presumes that a single pattern is dominant and that deviations are exceptional. The work presented in this book has taken a more expansive approach. Though possibly sacrificing some of the detail and precision of single case studies, this comparative orientation suggests a broader range of diffusion patterns of emerging innovations in America. It provides insights into how differences across historical eras, policy domains, interest-group carriers, and policy targets shape the pattern of diffusion. What has been sacrificed in detail is offset by gains in generalizability. The approach taken here is empirical. It involved examining a large set of innovations, first to evaluate how well the conventional explanation for the diffusion of innovations matched the actual record, and then to identify the triggers and causes of different patterns of diffusion. Although this book revives an older approach to studying policy diffusion, the questions it raises and the larger theoretical orientation of 169

170

Policy Diffusion Dynamics in America

this work invite a new interdisciplinary approach to modeling diffusion dynamics. This book borrowed from basic research in epidemiology to model the causes of diffusion dynamics. The point of this approach is not to explicitly argue that the process of policy diffusion is identical to the transmission of disease, but rather to point out that variation across the characteristics of the innovation, the internal dynamics of the states, and the unique attributes of policy carriers must be included into a comprehensive model of how policies are transmitted and spread across jurisdictions. Taken separately, each of these components is important for understanding how variation in any single link in the diffusion chain can alter the path of diffusion. Taken together, the interaction of these components is important for understanding why different innovations should produce such dramatically different patterns of diffusion. This chapter reviews the major findings of this research and evaluates how each component of innovation diffusion may interact to promote diffusion dynamics. It also points to some key areas where new inquiry can begin to provide an even more robust understanding of comparative policy diffusion. Decision Making and Diffusion Dynamics Models of decision making in interstate public-policy diffusion begin with an important observation about the demands of information processing in state government. State legislatures are constrained by a scarcity of time, resources, and political attention.1 Because elected officials must address a multitude of different issues on a daily basis – negotiating state budgets, designing regulatory policy, sitting in committee hearings, addressing local constituent concerns, etc. – state governments cannot dedicate the time and resources needed to engage in an optimal solution for each and every policy problem. To compensate, state legislators often emulate the innovations of their peers, making small adjustments to a neighbor’s policy in order to tailor it to their jurisdiction’s circumstances and needs. This is a conservative theory, as it suggests generally that states emulate policy successes but avoid costly policy failures. However, this theory only partially describes processes leading to public-policy diffusion. Although states certainly evaluate and adopt policy successes, they 1

This model of information processing has also been prominently used to understand Congressional decision making. However, it is likely that time and resource constraints are especially pronounced in American state legislatures, which generally have shorter legislative sessions than the national government and are afforded far less in the way of professional legislative assistance from expert staff.

Conclusion

171

also frequently (and sometimes rapidly) adopt policies that are either unsuitable for their immediate problems, or worse, are outright policy failures. In its emphasis on how limits of political attention shape the policy process, the incrementalist perspective is a useful starting point for thinking about the decision-making pressures leading to interstate policy diffusion. Constraints on the resources and time that state governments can devote to policy problems do not necessarily imply that innovations must spread through a gradual process of incremental evaluation and emulation. Studies of public-policy diffusion have recognized that information shortcuts help time-constrained decision makers identify accepted and popular solutions for preexisting and previously recognized political problems. This presents too neat an understanding of the American policy process. Sometimes, awareness of the innovation simultaneously reveals both the policy problem and the policy solution, leaving no time for policy evaluation. In these cases, a large number of states simultaneously adopt the same policy innovation, suggesting decision making driven by sudden policy imitation rather than incremental instrumental policy learning. State policy makers may look to neighboring states to address policy problems, but this process rarely entails a full evaluation of political, policy, and social costs and benefits. These incongruities need not be contradictory. Instead, they represent system-wide responses to very different signals. Attention and Decision Making in Policy Diffusion Chapter 2 argues that diffusion dynamics emerge because state governments prioritize and respond to problems differently based on perceptions of issue salience, issue importance, and issue complexity. State legislatures disproportionately respond to innovations that stimulate a sense of urgency. They give limited attention to the majority of issues that remain low priorities in the state political agenda. When such disproportionate information processing is common across state governments, it can trigger a positive feedback cycle, leading to abrupt policy diffusion across states. The key insight here is not that policy diffusion occurs at varying rates and to varying extents, but rather that distinct patterns of diffusion are triggered by the allocation of political attention in the federal system. Researchers have studied how the allocation of political attention contributes to policy stability and change across a range of American political institutions (B. Jones and Baumgartner 2005). This book connects

172

Policy Diffusion Dynamics in America

diffusion research to this larger tradition of agenda setting and decision making. To evaluate models of decision making in public-policy diffusion, the research presented here employed simple distributional analysis to compare how well empirical diffusion data matched a simulated incremental learning process. This technique made use of a common and persistent expectation that the patterns of diffusion follow a normal S-shaped cumulative distribution curve as units in the social system evaluate and adopt a new innovation over time (Ryan and Gross 1943; Gray 1973; Rogers 1983; Mooney and Lee 1999). Diffusion theory therefore provides a strong a priori expectation that diffusion data will be normally distributed and symmetrical, or at the very least that diffusion data will approximate a normal curve.2 Chapter 2 instead demonstrates that the standard indicator of incrementalism (a simple statistical measure of normality) was actually uncommon among the 133 policy innovations reviewed, indicating that decision making in policy diffusion generally does not match the expectations of policy incrementalism. Policy diffusion is generally characterized by considerable variation across measures of kurtosis, indicating that the diffusion of innovations is often punctuated by bursts of rapid state policy adoptions, and that over time, innovations spread more episodically than traditionally appreciated. Incrementalism may be an appealing and useful framework for describing some of the pressures leading to policy innovation and diffusion, but it leaves an incomplete understanding of the complexity of innovation and diffusion in America. These findings may seem unremarkable to those who have observed the rapid diffusion of select innovations over the past half-century. Political scientists have documented the uncharacteristically rapid diffusion of innovations like the death penalty, child abuse reporting requirements, gay marriage bans, three-strikes laws, or anti-lemon auto legislation. These cases have suggested a number of factors contributing to the rapid diffusion of innovations – for example, that morality policy will trigger faster innovation diffusion, or that the presence of organized opposition slows the rate of adoption across states. Unfortunately, the body of research on diffusion anomalies proposes no method for assessing the 2

It is worth noting that this methodology is itself a throwback to an earlier era of diffusion research. Although the use of distributional analysis to test models of decision making owes an immediate debt to Baumgartner and Jones’s research on punctuated equilibrium theory, it has older roots in diffusion research. Ryan and Gross (1943) made comparisons of normal simulated diffusion data to empirical data in their study of the diffusion of hybrid corn to draw inferences about the process of learning.

Conclusion

173

frequency of policy outbreaks, and no theory for evaluating what such rapid patterns of diffusion imply for our understanding of policy diffusion or policy making in federations. Comparing patterns of diffusion across historical eras and policy types presents a more nuanced picture of diffusion dynamics in America. Policy diffusion may be generally described as nonincremental, but this does not mean that innovations produce uniformly nonincremental patterns of policy change. Chapters 2 and 3 reveal considerable variation across the kurtosis and normality scores for different groups of innovations. For example, the emergence of mass media and communications technology has acted as an important trigger for positive feedback cycles in policy diffusion. Earlier policies diffused more slowly and most closely approximated the simulated incremental learning curve. With each successive generation, policy diffusion deviated more sharply from an incremental diffusion process. This finding confirms what political scientists have long expected. The speed of innovation diffusion is in no small part dictated by the speed of awareness of innovations. Of course, exposure to innovation is not the only determinant of policy adoption. Chapter 3 established that the characteristics of the innovations are themselves important determinants of diffusion dynamics. Policies targeting children are especially likely to trigger positive feedback cycles, whereas policies licensing professional associations appear much less prone to rapid innovation diffusion. More generally, variations in the cost, complexity, fragility, and salience of policies all contribute to the pattern of innovation diffusion. The implication of this distributional analysis is that public-policy diffusion represents a more dynamic process than has been previously appreciated. Policies may begin to spread gradually but then diffuse more rapidly in later generations of adopters. Diffusion can be episodic, spreading in regional or localized bursts over time, with appreciable periods of inactivity in between. At the extremes, policies may diffuse smoothly and gradually or in one sudden policy outbreak across states. Diffusion dynamics are the result of complex relationships, as an emerging innovation interacts with interest groups, receptive states, and the larger political environment to produce patterns of adoption over time. Modeling the Contagion of Innovation and the Dynamics of Diffusion To understand why innovations so frequently trigger nonincremental policy change across states, the second part of this book integrated research from political science and epidemiology to conceptualize how variation

174

Policy Diffusion Dynamics in America

among the agents, hosts, and carriers of public-policy innovation shape diffusion dynamics. Although this conceptual framework was inspired by epidemiology, the specific mechanisms of diffusion evaluated in this research fit neatly with agenda setting and public-policy process theory. This research explored how variation in the characteristics of innovations, the political/institutional attributes of states, and the organization of interest-group vectors worked to produce responses to policy innovation in the federal system. The comparative orientation of this work provided a good basis for modeling diffusion dynamics. First, it allowed isolation and testing of a series of hypotheses regarding how changes in the types of innovations, state political institutions, and interest-group carriers could shape publicpolicy diffusion. Second, it provided a larger theoretical context for analyzing how the key elements of innovation diffusion interact to produce policy change across states. Just as a disease spreads through a population through a combination of its own characteristics, the receptivity of hosts, the behavior of carriers, and the speed and extent of public-policy innovations are shaped by the interaction of receptive states, individual policies, and the unique attributes of interest-group carriers. These interactions help explain the tendency for innovations to deviate from incrementalism, and can be used to explain why both gradual incrementalism and sudden policy outbreaks remain uncommon. Together, the epidemiologic approach recognizes that it takes an entire system, not just a single factor, for policy diffusion of either pattern to occur. A change in any of the elements of this system can cause an abrupt positive feedback cycle, and any break in the chain can cause a policy diffusion to stop short. The variation in kurtosis scores and marked deviation from normality in Chapters 2 and 3 can be explained as extending from a series of interactions between innovations, interest groups, and states – some resulting in rapid policy outbreak across all states, some triggering localized outbreak across a subset of susceptible states, and some encouraging a gradual, propagated outbreak as one state transmits an innovation to another. Agents and Diffusion Dynamics A first implication to extend from the epidemiologic framework is that policy innovations possess distinct attributes that make them more or less prone to triggering positive feedback cycles of policy adoptions across states. Innovations vary according to issue complexity, salience,

Conclusion

175

cost, fragility, and target. Each of these attributes shapes the way state decision makers and mass publics process policy information when evaluating and responding to a new innovation. Changes in issue salience influence diffusion dynamics by shifting the way elected officials and attending publics prioritize information. Variation in issue complexity and cost determine the potential difficulty of policy analysis and implementation. The fragility and targets of innovations impact diffusion by shifting perceptions of the organized opposition to implementing a new policy. These characteristics are critical for understanding how state governments behave as they encounter and evaluate new innovations. Some policies require technocratic analysis of the expected costs and benefits of policy implementation, whereas other policies require minimal expertise and have seemingly obvious costs and benefits to politicians and mass publics. Chapter 3 traced patterns of diffusion for three classes of policy with different levels of salience, cost, and complexity: governance, morality, and regulatory policies. Both governance and morality policy are marked by high levels of issue salience and diminished levels of complexity. State regulatory policies are typically described by higher issue complexity, higher cost, higher fragility, and lower salience. Positive feedback cycles were most pronounced in the diffusion of governance policy, which is characterized by low cost and complexity, diminished levels of issue fragility, and high mass involvement in policy making through direct democracy. Although regulatory policies were marked by non-normality, they also had lower kurtosis scores than governance policy, suggesting that systemic responses to these policy forms are more uniform among the 50 state governments as they debate and evaluate innovations. These findings suggest that diffusion represents a complex interaction between susceptible states and innovations. It is not simply that policies with distinct characteristics spread at different rates, but more importantly that these patterns emerge because innovations engage very different decision-making processes by state governments. Because state regulatory policies often require sophisticated cost-benefit analysis, these policies may emerge first in states with the decision-making bandwidth to engage in and technically evaluate regulatory policies. States with less professional expertise may desire the benefits of a new regulatory policy, but depend on innovative states to first craft and modify these innovations. When these less expert states do adopt the policy, it may happen more rapidly and at later stages than a simple process of incremental policy learning would lead us to expect.

