Direct Democracy's Impact on American Political Institutions

Direct Democracy’s Impact on American Political Institutions

This page intentionally left blank

Direct Democracy’s Impact on American Political Institutions Edited by Shaun Bowler and Amihai Glazer

DIRECT DEMOCRACY’S IMPACT ON AMERICAN POLITICAL INSTITUTIONS

Copyright © Shaun Bowler and Amihai Glazer, 2008. All rights reserved. No part of this book may be used or reproduced in any manner whatsoever without written permission except in the case of brief quotations embodied in critical articles or reviews. First published in 2008 by PALGRAVE MACMILLAN™ 175 Fifth Avenue, New York, N.Y. 10010 and Houndmills, Basingstoke, Hampshire, England RG21 6XS. Companies and representatives throughout the world. PALGRAVE MACMILLAN is the global academic imprint of the Palgrave Macmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd. Macmillan® is a registered trademark in the United States, United Kingdom and other countries. Palgrave is a registered trademark in the European Union and other countries. ISBN-13: 978-0-230-60445-2 ISBN-10: 0-230-60445-5 Library of Congress Cataloging-in-Publication Data is available from the Library of Congress. A catalogue record for this book is available from the British Library. Design by Macmillan India Ltd. First edition: May 2008 10 9 8 7 6 5 4 3 2 1 Printed in the United States of America.

Contents

List of Figures

vii

List of Tables

ix

Notes on Contributors

xi

Preface

xv

Acknowledgments 1 “Hybrid Democracy” and Its Consequences Shaun Bowler and Amihai Glazer

xvii 1

2 Was Rove Right? Ohio’s Gay Marriage Ban and the 2004 Presidential Election Daniel A. Smith

21

3 Electoral Supply and Demand: Direct Democracy Campaigns, Political Interest, and Participation Caroline Tolbert and Daniel Bowen

35

4 Direct Democracy and the Public Agenda: Ballot Initiatives and Public Beliefs about Important Problems Stephen P. Nicholson

55

5

69

Direct Democracy’s Effects on Political Parties Shaun Bowler and Todd Donovan

6 The Initiative Process and Interest Group Attention to Legislative Activity Frederick J. Boehmke 7

Direct Democracy and the Executive Branch John G. Matsusaka

8 Direct Democracy and the Stability of State Policy Amihai Glazer and Anthony McGann

93 115 137

vi

9

CONTENTS

Direct Democracy and Good Governance: Does It Matter? Russell J. Dalton

149

Appendix A

169

Appendix B

171

Appendix C

172

References

173

Further Reading

189

Index

191

Figures

4.1 The agenda-setting effect of Prop. 8 on the public agenda, 1998

63

5.1 Percentage of all initiatives that concern governance issues by decade

72

5.2 Numbers of ballot proposals concerning elections and redistricting, 1990–2004

82

5.3 Number of reform proposals concerning elections and redistricting passing 1990–2004

83

7.1 Ideal point configurations of legislature (L), governor (G), and voter (V) to illustrate the impact of initiative on balance of power

117

9.1

The relationship between initiatives passed and government management

157

This page intentionally left blank

Tables

2.1 Statewide and county averages

26

2.2 Percent vote for Ohio’s Issue 1 in 2004, percent turnout in 2004, and percent vote for Bush in 2004

29

3.1 3.2

Impact of exposure to ballot initiatives and initiative campaigns on political interest, 2002 midterm election

44

What matters? Predicted effect of exposure to ballot initiative campaign spending on political interest—by education level

46

3.3 Impact of ballot initiative campaigns on political interest and the probability of voting 3.4

Impact of exposure to ballot initiatives and initiative campaigns on political interest, 2004 presidential election

4.1 Probit analysis of the agenda-setting effect of Proposition 8, 1998 4.2

Probit analysis of the agenda-setting effect of Proposition 8 by education, 1998

47 49 62 66

5.1 Number and pass rates of governance issues, 1911–2000

71

5.2 Bivariate regressions on measures of state parties and their environments

76

5.3 Effect of cultural variables and frequency of use of direct democracy on weakening of parties

78

5.4 Models of elite opinion toward direct democracy and electoral process

80

5.5

Term-limits repeals

84

5.6

Endorsements in Washington state voter’s guide on selected measures

87

6.1

Descriptive statistics

99

6.2

Bill involvement and average reasons for choice of high-attention bills by initiative possibility

101

x

TABLES

6.3 Tabulations of bill interest variables by initiative possibility

102

6.4 Tabulations of bill interest variables by organization type and initiative possibility

104

6.5

Bill involvement and average reasons for choice of high-attention bills by initiative involvement

105

6.6 Negative binomial analysis of number of bills that groups are involved with, by attention level, initiative possibility, and involvement

108

7.1 Comparison of executive branch institutions in initiative and noninitiative states, 2003

121

7.2 Comparison of executive office salaries between initiative and noninitiative states, 2003

123

7.3 Regressions of incumbent party vote share on economic performance and government spending for gubernatorial elections, 1950–1988

130

8.1 Effect on budget allocations, annual change (all states)

142

8.2

Effect on budget allocations, annual change (Alaska excluded)

143

8.3

Effect on budget allocations, mean deviation (all states)

143

8.4 Effect on budget allocations, mean deviation (Alaska excluded)

144

8.5

145

Effect on taxes (all states)

8.6 Effect on taxes (Alaska excluded)

145

8.7

146

Effect on total spending (all states)

8.8 Effect on total spending (Alaska excluded)

146

9.1

State rankings in average governmental performance

155

9.2 Correlation between initiative usage and measures of government performance

156

9.3 Correlation between initiative usage and measures of educational performance

158

9.4 Correlation between initiative usage and measures of tax performance

161

9.5 Predicting government performance

164

Notes on Contributors

Frederick J. Boehmke of the University of Iowa is a leading expert on the relationship between interest groups and the initiative process. He is a Robert Wood Johnson Scholar in Health Policy (2005–2007) in the School of Public Health at the University of Michigan. In addition to a number of articles on such topics as the link between Indian gaming and the initiative, he is also the author of The Indirect Effect of Direct Legislation: How Institutions Shape Interest Group Systems, published by Ohio State University Press (2005). Daniel Bowen is a graduate student at the University of Iowa with interests in state and local politics. He has written on the initiative process and electoral politics. Shaun Bowler, of the University of California, Riverside, coauthored Demanding Choices: Opinion, Voting, and Direct Democracy, published by University of Michigan Press (1998), and coedited Citizens as Legislatures. He is a member of the Board of Scholars of the Initiative and Referendum Institute at the University of Southern California (USC). Russell J. Dalton is the author or editor of over twenty books, including Democratic Challenges, Democratic Choices: The Erosion of Political Support in Advanced Industrial Democracies (2004) and Democracy Transformed? (2003), both published by Oxford University Press. A previous Fulbright Fellow and Marshall Fund Fellow, he was director of the Center for the Study of Democracy at the University of California, Irvine. He is currently on the editorial boards of six journals and has previously been a member of the editorial boards of American Political Science Review, American Journal of Political Science, and Journal of Politics. He has also been a member of the Board of Overseers of the American National Election Study and is a current member of the planning committee of the Comparative Study of Electoral Systems project. Todd Donovan coauthored Demanding Choices: Opinion, Voting, and Direct Democracy, published by the University of Michigan Press (1998), and coedited Citizens as Legislatures. Donovan’s current book project is a

xii

NOTES ON CONTRIBUTORS

comparative treatment of state politics. He is a member of the board of scholars of the Initiative and Referendum Institute and a senior research associate of the University of Washington’s Center for American Politics. Amihai Glazer, besides his many publications, is the coauthor of Why Government Succeeds and Why It Fails, published by Harvard University Press (2001). He is the coeditor of Economics of Governance and is an editorial board member of Public Choice. He has held visiting appointments at Stanford and Carnegie Mellon and is past chair of the Department of Economics, UCI, and director, Focused Research Program in Public Choice. John G. Matsusaka is a professor in the Marshall School of Business and Law School at USC and president of the Initiative and Referendum Institute. Matsusaka received his PhD in economics from the University of Chicago and has held visiting appointments at the Hoover Institution at Stanford University; University of California, Los Angeles; California Institute of Technology; and the University of Chicago. His research focuses on the financing, governance, and organization of corporations and governments. He has published numerous scholarly articles, has served as a consultant for the White House Council of Economic Advisors, and is the author of the recently published book For the Many or the Few: The Initiative, Public Policy, and American Democracy (University of Chicago Press, 2004). Anthony McGann of the University of California-Irvine is the author of The Logic of Democracy, published by the University of Michigan Press, and has published widely on parties and democratic decision making. McGann is an affiliated faculty member of UCI’s Center for the Study of Democracy. Stephen P. Nicholson is the author of Voting the Agenda, published by Princeton University Press (2005). In his book, Nicholson puts forward the argument that ballot propositions may set the electoral agenda in candidate races. Currently, he is researching the role of group stereotypes in voting on ballot propositions. In addition to studying direct democracy, his research focuses on divided government, approval of governmental institutions, and Latino political behavior. Nicholson is the recipient of the 2006 Emerging Scholar Award given by the American Political Science Association’s organized section on Elections, Public Opinion, and Voting Behavior. Recently, he became a founding member of the political science program at the University of California, Merced. Nicholson’s research has been published in American Journal of Political Science, Journal of Politics, British Journal of Political Science, and Political Research Quarterly.

NOTES ON CONTRIBUTORS

xiii

Daniel A. Smith is the coauthor of Educated by Initiative: The Effects of Direct Democracy on Citizens and Political Organizations in the American States, published by the University of Michigan Press (2004). He has published more than two dozen scholarly articles on the politics, processes, and campaign financing of ballot initiatives, as well as on the workings of American political parties and interest groups. He has also written several articles examining Ghanaian politics. His book Tax Crusaders and the Politics of Direct Democracy (Routledge, 1998) examined the financial backing and the populist-sounding rhetoric of three antitax ballot initiatives—Proposition 13 in California (1978), Proposition 2 1/2 in Massachusetts (1980), and Amendment 1 in Colorado (1992)—that did much to transform the ideological debates over state and market in the 1970s and 1980s. Professor Smith serves on the board of directors of the Ballot Initiative Strategy Center Foundation, a nonprofit organization based in Washington, D.C., and is a senior research fellow at the Initiative and Referendum Institute. Caroline Tolbert is a leading scholar on the educative effects of the initiative. She is the coauthor of Educated by Initiative: The Effects of Direct Democracy on Citizens and Political Organizations in the American States (2004), published by the University of Michigan Press, and the coeditor of Citizens as Legislators. She is also the coauthor of Virtual Inequality: Beyond the Digital Divide (Georgetown University Press, 2003). She serves on the editorial board of State Politics and Policy Quarterly and is a member of the board of scholars of the Initiative and Referendum Institute. She is the recipient of national grants from the U.S. Department of Housing and Urban Development and the Smith Richardson Foundation.

This page intentionally left blank

Preface

At each election, dozens of proposals are put directly to voters by the process of direct democracy. In November 2006 alone there were over 200 proposals on the ballot ranging in topic from the legality of gay marriage to stem cell research, from eminent domain to the regulation of tobacco smoking. Past years have seen similarly controversial issues put to voters; abortion law, immigrant rights, drug use, animal rights, and “official English” language policy have all appeared on ballots. Critics are quick to stress the downside of a process that allows so many issues to be voted on directly. Over and above the number of proposals— the dozens of policy changes proposed in each election—there may also be proposals that seek to regulate government and politics directly. Alongside hot-button issues of morality, voters may also be asked to vote on questions of term limits, tax and expenditure limits, campaign finance laws, and the regulation of primaries. All of these proposals have appeared on ballots and all sought to directly regulate the state’s political and constitutional structure. As a consequence both of the number of the proposals and also of their content, critics argue that direct democracy is just too disruptive of “normal” politics and political life, either because there are simply so many policies introduced this way or because the proposals often seek to regulate government itself. Critics therefore argue that direct democracy is bad for representative democracy and, in a wider sense, for democracy itself. To date, however, there has been little or no systematic assessment of the claim that direct democracy is bad for the workings of representative democracy. Criticism of the initiative, the referendum, and the recall, which together comprise the processes of direct democracy, is often very vocal. But that criticism is often grounded in a few specific examples that may well be taken out of context or are exceptional. This volume marks the first attempt to examine systematically the impact of direct democracy on representative democracy. It studies how direct democracy affects the ways in which representative democracy works or disrupts it. The United States provides an ideal testing ground for this study because roughly half the states have some form of direct

xvi

PREFACE

democracy and half do not. We can therefore compare the workings of representative democracy across states both with and without direct democracy. The volume comprises a series of chapters that examine the impact of direct democracy on the different elements of representative democracy. Each chapter in the volume is written by a leading scholar in the field, and the chapters themselves concern specific topics that are tied to the overall theme of the volume.

Acknowledgments

The editors would like to thank the Center for the Study of Democracy at UC Irvine and in particular, Russell Dalton for the financial support and encouragement that made the original conference on this topic possible. We would also like to thank John Matsusaka and the Initiative and Referendum Institute for the material presented in the Appendices. We would also like to thank Reuben Kline for help in preparing the index and the editors and copy editors at Macmillan for their hard work in making the book a finished product.

This page intentionally left blank

1

“Hybrid Democracy” and Its Consequences Shaun Bowler and Amihai Glazer

ost Americans live in a “hybrid democracy.” That is, they live where processes of both direct democracy and representative democracy coexist as part of state government (Garrett 2005, 2006). Yet, as Garrett notes, while studies of direct democracy and of state government are common, few authors examine the interaction between the two and their effects on each other. This volume systematically looks at the interrelationship between direct and representative democracy and the impact of one on the other. We show that while there do exist arguments concerning the effects, few studies assess these arguments. Some aspects of direct democracy, such as voting behavior, are reasonably well understood, whereas other aspects, such as how parties are shaped by direct democracy or what direct democracy means for governability, are poorly understood. The chapters in this volume further our understanding of these important yet understudied areas. Moreover, whereas much work takes one dimension of direct democracy and draws some conclusions about that dimension (e.g., whether turnout or campaign spending increases), this book looks at a broader range of outcomes. Although our primary purpose is to further our understanding of government, the findings in the book may also prove relevant to the evaluation of direct democracy—is it good or bad? Here, in one volume, the trade-offs inherent in direct democracy become clear. It increases participation, but also makes it harder for government to package policies. It can make government better respond to the voters’ preferences on some issues, but may also make government less efficient or effective. These trade-offs may remain hidden to the reader who encounters articles one at a time, with each focusing on only one aspect of direct democracy.

M

2

SHAUN BOWLER AND AMIHAI GLAZER

We do not intend to compare direct democracy with representative democracy. Perhaps ancient Athens had the first without the second; in the United States we observe no pure forms of direct democracy, but rather different degrees of direct democracy coupled with a large dose of representative democracy. Our focus is then on how increased presence of direct democracy affects the behavior of voters, parties, and politicians and the performance of government. Of course, within the United States, increased presence of direct democracy implies reduced presence of representative democracy, and we cannot distinguish between an increase in one and a decrease in the other. This introductory chapter outlines what a hybrid democracy looks like and the ways in which direct democracy may affect the workings of representative democracy. Subsequent chapters assess how changes in the level of hybrid democracy affect the workings of government and of political actors. Throughout the volume the emphasis is on two issues. First, the chapters develop a clearer and more precise delineation of how the initiative affects other institutions. Second, the chapters develop clear, and often new, empirical tests of the impact of direct democracy. This volume aims to arrive at a clearer understanding both of what the effects of direct democracy are supposed to be and of what the effects actually are. One of the main lessons of this volume is that hybrid democracy involves trade-offs in democratic governance. On the one hand, the process, especially on participatory politics, often has the intended effects. For example, the process does—as was intended—weaken political parties. On the other hand, we also see that direct democracy can have unintended effects. For example, governments seem to perform worse in the presence of direct democracy than they do in its absence: more and wider participation may, then, be bought at the expense of performance. While critics and supporters may view the initiative in black-and-white terms, the overall picture of the effects of direct democracy are better painted in shades of gray. Before introducing the chapters themselves, we begin by outlining the different kinds of direct democracy processes and how they contribute to a hybrid democracy. The Development and Types of Direct Democracy Twenty-four of the U.S. states encompassing some seventy percent of the U.S. population allow some form of direct democracy. There are three main forms of direct democracy: the initiative (which can be either direct or indirect), the referendum, and the recall.1

“HYBRID DEMOCRACY” AND ITS CONSEQUENCES

3

Under the direct initiative, voters bypass the legislature and make a proposal directly and vote on a piece of legislation after circulating a petition. Under the indirect initiative, proposals are submitted first to the legislature, which may act on them. Only if the legislature does not act on the proposal it goes to the ballot. In some states with the indirect process, the legislature may submit a competing measure that appears on the ballot along with the original proposal. Under the referendum, voters vote on a proposal put on the ballot by the legislature. The referendum is most familiar from Europe, where referenda are sometimes held on matters relating to the European Union. For the most part, referenda are held on questions and at times that politicians choose, though a referendum can be mandated under the state constitution. In the United States, state constitutional changes and some tax or bond measures are subject to mandatory referendum. The mandates necessarily limit the scope of the legislature to craft or time the proposal in these cases. Another version of the referendum appears when the legislature passes a piece of legislation to which many voters object. A petition may be circulated requesting that this piece of legislation be put to a popular vote. Should enough signatures be gathered, voters get a chance to support or reject the legislation. Under the recall process, voters can decide to hold a special election on removing from office an elected official (governor, legislator, or judge) before the end of his or her term of office. A famous example of the recall is the removal from office of Gray Davis, governor of California, and his replacement by Arnold Schwarzenegger in 2003. Generally speaking, the direct initiative attracts the most controversy because it grants both agenda-setting and veto power to the voters themselves. For many critics the recall runs a close second because it allows for near-constant churning in political leadership. But most critics reserve their real ire for the initiative because, at least according to them, it detracts from and damages the “republican” or representative form of government. The initiative, then, represents a challenge to the system of government. States differ in their requirements for the initiative and, most especially, for the process that sees initiatives qualified to be placed on the ballot.2 Generally, however, the process includes these steps: (1) preliminary filing of a proposed petition with a designated state official (for example, the state attorney general or secretary of state); (2) review of the petition for conformance with statutory requirements and, in several states, a review of the language of the proposal; (3) preparation of a ballot title and summary;