176

Policy Diffusion Dynamics in America

Perhaps more interesting are the ways that both morality policy and governance policy interact with susceptible states. For morality policy, an innovation may well trigger a series of adoptions rapidly across states. However, social regulatory preferences differ dramatically across the federation. A mandatory waiting period for abortion may immediately appeal to some states but face staunch resistance in others. For governance policy, the opposition arrives not from liberal or conservative states, but from elected governments themselves. Governance reforms are frequently adopted in states with initiative processes, but fail to gain traction in states without the tradition of direct democracy. The diffusion of both governance and morality policy suggests that a mixture of highly susceptible and highly resistant states produces patterns of diffusion that are neither smooth and gradual, nor sudden and extensive. Instead, diffusion patterns are interrupted by the process of decision making in both receptive and resistant states. Host Characteristics and State Receptivity to Innovation Chapter 4 complemented these findings by measuring state receptivity to morality, governance, and regulatory policy. Rather than being uniformly responsive to each policy type, the speed of state policy adoption of morality, governance, and regulatory policy is related to a state’s capacity to engage in different types of information processing. To measure a state’s generic receptivity to innovations, this chapter returned to an older method of diffusion research, constructing a standardized measure for the general speed with which a state adopts innovations. This research cast new insights on state “innovativeness” by showing that state receptivity to innovation is not consistent across all types of policy, but instead that receptivity shifts in response to the unique makeup of each state. Chapter 4 established how the internal political, institutional, ideological, and sociodemographic characteristics of states shape their receptivity to policy innovations. States with the policy-making institution of direct democracy are most receptive to governance policy reform, as the tradition of direct democracy permits states to quickly adopt innovative regulations on the behavior of government and elected officials. State receptivity to morality policy is shaped by the electoral connection between elected government and mass opinion. States with high levels of electoral competition proved especially receptive to morality innovation. Finally, state responsiveness to regulatory policy is related to levels of legislative professionalism in state governments, as the speed with which

Conclusion

177

a state is able to react to emerging regulatory innovation is shaped by the capacity of a state to engage in complex policy analysis required for regulatory design and implementation. A key insight here is that there are multiple determinants of state receptivity to innovation. Just because a state is susceptible to one form of innovation does not imply that it is susceptible to another. In fact, the factors that make a state receptive to one innovation are often completely unrelated to those that make states receptive to another. This is highlighted in the interesting relationship that emerges between state legislative professionalism and receptivity to morality and regulatory policy. States with more professionalized legislatures appear to be generally more susceptible to regulatory policy innovations, but are modestly more resistant to morality policy innovation. The opposite dynamic emerges in states with citizen legislatures, which respond more rapidly to morality policies than to economic, environmental, and professional regulatory policies. A second implication is that the number of susceptible states dictates the speed and scope of policy diffusion. There is clearly a ceiling effect, as certain states are immune or resistant to types of innovation. Consider the example of legislative term limits, which were adopted rapidly by a number of states in the early 1990s. States with the initiative process provided a tool for citizens to vote directly on the popular legislative reform. However, those states without the initiative process were forced to rely on their legislatures, and only in Louisiana did politicians choose to limit their own terms in office. Policies therefore spread rapidly through a susceptible fraction of states but then stop abruptly. A similar dynamic can occur with morality policies, in which several states will move to embrace a social regulatory restriction, whereas other states, perhaps because of strong religious or historical traditions in the electorate, will be far more resistant to these innovations. Policy Vectors and Diffusion Dynamics A final implication of the epidemiologic model is that the carriers of innovation are important in determining the speed and scope of policy innovation. Chapter 5 considered how variation in the organization and behavior of interest-group carriers of innovation shapes diffusion dynamics. This chapter demonstrated that important differences in the organizational structure and the strategic choices of interest groups have a profound impact on the speed and extent of public-policy diffusion.

178

Policy Diffusion Dynamics in America

Wealthy and centrally organized interest groups possess the resources and mass membership needed to coordinate a rapid response following changes in mass political attention. These interest groups are equipped to mobilize a diffusion campaign in order to simultaneously affect policy change across the states, as they possess the hierarchical structure needed to develop a strategic diffusion campaign, and the financial and human resources needed to concurrently pressure for policy adoption across multiple state governments. The choices and behaviors of smaller, less-organized interest groups also contribute to diffusion dynamics. Interest groups engage in issue framing and venue shopping to shift public support for innovations. When demands for innovation are met with opposition, interest-group activists experiment with the framing and reframing of policy proposals in order to expand mass political support for innovations. Likewise, interest groups engage in venue shopping, pressuring for political change and multiple venues of government in the hopes of elevating issue attention and legitimizing a policy reform. On rare occasions, these strategic choices work to elevate wide political attention across the federation, legitimizing a new policy proposal and leading to an increase in public demands for policy change. Although interest groups generally affect change across the states gradually, in these rare cases, interest-group strategies precipitate a positive feedback cycle. Again, there are several key implications for how interest-group carriers interact with states and innovations to shape diffusion dynamics. First and most obvious is that the size, resources, and organization of interest groups has clear implications for the speed, scope, and extent of diffusion. Smaller and poorer interest groups may need to be initially selective in which jurisdiction they pursue policy change, whereas larger and wealthier groups can push more expansively and immediately for interstate policy adoption. The process of policy diffusion is not simply determined by an interaction of susceptible states and innovations, but also by the capacity of interest-group carriers of innovation to pressure for policy change across the country. The way that interest groups can shape the tenor and dimensions of a policy over time presents another insight into the causes of diffusion dynamics. As the case study on the prohibition movement in Chapter 5 demonstrates, interest groups are extremely active in framing and reframing innovations as they appeal for policy change. In many cases, interest groups alter the fundamental dimensions of a policy innovation, shifting

Conclusion

179

the dominant attributes of a policy over time. Although the prohibition of alcohol is generally perceived as a morality policy reform, prohibitionists at various periods packaged alcohol reform as economic regulatory policy, linking alcohol consumption to lower worker productivity or governance policy, arguing that alcohol regulation should be a choice granted to local governments. These efforts to reshape the policy debate were intended to make prohibition more palatable to states that had previously been resistant. When a shift in the tone or frame of policy succeeds, it can spread over a set of states that are newly susceptible to the innovation. Tracing the Anatomy of an Outbreak: Interactions and Diffusion Dynamics Up to this point, the triggers of diffusion dynamics have largely been treated in isolation. Independently, changes in the attributes of the innovation, the sociopolitical makeup of the state, and the organization and relative strengths of interest groups can dramatically shape diffusion dynamics, triggering a positive feedback cycle and escalating the rate of innovation diffusion, or diminishing the rate of diffusion – in either case by requiring extended and specialized policy analysis. Policy diffusion is rarely as smooth as theory anticipates, in large part because the factors shaping innovation diffusion rarely interact to overcome the considerable friction-slowing rates of policy change as 50 state governments consider an innovation. One final insight from epidemiology is that it takes interaction of the four general factors associated with diffusion to trigger an epidemic. Domestic and international watchdogs of public health (organizations such as the Centers for Disease Control and Prevention or the World Health Organization) monitor a broad range of factors as they attempt to anticipate and respond to epidemics. Rapid and extensive outbreaks occur in those rare instances when the hosts, carriers, agents, and environments intersect to create conditions for the rapid proliferation of disease throughout a population. Policy outbreaks are likewise triggered by the rare interaction of environment, policy innovation, carriers, and hosts. A change in any single dimension may disrupt a gradual process of incrementalism, but is generally insufficient to trigger a positive feedback cycle leading to the nearly immediate adoption of innovation across states. It is the occasional alignment of rising issue salience, a widely appealing innovation,

180

Policy Diffusion Dynamics in America

well-organized interest groups, and a large number of susceptible states that can produce periods of extremely rapid policy diffusion. In this regard, the Amber Alert represented something of a perfect storm for rapid policy diffusion – a cheap, appealing innovation that emerged during a period when Americans were disproportionately concerned with anti-crime legislation, and which required interstate coordination to be most effective. The Amber Alert was sponsored by child welfare and anti-crime advocates, who faced little if any opposition to the policy and had learned how to effectively pressure for interstate policy adoption from a series of previous policy successes. It was the intersection of each of these facets – policy environment, policy characteristic, carrier, and susceptible hosts – that triggered the outbreak of the Amber Alert. Conversely, two anti-crime policies introduced within the same period – the three-strikes law and needle-exchange programs – produced far different patterns of diffusion. The three-strikes law appealed to a large subset of states, but stopped abruptly. Although the policy represented a simple and relatively easy to understand prescription for addressing crime control, benefited from a period of widespread public interest in anti-crime policy, and had a strong set of interests championing the reform across states, it stopped short of full diffusion because a significant subset of states determined that the anticipated costs outweighed the benefits. In an even more extreme case, innovations like needle-exchange laws have faced considerable opposition to diffusion. Although these policies were also introduced in a period when Americans were concerned with crime, they involved relatively complex arguments and did not have an immediately appealing policy frame. These policies have spread very gradually since their introduction. These factors associated with the rapid and extensive diffusion of innovations are nearly identical to the triggers of sudden policy change described in the classic studies of agenda setting. Kingdon (1984) argued that sudden policy change occurred when the three streams of the policy process (problem, politics, and policy) converge to open a window of opportunity. Likewise, Baumgartner and Jones (1993) observed that policy punctuations were triggered when a focusing event allowed a fundamental shift in the way that decision makers described and understood a policy problem. Rapid and extensive diffusion similarly occurs when an event causes an innovation to rise to prominence across states, changing state susceptibility to the given innovation, and leading to widespread policy imitation across the states.

Conclusion

181

Old Approaches and New Questions in Diffusion Research Although this research provides insight into the causes of diffusion dynamics, it also presents a number of intriguing directions for future research on the comparative contagion of public-policy innovations. The study of innovation diffusion is complex, and the attributes identified in the discussions of the nature of the policy idea, the characteristics of states, and the capacity of interest groups do not exhaustively explain the sources of diffusion dynamics. Certainly, other attributes beyond issue salience, cost, complexity, and fragility shape the contagion of innovation. Likewise, research exploring the diffusion of policy types beyond morality, governance, and regulatory policy can only improve our understanding of how different innovation attributes shape diffusion dynamics. In each of the core components of the model, new questions have emerged that may provide a more comprehensive understanding of the process of policy innovation and diffusion in America. The next section considers future directions in comparative diffusion research. First and foremost, future research can cast light on how the organization and behavior of interest groups drive the diffusion of innovations. Chapter 5 presented a series of illustrative case studies to demonstrate how differences in the organization, resources, and structure of interest groups could explain patterns of policy diffusion across the states. An even greater understanding of the role interest groups play as carriers of policy innovation could be gained from a systematic analysis of interest-group characteristics organized around each of the interest-group traits identified in Chapter 5. Indicators of interest-group resources could be collected through the endowments and annual operating budgets of advocacy associations, the size and membership of associations, and the degree of the number of professional paid staff members in federal and state chapters. To measure the organization and structure of interest groups, researchers could count the number and size of state-level chapters, and evaluate the strength of central and state interest-group representation in Washington, D.C., and state capitols. Such data on the membership, resources, and organization of interest groups would allow for a much more rigorous understanding of how the relative resources and makeup of interest-group organizations contributes to the diffusion of innovations. The interesting insight here may come less from research linking overall resources to the speed and scope of innovation diffusion, but rather on how the organization and hierarchy

182

Policy Diffusion Dynamics in America

of interest groups facilitate diffusion campaigns. Venue-shopping theory suggests that interest groups have specialized in how they pressure for policy change across the myriad jurisdictions in America. It would be interesting to test whether interest groups that have adopted a specifically federated organizational form – choosing to operate at the national, state, and municipal levels of government – are better equipped to sponsor innovation diffusion campaigns. A second question of interest revolves around how interest groups evaluate and draw lessons from their strategic choices. Political scientists have speculated that the selection of venues and issue frames is often haphazard, but have overlooked how interest groups develop their rhetorical strategies, select venues, and evaluate and act upon policy successes during diffusion campaigns. Research into the development of framing and venue-shopping strategies could take the form of a detailed case study approach, perhaps involving interviews with leaders among state and national interest-group activists, or could make use of mass media communications in the news and on the Internet, as well as expert testimony in state legislatures to trace changes in the tone and content of interest-group-sponsored legislation. A more detailed understanding of the factors leading activists to adopt or modify strategies as they work to influence state legislative decisions would provide valuable insights into how interest-group activists read and interact with changes in the larger political environment during issue campaigns. Similarly broad questions emerge from the idea that innovation characteristics shape patterns of diffusion. As Karch (2007b) observes, policy diffusion research has largely overlooked how the characteristics of innovations themselves shape the process of innovation and diffusion. This research proposed that policy salience, cost, complexity, fragility, and the target of innovations shaped patterns of diffusion. These are important characteristics, but they are far from exhaustive. For example, Mossberger (2000) proposed the notion of policy flexibility, arguing that the relative ability of state governments to shape policy to fit their local preferences and needs is an important determinant of diffusion. Glick and Hays (1991) have noted how policies change in scope and content over time as innovations evolve in response to arguments and information that emerge over a diffusion cycle. Understanding how each characteristic shapes the speed and scope of diffusion will provide an even better insight into how states evaluate and adopt policies with distinct characteristics. More importantly, the study of diffusion could be advanced with a much more expansive study of how the target of innovations shaped

Conclusion

183

comparative patterns of diffusion. As A. Schneider and Ingram (1993) argue in “The Social Construction of Target Populations,” the way that society perceives the target of policy intervention shapes political conflicts that emerge around the policy. A comparison of innovations addressing a common issue but targeting very different populations could show how targeted populations shape patterns of diffusion in America. This body of research would not only extend our understanding of how innovation characteristics shape patterns of diffusion but would provide a valuable test of the generalizability of social constructivist theory in public-policy research. If publics do, in fact, respond differently to innovations depending on how they confer benefits or burdens to targeted groups, then this should emerge in patterns of diffusion. Policies that meet with the expectations of social constructivist theory – those that proscribe benefits to positively constructed groups or policy burdens to negatively constructed groups – should diffuse more rapidly and extensively than policies that challenge these expectations. Future research should also explore how the characteristics of innovations evolve over the course of diffusion. This research largely treats the characteristics of innovation as static, however, as the case studies in Chapter 5 demonstrate, this is a simplification of how publics respond to policies as they spread across the states. Policies change in scope, complexity, and target as they spread from one state to another. Research documenting how changes in public perception of dominant policy dimensions alter state responses to innovation could provide an even more precise understanding of why incremental diffusion processes can become short-circuited in response to changes in the innovation. Finally, research linking the characteristics of innovation to diffusion dynamics could address the common complaint that studies of policy diffusion are of limited generalizability because they do not examine the large (and largely unknown) set of innovations that fail to diffuse across states. If cost, complexity, fragility, and salience are associated with the speed and scope of policy diffusion, then researchers should be able to use these characteristics to model the nondiffusion of innovations. For example, the Oregon Health Plan emerged as a promising approach for states to improve access to medical care, but was ultimately too complex and controversial to gain traction across state governments. California’s Proposition 187 – prohibiting undocumented immigrants from access to state social services – certainly appealed to a large set of voters across the Southwest, but was ultimately too controversial (and not incidentally, was ruled unconstitutional). Examining how policy characteristics explain the

184

Policy Diffusion Dynamics in America

nondiffusion of innovations will provide additional perspective on the triggers of diffusion itself. A third set of questions emerges from research on the relative receptivity and resistance of states to innovation. In this regard, basic improvements can provide additional understanding of how state characteristics shape diffusion dynamics. First, researchers could continue to identify correlates of state receptivity to different classes of innovation. Many of the measures employed in this research were imperfect proxies for complex features of state sociopolitical characteristics. Better measures of ideology and political culture would provide an even better test of theory. Adding additional measures of state capacity to engage in political decision making – including measures of state bureaucratic expertise – or integrating study on state legal systems into the research could provide additional insight into how the relative differences of states shape receptivity to innovation. A final implication of the epidemiologic model is that political environment shapes patterns of diffusion. This book largely folded the question of environment into the discussion of innovation characteristics, state characteristics, and interest-group activity. An important direction of future research would be to tease out the independent effects of the public’s political awareness on the comparative diffusion of innovations. Although beyond the immediate scope of this research, students could link the ebbs and flows of the focus of public political interest to patterns of diffusion, exploring whether the national or state level of political attention on the subject area of an innovation accelerated its diffusion. A related study could link media agenda-setting effects to diffusion dynamics, connecting local and national media coverage to state policy adoption. These approaches have proved difficult but would nonetheless provide more compelling and direct insight into how the allocation of mass political attention shapes innovation diffusion. A second breakthrough could emerge in expanding research in diffusion to document evolution of an innovation across the decision agendas of state governments. The overwhelming majority of diffusion research narrowly explores the timing of state policy adoptions. By focusing on a single rough indicator of whether a state has adopted an innovation at a given date, this orientation overlooks whether an innovation was formally considered and rejected by state governments and provides no insight into the intensity (number of hearings, votes, etc.) a given innovation engendered across states. By expanding the scope of diffusion research to more realistically capture the agenda-setting dynamic, researchers will be able

Conclusion

185

to provide greater insight into the dynamics of attention and the larger factors leading to state policy adoption that are of central interest to diffusion researchers. Conclusions Although this research addresses questions that have long stimulated the curiosity of students of policy making in the United States, it provides new insights into the processes driving the diffusion of innovation from one state to another. This book developed a policy-process theory for understanding the diffusion of innovations based on recent developments in the study of policy dynamics. It has introduced a model for understanding the contagion of policy innovations, demonstrating how variations in the agents, hosts, and carriers of innovation shape the movement of public policy across states. This book applied the epidemiologic framework to develop a theoretical model explaining the causes of diffusion dynamics. There are clearly limitations to the comparison between epidemiology and public-policy diffusion. Political decision making and policy diffusion are shaped by social choices. Interest groups can freely choose to embark on a campaign for policy diffusion, and interest-group activists can attempt to shape the dimensions of innovations through policy targeting and issue framing. State receptivity to innovation can shift suddenly, either in response to exogenous events or because of changes in the composition and makeup of governments. Even the characteristics of policy evolve, and in so doing can be viewed differently by mass publics from one moment to the next. Yet despite these differences, the diffusion of innovations, like the spread of an infectious disease, a new product, or a social behavior – is shaped by a series of common factors. In identifying how differences across the agents, host, and carriers shape patterns of diffusion, this research provides a foundation for thinking about the process of innovation diffusion in America. As is the case with the approach taken in this book, future advances in the study of public-policy diffusion will be driven by comparative research into the causes of diffusion dynamics. Only an explicitly comparative approach will permit us to understand the processes driving both the gradual, incremental diffusion of innovations and sudden, sweeping moments of policy change. Building upon this approach will permit students of public policy to develop a theory connecting the behavioral model of political choice to the political and institutional factors driving the diffusion of innovations across the laboratories of democracy.

Appendix A List of Innovations Collected

Innovations that served as the basis for this study were collected using the following protocol: To ensure a balanced representation of state public policies by historical era, policy type, and speed of diffusion, this research followed the sampling procedures proposed by Walker (1969) and Savage (1978), sampling from a discrete list of state issue areas representing welfare, health and public safety, crime and corrections, taxes, licensing and professional regulation, education, elections, sexuality, state economic development, and environmental policy. Following Walker’s definition of innovation as “a program or policy which is new to the state adopting it, no matter how old the program may be or how many other states have adopted it,”(1969, 881),1 this research included only innovations that were formally enacted into law by state governmental institutions. In keeping with common selection criteria in innovation and diffusion research, the scope of innovation adoption was considered for identifying diffusing innovations (Walker 1969; Savage 1978; Canon and Baum 1981). Only those innovations adopted by at least 10 states before 2007 were included for analysis.2 Below are two lists of innovations that served as the basis for this study. The first list represents innovations provided from data compiled 1

2

There are two notable exceptions included in the study. Given their prominence in the policy diffusion literature, both state lottery programs and the readoption of the death penalty are included as innovations for study, even though states had previously experimented with both throughout U.S. history. The scope criteria used in this chapter is more lenient than the approach taken by Walker (1969) who required a policy be implemented in at least 20 states, or Canon and Baum, who included innovations in only 18 states.

187

188

Appendix A: List of Innovations Collected

by Jack Walker for his study “The Diffusion of Public Policy Innovation Among the American States” (1972). These data are made available for download by the Inter-University Consortium for Political and Social Research (ICPSR) and can be accessed on the Internet by requesting ICPSR Study No: 66 on the ICPSR website (http://www.icpsr.umich. edu/) or through other correspondence.3 The second list of innovations was collected directly by the author from keyword searches of state legislative statutes and state news agencies, as well as from information provided by interest groups, professional organizations, and governmental agencies. Innovations were also collected from a number of academic studies of state policy innovation using keyword searches in JSTOR and Expanded Academic Index. The sources for data collection are indicated with text citations. Innovations that are followed by an asterisk symbol indicate that state years of adoption were constructed or updated from multiple sources by the author. Policies Compiled by Walker Equal Pay for Females Controlled Access Highways Soil Conservation Districts – Enabling Fish Agency Conservation of Gas and Oil Air Pollution Control Forest Agency Normal Schools – Enabling Act School for the Deaf Superintendent of Public Instruction Workmen’s Compensation Utility Regulation Commission Compulsory School Attendance Probation Law Real Estate Brokers Licensing Integrated Bar Slaughterhouse Inspection Tax Commission 3

The full citation for this study is as follows: Walker, Jack L. 1972. “Diffusion of Public Policy Innovation Among the American States.” [Computer file]. Compiled by Jack L. Walker, University of Michigan, Institute of Public Policy Studies. ICPSR, ed. Ann Arbor, MI: Inter-University Consortium for Political and Social Research [producer and distributor].

Appendix A: List of Innovations Collected Seasonal Agricultural Labor Standards Agricultural Experiment Stations Fair Trade Laws Highway Agency State Program Accountants Licensing Gasoline Tax Welfare Agency Park System Nurses Licensing Merit System Engineers Licensing Board of Health Budgeting Standards State Police or Highway Patrol Pharmacists Licensing Automobile Registration Beauticians Licensing Architects Licensing Australian Ballot System Anti-Age Discrimination Anti-Injunction Laws Juveniles Supervision Compact Legislative Pre-Planning Agencies Dentists Licensing Migratory Labor Committee Mental Health Standards Committee Fair Housing – Private Housing Fair Housing – Public Housing Fair Housing – Urban Renewal Areas Human Relations Commission Automobile Safety Compact Municipal Home Rule Junior College – Enabling Legislation Labor Agencies Committee on the Aged Teacher Certification – Secondary Teacher Certification – Elementary Legislative Research Agencies Retirement System for State Employees

189

190

Appendix A: List of Innovations Collected

Educational Television Parking Agencies – Enabling Act for Cities Council on the Arts Library Extension System Zoning in Cities – Enabling Legislation Public Housing – Enabling Legislation State Planning Board Aid to Dependent Children (Social Security) Initiative and Referendum Direct Primary Advertising Commissions Aid to the Blind (Social Security) Minimum Wage Law Debt Limitation Urban Renewal – Enabling Legislation Planning and Development Agencies Cigarette Tax Chiropractors Licensing Child Labor Standards Retainer Agreement Reciprocal Support Law Parolees and Probationers Supervision Old Age Assistance (Social Security) Aid for Roads and Highways Alcoholic Beverage Control Court Administrators Alcoholic Treatment Agencies Aid to Permanently and Totally Disabled Policies Compiled by the Author Child-Abuse Reporting Requirement (Hays 1996) Crime Victims’ Compensation Fund (Hays 1996) Prohibition of Alcohol4 Ban Recognition of Out-of-State Same-Sex Marriage (Eskridge 1999)

4

Pre-Eighteenth Amendment legislation was provided through correspondence with the librarian at the Anti-Saloon League Museum at the Westerville Library. For information on the Westerville collection, see http://www.westervillelibrary.org/AntiSaloon/; accessed August 21, 2007.

Appendix A: List of Innovations Collected

191

Smoking Ban in Government Buildings (Volden and Shipan 2006) Bans of Out-of-Package Cigarette Sales (Volden and Shipan 2006) Restaurant Smoking Ban (Volden and Shipan 2006) Credit Security Breach Notification (State PIRG 2006) DUI .08 Per Se Legislation (National Conference of State Legislatures 2004) Amber Alert5 Mandatory Bicycle Helmets for Children (Bicycle Helmet Safety Institute) Child Access Prevention Laws Requiring Gun Locks (Brady Campaign) Charter Schools Enabling Legislation (Rincke 2004)6 Death Penalty Reenactment (Mooney and Lee 1999) State Legislative Term Limits (U.S. Term Limits) Living Will Legislation (Glick and Hays 1991) Lottery Enabling (F. Berry and Berry 1999) Medical Marijuana (Drug Policy Institute) Medical Savings Accounts (Bowen 2005) Mandatory Motorcycle Helmet (Insurance Institute for Highway Safety 2007) Parental Notification/Consent for Abortion (Center for Reproductive Rights) Primary Seat Belt Laws – Ticket for Not Wearing Seatbelt. (National Highway Traffic Safety Administration) Seat Belt Laws (Insurance Institute for Highway Safety) Security Credit Card Freeze Laws (U.S. PIRG) State Supermajority Tax Rule (National Conference of State Legislatures 2005) Victims’ Rights Amendments (National Victims’ Constitutional Amendment Passage) Tax and Expenditure Limit (National Conference of State Legislatures 2005) Three-Strikes Sentencing (PBS POV) No-Fault Divorce (Nakonezny, Shull, and Rogers 1995) Statutory Rape Legislation (Cocca 2002) Sodomy Laws – Repeal (Disarro 2005) 5

6

Information on state Amber Alert Plans were collected through correspondence with the National Center for Missing and Exploited Children, and by keyword Internet searches for specific state Amber Plans. Rincke’s data were updated and cross-referenced from information linked at http://www .uscharterschools.org/pub/uscs docs/sp/index.htm; accessed February 14, 2006.