4

SHAUN BOWLER AND AMIHAI GLAZER

(4) circulation of the petition to obtain the required number of signatures of registered voters. This threshold is usually expressed in terms of a percentage of the votes cast for a statewide office in the preceding general election; (5) submission of the petitions to the state elections official, who must verify the number of signatures. If enough valid signatures are obtained, the question goes on the ballot or, in the indirect process, the proposal is sent to the legislature. Depending upon the state, any issue on any topic, either constitutional or statutory, may be placed on the ballot for vote approval and, typically, a simple majority of votes cast is needed to pass the measure into law or into the constitution. In most states direct democracy—in whatever form—was introduced just before World War I as part of a package of reforms associated with the Progressive and Populist movements. Part of the reform effort was an attempt to clean up massively corrupt state government. In California, for example, railroad interests literally bought legislative votes. The initiative was intended to act as the “spur in the flank” or the “bit in the teeth” of state legislatures. Or, as a more menacing phrase had it, the initiative promised to be the “gun behind the door” that would persuade legislators to behave themselves. “Hybrid” Democracy: Initiative Politics and the Reshaping of Democracy The initiative raises a series of concerns. Some of these concerns relate to the specific issues taken up by proponents, and many more relate to the consequences of granting agenda-setting power to players outside the “regular” political process. After all, were the initiative a matter of a handful of propositions on, say, the topic of animal welfare, it would hardly be seen as challenging representative democracy or threatening legislators: the gun behind the door would be a water pistol. But for critics the problem with the initiative is that there are not one or two proposals but hundreds; and the proposals are not always concerned about narrow topics but may introduce sweeping constitutional changes to budgetary processes. Since the first appearance of initiatives on the Oregon ballot in 1904 with two proposals—one proposal on primary elections and one on alcohol regulation—over 2,100 proposals have appeared on ballots across the country, with about fourty percent of them adopted. The period from its introduction through to the 1930s saw a flurry of initiative proposals put to voters on topics ranging from women’s suffrage (Arizona and Oregon in 1912) and Prohibition (Arkansas 1912) to the establishment of a state fair

“HYBRID DEMOCRACY” AND ITS CONSEQUENCES

5

(Colorado 1912) to licensing chiropractors, prohibiting vivisection, and making sure that vaccination not be made a condition for admission to higher education (California 1920). Use of the process declined around the time of World War II, dipping to a low point in the 1960s. Beginning in the late 1970s, use of the process revived and has continued to climb until the present day. The post-1970s period has seen a similarly broad set of proposals put to voters, ranging from California’s Proposition 13 property tax limitation (1978), to attempts to regulate the treatment of nuclear waste (Oregon and South Dakota 1980), more regulation and taxation of alcohol (Colorado 1982, Ohio 1983), and banning cockfighting (Arizona 1998). In all these examples the power to propose and pass legislation is taken from the hands of those who are skilled at making policy (legislators) and put into the hands of the unskilled (voters). For critics, the consequences are plain. Speaking of California, Peter Schrag argues that the accumulating effects of 25 years of initiatives—from the tax limitations of Proposition 13 in 1978, to Proposition 98, the school spending formula passed in 1988, to term limits (1990), to the latter-day ballot-box budgeting that mandates spending on everything from parklands to roads to afterschool day care—have so hamstrung both state and local governments that elected legislators, county supervisors and school board members have become the handmaidens, not the leaders, of policymaking in California. Because of it they’ve become increasingly unable (and sometimes unwilling) to set priorities and respond to problems when they occur. (Schrag 2003)

The National Conference on State Legislatures (NCSL) Task Force on direct democracy made the point more broadly when it concluded that the initiative has evolved from its early days as a grassroots tool to enhance representative democracy into a tool that too often is exploited by special interests. The initiative lacks critical elements of the legislative process and can have both intended and unintended effects on the ability of the representative democratic process to comprehensively develop policies and priorities. (NCSL 2001, vii)

Embedded within these assessments is the idea that the initiative has broad and deep implications for the broader workings of the rest of the political process. The argument of the NCSL report, of Schrag, and of others is that the initiative changes the way government performs. Furthermore, the impact of direct democracy goes beyond policy outcomes. Yes, the initiative will shape taxes or determine which animals may be protected, but the

6

SHAUN BOWLER AND AMIHAI GLAZER

NCSL report concluded that the democratic process as a whole is affected by the workings of the initiative. It is more than a matter of some specific policy outcomes. Direct democracy may also affect how both parties and candidates function and campaign, what strategies interest groups adopt, and whether government can govern effectively. These are all very broad claims to make. In this volume we ask, and answer, whether those—and similar—claims hold up to systematic analysis. We also inquire into the implications of direct democracy on political parties, interest groups, and candidates. It may be that once the initiative is introduced in the system, then other actors in the system may not, or simply cannot, ignore it. These kinds of broad claims are little studied in the literature. Much of the empirical work on the initiative since its introduction has considered the process more or less in isolation from other kinds of democratic processes. Studies of voting behavior and the initiative, for example, tend to examine individual microlevel models with little institutional structure. Similarly, studies of state legislatures tend to leave to one side the role of the initiative. There is nothing terribly unusual about this approach: it is one way to make studies of complex problems more manageable. But, more than most institutions, the initiative may show ripple effects throughout the rest of the political system, in part because it is a process that may be used to force changes on other institutions of government. To date—with the notable exception of work by Gerber—little work considers the effect of direct democracy on these other institutions and processes of government (Gerber 1996a, 1996b; Gerber et al. 2000). This collection of chapters marks the first sustained and systematic attempt to assess the impact of the initiative on the other institutions, actors, and processes of government. We begin by outlining the ways in which the initiative may change the incentives and opportunities facing actors within the political system—voters, interest groups, candidates, and political parties. Effects of the Initiative on Voters and Voting Voters are the starting point in any democratic system of government and, to date, much of the work on the initiative concerns voters and their ability to cope with the demands of initiative elections. In some ways the original proponents of the initiative—the Progressives—foreshadowed academic work published nearly a century later. Many Progressives argued that the cure for the ills of democracy was more democracy. Direct democracy offered a way to end the massive corruption that characterized turn-of-the-century representative democracy. The initiative would seem to require an active, interested, and participatory

“HYBRID DEMOCRACY” AND ITS CONSEQUENCES

7

electorate. These qualities seem at odds with the restricted franchises and limited education of pre–World War I electorates, but they do seem to be qualities that are realized by the expansive franchise of highly educated electorates after World War II. Norris (1999) and Dalton (2004) document the rise of “critical citizens” as a consequence of changing social trends following World War II. These postwar generations of citizens are both more highly educated and, also, more willing to be (thoughtfully) critical of government and politics. The tool of direct democracy would seem especially fitting for new generations of these “cognitively mobilized” voters, who are much less willing than were previous generations to defer to or to be guided by the established parties (Dalton 2004). It would seem, then, that modern voters may be especially suited to the demands of direct democracy. But it is far from clear that even the “new” and changing electorate of the post–World War II period is suited to the demands of direct democracy: with democracy it may be possible to have too much of a good thing, even for the new generation of critical citizens. Precisely because direct democracy elections lack the familiar signposts of party labels and of incumbency, and because the issues generated may be complex or simply confusingly drafted and presented, some worry that initiative elections can and do confuse voters. In consequence, election results in initiative elections may be more susceptible to spending effects and slick advertising campaigns than are most other elections. In initiative elections voters may even be fooled into voting against their own interest. Some then, have serious misgivings about direct voting on policies. And though the most educated and most aware sections of the electorate can best cope with the demands of direct democracy, they are often the most dubious about the benefits of direct democracy (Dalton 2004), perhaps in part because of the role of the uninformed in determining majority decisions. Many commentators are disturbed by the way in which the initiative often sees issues brought to the fore, and decided on. The series of propositions concerning race and immigration that surfaced in California in the 1990s are clear examples of proposals that many observers, but a minority of voters, found distasteful. Much of the scholarly work on voters and voting, however, shows that voters can figure out how to vote on issues of interest to them (Bowler and Donovan 1998; Nicholson 2005). The initiative, then, may not make it difficult for voters to figure out how to vote their true preferences, but rather that the true preferences of voters strike many as unpleasant, shortsighted, narrow, or all three. But there are other, clearer cut, effects from the initiative so far as the electorate as a whole is concerned. Voters may find their interest piqued by the presence of a particularly controversial topic, even as they find candidates in a given race dull and uninspiring. The initiative is therefore likely