192

Appendix A: List of Innovations Collected

Hate Crimes – Include Protections for Homosexuals (Disarro 2005) High School Exit Exams7 State Auto Lemon Laws (Savage 1985a) Mandatory Child Passenger Restraints (Savage 1985a) Guaranteed Renewal – Health Insurance (Stream 1999) Portability – Health Insurance (Stream 1999) Guaranteed Issue – Health Insurance (Stream 1999) Preexisting Condition Limits – Health care (Stream 1999) Eminent Domain Restrictions (National Conference of State Legislatures 2006) Paper Terrorism Lien Legislation (Chamberlain and Haider-Markel 2000). Identity Theft Legislation∗ (GAO 2002) Post-Conviction DNA Access for Exoneration (National Conference of State Legislatures 2003) Prescription Drug Registrations (GAO 2002; National Alliance for Model State Drug Laws) Anti-Stalking Laws (National Institute of Justice 1996) Legalization of Syringes for IV Drug Addicts (Burris, Vernick, Ditzler, and Strathdee 2002) Adoption of Direct Citizen Initiative (Cronin 1989) 7

Ashley Watson generously shared state years of adoption for state exit exams.

Appendix B Policies Collected by Historical Era

This research grouped the innovations listed in Appendix A into four different historical eras: Pre-Twentieth Century, Early Twentieth Century (1900–1929), Mid-Twentieth Century (1930–1959), and Late Twentieth and Early Twenty-First Centuries (1960–2006). To classify policies by historical era, this research employed the procedure suggested by Walker (1969), placing a policy in the period when the tenth state adopted the innovation. For example, if the first nine states adopted an innovation in the 1920s, but the tenth state implemented a policy in 1931, the policy was included in the groups of policies for the mid-twentieth century. To confirm the reliability of this sorting procedure, this research performed a secondary sorting of policies by first taking the mean year of adoption for all states for a given innovation, and then subtracting one standard deviation from this mean year to find the period of early adoption. This procedure produced a grouping of policies by era almost identical to the protocol suggested by Walker. This process identified 17 policies that diffused prior to 1900, 28 policies in the 1900–1929 period, 35 policies in the 1930–1959 period, and 52 policies in the 1960–2006 period. Policies in each of these groupings are listed below. Early Twentieth Century: 1900–1929 Conservation of Gas and Oil Workmen’s Compensation Real Estate Brokers Licensing Slaughterhouse Inspection Tax Commission 193

194

Appendix B: Policies Collected by Historical Era

Highway Agency Accountants Licensing Gasoline Tax Park System Nurses Licensing Engineers Licensing Budgeting Standards State Police or Highway Patrol Automobile Registration Beauticians Licensing Architects Licensing Anti-Injunction Laws Municipal Home Rule Junior College – Enabling Legislation Teacher Certification – Secondary Legislative Research Agencies Zoning in Cities – Enabling Legislation Direct Primary Cigarette Tax Chiropractors Licensing Child Labor Standards Aid for Roads and Highways Direct Citizen Initiatives Mid-Twentieth Century: 1930–1959 Probation Laws Integrated Bar Equal Pay for Females Controlled Access Highways Soil Conservation Districts – Enabling Legislation Seasonal Agricultural Labor Standards Fair Trade Laws Merit System Anti-Age Discrimination Juveniles Supervision Compact Legislative Pre-Planning Agencies Migratory Labor Committee Mental Health Standards Committee Fair Housing – Public Housing

Appendix B: Policies Collected by Historical Era Human Relations Commission Automobile Safety Compact Committee on the Aged Teacher Certification – Elementary Retirement System for State Employees Parking Agencies – Enabling Act for Cities Public Housing – Enabling Legislation State Planning Board Aid to Dependent Children (Social Security) Advertising Commissions Aid to the Blind (Social Security) Minimum Wage Law Urban Renewal – Enabling Legislation Planning and Development Agencies Reciprocal Support Law Parolees and Probationers Supervision Old Age Assistance (Social Security) Alcoholic Beverage Control Court Administrators Alcoholic Treatment Agencies Aid to Permanently and Totally Disabled Late Twentieth and Early Twenty-First Centuries: 1960–2006 Air Pollution Control Fair Housing – Private Housing Automobile Safety Compact Educational Television Council on the Arts Retainers Agreement Mandatory Motorcycle Helmet Child-Abuse Reporting Requirement Crime Victims’ Compensation Fund Ban Recognition of Out-of-State Same-Sex Marriage Smoking Ban in Government Buildings Bans of Out-of-Package Cigarette Sales Restaurant Smoking Ban Credit Security Breach Notification DUI. 08 Per Se Legislation Amber Alert

195

196

Appendix B: Policies Collected by Historical Era

Mandatory Bicycle Helmets for Children Child Access Prevention Laws Requiring Gun Locks Charter Schools Enabling Legislation Death Penalty Reenactment State Legislative Term Limits Living Will Legislation Lottery Enabling Medical Marijuana Medical Savings Accounts Mandatory Motorcycle Helmet Parental Involvement Abortion Primary Seat Belt Laws – Ticket for Not Wearing Seatbelt Seat Belt Laws Security Credit Card Freeze Law State Supermajority Tax Rule Victims’ Rights Amendments Tax and Expenditure Limits Three-Strikes Sentencing No-Fault Divorce Statutory Rape Legislation Sodomy Laws – Repeal Hate Crimes – Include Protections for Homosexuals High School Exit Exams State Auto Lemon Laws Mandatory Child Passenger Restraints Guaranteed Renewal – Health Insurance Portability – Health Insurance Guaranteed Issue – Health Insurance Preexisting Condition Limits – Health Care Eminent Domain Restrictions Paper Terrorism Lien Legislation Identity Theft Legislation Post-Conviction DNA Access for Exoneration Prescription Drug Registrations Anti-Stalking Laws Legalization of Syringes for IV Drug Addicts

Appendix C Innovations Collected by Policy Type and Target

To construct a list of morality, governance, and regulatory policy innovations, a simple coding procedure was used to classify innovations by policy type. Three graduate student coders familiar with the policy typologies literature were independently asked to identify three types of policies: (a) morality policies – a form of social regulatory policy in which the government practices “the exercise of legal authority to affirm, modify, or replace community values, moral practices, and norms of interpersonal conflict” (Tatalovich and Daynes 1998, xxx); (b) governance policies – a category of regulatory policy that “modifies the behavior of the public sector and government officials” by changing the rules of the political system to “regulate how the state should proceed to govern” (Tolbert 2002, 80); and (c) state regulatory policies,– economic, environmental, and professional regulatory policy innovations in which government uses its coercive authority to shape the behavior of private industry or citizens in order to achieve policy goals. Coders were given a fourth option for policies that did not match any of these criteria. Because policies are often overlapping (involving both regulatory and morality components), coders were instructed to classify policies by their dominant characteristics. Only policies with perfect intercoder agreement were included in the aggregated measure. This method identified 40 regulatory policies, 22 morality policies, and 10 governance policies. These policies are listed in pages that follow. Appendix C also presents a list of public policies targeting children and policies imposing licensing regulations of specific groups. Innovations were included in the children’s policy measure if the innovation directly addresses the welfare or well-being of children. Innovations that 197

198

Appendix C: Innovations Collected by Policy Type and Target

indirectly address child welfare – such as teacher’s licensing policies or charter schools enabling legislation – were excluded from this list. Licensing policies represent public policies that specifically grant entry to and maintain professional standards for the conduct and practice of specialized labor in the states. Morality Policies Equal Pay for Females Anti-Age Discrimination Child-Abuse Reporting Requirements Crime Victims’ Compensation Prohibition of Alcohol Ban Recognition of Out-of-State Same-Sex Marriage DUI .08 Per Se Legislation Amber Alert Child Access Prevention Laws (Gun Locks) Death Penalty Reenactment Medical Marijuana Parental Notification/Consent for Abortion Victims Rights’ Amendments Three-Strikes Sentencing Laws No-Fault Divorce Laws Statutory Rape Age Span Laws Sodomy Laws – Repeal Hate Crimes – Include Protections for Homosexuals Anti-Stalking Laws Needle Sales for IV Drug Users Post-Conviction DNA Bank Access for Exoneration Regulatory Policies Fish Agency Conservation of Gas and Oil Air Pollution Control Forest Agency Utility Regulation Commission Real Estate Brokers – Licensing Integrated Bar Slaughterhouse Inspection Seasonal Agricultural Labor Standards

Appendix C: Innovations Collected by Policy Type and Target Accountants Licensing Nurses Licensing Engineers Licensing Board of Health Pharmacists Licensing Automobile Registration Beauticians Licensing Architects Licensing Dentists Licensing Migratory Labor Committee Mental Health Standards Committee Teacher Certification – Secondary Teacher Certification – Elementary Zoning in Cities – Enabling Legislation Public Housing – Enabling Legislation Minimum Wage Law Chiropractors Licensing Parolees and Probationers Supervision Alcoholic Beverage Control Bottle Bills (Recycling Deposit) Charter Schools – Enabling Legislation Primary Seat Belt Laws Seat Belt Laws (Required) Ban Sale of Out-of-Package Cigarettes Auto Lemon Laws Mandatory Child Passenger Restraints Guaranteed Renewal Health Insurance Portability – Health Insurance Guaranteed Issue – Health Insurance Preexisting Condition Limits – Health care Governance Policies Budgeting Standards Australian Ballot System Municipal Home Rule Initiative and Referendum Direct Primary Planning and Development Agencies State Legislative Term Limits

199

200

Appendix C: Innovations Collected by Policy Type and Target

Supermajority Tax Rule Tax and Expenditure Limits Limit Powers of Eminent Domain Licensing Policies Real Estate Brokers – Licensing Integrated Bar Accountants Licensing Nurses Licensing Pharmacists Licensing Beauticians Licensing Architects Licensing Dentists Licensing Teacher Certification – Secondary Teacher Certification – Elementary Chiropractors Licensing Policies Targeting Children Aid to Dependent Children (Social Security) Child Abuse Reporting Requirement Amber Alert Child Access Prevention Laws (Gun Locks) Child Labor Standards Parental Notification/Consent for Abortion Statutory Rape Age Span Law

Appendix D State Receptivity to Innovation Ranked by Policy Type

Appendix D presents three different tables ranking state receptivity to morality, governance, and regulatory policy innovation from 1960 to 2006. Here, the concept of state receptivity is used to measure the speed with which states typically adopt an innovation. The receptivity scores were calculated using morality, governance, and regulatory policy innovations that diffused between 1960 and 2006. To identify those states that are the most receptive to each type of innovation, the author constructed a composite innovation score for each state by taking the ratio measuring the time between the state’s adoption of a specific program and the last state adoption of that innovation, against the total time elapsed between the first and last state adoption of the innovation. For each state, the receptivity score was arrived at using the following equation: IS = [(LY + 1) − SY]/[(LY + 1) − FY] Where IS is the innovation score for the state, LY + 1 is one year after the last state adopted the policy innovation, SY represents the year the state adopted the innovation, and FY is the first year a state adopted the policy innovation. Following this equation, the first state to adopt a policy innovation receives a score of 1.000, whereas those states that do not adopt a given innovation within the observation period receive a score of 0.000. The last state to adopt an innovation receives a score slightly above 0.000, differentiating it from nonadopting states by assigning it a small positive value as a late adopter of innovation. The composite innovation scores of state receptivity to each policy type were calculated by taking the average of individual state receptivity scores for each group of innovations. 201

Appendix D: State Receptivity to Innovation

202

State Responsiveness to Regulatory Policy, 1960–2006 State

Index Score

State

Index Score

Connecticut New York Oregon New Jersey North Carolina Massachusetts Minnesota California Delaware Texas Iowa Kansas Wisconsin Rhode Island Ohio Maine Florida

0.731 0.711 0.631 0.609 0.575 0.567 0.558 0.548 0.530 0.525 0.498 0.490 0.487 0.485 0.473 0.458 0.457

South Carolina Hawaii Tennessee Alaska Virginia Michigan New Mexico Oklahoma Maryland Idaho Colorado Illinois Louisiana New Hampshire Vermont Missouri North Dakota

0.457 0.448 0.438 0.433 0.429 0.418 0.418 0.413 0.400 0.399 0.397 0.396 0.395 0.386 0.385 0.383 0.378

State

Index Score

Washington Arizona Wyoming West Virginia Montana Pennsylvania Mississippi Nebraska Indiana Kentucky Utah Arkansas Nevada Georgia Alabama South Dakota

0.378 0.356 0.350 0.339 0.334 0.330 0.299 0.298 0.289 0.283 0.276 0.270 0.264 0.254 0.200 0.189

State Responsiveness to Morality Policy, 1960–2006 State

Index Score

State

Index Score

California Oregon Florida Nevada Connecticut Colorado Louisiana New Mexico Washington Maryland Delaware Virginia Minnesota New Jersey Illinois North Carolina Arizona