8

SHAUN BOWLER AND AMIHAI GLAZER

to raise turnout (Smith and Tolbert 2004). Voters as a whole may also sense a deeper efficacy as a consequence of being allowed to participate via direct democracy (Bowler and Donovan 2002). Some observers argue that the majority-rule aspect of direct democracy means that voters in a minority (and for many this means ethnic minority voters) will be relegated to the losing side in election after election. While voters in the majority may benefit from the initiative, some voters may be less sanguine about the untrammeled exercise of the majority will. This concern over majority oppression of minorities, too, has been challenged on the grounds that majorities may form and re-form in a classic pluralist sense (Miller 1983; Gerber, Hajnal, and Louch 2002). Furthermore, so long as the courts can block discriminatory legislation, the rights of the minority are protected from both direct and representative democracy. On balance, then, doubts about the capacity of voters seem related more to ideological disagreements over outcomes between observer and observed than to any innate voter inability. That is, outcomes in direct democracy elections may well reveal that voters have policy preferences different from analysts and academics and journalists, but the outcomes do not reveal that voters are confused or baffled. While the effects of direct democracy on voters and voting behavior are well established, the effects of direct democracy on other actors (interest groups, candidates, and parties) are less well established. Effects of the Initiative on Interest Groups, Candidates, and Parties In principle, direct democracy opens up ways for citizens to bypass the standard vehicles of mobilization—interest groups and political parties. This ability is especially pronounced if the arguments about the development of new generations of capable and independent-minded voters hold true. The “new” generations of voters (the boomers, Generation X, and the others) do not necessarily require the established political parties or existing interest groups to organize them to participate in politics. Rather, these generations of voters could either organize themselves or, more realistically, organize fresh parties or new organizations to take advantage of the opportunities afforded by direct democracy. In short, if direct democracy opens up scope for participation in a general sense, it should also open up participation in the specific sense of changing the identity of the (organized) players. Of course, the existing players—the established parties and the existing interest groups—are not passive observers: they can adapt and respond both to the opportunities opened up by direct democracy and to the challenge of new rivals. The presence of the initiative may therefore change the strategies of both existing interest groups and existing political parties.

“HYBRID DEMOCRACY” AND ITS CONSEQUENCES

9

A common criticism of direct democracy is that interest groups use the process to twist public policy outcomes to their advantage and serve vested interests at the expense of the public good. By appealing to the voters directly and over the heads of legislators, the initiative may spur not simply interest-group politics in general, but special interest politics in particular. Although systematic research casts considerable doubt on the claim that special interests dominate the process (Gerber 1998), the presence of the initiative nevertheless likely leads groups (and candidates) to shift, or at least to reconsider, political strategies such as donations to candidates or campaign effort. The initiative provides an opportunity for interest groups to sidestep a reluctant legislature and pursue an agenda directly with voters. This gives groups a wider choice of venues in which to engage in political action. Groups can put on the ballot issues of concern to them but that have a hard time gaining traction in the legislature. So, for example, we may see initiatives that pursue particular goals such as the protection of mountain lions, or the banning of cockfights, or simply outlawing the consumption of horsemeat. Each of these issues—which were initiatives that passed state ballots—represents the agenda of specialized groups that would have little influence inside a legislature. The initiative, then, allows the pursuit of interests and preferences that may be squeezed out of normal politics. But the initiative can threaten some interest groups, as competing groups try to use it to undercut each other’s base or organizational structure. Labor groups, for example, were threatened by initiatives in both Washington and California that restricted their ability to raise and spend money on political campaigns by requiring that union members make a conscious decision to pay political levies. The attempt to regulate union political levies is just one example of jockeying for partisan advantage via the initiative. But, as with the potential for changing the behavior of political parties, some implications reach beyond partisanship. The pluralist model of politics requires that groups be allowed to participate in the political process. Like political parties, interest groups provide a key linkage organization between citizens and the wider political system. The initiative may be used to expand the ways in which groups can participate in politics and provide them with new opportunities. The initiative, then, could encourage interest groups to form and act, and so enhance their linkage functions. Policy outcomes with the initiative can therefore differ from those absent the initiative because the set of interests and interest groups taking part is different (Gerber 1998). The example of the political levies, however, reveals another possibility—the initiative being used, possibly for partisan reasons, to restrict the abilities of groups to take part in the process. The initiative therefore has the potential for dramatically affecting group activity by

10

SHAUN BOWLER AND AMIHAI GLAZER

either encouraging its expansion or restricting it. Similar kinds of arguments may be seen in relation to both candidates and political parties. The most prominent finding so far on how direct democracy affects candidates is that when a candidate race is held at the same time as an initiative vote on a controversial issue such as gay marriage, then that issue will color the campaign in the candidate race and either be incorporated into voter assessments or shape the positions adopted by candidates or both (Nicholson 2005). In the 2004 U.S. presidential election, many believed that the initiatives on gay marriage had significant effects. Eleven states—including the crucial state of Ohio—had defense of marriage (anti–gay marriage) initiatives on the ballot. The initiatives had two consequences for the presidential race. First, socially conservative voters—a core vote base of George Bush—had reason to turn out for the polls in those states regardless of how lukewarm they felt about the president. The general bump in turnout associated with having initiatives on the ballot thus had a partisan effect in this instance and a potentially decisive one in certain key states and, hence, the election as a whole (see chapter 2 by Daniel Smith, which outlines this argument in greater detail). Second, in those states—and in line with Stephen Nicholson’s work—candidates were more likely to be judged on their stance on gay marriage (Donovan, Tolbert, and Smith 2005). The example of defense of marriage initiatives on the ballot illustrates a general strategy: candidates and parties may use the initiative to advance wedge issues to split the base of their opponents. California Governor Wilson’s use of a series of propositions in the 1990s provides a clear example of this strategy, as the Republican Party used these issues concerning immigration and affirmative action that helped split the Democratic vote. Some scholars see in these examples the way in which political parties act as the central players in the initiative process (Smith and Tolbert 2004; Smith, this volume). Schwarzenegger’s use of the initiative, noted at the outset, is just the most recent, albeit an extreme, example of how candidates and parties use the process to their advantage. The well-known model of party government outlined in the American Political Science Association report a generation ago looks more tenuous today than it did 50 years ago. In part this is because a series of changes—in the electorate, in parties themselves, and in governance and administration—have changed the ways in which parties operate within the system (Dalton and Wattenberg 2002). Parties are less central players in the game of democratic politics than they once were, yet they still remain central in both government and the definition of the party government model. For the party government model to work, parties need to be long-lasting organizations that retain the capacity to nominate

“HYBRID DEMOCRACY” AND ITS CONSEQUENCES

11

candidates and to fight elections. In principle the initiative can weaken parties by undercutting their ability to organize. Initiatives that adopt open primaries can take away the crucial power of nomination from the party. By weakening parties, therefore, the initiative may well weaken the party government model. Beyond identifying this possibility for strategic behavior, little is known, however, about the relationship between the initiative and political parties and, in particular, about the impact of the initiative upon parties. Yet the initiative itself was introduced with the aim of restraining and disciplining parties and legislatures; for that reason alone it is reasonable to think that parties are as much subject to the initiative as masters of it. As one early supporter of the initiative put it, “The American people despise legislatures, not because they are averse to representative government, but because legislatures are in fact despicable” (Ford 1912, 72), and the initiative allows voters to give vent to that scorn. To some extent, whether the initiative weakens parties and strengthens interest groups, it does generate debates over differing values. For some, the weakening of parties weakens the party government model and therefore damages a working democratic process. For others, as in the comment by Ford, it is good that parties are weakened. Regardless of that value judgment, the question is whether the initiative does, in fact, have this effect or indeed many of the other effects we discussed in this section. If the empirical work on the impact of the initiative on actors such as political parties or interest groups is sparse, it seems lush and tropical compared with the research on the impact of the initiative on governing and governability. Effects of Direct Democracy on Governing and Governability We see that the initiative can change the behavior of political actors: it can motivate voters to turn out when they would otherwise stay at home, give “outsider” groups the chance to influence policy, or force candidates and special interest groups to confront issues they would rather not. All these suggest that the initiative can make public discourse broader and more open. The initiative can also unsettle and disrupt the “normal” operation of political institutions, restricting the ability of parties and groups to organize and mobilize. In the turbulent 1970s and 1980s, political science considered the questions of governability and the overload of government. Modern nation states were held to be increasingly hard to govern, in part because of the many expectations and the many demands made on them by citizens: if too many citizens make too many demands, governments become