0.676 0.563 0.531 0.514 0.493 0.478 0.468 0.466 0.458 0.455 0.451 0.445 0.444 0.444 0.433 0.433 0.425

Texas Oklahoma Rhode Island Maine Kansas Utah Alaska Indiana Kentucky Tennessee Idaho Missouri Montana Wisconsin Iowa New York Nebraska

0.423 0.416 0.412 0.406 0.405 0.402 0.400 0.399 0.391 0.383 0.382 0.368 0.366 0.366 0.362 0.362 0.349

State

Index Score

Georgia Pennsylvania Ohio Alabama Hawaii West Virginia Arkansas North Dakota Michigan Vermont New Hampshire Massachusetts Wyoming South Dakota Mississippi South Carolina

0.347 0.347 0.345 0.342 0.340 0.328 0.324 0.318 0.316 0.312 0.305 0.297 0.279 0.270 0.259 0.234

Appendix D: State Receptivity to Innovation

203

State Responsiveness to Governance Policy, 1960–2006 State

Index Rank

State

Index Rank

Colorado Idaho Alaska Missouri Nevada Tennessee Michigan South Carolina Florida Oklahoma Texas California Massachusetts Louisiana Oregon Arizona Montana

0.634 0.613 0.544 0.541 0.504 0.500 0.500 0.496 0.485 0.485 0.477 0.474 0.470 0.468 0.398 0.391 0.388

Ohio Utah Wisconsin South Dakota Maine Washington Mississippi Alabama North Carolina Delaware Rhode Island Hawaii Arkansas Iowa Connecticut Kansas Pennsylvania

0.364 0.356 0.336 0.332 0.330 0.327 0.279 0.277 0.276 0.263 0.254 0.250 0.246 0.246 0.246 0.238 0.238

State

Index Rank

Wyoming West Virginia Indiana New Mexico New Hampshire Georgia New Jersey Kentucky Virginia Illinois Nebraska Minnesota Vermont North Dakota New York Maryland

0.225 0.208 0.205 0.200 0.192 0.177 0.176 0.165 0.154 0.146 0.134 0.115 0.115 0.100 0.092 0.077

References

Aaron, Paul, and David Musto. 1981. “Temperance and Prohibition in America: A Historical Overview.” In Mark Moore and Dean Gerstein, eds. Alcohol and Public Policy: Beyond the Shadow of Prohibition (pp. 125–181). Washington, DC: National Academy Press. Anderson, J. E. 1997. Public Policy-Making. New York: Houghton Mifflin. Aunger, Robert. 2002. The Electric Meme: A New Theory of How We Think. New York: Free Press. Balla, Steven J. 2001. “Interstate Professional Associations and the Diffusion of Policy Innovations.” American Politics Research 29(3): 221–245. Bass, Frank M. 1969. “A New Product Growth Model for Consumer Durables.” Management Science 15: 215–227. Baumgartner, Frank, Christian Breunig, Christoffer Green-Pedersen, Bryan D. Jones, Peter B. Mortensen, Michiel Neytemans, and Stefaan Walgrave. 2009. “Punctuated Equilibrium in Comparative Perspective.” American Journal of Political Science, 53(3): 602–619. Baumgartner, Frank, and Bryan D. Jones. 1993. Agendas and Instability in American Politics. Chicago: University of Chicago Press. Baumgartner, Frank, and Bryan Jones, eds. 2002. Policy Dynamics. Chicago: University of Chicago Press. Baumgartner, Frank, and Beth Leech. 1998. Basic Interests: The Importance of Groups in Politics and in Political Science. Princeton, NJ: Princeton University Press. Baumgartner, Frank, and Beth Leech. 2001. “Interest Niches and Policy Bandwagons: Patterns of Interest Group Involvement in National Politics,” Journal of Politics 63(4): 1191–1213. Benjamin, Gerald, and Michael Malbin, eds. 1992. Limiting Legislative Terms. Congressional Quarterly Press. Berry, Frances S, and William D. Berry. 1990. “State Lottery Adoptions as Policy Innovations: An Event History Analysis.” American Political Science Review 84(2): 395–415. 205

206

References

Berry, Frances S., and William D. Berry. 1992. “Tax Innovation by American States: Capitalizing on Political Opportunity.” American Journal of Political Science 36: 715–742. Berry, Frances S., and William D. Berry. 1999. “Innovation and Diffusion Models in Policy Research.” In Paul A. Sabatier, ed. Theories of the Policy Process (pp. 169–200). Boulder, CO: Westview Press. Berry, William, and Brady Baybeck. 2005. “Using Geographic Information Systems to Study Interstate Competition.” American Political Science Review 99: 505–519. Berry, William, Evan Ringquist, Richard Fording, and Russell Hanson. 1998. “Measuring Citizen and Government Ideology in the American States, 1960– 93.” American Journal of Political Science 42(1): 327–345. Berry, William, Evan Ringquist, Richard Fording, and Russell Hanson. 2007. “The Measurement and Stability of State Citizen Ideology.” State Politics and Policy Quarterly 7: 111–132. Bibby, John F., and Thomas Holbrook. 2004. “Parties and Elections.” In Virginia Gray and Russell Hanson, eds. Politics in the American States: A Comparative Analysis (pp. 62–99). Washington, DC: Congressional Quarterly Press. Bicycle Helmet Safety Institute. “Helmet Laws for Bicycle Riders.” http://www. helmets.org/mandator.htm. Accessed August 20, 2007. Billings-Yun, Melanie, David Brittan, and Anna Donahue. 1983. “Prohibition A: Enactment.” Kennedy School of Government Case Study 492.0. Boehmke, Frederick J. 2005. “Sources of Variation in the Frequency of Statewide Initiatives: The Role of Interest Group Populations.” Political Research Quarterly 58(4): 575–585. Boehmke, Frederick J., and Richard Witmer. 2004. “Disentangling Diffusion: The Effects of Social Learning and Economic Competition on State Policy Innovation and Expansion.” Political Research Quarterly 57(1): 39–51. Bowen, William R. 2005. “Policy Innovation and Health Insurance Reform in the American States: An Event History Analysis of State Medical Savings Account Adoptions (1993–1996).” PhD diss., Florida State University. Brady Campaign to Prevent Gun Violence. “Child Access Protection (Cap) Laws: State Summaries.” http://www.bradycampaign.org/facts/issues/?page= capstate. Accessed August 20, 2007. Brandeis, Louis Dembitz. 1932. “Dissenting Opinion.” New State Ice Co. v. Liebmann, 28 U.S. 262, 311. Breunig, Christian. 2006. “The More Things Change, the More Things Stay the Same: A Comparative Analysis of Budget Punctuations.” Journal of European Public Policy 13(7): 1065–1081. Burris, S., Jon S. Vernick, Alyssa Ditzler, and Steffanie Strathdee. 2002. “The Legality of Selling or Giving Syringes to Injection Drug Users.” Journal of the American Pharmaceutical Association 42(6), Supp. 2: S13–18. Campbell, John. 2002. “Ideas, Politics, and Public Policy.” Annual Review of Sociology 28: 21–38. Canon, Bradley C., and Lawrence Baum. 1981. “Patterns of Adoption of Tort Law Innovations: An Application of Diffusion Theory to Judicial Doctrines.” The American Political Science Review 75(4): 975–987.

References

207

Carter, Larry, and James LaPlant. 1997. “Diffusion of Health Care Policy Innovation in the United States.” State and Local Government Review 29: 17–26. Centers for Disease Control and Prevention. 2004. “Malaria: Epidemiology.” National Center for Infectious Diseases, Division of Parasitic Diseases. http:// www.cdc.gov/malaria/distribution epi/epidemiology.htm. Accessed July 28, 2007. Centers for Disease Control and Prevention. 2009. “Constructing an Epidemic Curve.” http://origin.cdc.gov/globalhealth/FETP/modules/MiniModules/ Epidemic Curve/page01.htm. Accessed August 20, 2009. Chamberlain, Robert, and Donald Haider-Markel. 2000. “‘Lien on Me’: State Policy Innovation in Response to Paper Terrorism.” Political Research Quarterly 58(3): 449–460. Chiang, Harriet. “Court Rejects Term Limit Measure.” The San Francisco Chronicle, July 9, 1999, A18. Clark, Jill. 1985. “Policy Diffusion and Program Scope: Research Directions.” Publius 15(4): 61–70. Cocca, Carolyn E. 2002. “The Politics of Statutory Rape Laws: Adoption and Reinvention of Morality Policy in the States, 1971–1999.” Polity 35(1): 51–72. Conover, W. J. 1999. Practical Nonparametric Statistics, 3rd ed. New York: Wiley. Cronin, Thomas E. 1989. Direct Democracy: The Politics of Initiative, Referendum and Recall. Cambridge, MA: Harvard University Press. Dawkins, Richard. 1989. The Selfish Gene. Oxford: Oxford University Press. Daynes, Byron. 1998. Pornography: Freedom of Expression or Sexual Degradation? In Raymond Tatalovich and Bryan Daynes, eds. Moral Controversies in American Politics (pp. 221–257). Armonk, NY: M. E. Sharpe. DeCarlo, Lawrence T. 1997. “On the Meaning and Use of Kurtosis.” Psychological Methods, 2(3): 292–307. Disarro, Brian. 2005. “Policy Diffusion Across States: The Case of Gay Rights.” Presented at the annual meeting of the Northeastern Political Science Association, Philadelphia, PA, November 17–19, 2005. Donovan, Mark. 2001. Taking Aim: Target Populations and the Wars on Aids and Drugs. Washington, DC: Georgetown University Press. Edelman, Murray. 1985. The Symbolic Uses of Politics. Champaign, IL: University of Illinois Press. Elazar, Daniel J. 1984. American Federalism: A View from the States, 3rd ed. New York: Harper and Row Erikson, Robert, Gerald C. Wright, and John P. McIver. 1993. Statehouse Democracy: Public Opinion, and Policy in the American States. Cambridge, U.K.; New York: Cambridge University Press. Eshbaugh-Soha, Matthew. 2006. “The Conditioning Influence of Policy Salience and Complexity on American Political Institutions.” Policy Studies Journal 34: 223–243. Eskridge, William. 1999. Gaylaw: Challenging the Apartheid of the Closet. Cambridge, MA: Harvard University Press. Eyestone, Robert. 1977. “Confusion, Diffusion and Innovation.” American Political Science Review 71(2): 441–447.

208

References

Fisher, Jim. “Let Lance Move to Erase Term Limits Scarlet Letter.” Lewiston Morning Tribune, November 7, 1996, 12A. Florida, Richard. 2003. The Rise of the Creative Class: And How It’s Transforming Work, Leisure, Community and Everyday Life. New York: Basic Books. Gerber, Brian J, and Paul Teske. 2000. “Regulatory Policymaking in the American States: A Review of Theoretical Evidence.” Political Research Quarterly 53(4): 849–886. Gerber, Elisabeth R. 1999. The Populist Paradox: Interest Group Influence and the Promise of Direct Legislation. Princeton, NJ: Princeton University Press. Glick, Henry R, and Scott P. Hays. 1991. “Innovation and Reinvention in State Policymaking: Theory and the Evolution of Living Will Laws.” The Journal of Politics 53(3): 835–850. Glick, Henry R, and Scott P. Hays. 1997. “The Role of Agenda Setting in Policy Innovation: An Event History Analysis of Living-Will Laws.” American Politics Quarterly 25(4): 497–516. Gordis, Leon. 2004. Epidemiology. Philadelphia, PA: Elsevier Saunders. Gormley, William T. 1986. “Regulatory Issue Networks in a Federal System.” Polity 18(4): 595–620. Grant, Alan. 1995. “The Term Limitation Movement in the United States.” Parliamentary Affairs 48(3): 515–530. Grattet, Ryken, Valerie Jenness, and Theodore Curry. 1998. “The Homogenization and Differentiation of ‘Hate Crime’ Law in the U.S., 1978–1995: An Analysis of Innovation and Diffusion in the Criminalization of Bigotry.” American Sociological Review 63: 286–307. Gray, Virginia. 1973. “Innovation in the States: A Diffusion Study.” American Political Science Review 67(4): 1174–1185. Gray, Virginia. 1974. “Expenditures and Innovation as Dimensions of ‘Progressivism’: A Note on the American States.” American Journal of Political Science 18: 693–699. Gray, Virginia, and David Lowery. 1996. The Population Ecology of Interest Representation: Lobbying Communities in the American States. Ann Arbor, MI: University of Michigan Press. Grossback, Lawrence J., Sean Nicholson-Crotty, and David A. Peterson. 2004. “Ideology and Learning in Policy Diffusion.” American Politics Research 32(5): 521–545. Haider-Markel, Donald P. 2001. “Policy Diffusion as a Geographical Expansion of the Scope of Political Conflict: Same Sex Marriage Bans in the 1990s.” State Politics and Policy Quarterly 1: 5–26. Hamm, Keith, and Gary Moncreif. 2004. Legislative Politics in the States. In Virginia Gray and Russell Hanson, eds. Politics in the American States: A Comparative Analysis (pp. 159–193). Washington, DC: Congressional Quarterly Press. Hays, Scott P. 1996. “Patterns of Reinvention: The Nature of Evolution During Policy Innovation.” Policy Studies Journal 24(4): 551–566. Henrich, Joseph. 2001. “Cultural Transmission and the Diffusion of Innovations: Adoption Dynamics Indicate That Biased Cultural Transmission Is the Predominate Force in Behavioral Change.” American Anthropologist 103(4): 992–1013.

References

209

Holyoke, Thomas T. 2003. “Choosing Battlegrounds: Interest Group Lobbying Across Multiple Venues.” Political Research Quarterly 56: 325–336. Hosking, Jonathan R. M. 1990. “L-Moments: Analysis and Estimation of Distributions Using Linear Combinations of Order Statistics.” Journal of the Royal Statistical Society Series B 52: 105–124. Howlett, Michael. 2002. “Do Networks Matter? Linking Advocacy Structure to Policy Outcomes: Evidence From Four Canadian Policy Sectors: 1990–2000.” Canadian Journal of Political Science 35(2): 235–267. Insurance Institute for Highway Safety. 2007. “History of U.S. Motorcycle Laws and Changes in Coverage.” http://www.iihs.org/laws/state laws/helmet history.html. Accessed August 20, 2007. Insurance Institute for Highway Safety. 2007. “Safety Belt Use Laws.” http://www.iihs.org/laws/state laws/restrain3.html. Accessed August 20, 2007. John, Peter. 2003. “Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations? Using Evolutionary Theory to Explain Policy Change.” Policy Studies Journal 31(4): 481–498. Jones, Bryan D. 1994. Reconceiving Decision-Making in Democratic Politics: Attention, Choice, and Public Policy. Chicago: The University of Chicago Press. Jones, Bryan D. 2001. Politics and the Architecture of Choice: Bounded Rationality and Governance. Chicago: The University of Chicago Press. Jones, Bryan D., and Frank Baumgartner. 2005. The Politics of Attention: How Government Prioritizes Problems. Chicago: The University of Chicago Press. Jones, Bryan D., Frank Baumgartner, and James True. 1998. “Policy Punctuations: U.S. Budget Authority, 1947–1995.” Journal of Politics 60(1): 1– 30. Jones, Bryan D., Graeme Boushey, and Samuel Workman. 2006. “Behavioral Rationality and the Policy Process.” In Handbook of Public Policy, Guy Peters and Jon Pierre, eds. London: Sage Press. Jones, Bryan D., Tracy Sulkin, and Heather Larsen. 2003. “Policy Punctuations in American Political Institutions.” American Political Science Review 97: 151– 169. Jost, Kenneth. 1994. “Testing Term Limits.” The CQ Researcher Online 4(43). http://library.cqpress.com/cqresearcher/cqresrre1994111800. Accessed January 26, 2005. Kahneman, Daniel, and Amos Tversky, eds. 2000. Choices, Values and Frames. Cambridge, U.K.: Cambridge University Press. Karch, Andrew. 2006. “National Intervention and the Diffusion of Policy Innovations.” American Politics Research 34(4): 403–426. Karch, Andrew. 2007a. Democratic Laboratories: Policy Diffusion among American States. Ann Arbor, MI: University of Michigan Press. Karch, Andrew 2007b. “Emerging Issues and Future Directions in State Policy Diffusion Research.” State Politics and Policy Quarterly 7(1): 54–80. Karp, Jeffrey. 1995. “Explaining Public Support for Legislative Term Limits.” Public Opinion Quarterly 59(3): 373–391. Kerr, Austin K. 1985. Organized for Prohibition: A New History of the AntiSaloon League. New Haven, CT: Yale University Press. King, James D. 2000. “Changes in Professionalism in U.S. State Legislatures.” Legislative Studies Quarterly 25(2): 327–343.

210

References

Kingdon, John. 1984. Agendas, Alternatives, and Public Policies. Boston: Little, Brown. Koch, Kathy. 1999. “Medical Marijuana.” The CQ Researcher Online 9(31). http://library.cqpress.com/cqresearcher/cqresrre1999082000. Accessed January 26, 2005. Koski, Chris. 2007. “Requests and Reality: The Interplay of Environmental Regulatory Policy Design and Implementation.” PhD diss., University of Washington. Koski, Chris, and Christian Breunig. 2006. “Punctuated Equilibria and Budgets in the American States.” Policy Studies Journal 34(3): 363–379. Lacayo, Richard. “Marijuana: Where There’s Smoke, There’s Fire.” Time. October 28, 1996, 36–38. http://www.time.com/time/magazine/article/0,9171, 985373,00.html. Accessed May 3, 2010. Larsen, Heather. 2006. “The Dynamics of Presidential Policy Attention.” PhD diss., University of Washington. Last, J.M. 2001. A Dictionary of Epidemiology, 4th ed. New York: Oxford University Press. Lee, Mei-Hsien, and Christopher Z. Mooney. 1999. “The Temporal Diffusion of Morality Policy: The Case of Death Penalty Legislation in the American States.” Policy Studies Journal. 27(4): 766–780. Lindblom, Charles E. 1959. “The Science of ‘Muddling Through.’” Public Administration Review 19: 79–88. Lowi, Theodore J. 1964. “American Business and Public Policy. Case Studies and Political Theory,” World Politics 16(4): 677–715. Lowi, Theodore J. 1972. “Four Systems of Policy, Politics and Choice.” Public Administration Review 32(4): 298–310. Lowi, Theodore J. 1998. “New Dimensions in Policy and Politics.” In Raymond Tatalovich and Bryan Daynes, eds. Moral Controversies in American Politics (pp. xiii–xxviii). Armonk, NY: M. E. Sharpe. Mahajan, Vijay, and Robert A. Peterson. 1985. Models for Innovation Diffusion. California: Sage. May, Peter. 1992. “Policy Learning and Failure.” Journal of Public Policy 12(4): 331–354. Mintrom, Michael. 1997. “Policy Entrepreneurs and the Diffusion of Innovation.” American Journal of Political Science 41(3): 738–770. Mintrom, Michael. 2002. Policy Entrepreneurs and School Choice. Washington, DC: Georgetown University Press. Mintrom, Michael, Mark Schneider, and Paul Teske. 1995. Public Entrepreneurs: Agents for Change in American Government. Princeton, NJ: Princeton University Press. Mintrom, Michael, and Sandra Vergari. 1998. “Policy Networks and Innovation Diffusion: The Case of State Education Reforms.” Journal of Politics 60(1): 126–148. Mooney, Christopher. 1995. “Citizens, Structures, and Sister States: Influences on State Legislative Professionalism.” Legislative Studies Quarterly 20(1): 47– 67. Mooney, Christopher Z. 1999. “The Politics of Morality Policy: Symposium Editor’s Introduction.” Policy Studies Journal 27(4): 675–680.

References

211

Mooney, Christopher, ed. 2001. The Public Clash of Private Values. New York: Chatham House. Mooney, Christopher, and Mei-Hsien Lee. 1995. “Legislating Morality in the American States: The Case of Pre-Roe Abortion Regulation Reform.” American Journal of Political Science 39: 599–627. Mooney, Christopher, and Mei-Hsien Lee. 1999. “Morality Policy Reinvention: State Death Penalties.” The Annals of the American Academy of Political and Social Science 566(1): 80–92. Mooney, Christopher, and Richard G. Schuldt. 2008. “Does Morality Policy Exist? Testing a Basic Assumption.” Policy Studies Journal 36(2): 199–218. Mossberger, Karen. 1999. “State-Federal Diffusion and Policy Learning: From Enterprise Zones to Empowerment Zones.” Publius 29(3): 31–50. Mossberger, Karen. 2000. The Politics of Ideas and the Spread of Enterprise Zones. Washington, DC: Georgetown University Press. Nakonezny, Paul, Robert Shull, and Joseph Rogers. 1995. “The Effect of NoFault Divorce Law on the Divorce Rates Across the 50 States and Its Relation to Income, Education, and Religiosity.” Journal of Marriage and the Family 57(2): 477–488. National Alliance for Model State Drug Laws. “Prescription Drug Monitoring Programs.” http://www.natlalliance.org/prescription drug.asp. Accessed August 21, 2007. National Conference of State Legislatures. 2004. “State .08 Dui Laws.” http:// www.ncsl.org/programs/lis/Dui/bac08.htm. Accessed April 5, 2006. National Conference of State Legislatures. 2005. “State Supermajority Tax Laws.” http://www.ncsl.org/programs/fiscal/tels2005.htm. Accessed April 5, 2006. National Conference of State Legislatures. 2005. “State Tax and Expenditure Limits.” http://www.ncsl.org/programs/fiscal/tels2005.htm. Accessed April 5, 2006. National Conference of State Legislatures. 2006. “Eminent Domain.” http:// www.ncsl.org/programs/natres/emindomain.htm. Accessed August 20 2007. National Conference of State Legislatures. 2007. “Post-Conviction DNA Motions.” http://www.ncsl.org/programs/cj/postconviction.htm. Accessed August 21, 2007. National Institute of Justice. 1996. “Domestic Violence, Stalking, and Antistalking Legislation: An Annual Report to Congress under the Violence Against Women Act.” National Institute of Justice Research Report. http://www.ncjrs. gov/pdffiles/stlkbook.pdf. Accessed August 21, 2007. National Victims’ Constitutional Amendment Passage. “State Victim Rights Amendments.” http://www.nvcap.org/stvras.htm. Accessed August 20, 2007. Nelson, Barbara J. 1984. Making an Issue of Child Abuse: Political Agenda Setting for Social Problem. Chicago: University of Chicago Press. Newell, Allen. 1990. Unified Theories of Cognition. Cambridge, MA: Harvard University Press. Nice, David C. 1994. Policy Innovation in State Government. Ames, IA: Iowa State University Press. Odegard, Peter. 1928. Pressure Politics: The Story of the Anti-Saloon League. New York: Columbia University Press.

212

References

Padgett, John F. 1980. “Bounded Rationality in Budgetary Research.” American Political Science Review 74: 354–372. PBS POV. 1999. “The Legacy: Murder & Media, Politics & Prisons.” http://www. pbs.org/pov/pov1999/thelegacy/timeline.html. Accessed August 20, 2007. Phyllis Schlafly Report. 1996. Online. Pralle, Sarah. 2003. “Venue Shopping, Political Strategy, and Policy Change: A Case Study of Canadian Forest Advocacy.” Journal of Public Policy 23: 233– 260. Quaile-Hill, Kim, and Patricia A. Hurley. 1988. “Uniform State Law Adoptions in the American States: An Explanatory Analysis.” Publius: The Journal of Federalism 18(1): 117–126. Ranney, Austin. 1976. “Parties in State Politics.” In Politics in the American States: A Comparative Analysis, Herbert Jacob and Kenneth Vines, eds. Boston: Little, Brown. Rincke, Johannes. 2004. “Neighborhood Effects in the Diffusion of Policy Innovations: Evidence from Charter School Legislation in the US.” Retrieved online at http://cournot2.u-strasbg.fr/sew/papers sew/Johannes Rincke.pdf. Accessed May 3, 2010. Ringquist, Evan J. 1993. Environmental Protection at the State Level: Politics and Progress in Controlling Pollution. Armonk, NY: M. E. Sharpe. Rogers, Everett M. 1976. “New Product Adoption and Diffusion.” Journal of Consumer Research 2: 290–301. Rogers, Everett M. 1983. Diffusion of Innovations. New York: Free Press. Rogers, Everett M. 2003. Diffusion of Innovations, 5th ed. New York: Free Press. Rothenberg, Stuart. 1992. “Transplanting Term Limits: Political Mobilization and Grass Roots Politics.” In Limiting Legislative Terms, Gerald Benjamin and Michael Malbin, eds. Washington, DC: Congressional Quarterly Press. Ryan, Bryce, and Neil Gross. 1943. “The Diffusion of Hybrid Seed Corn in Two Iowa Communities.” Rural Sociology 8: 15–24. Sabatier, Paul A., and Hank Jenkins-Smith. 1999. The Advocacy Coalition Framework: An Assessment. In Paul A. Sabatier, ed. Theories of the Policy Process (pp. 117–168). Boulder, CO: Westview Press. Savage, Robert. 1978. “Policy Innovativeness as a Trait of American States.” The Journal of Politics 40(1): 212–224. Savage, Robert. 1985a. “When a Policy’s Time Has Come: Cases of Rapid Policy Diffusion, 1983–1984.” Publius: The Journal of Federalism 15(3): 111–126. Savage, Robert. 1985b. “Diffusion Research Traditions and the Spread of Policy Innovations in a Federal System.” Publius (15)4: 1–27. Schattschneider, Elmer E. 1975. The Semi-Sovereign People. Hinsdale, IL: Dryden Press. Schneider, Anne Larason. 2006. “Patterns of Change in the Use of Imprisonment in the American States: An Integration of Path Dependence, Punctuated Equilibrium and Policy Design Approaches.” Political Research Quarterly 59(3): 457–470. Schneider, Anne L., and Helen Ingram. 1993. “The Social Construction of Target Populations: Implications for Politics and Policy.” American Political Science Review 87(2): 334–347.