12

SHAUN BOWLER AND AMIHAI GLAZER

overloaded, causing ungovernability (see Birch 1984 for a critical review of leading theories). Current critics of the initiative echo many of the arguments from this literature. The initiative allows citizens (and groups) to make too many, and too many unreasonable, demands on government. In a purely representative democracy, legislators can exercise trusteeship and so moderate the demands of groups, or broker compromises between rival interests, or act as a gatekeeper to keep out the more extreme proposals. Legislators— within this view—perform a useful service when translating public preferences into policy by sorting out good ideas from bad, or reasonable ones from unreasonable ones. Direct democracy, however, provides no space for this useful filtering of ideas. Groups and individual candidates can shortcut or do an end run around the legislature and, in consequence, too many bad ideas are put directly into practice with the end result of overload and ungovernability. We know that some policy outcomes under the initiative differ from those produced by legislatures (Gerber 1998). The potential for the initiative to affect parties and their behavior, especially the behavior of their members in the legislature, has attracted much rhetoric over the years but little analysis. One exception is found in the work of Gerber, who shows that the initiative pushes legislators to pass laws closer to the median voter (Gerber 1996b). In part because of such low regard for politicians, the initiative has long been associated with a series of reforms such as term limits, redistricting, or open primaries—reforms that directly shape the institutions of government themselves. These reforms pose particular troubles for parties. Not only is policy making taken out of their hand by the initiative, but the parties and party organizations themselves become targets for proposals such as term limits. Moreover—regardless of the actual content of the reform—changing the very rules of the game for parties every few years constantly forces the parties to adjust their behavior and organization. While parties may, then, try to use or at least shape specific initiatives— especially ones that are suitable to mobilizing support or that create wedges—parties may also find their workings disrupted in fundamental ways, with important consequences for their capacity to govern. A clear example of this potential for wide effects comes from votersponsored initiatives on budgetary matters—often generically known as tax and expenditure limits, or TELs—which bind the hands of legislators. California’s Proposition 13 (1978), Oregon’s Ballot Measure 5 (1990), or Washington state’s I-601 (1993) are all examples of the ways in which states have their ability to raise revenue constrained by proposals introduced and passed by way of the initiative (Steel and Lovrich 1998). These kinds of proposals may severely limit the ability of legislators to raise

“HYBRID DEMOCRACY” AND ITS CONSEQUENCES

13

revenue, and often limit their flexibility to respond to crisis when state revenues drop during economic downturns. It is worth remembering, however, that supporters of direct democracy argued that binding the hands of legislators was precisely what should happen: untrustworthy and undisciplined legislators should be disciplined by voters; the initiative is the tool, and possibly the only tool, that gives voters such power. If reluctant legislators won’t spend enough on education, then the initiative allows voters to mandate that (Proposition 98 referred to by Schrag). On the other hand, if legislators tax and spend excessively, then voters can simply restrain the ability of politicians to raise taxes (Proposition 13). The more general point illustrated by the specific example of TELs is that initiatives limit flexibility in policy making. It is a point that is consistent with the theoretical work on direct democracy of Tsebelis (2002), who argues that the institution of direct democracy adds another veto player to the policy-making process, thereby making policy change much harder. Loosely speaking, it is hard enough to get policy changed if it has to pass through two houses of a legislature and then an executive. If we have the additional hurdle of a direct democracy process that could either overturn the policy or force some reaction by the legislature (Gerber 1996b), then change becomes even less likely: herding cats is not made any easier by adding more cats to the herd. Little is known, however, about the actual effect of the initiative on the rest of the system. Take a seemingly straightforward example of the kind of claim made by Schrag, referred to earlier. It is widely claimed that the initiative constrains the freedom of action of budget makers in many and serious ways, often through tax and expenditure limits. In one of the few systematic evaluations of the restrictions—on the California budget—John G. Matsusaka shows that they are much less severe than critics tend to make out. Approximately 32 percent of California’s general fund, for example, is tied up by restrictions imposed by the initiative (2005a). This figure is lower than we might be led to expect, given the comments of Schrag and others. Nevertheless, even one-third of the budget constrained in this way might limit policy flexibility and damage government’s ability to respond to changed fiscal circumstances. Taken together, the arguments of Tsebelis and of Matsusaka imply that in states that have and use the initiative, fiscal policy should respond more sluggishly and change more slowly than in states that do not have the initiative. Existing political institutions and policy-making processes may thus be faced with two quite different pressures stemming from the initiative. On the one hand are constraints that impose excessive stability on policy making; on the other are a series of changes that unsettle the institutions themselves of the kind noted earlier in relation to political parties.

14

SHAUN BOWLER AND AMIHAI GLAZER

These pressures can have several consequences. First, we are likely to see different sets of institutional arrangements in initiative states—produced in large part by the initiative. The differences are likely to be especially large in political parties, which seem to be a favorite target of reformers. Many of the reforms passed by the initiative have been aimed at parties and their members in the legislature through introducing open primaries and term limits, changing redistricting, and implementing other reforms. Second, there is an overall effect on policy and policy making. The discussion so far offers a sizable list of impacts that the initiative may have on the other institutions of government, ranging from the restrictions on legislatures; the series of “reforms” that lead to a series of changes in government; the weakening of parties (and the possible weakening of a party government model); the change in incentives presented to voters, candidates, parties, and special interests; and the direct restraints placed on policy and policy making itself. The overall consequence is that states that use the initiative may be more poorly governed—and possibly by a large measure. The initiative can be seen to have wide-ranging effects and has yet wider implications still for our understanding of democratic governance. But while there exists a series of statements and conjectures concerning the impact of direct democracy, little hard evidence speaks to this question: how are initiatives and referenda changing the process of democratic governance? The Plan of the Book Our analyses were developed through a series of planning meetings and conferences hosted by the Center for the Study of Democracy at the University of California, Irvine, and by the Initiative and Referendum Institute at the University of Southern California. We first agreed to focus on the comparative experience of the U.S. states. The experiences of California and Oregon, for example, with extensive use of initiatives and referendums, should provide a natural contrast to states such as Mississippi and Illinois, which have the process but rarely—if ever—use it. We also agreed to focus on an empirical social science approach rather than a normative or legal one in comparing across the fifty states. Beginning with this common framework, the project assembled leading scholars on direct democracy to collect and analyze the empirical evidence. These analyses led to the research collected in this volume. We divide our answer to the impact of the initiative into two broad sections. The first looks at the impact of the initiative on the actors in democratic politics—voters, candidates, parties, and interest groups. The second part examines policy making. As we will see, the kind and size of impact that direct democracy has differs across the two sections, implying that there are trade-offs inherent within “hybrid democracy.”

“HYBRID DEMOCRACY” AND ITS CONSEQUENCES

15

Citizens, Interest Groups, Candidates, and Parties The first chapters examine the impact on actors in the democratic system—voters, candidates, parties, and interest groups. Caroline Tolbert’s study of voters shows that exposure to ballot measures produces citizens who are more interested in the election, and are more likely to vote. Part of the reason for the increased participation is that ballot campaigns provide an extra source of information and emotional content to voters, encouraging civic engagement and participation. This increase in civic engagement is especially important in an era of generally declining interest in and regard for the political process. Fred Boehmke considers how the initiative affects the behavior of special interest groups. He finds that groups involved with initiatives are also involved in legislative lobbying. The initiative does not, therefore, cause groups to substitute action in one arena for action in another. The result contradicts a standard assumption that special interest groups face fixed budgets or large opportunity costs for lobbying. Instead, increased opportunities for political action just increase spending by special interest groups, resulting in lobbying of legislators remaining at the same level. For candidates and parties the message is a little more muddled. Smith examines a key example of the way in which initiative campaigns can be used to mobilize support. During the 2004 election the Republican Party used the issue of gay marriage to generate enthusiasm and turnout among conservative groups that seemingly stayed at home in 2000. The 2004 election promised to be another close one, whose outcome could depend on turnout among the vote base in a handful of states. Smith examines whether this strategy worked in the key state of Ohio as Republicans relied on gay marriage initiatives to carry the state. Note that the chapter by Smith differs from many of the others in this volume as it asks not how outcomes in states with the initiative differ from outcomes in other states, but how a political party may use the initiative for electoral purposes. Nicholson addresses the reverse effect: the impact of initiative contests on the public image and campaigns of candidates. While candidates may hope—as Smith discusses—to use the initiative for their own purposes, the potential for ballot propositions to frame contests means that candidates may be painted into a corner by the issues raised in initiative contests. Ballot propositions can be a two-edged sword; between them the chapters by Smith and Nicholson discuss both those edges. From here, Shaun Bowler and Todd Donovan examine political party systems to show that direct democracy alters the institutional context for parties, sometimes leading to changes that weaken them in substantial ways, thereby disrupting the “party government” model. Despite a decline in the importance of political parties around the world (Dalton and Wattenberg