References

213

Schneider, Mark, Paul Teske, and Michael Mintrom. 1995. Public Entrepreneurs: Agents for Change in American Government. Princeton, NJ: Princeton University Press. Sharkansky, Ira. 1969. “The Utility of Elazar’s Political Culture.” Polity 2: 66–83. Smith, Kevin B. 2002. “Typologies, Taxonomies, and the Benefits of Policy Classification.” Policy Studies Journal 30(3): 379–395. Smith, Mark A. 2007. The Right Talk: How Conservatives Transformed the Great Society into the Economic Society. Princeton, NJ: Princeton University Press. Soule, Sarah A. 2004. “Going to the Chapel? Same-Sex Marriage Bans in the United States, 1973–2000.” Social Problems 51(4): 453–477. Squire, Peverill. 1992. “Legislative Professionalization and Membership Diversity in State Legislatures.” Legislative Studies Quarterly 17: 69–79. Srinivasan, V. Seenu, and Charlotte H. Mason. 1986. “Nonlinear Least Squares Estimation of New Product Diffusion Models.” Marketing Science 5(2): 169– 178. State PIRG. 2006. “State PIRG Summary of State Security Freeze and Security Breach Notification Laws” http://www.pirg.org/consumer/credit/statelaws. htm. Accessed August 20, 2007. Stream, Christopher. 1999. “Health Reform in the States: A Model of Small Group Health Insurance Market Reforms.” Political Research Quarterly 52(3): 499–525. Tatalovich, Raymond, and Bryan Daynes. 1998. Moral Controversies in American Politics. Armonk, NY: M.E. Sharpe. Tolbert, Caroline. 2000. “Rethinking Lowi’s Constituent Policy: Governance Policy and Direct Democracy.” Environment and Planning C: Government and Policy 20(1): 75–93. True, James, Frank Baumgartner, and Bryan Jones. 1999. “Punctuated Equilibrium Theory: Explaining Stability and Change in American Policymaking.” In Theories of the Policy Process, Paul Sabatier, ed. Boulder: Westview Press, 1999, pp. 97–115. True, James, and Glenn Utter. 2002. “Saying ‘Yes,’ ‘No,’ and ‘Load Me Up’ to Guns in America.” The American Review of Public Administration 32(2): 216–241. Truman, David B. 1951. The Governmental Process. New York: Alfred A. Knopf. US English, Inc. “States with Official English Laws.” http://www.us english.org/ inc/official/states.asp. Accessed August 20, 2007. US PIRG. “Summary of State Security Freeze and Security Breach Notification Laws.” http://www.uspirg.org/financial-privacy-security/identity-theftprotection/summary-of-state-laws. Accessed August 20, 2007. U.S. Term Limits. “State Legislative Term Limits.” http://www.termlimits.org/ content.asp?pl=18&sl=19&contentid=19. Accessed May 3, 2010. Valente, Thomas. 1993. “Diffusion of Innovations and Policy Decision-Making.” Journal of Communication 43(1): 30–45. Volden, Craig. 2006. “States as Policy Laboratories: Emulating Success in the Children’s Health Insurance Program.” American Journal of Political Science 50(2): 294–312.

214

References

Volden, Craig, and Charles R. Shipan. 2006. “Bottom-up Federalism: The Diffusion of Antismoking Policies from U.S. Cities to States.” American Journal of Political Science 50(4): 825–843. Walker, Jack L. 1969. “The Diffusion of Innovation Among the American States.” American Political Science Review 63(3): 880–889. Walker, Jack L. 1972. “Diffusion of Public Policy Innovation among the American States.” [Computer file]. Compiled by Jack L. Walker, University of Michigan, Institute of Public Policy Studies. ICPSR, ed. Ann Arbor, MI: Inter-University Consortium for Political and Social Research [producer and distributor]. Walker, Jack L. 1991. Mobilizing Interest Groups in America: Patrons, Professionals, and Social Movements. Ann Arbor, MI: University of Michigan Press. Wallison, Ethan. 2000. “The Legislative Shuffle.” Washington Monthly 32(9): 39. Weyland, Kurt. 2005. “Theories of Policy Diffusion.” World Politics 57: 262– 295. Wood, Jennifer. 2005. “In Whose Name? Crime Victim Policy and the Punishing Power of Protection.” NWSA Journal 17(3): 1–17. Zaller, John R. 1992. The Nature and Origins of Mass Opinion. Cambridge, U.K., New York: Cambridge University Press.

Index

18th Amendment to the U.S. Constitution, 154 AARP, 143 abortion, 135, 154n6, 162, 166, 176 ACT UP!, 164 activists, 29, 30, 76, 139, 142, 144, 146 agenda setting, 8, 11, 22–24, 29, 34, 60 and attention allocation, 29 and diffusion dynamics, 34 and distributional tests, 60 agents and diffusion dynamics, 174–176 and disease, 13 See innovation agents AIDS, 163–164, 166 Alaska, 111n13 Amber Alert, ix, 1n1, 1n2, 1n3, 1n4, 1–2, 5, 15, 16, 17, 24, 27, 42, 60, 66, 68, 100, 156, 157, 158, 159, 165, 180, 191, 191n5, 195, 198, 200 American Cancer Society, 166 American Society for Temperance, 149, 151, 153 Americans Back in Charge (ABC), 160 Americans for Democratic Action, 148 Americans to Limit Congressional Terms (ALCT), 160 Anderson-Darling test, 53, 55, 57, 82, 85 anomalies models, 40–46 Anopheles mosquito, 12, 13 Anti-Saloon League, 149, 152–153, 154–156, 165

Arizona, 164 Arkansas, 101 attention allocation, 8 and agenda setting, 29 and child-protection policies, 156–159 and decision making, 22, 173 and emotion, 8, 18, 24 and policy change, 7 See focusing event bandwagon effect, 42 Bass diffusion model, 44–45, 45n7 Bass, Frank M., 44–45 Baumgartner, Frank, 2n6, 7, 37, 41, 42, 46, 180 Berry, Frances S., 25, 26, 95–96, 108n10 Berry, William D., 25, 26, 95–96, 96n2, 108n10, 110–111 bounded rationality, 34n5 Brandeis, Louis D., 24 California, 101, 107, 120, 130n26, 134, 146n4, 157, 162, 163, 164, 167, 183 Proposition 187, 146n4, 183 Canada, 1 carriers of innovation, 16 Carter, Larry, 97, 100n8, 107 Centers for Disease Control, 12, 179 charter school legislation, 27 child-protection policies, 1–3, 68, 79, 86, 148, 159 1990–2005, 159 See Amber Alert; Megan’s Law

215

216 Child Rescue Alert, 1 Children’s Health Insurance Program (CHIPs), 33, 98 citizen advocacy groups, 109–110, 117 Citizens for Congressional Reform (CCR), 160 Civil War, 149, 151, 153 cognitive band, 35–36 Common Cause, 148 communications technology, 23, 49–50, 60–61 conservatism, 110–111, 115, 128–129, 131, 133 crime control legislation, 7, 15–17 death penalty, 2, 24, 27, 28n4, 31, 41, 42, 49, 51, 60, 74, 81, 90, 129, 172, 187n1 decision making and agenda setting, 22–24 and attention allocation, 173 and diffusion dynamics, 170–171 and the diffusion of innovations, 7–9 incremental, 5 models of, 34–36, 46–47 and morality policy, 133 and social fads, 8n14 and state receptivity, 99 Democratic party, 108 diffusion dynamics anatomy of, 179–180 and agenda setting, 34 causes of, 8 and decision making, 170–171 differences in, 8 and disease, 10–13, 173–174 epidemiologic framework, 9–10 in the future, 181–185 and innovation attributes, 17–19, 62–70 and innovation characteristics, 17–19 and interest groups, 139–141 and issue fragility, 78–80 modeling, 9–10 and policy targets, 78–80 and policy vectors, 20–21, 179 and public policy, 99 and state characteristics, 19–20, 92–94 See attention allocation diffusion research, 4n12, 5, 10n16, 17n18, 62–63, 63n3, 94–98, 181–185 in the future, 181–185 and state receptivity, 94–98

Index direct democracy, 4, 17, 74, 75, 75n9, 76, 78, 89, 90, 109, 127, 129, 130, 135, 137, 155, 160, 175, 176 disease diffusion, factors of, 10–13 distributional tests, 53–55 Donahue, Anna, 155 drug crime control, 68–69 Drug Policy Institute, 164 early adopters, 40, 47, 92n1, 93, 94, 133 early majority, 40 economic competition, 3, 4, 4n12, 25, 26, 90, 96n2 economic regulatory policy, 18–20, 64, 76–77, 78, 118–129, 138, 175 determinants of, 124–125 and diffusion dynamics, 175 and nongovernmental actors, 127–128 and political ideology, 128–129 political institutional correlates of, 125–127 state rankings in, 118–124 state responsiveness to, 129–136 Edelman, Murray, 70 Elazar, Daniel, 111 emotional reasoning, 34, 35, 36 empirical cumulative distribution function (ECDF), 53 English Only language legislation, 2, 28, 28n3 environmental factors, and disease, 12 environmental policy, 7, 72 epidemiologic framework, 6, 11 and agents, 174–176 and diffusion dynamics, 9–10, 170, 173–174 and disease diffusion, 10–13 and host characteristics, 177 implications of, 17–21 and innovation attributes, 17–19 and policy diffusion, 11–17, 89 and policy vectors, 179 epidemiologists, 36 epidemiology, 9, 10n16 Erikson, Robert, 110, 111n13 Eshbaugh-Soha, Matthew, 67, 72 event history framework, 4n12, 10n16, 25, 96n2, 96–97, 98–99 and internal dynamics model, 98–99 external influence model, 43–44

Index federalism and complexity of policy making, 4 and diffusion dynamics, 91 and rapid policy change, 2 and venue shopping, 29 Florida, 101, 103 focusing event, 7, 7n13, 29, 43, 180 framing, 78–79, 139, 141, 142, 144, 146–147, 148, 154, 158, 162, 163, 166–167, 178, 182 and medical marijuana reform, 163 See also reframing Gaussian curve, 47, 53, 82 gay marriage, 89n12 Georgia, 120 Gerber, Brian J., 77, 81 Glick, Henry R., 182 Gormley, William T., 72–73 governance policy, 18–20, 74–76, 77–78, 82–84, 88, 89, 90–91, 93, 118–138, 176 determinants of, 124–125 and nongovernmental actors, 127–128 and political ideology, 128–129 political institutional correlates of, 125–127 and state receptivity, 93 state rankings in, 118–124 state responsiveness to, 129–136 Gray, Virginia, 3n10, 47n8, 71, 71n7, 95, 110 greenhouse gas regulations, 66 Gross, Neil, 41, 42 gun laws, 7 H1-N1 influenza, 62 Hagerman, Amber, 157, 158, 165 hate crimes legislation, 38 Hawaii, 100, 111n13 Hays, Scott, 31 Henrich, Joseph, 42 hosts and disease, 13 and state receptivity to innovation, 176–177 hybrid corn, 37, 41–42, 172n2 hyperincrementalism, 47–49 Idaho, 162 Illinois, 101 imitation. See policy imitation

217 incremental learning, 4–5, 9, 21, 22, 23, 24, 25, 26, 27, 29, 33, 34, 36, 46–47, 49, 50, 53, 55, 59–61, 76, 77, 78, 82–85, 87, 89, 134, 135–137, 172, 173 incremental learning curve, 47–49, 135–137 incremental learning model, 46–47 incrementalism, 3–5, 8, 9, 22–26, 28, 33, 45, 46, 47–49, 51, 59, 60, 61, 64, 67, 69, 76, 78, 136, 171, 172, 174, 179 challenges to, 28 and diffusion models of policy adoption, 24–26 expectations of, 49 and nonincremental patterns of policy diffusion, 47–49 and policy outbreaks, 22–24 See incrementalism distributional analysis; negative feedback cycles; “S-shaped” patterns of adoption incrementalism distributional analysis and data, 50–52 and distributional tests, 53–55 discussion of, 59–61 expectations of, 49 methodology for, 52–53 results of, 55–59 individualist states, 111, 115 influenza, 11, 62, 62n1, 64 informed voter ballots, 161–162 Ingram, Helen, 80n10, 183 innovation adopters, 38, 40 innovation adoption and common external parameters, 43 decision-making models of, 34–36 five stages of, 34 as incremental learning model, 46–47 and state receptivity, 94–98 innovation agents, 16–19 innovation attributes, 62–70 innovation carriers, 16 innovation characteristics, 17–19, 182–184 innovation cost, 63, 65–66, 69 innovation diffusion anomalies models in, 40–46 and decision making, 7–9 decision-making models in, 34–36 empirical models of, 37–40 and epidemiologic framework, 9–10 external influence model of, 43–44 in the future, 185

218 innovation diffusion (cont.) and geography, 96n2, 124n21 as incremental learning model, 46–47 and innovation characteristics, 17–19 and interest groups, 164–168 and modernization, 60–61 “neighborhood effect” in, 96n2 and policy types, 71 and policy vectors, 21 and professional networks, 14 and “S-shaped” patterns of adoption, 37–40 and state characteristics, 98–99 innovation flexibility, 65n5 innovation index, 99–100 innovation, defined, 24n2, 50 innovations list of, 192 by policy type, 197–200 state receptivity rankings list, 201–203 innovators, 40 insider lobbying, 142 instrumental policy learning, 32–33 intendedly rational band, 35–36 interest groups, 6, 14, 16, 26, 73, 76, 93, 115, 115n16, 117, 127–128, 131, 137–138 and agenda setting, 30 behavior of, 144–145 case selection of, 148–149 and child-protection policies, 156–159 and diffusion dynamics, 139–141, 177–179 and framing, 146–147 future study of, 181–182 and governance policy, 89 and informed voter ballots, 161–162 and innovation diffusion, 164–168 as key carriers of policy reform, 127–128 and medical marijuana reform, 162–164 organization and resources of, 142–144 and policy entrepreneurs, 141–142 as policy vectors, 21 and prohibition, 149–156 and state receptivity, 108–110 and term-limit movement, 159–160 variation in organization and strategic behavior of, 150 and venue shopping, 145–146

Index See child-protection policies; medical marijuana reform; prohibition; term limits internal diffusion model, 42 internal dynamics model, 98–99 internal influence diffusion model, 38–40 issue complexity, 18, 63, 65–66, 69 issue fragility, 63, 65, 68, 69, 77, 78–79, 80 issue salience, 6, 8, 18, 31, 63, 65, 66–67, 69, 72, 73, 74, 76, 78, 80, 89, 90, 117, 134, 147, 159, 160, 166, 167, 171, 175, 179, 181 defined, 63 “issues of the day,” 147 Jessica’s Law, 68 Jones, Bryan, 2n6, 7, 37, 41, 42, 46, 180 Kanka, Megan, 157, 165 Karch, Andrew, 17n18, 31, 63n3, 182 King, James D., 107 Klaas Kids Foundation, 157 Klaas, Polly, 157, 165 Kolmogorov-Smirnov test, 53 kurtosis tests, 53–55, 55n13, 85 laboratories of democracy, 5, 24, 29, 185 laggards, 40, 94, 100, 103, 120, 134 LaPlant, James, 97, 100n8, 107 late majority, 40 Lee, Mei-Hsien, 31, 41, 42, 74 liberalism, 110–111, 115, 128–129, 133 licensing policies, 86–87, 88 Lindblom, Charles E., 25 living will legislation, 24, 26, 31, 51 lobbying, 142–143, 148 Louisiana, 103, 136, 160, 177 Lowery, David, 110 Lowi, Theodore J., 70–71 Lowi’s policy typology, 70–71, 72 Mahajan, Vijay, 38–39, 43–45 Maine, 164 malaria, 13 Marijuana Policy Project, 164 Massachusetts, 120 material policies, 70 May, Peter, 31–33 McIver, John P., 110, 111n13 medical marijuana reform, 52, 129, 141, 148, 162–164, 165, 167

Index Megan Kanka Foundation, 157 Megan’s Law, 16, 68, 156–159, 167 methodology, 5–7, 169–170 Minnesota, 101 Mississippi, 101 mixed influence diffusion model, 44–45 modernization, 60–61 Montana, 107, 134 Mooney, Christopher, 31, 41, 42, 74 moralist states, 111, 115 morality policy, 18–20, 31, 73–74, 74n8, 78, 84–85, 88, 89n12, 90–91, 93, 117, 118–138, 172, 176 determinants of, 124–125 and nongovernmental actors, 127–128 and political ideology, 128–129 political institutional correlates of, 125–127 state legislative professionalism, 134–135 state rankings in, 118–124 and state receptivity, 93 state responsiveness to, 129–136 Mossberger, Karen, 33, 65n5, 182 Mothers Against Drunk Driving, 16, 30, 166 National Center for Missing and Exploited Children, 16, 157 National Organization for the Reform of Marijuana Laws (NORML), 148, 164 National Rifle Association (NRA), 143, 148, 166 Nebraska, 120, 160n11 negative feedback cycles, 7, 8 Nevada, 111n13, 120, 162, 164 New Hampshire, 159n9 New York, 101, 124 Newell, Allen, 35 nongovernmental actors, 4, 108–110, 116, 133 normality tests, 53–55 North Carolina, 103 North Dakota, 120 nuclear energy policy, 7 Oklahoma, 100 Oregon, 120, 164 Oregon Health Plan, 146n4, 183 outsider lobbying, 142

219 P. falciparum malaria parasite, 13 Parents for Megan’s Law, 157 Pennsylvania, 101, 105 person-to-person outbreaks, 36 Peterson, Robert A., 38–39, 43–45 “point source” outbreaks, 36 policies collected, by historical era, 193–196 policy adoption, 24–26 See state receptivity policy agents distributional analysis and data, 80–81 discussion of, 87–89 implications of, 89–91 methods of, 82 results of, 82–87 policy contagion model. See epidemiologic framework; policy outbreaks; policy vectors policy diffusion challenges to incrementalism in, 26–28 and communications technology, 23 comparative approach to, 3n10, 5, 15, 23 and decision making, 106, 173 decision-making models in, 34–36 defined, 3, 24n2 and demographics, 112 and distributional analysis, 23n1 and epidemiologic framework, 11–17 evaluation of patterns of, 50–61 historical versus simulated patterns of, 23 and incrementalism, 24–26 national interaction effects in, 2n7 nonincremental patterns of, 47–49 patterns of, 9–10 and political competition, 107–108 and political culture, 111–112 and punctuated equilibrium, 36–37 See incrementalism; policy outbreaks policy entrepreneurs, 14, 21, 30, 97n3, 106, 139, 140, 141–142, 142n1, 161, 162, 163 and interest groups, 141–142 policy imitation, 4, 8, 10n16, 24, 25–26, 31–34, 42, 49, 60, 61, 64, 89, 90, 96, 117, 140, 146n4, 162, 170 and policy learning, 31–34 and rapid policy diffusion, 140 and states, 25–26

220 “policy innovativeness,” 94–95 policy learning, 31–34 policy outbreaks, 9, 77, 90, 167 anatomy of, 179–180 and attention allocation, 171–173 and contagion, 1–5, 173–174 curve, 47–49 and imitation, 140 and incrementalism, 22–24 innovation attributes of, 17–19 policy outbreaks curve, 47–49 See positive feedback cycles policy reinvention, 25 policy stagnation, 74 policy targets, 63, 70–78, 147 policy traits, 65–70 policy types, 70–78 approaches to developing, 70–71 and policy dynamics, 77–78 See economic regulatory policy; governance policy; morality policy policy vectors, 20–21, 177–179 political competition, 107–108, 116, 124, 126–127, 133, 176 political culture, 111–112 political decision making. See decision making political environment, 15–16 political ideology, 110–111, 115–116, 128–129, 131 political learning, 32 Polly Klaas Foundation, 157 positive feedback cycles, 2, 5, 6, 7, 8, 18, 21, 23, 37, 41, 42, 50, 61, 65, 70, 74, 78, 80, 86, 88, 91, 93, 146, 147, 149, 171–173, 174, 180 and attention allocation, 171–173 problem identification, stages of, 34 procedural policies, 70 professional licensing policies, 79–80 prohibition, 2, 141, 148, 149–156, 157, 164, 165, 178 Prohibition, 165, 166, 167 Prohibition Party, 149, 151–152 public opinion, 11, 29, 31, 74, 97n4, 106, 117, 141, 143, 163 punctuated equilibrium, 36–37 Ranney, Austin, 108 Ranney index, 108, 108n11 rapid policy diffusion. See policy outbreaks

Index reframing, 139, 141, 144, 147, 166–167, 178 regulatory policy, 20, 72–74, 85, 118 and state receptivity, 93 Republican party, 108, 161 Rockefeller, John D., 154 Rogers’ diffusion-of-innovations model, 34–36, 34n6, 47n8 Rogers, Everett M., 34–36, 37, 40, 47 Rothenberg, Stuart, 160 “R-shaped” patterns of adoption, 24, 41, 44, 46 Ryan, Bryce, 41, 42 same-sex marriage, 51, 73, 126, 128, 135, 172 San Francisco Aids Foundation, 164 satisficing, 6, 25 Savage, Robert, 50–52, 68, 71n7 Scarlet Letter initiatives, 161–162, 165, 167 Schneider, Anne, 80n10, 183 Scott P. Hays, 182 sex offender registries, 2, 68, 156, 157, 158 Shapiro-Wilk test, 53, 55, 57, 82, 85 Sharkansky, Ira, 111 Smith, Kevin B., 71, 74n8 social constructivism, 67 social fads, 8n14 social policy learning, 25, 26, 32 social regulatory policy. See morality policy Soros, George, 164 South Dakota, 100–101 Southern states, 101–103 Squire, Peverill, 107 “S-shaped” patterns of adoption, 23n1, 23–24, 28, 31, 37–46, 47–49, 49n10, 53, 60, 74, 82, 172 and anomalies, 40–46 and innovation diffusion, 37–40 and policy diffusion, 47–49 and punctuated equilibrium model, 37 See internal influence diffusion model state characteristics and diffusion dynamics, 19–20, 92–94 as predictors of state responsiveness, 105 state hosts, 17 “state innovativeness,” 98 state legislative professionalism, 93, 97, 106–108, 110–111, 115, 117, 118,

Index 126n22, 125–127, 130n26, 131, 134–135, 137, 177 indexes for, 107–108 and morality policy, 134–135 and policy types, 125–127 and political ideology, 110–111 and state policy receptivity, 106–108 and state receptivity, 130, 131, 177 state lottery, 4n12, 5, 24, 26, 27, 28n4, 51, 60, 96, 96n2, 98, 187n1 state ranks, 100–105, 118–124 1960–2006, 100–103, 114 by historical era, 210–14, 104 by policy types, 118–124 regional clustering in, 100–103 state receptivity, 19–20, 93–94, 98–99, 185 analysis of, by policy type, 129–136 analysis of scores in, 112–116 calculating, 99–100 demographics and economics in, 112 determinants of, 124–125 as a general trait, 118 and host characteristics, 176–177 modeling, 94–98, 106 and nongovernmental actors, 108–110, 128 and political characteristics, 110–111 and political competition, 107–108, 118–124, 126–127 and political culture, 111–112 and political ideology, 128–129 political institutional correlates of, 125–127 predictors of, 131, 132 rankings in, 100–105 and state characteristics, 105 See state ranks states, 6, 8 and borrowing, 25–26 incrementalism and policy outbreaks in, 22–24 as laboratories of democracy, 30 and policy adoption, 24–26, 98 See laboratories of democracy

221 stem cell research, 128 stochastic process model, 8, 46, 52, 65 substantive policies, 70 superstitious instrumental policy learning, 32, 33 symbolic policies, 70 target populations, 67–69 term limits, 3n9, 2–5, 24, 28, 30, 52, 60, 81, 89, 136, 141, 148, 159n8, 159n9, 160n11, 159–162, 165, 167, 177 and informed voter ballots, 161–162 movement (1990–1998), 160 Teske, Paul, 77, 81 Texas, 101, 103 “three strikes” sentencing laws, 2, 15, 16, 24, 29–30, 52, 68, 157, 167, 172, 180 Tolbert, Caroline, 74–76, 81 traditionalist states, 111–112, 115, 128 trial and error learning, 32, 34, 35, 46, 47, 140, 146, 164, 165, 167 true instrumental policy learning, 32, 33 United Kingdom, 1 US Term Limits, 160, 161, 162, 165 Valente, Thomas, 41 vectors, and disease, 13 See carriers of innovation; policy vectors venue shopping, 29, 145–146, 146n4, 148, 162, 167, 182 Vermont, 120 virulence, of policy ideas, 14 Volden, Craig, 98 Walker, Jack L., 3n10, 25, 26, 50–51, 52n12, 94–95, 98, 100n8, 110 Washington, 164 Washington, D.C., 100 Wisconsin, 101, 105 Women’s Christian Temperance Union, 149, 154 World Health Organization, 179 Wright, Gerald C., 110, 111n13