16

SHAUN BOWLER AND AMIHAI GLAZER

2001), parties and party government remain central to the model of representative government. For the representative and party government model to work, parties need to be standing organizations that endure in their capacity to nominate candidates and fight elections. Yet, as we noted earlier, direct democracy was introduced with the aim of restraining and disciplining parties and legislatures. For that reason alone it is reasonable to think that parties are as much subject to the initiative process as masters of it. The Executive and Policy In the second section the collection moves away from examining the impact of the initiative on political actors to look at the ways in which the initiative shapes policy and policy making. In his study of the executive, Matsusaka shows that direct democracy can make the electorate more sensitive to the performance of governors at the time they are up for reelection. If voters are well informed and have good choices among candidates, then the introduction of referenda should make little difference to how they behave when a governor is running for reelection. But Matsusaka finds that voters punish an incumbent more strongly for poor performance when referenda are available. The evidence is consistent with his argument that this difference arises because referenda remove some issues from the reelection campaign, allowing voters to focus on the incumbent’s performance. Thus, the study of referenda sheds new light of gubernatorial elections, both in states with and without referenda, suggesting that the multiplicity of issues in a campaign makes it difficult for voters to hold elected officials responsible for their performance Amihai Glazer and Tony McGann examine stability of public policy. Using detailed analysis of changes in budget patterns, they find that states that use direct democracy show about the same stability in spending policy as states that do not. The initiative does not, then, cause instability in the budgetary process, much in line with the predictions of Tsebelis and others. Lastly, Russell Dalton assesses the claim that initiative states should be more poorly governed than noninitiative states. Across a range of measures—including reputational rankings, tax performance, and educational performance—he finds that the initiative can result in less effective government. A Note about Methodology One of the arguments made by this volume as a whole is that—in studying direct democracy—the temptation is to examine one dimension of

“HYBRID DEMOCRACY” AND ITS CONSEQUENCES

17

direct democracy and rest an argument about direct democracy as a whole on findings from that single dimension. We argue that the relationship of direct democracy to governance is really multidimensional and that there may be trade-offs across the dimensions for the conduct of governance. We must, therefore, be cautious before making too many sweeping claims about direct democracy in general. A broadly similar argument of caution also applies to the means of studying the impacts of direct democracy and, in this section, we outline some of the reasons for urging caution in interpreting results. Most of the chapters in this volume look at the connections between the initiative and behavior by comparing outcomes in states with the initiative and in states without the initiative. It is simple and direct to speak of the initiative causing such and such an outcome. A different approach that could be useful, but that this book does not pursue, is the study of interrupted time series. In this form of a test we observe subjects in a group who, at various times, adopt direct democracy. If we can predict outcomes at time t⫹1 in a time series (from various observations of t, t⫺1, and so forth, plus the interjection of notable events), we can predict the level of our dependent variable in each instance for all those subjects who did and did not adopt direct democracy, and see if there was a significant change in the trend line for those who got direct democracy. A difficulty with this approach is that if our time-series predictive ability is not almost perfect, then we have little confidence that we are finding a difference owing to direct democracy rather than just error. We are aware that cross-sectional studies may be unable to determine causation. But, for our purposes, some of the perils of cross-sectional inferences are absent. The constitutional rules of direct democracy were often established at the beginning of the twentieth century, and changes in rules are slow to respond to changes in political preferences. Thus, we think it more likely that the initiative results in low spending than that low spending (or a desire for low spending) was caused by adoption of the initiative. True, the causation can still be unclear for a variety of reasons. States with the initiative may differ from other states in several ways that we do not capture in our models, although the models presented in this volume try hard to capture the “usual suspects” of those differences. Second, though our thinking is that constitutional rules largely reflect historical preferences rather than current preferences, that may not always hold. If, for example, the initiative really does control spending, then a current majority that wishes to reduce spending may vote for a constitutional change. As we noted earlier, however, in general the institutional rules of direct democracy precede many of the patterns of policy that the authors examine. Furthermore, the rules have been largely unchanged since their inception.

18

SHAUN BOWLER AND AMIHAI GLAZER

We do believe that the patterns we examine can speak to some of the claims advanced by advocates and critics. If the claim of, say, a critic is that the initiative causes X, then a necessary condition for that argument to hold is that the initiative and X should be correlated. If we find no such correlation, then the criticism may be invalid, regardless of claims of causality. That is, though we must be cautious in advancing causal claims on the basis of cross-sectional evidence, we can nevertheless advance our understanding of the effects of the initiative within the limitations of the data at hand. Of course, in addition to being cautious about causal claims, we should also be cautious about measures. Measurement in social science is often an imperfect art—perhaps nowhere more than in examining the relatively new areas of governance that we address. We are aware that the measures employed may be imperfectly related to constructs discussed in the hypotheses. For example, it may be that state A has strong constitutional provisions for the initiative, whereas state B has weaker provisions. Nevertheless, state B may make greater use of the initiative, and so looking at constitutional provisions may be the wrong measure. But this problem may be mitigated in several ways. First, a bad measure of direct democracy will often find little effects on outcomes—a better measure may find even stronger effects than those found using a noisy measure. Second, if a measure of direct democracy finds no effect on outcome X, but does for outcome Y, then we have some confidence that the measure has some meaning (it did after all affect Y), and that the small effect on outcome X is not merely the artifact of poor measurements. Lastly, one way in which we can address the limitations of individual measures is by advancing several measures of the same construct. In the chapters that follow, the arguments of the authors are generally not carried by reference to a single measure but by reference to several measures of the same underlying concept. The limitations noted above are worth bearing in mind, but still allow the research reported in these chapters to be valuable. Causation is not the only matter of interest. We may, for example, be interested in predicting spending in different states, and for that purpose a finding that states with the initiative have lower spending is most useful, regardless of causation. Initiatives and the Democratic Process These chapters yield an overall picture of the initiative that is more measured than is seen in the literature to date. In some ways the picture that evolves is not a straightforward one. Greater participation and the breaking down of standard means of mobilizing and organizing voters is bought at the expense of some great complication in governability: greater participation is bought

“HYBRID DEMOCRACY” AND ITS CONSEQUENCES

19

at the expense of performance. That said, the consequences of direct democracy are not as dramatic or as catastrophic as the language of alarmists or the language of boosters suggests. Overall, the results are somewhat encouraging so far as representative democracy is concerned, since they suggest that representative democracy—even when changed by the presence of the initiative—is a robust creature. Taken as a whole, our findings suggest that it is harder to destabilize or disrupt the workings of representative government than reformers might like and institutional conservatives fear. Notes 1. Succinct accounts of direct democracy may be found on the Web sites of the Initiative and Referendum Institute (http://www.iandrinstitute.org/) and the National Conference of State Legislatures (http://www.ncsl.org/programs/legman/ elect/initiat.htm). The following section draws heavily on the accounts on both those two Web sites and Bowler and Donovan (2004) for state-level variation in the process. 2. This statement and the following section draw heavily on the accounts from an especially clear exposition found on the National Conference of State Legislatures (http://www.ncsl.org/programs/legman/elect/initiat.htm).

This page intentionally left blank

2

Was Rove Right? Ohio’s Gay Marriage Ban and the 2004 Presidential Election Daniel A. Smith

id ballot measures banning same-sex marriage swing the 2004 general election for President George W. Bush? In 2004, activists and state legislators placed anti–gay marriage questions on the general election ballots of 11 states. All of the ballot measures passed easily, receiving on average roughly 70 percent support.1 Pundits argued that the marriage measures on the November ballot would be a major motivating factor in the election and would help to ensure Bush’s reelection. The measures, so the logic went, would receive broad support from social conservatives who would be mobilized to go to the polls and be primed to vote for Bush, who was firmly aligned with the issue. Merging county-level religious, socioeconomic, and political data with the 2004 election results, I examine the electoral effects of same-sex marriage ballot measures in the presidential battleground state of Ohio. The anti–gay marriage measure passed easily and statewide turnout was higher in 2004 than in 2000. To test the impact of the anti–gay marriage measure in Ohio, I examine first the county-level voting patterns on the gay marriage ban, then the patterns of county-level turnout in 2004, and finally the county-level support for Bush in 2004.

D

The Mobilizing and Priming Effects of Ballot Measures Following the election, numerous opinion-makers claimed that the anti–gay marriage measures—especially the initiative on the ballot in the swing state of Ohio—were the key to Bush’s win (Davies 2004). Tony Perkins,

22

DANIEL A. SMITH

the president of the Family Research Council, asserted after the election that gay marriage was “the hood ornament on the family values wagon that carried the president to a second term” (Cooperman and Edsall 2004). The director of the conservative Culture and Family Institute think tank, Robert Knight, claimed the anti–gay marriage measures “galvanized millions of Christians to turnout and vote, and George Bush and the GOP got the lion’s share of that vote” (Lochhead 2004). For his part, Karl Rove, Bush’s principal advisor, told reporters at the Republican National Convention in August 2004 that “to the degree it energizes people who might otherwise not vote,” a ballot measure banning gay marriage “tends to help us” (Korte 2004). Most of the Monday morning quarterbacking focused on Ohio’s samesex measure, Issue 1. Reporters for many national and local papers, including the venerable New York Times, reported matter-of-factly that Issue 1 “helped turn out thousands of conservative voters on Election Day,” and that their support was “widely viewed as having been crucial to Bush’s narrow victory in that swing state” (Dao 2004. Opining in his postelection USA Today syndicated column that “gay marriage was probably the issue that spelled the difference in hard-fought Ohio” and that Bush “directly benefited from the outpouring of social conservatives” who turned out in support of Issue 1, Walter Shapiro (2004) approvingly quoted an Ohio Democratic Party strategist who claimed that “if Issue 1 had not been on the ballot, John Kerry would have won Ohio.” A postelection analysis of the county-by-county presidential vote conducted by the Wall Street Journal found that “Bush won thanks to a pitch on morals that went beyond evangelicals to Roman Catholics” and “a strong effort to turn out rural voters” (Cummings 2004). Bush did particularly well, the paper reported, in small and conservative counties, including 12 of 14 with the most heavy job losses since 2000. A story in USA Today reinforced this conventional wisdom that smaller, more conservative, exurban or “micropolitan” counties in Ohio were central to Bush’s victory, with the Republican winning 27 of 29 such counties with “rural sensibilities” (El Nasser 2004). In contrast to the conventional wisdom of the media following the election, recent scholarship on the electoral effects of the anti–gay marriage amendments is decidedly mixed. Relying on aggregate national turnout data and statewide presidential votes, Abramowitz (2004) and Burden (2004) find that the 11 states with gay marriage bans in November did not have higher voter turnout or support for Bush than states without the bans, after controlling for the state’s rate of turnout in 2000, if the state had a 2004 U.S. Senate race, or if it was a “swing” state in 2004. By extension, they contend that the same-sex marriage bans did not contribute to the

WAS ROVE RIGHT?

23

reelection of Bush. Preliminary analyses using county-level data are more mixed. A few indicate that support for the gay marriage bans did not lead to higher turnout or a positive effect of support for Bush in Ohio (Jackman 2004; Freedman 2004), but others suggest that the measures may have had a positive effect on Bush’s victory across the 11 states with the gay marriage bans, though they did not boost turnout (Campbell and Monson 2005). Studies using individual-level data national surveys are also mixed. Some conclude that the issue of “moral values” (including support for banning gay marriage) was not central to Bush’s reelection (Hillygus and Shields 2005; Lewis 2005). Others, though, find that the issue of gay marriage was more important to voters living in states where the measures were on the ballot and, more significantly, may have helped to prime voters’ support for Bush (Donovan, Tolbert, and Smith 2005). Theoretically, there are good reasons why ballot measures—particularly ones so polarizing as those banning same-sex marriage—might be expected to mobilize certain citizens to the polls and prime them to vote for candidates. First, there is solid scholarly evidence supporting the operating assumptions about the potential electoral effects of ballot measures (Smith and Tolbert 2004; Nicholson 2005). Researchers have shown that ballot measures increase voter turnout, particularly partisans (Smith 2001; Tolbert, Grummel, and Smith 2001; Tolbert and Smith 2005). Scholars have also shown that voters tend to draw on elite and partisan cues to inform their decisions on how to cast their ballots on ballot measures (Lupia 1994; Bowler and Donovan 1998, and that the ensuing votes on ballot measure tend to be split along partisan lines (Branton 2003; Smith and Tolbert 2001). Finally, there is new scholarly evidence showing that ballot measures may have priming effects on candidate races (Nicholson 2005; Donovan, Tolbert, and Smith 2005). The Political Context In the buildup to the 2004 election, there were many reasons to expect that the gay marriage amendments might create a perfect storm in the reelection of the president. In late February 2004, Bush expressed his support for a U.S. constitutional amendment banning gay marriage. Citing the efforts of activist judges in Massachusetts and local officials in San Francisco to redefine marriage, the president claimed that the 1996 Defense of Marriage Act (DOMA), signed into law by President Bill Clinton, could be undermined by the courts and did not protect marriage in the states. Nearly 40 states already had their own versions of DOMA on the books, including Ohio (Williams 2004). Over the summer, though, the U.S. Senate rejected a federal marriage constitutional amendment, which many media

24

DANIEL A. SMITH

outlets reported as “a big election-year defeat” for the White House (Cooperman and Edsall 2004). Meanwhile, in roughly a dozen states already with statutory DOMAs, there were efforts underfoot in early 2004 to place amendments on statewide ballots that would constitutionally define marriage as between a man and a woman (O’Connell 2006). By April, proponents in Ohio announced they would begin circulating petitions to place amendments with traditional definition of marriages on the November ballots. The gay and lesbian community announced a countermobilization strategy and initiated court challenges to both the validity of the petition’s signatures and the amendment’s constitutional language. State courts, however, rejected these challenges, and the measure was subsequently placed on the ballot for citizen approval. In early August, the Ohio Campaign to Protect Marriage (OCPM), an evangelical group, submitted 575,000 signatures to the Ohio Secretary of State, easily surpassing the 316,888 valid signatures needed to place a constitutional amendment on the ballot, and the Ohio Ballot Board unanimously approved the measure’s title and two-sentence ballot summary. Issue 1 would amend Section 11 of Article XV of the state constitution, stipulating that “only a union between one man and one woman may be a marriage valid in or recognized by this state and its political subdivisions,” and that the “state and its political subdivision shall not create or recognize a legal status for relationships of unmarried individuals that intends to approximate the design, qualities, significance or effect of marriage.” On September 29, Ohio’s Secretary of State certified the initiative for placement on the ballot. Research Design I compiled a dataset of Ohio’s 88 counties to probe where the ballot measure banning same-sex marriage gained its electoral support as well as to test whether it helped to increase turnout and aid in the reelection of Bush. Aggregate-level data, of course, cannot be used to infer individual patterns of political behavior without risking an ecological fallacy (King 1997). However, while not well suited for making inferences about individuallevel voting behavior, county-level data are useful when exploring the political, economic, and social environment in which voters made their electoral decisions (Key 1949; Giles 1977; Oliver and Mendelberg 2000; Hero 1998; Tolbert and Hero 2001). These data are especially useful when trying to explain turnout and voting patterns, as survey data may contain nonresponse and other sampling errors, or may accentuate the social desirability of participation and vote choice (Belli et al. 1999). Finally, in Ohio, counties are particularly a useful unit of analysis to understand voter

WAS ROVE RIGHT?

25

turnout, as they are the administrative arms of the state and are key electoral arenas in which candidates and parties wage their campaigns. Previous research utilizing aggregate-level analyses of political behavior demonstrates that partisan, religious, and socioeconomic compositions often influence the overall political behavior of communities or counties (Putnam 1966; Foladare 1968; Huckfeldt, Plutzer, and Sprague 1993). For example, Foladare (1968) demonstrated that Protestants who lived in predominantly Catholic communities had significantly different partisan identification than those Protestants who lived in communities that had a large Protestant population, and vice versa. Paralleling religious identification, scholars also have demonstrated that the racial composition of a state influences the type of racial policies they enact (Giles and Evans 1986; Hero and Tolbert 1996; Hero 1998; Branton 2004). My research design combines elements from these perspectives in assessing the influence of various communities on the voting patterns on the gay marriage issue and changes in the level of turnout and Bush’s vote share in Ohio. Expectations and Specifications of Models This investigation into the electoral and turnout effects of Issue 1 lends itself to three testable hypotheses. H1: Counties with greater densities of Republicans, rural and exurban populations, and evangelical Protestants and Catholics will have higher levels of support for the ballot measures banning same-sex marriage, compared with other counties. H2: Counties with greater support for the measures banning same-sex marriage will have higher levels of turnout in 2004, compared with other counties. H3: Counties with greater support for the measures banning same-sex marriage will have higher levels of support for Bush in 2004, compared with other counties. The first hypothesis provides insight into the forces at play in the passage of Issue 1, while the second and third hypotheses capture the impact the gay marriage ban may have had on turnout and support for Bush in 2004.2 A series of weighted least squares multivariate regression models estimate the aggregate-level patterns of voting on Issue 1, turnout, and support for Bush in Ohio in 2004, controlling for a host of other factors.3 In Ohio, turnout across the state’s 88 counties was 8.2 percentage points higher in 2004 than in 2000, and Bush’s vote share increased 2.0 percentage points after his four years in office. Table 2.1 details the statewide averages and the

26

DANIEL A. SMITH

Table 2.1 Statewide and county averages Ohio Variables

Statewide average

Vote in favor of initiative banning gay marriage Percentage point change in turnout, 2000–2004 Percentage point change in vote for Bush, 2000–2004

61.7% 8.2 0.8

County average 68.5% 8.2 2.0

county averages of the vote for Issue 1, the percentage point change in voter turnout, and percentage point change in vote for Bush. Expected Support for Issue 1 The first model tests whether Ohio counties with greater densities of religious conservatives, Republicans, and rural and exurban populations exhibited increased levels of support for the initiatives banning gay marriage, holding other factors constant. The dependent variable is the percent vote in favor of Issue 1 in each of Ohio’s 88 counties. Counties with dense populations of evangelical Protestants are expected to have stronger support for the measures in the two states. Evangelicals tend to view same-sex ballot measures, which restrict the rights of gays, as a tactic to battle against a way of life they perceive as threatening to their own social and cultural values (Button, Rienzo, and Wald 1997; Wald, Owen, and Hill 1990). There is also evidence at the aggregate level that communities with higher percentages of evangelicals have a great likelihood of adopting ordinances limiting the rights of gays and lesbians (Button, Rienzo, and Wald 1997, 80–82). For these reasons, then, there is good reason to expect that Issue 1 will have higher levels of support in counties with large percentages of evangelicals. The model includes the county population belonging to the Catholic Church. While in some states the backing for the anti–gay marriage measures came largely from the Catholic Church, in Ohio, most financial support for Issue 1 came from the OCPM. Since the Catholic Church did not play a central role in the Issue 1 campaign, the percentage of Catholics in a county is not expected to have a positive effect on county-level support for the initiative. The model also controls for the percentage of a county’s population belonging to other denominations that did not officially oppose the gay marriage bans. As with evangelicals, the expectation is that as the density of a county’s adherents to conservative denominations increases, so too will the support for the measures banning gay marriage. By default, the excluded (reference) category is the percentage of a county’s population

WAS ROVE RIGHT?

27

belonging to liberal denominations, combined with the percentage of a county’s population with no official religious adherence.4 Backers of the measure also hoped that the initiatives would mobilize socially conservative Republicans, priming their support for Bush. Scholars have shown partisanship to be the most predicable indicator of support or opposition for ballot measures (Smith and Tolbert 2001; Branton 2003). In 2004, the issue of gay marriage was defined along partisan lines, with the Republican Party generally supporting (if sometimes tacitly) the ban, and the Democratic Party generally opposing it. Expecting greater support for Issue 1 in counties with larger percentages of Republicans, partisanship is controlled with a measure of the county percent vote for Bush in 2000, as Ohio does not officially record the party affiliation of registered voters.5 Proponents of Issue 1 also targeted support from voters living in rural and exurban counties. Using a 9-point scale created by the U.S. Department of Agriculture (ranging from 1 = metropolitan area of 1 million population, to 9 = completely rural) to control for the rural/urban nature of a county, the expectation is that Ohio’s more rural counties would exhibit greater levels of support for the same-sex marriage bans. Included as well in the model are traditional socioeconomic control variables that researchers have found to be significant predictors in the support or opposition of gay rights measures (Button, Rienzo, and Wald 1997; Donovan and Bowler 1998). Advocates of Issue 1 had hoped to garner the support of socially conservative African Americans, envisioning that the wedge issue would pry them away from the Democratic Party’s candidate. While many African Americans self-identify as evangelicals and have a theology that often reflects that of the Christian Right, their voting habits tend to be defined by liberal economic issues (Wilcox 2000). As such, counties with dense African American populations, measured by the percentage of African Americans residing in each county in 2000, are likely to be cued by the Democratic Party to vote against the anti–gay marriage initiatives. Educational attainment and age have been found to be major cleavages in the larger gay rights debate, as individuals with lower levels of education and older individuals generally oppose increasing the rights of gays and lesbians (Button, Rienzo, and Wald 1997). Counties with a higher median age and lower level of education (as measured by the percentage of the county’s population over 25 with at least a high school education) are expected to have greater support for the ban on same-sex marriage. Finally, the economic well-being of a county is controlled with a measure of the percent change in a county’s rate of unemployment between 2000 and 2003.6 Many counties in Ohio were hit hard by rising unemployment levels during Bush’s first term. Organized labor in this heavily unionized state decided to stay on the sidelines in the pitched battle over the same-sex

28

DANIEL A. SMITH

marriage, focusing its attention instead on economic issues. Citizens in counties with increasing rates of unemployment, then, may have grown frustrated with the Bush campaign’s symbolic crusade on moral issues. Motivated instead by job growth and more immediate material relief, counties experiencing increasing levels of unemployment between 2000 and 2003, controlling for other factors, are expected to have lower support for Issue 1 relative to other counties. Expected Voter Turnout in 2004 The second model estimates the possibility of a mobilization effect of Issue 1. The dependent variable is the percent turnout in each of Ohio’s 88 counties in 2004. As mentioned previously, supporters of the same-sex marriage ban claimed that the initiative would mobilize conservative evangelicals and rural voters. Most notably, evangelicals were considered to be a bloc of voters that had allegedly stayed home four years earlier (see Teixeira 2004). Previous studies find the mobilization of evangelical Protestants is maximized when issues such as abortion, the rights of gays and lesbians, and school prayer are either on the ballot or are major components of a political campaign (Wilcox 1989; Jelen 1991; Rozell and Wilcox 1996; Carmines and Layman 1997; Wilcox 2000), and gay rights ballot measures are often seen as polarizing issues with clear partisan cleavages (Donovan, Wenzel, and Bowler 2000; Donovan and Bowler 1998; Witt and McCorkel 1997). Of course, the strategy to mobilize evangelicals in support of the anti–gay marriage measures was vulnerable to countermobilization strategies (Reichley 1986; Green 1994; Rozell and Wilcox 1996; Haider-Markel and Meier 1996; Wilcox 2000). In estimating the 2004 county-level turnout in Ohio, the second model uses the same control variables as the previous model, but adds percent turnout in 2000 and the percent vote in favor of the anti–gay measures in 2004. Both variables are expected to have positive impacts on county-level turnout in 2004, all else held equal. Expected Vote Share for Bush in 2004 The final model, also essentially a panel model, estimates the effect the same-sex ballot measures had on presidential vote choice in 2004 in Ohio. The dependent variable is Bush’s share of the county-level presidential vote in 2004. The main control variable estimating the vote for Bush in 2004 is the level of support Bush received in 2000; it is expected to be strongly positive. Even though evangelical Protestants are not always a cohesive voting bloc and are sometimes apolitical (Smidt 1988; Wilcox

WAS ROVE RIGHT?

29

1989; Jelen and Wilcox 1992; Green 1994; Rozell and Wilcox 1996; Shibley 1998; Wilcox 2000), the density of evangelicals in a county is expected in Ohio to help explain support for Bush, as Issue 1 was vigorously pushed by evangelical leaders. The other independent variables are identical to those in the previous model with the same expected directions. Findings The results of the weighted least squares regressions for Ohio’s counties are presented in table 2.2, Models A, B, and C. What is most striking about the results is that support for the anti–gay marriage measures affects the vote

Table 2.2 Percent vote for Ohio’s Issue 1 in 2004, percent turnout in 2004, and percent vote for Bush in 2004

County-level variables

Model A Percent vote for Issue 1, 2004
(se)

Percent vote for same-sex marriage bans Proportion evangelical Protestant, 2000 Proportion Catholic, 2000 Proportion other conservative denominations, 2000 Percent African American, 2000 Urban/rural continuum, 2003 (1 ⫽ urban; 9 ⫽ rural) Median age, 2000 Change in percent unemployment, 2000–2003 Percent population with high school degree, 2000 Percent vote for Bush, 2000 Percent voter turnout, 2000 Constant Adjusted R2 F N

Model B Percent turnout, 2004
(se)

Model C Percent vote for Bush, 2004
(se)

⫺.220 (.156)

.212 (.053)

.245 (.110) .000 (.036)

.087 (.156) .008 (.050)

.105 (.053) .015 (.017)

.072 (.082)

⫺.029 (.112)

⫺.019 (.038)

⫺.257 (.049)

⫺.079 (.079)

⫺.038 (.027)

.006 (.003) .012 (.002)

.000 (.004) .010 (.003)

.000 (.001) .001 (.001)

⫺.778 (.401)

⫺.835 (.561)

⫺.068 (.191)

⫺.511 (.101) .312 (.041) — .468 (.114) .678 21.360 88

⫺.108 (.164) .089 (.083) .594 (.153) .199 (.173) .402 6.323 88

.054 (.054) .957 (.025) — ⫺.180 (.059) .989 797.993 88

—

Note: Weighted least squares regression, weighted by total population of county. Sources: U.S. Census Bureau (2000); Ohio secretary of state, 2004; U.S. Department of Agriculture (2004); Glenmary Research Center (Jones et al. 2002); and Ohio Department of Job and Family Services, Bureau of Labor Market Information (2004). The model reports unstandardized regression coefficients with standard errors in parentheses. Bold coefficients indicate observed statistical reliability of 95 percent confidence intervals bounded away from zero, p-values ⬍ .1.

30

DANIEL A. SMITH

for Bush in 2004, even when controlling for his level of support in 2000. A county’s percentage of evangelicals is also positively significant in Ohio when estimating support for Bush in 2004. County-level turnout in 2004, however, is not affected by the level of support for the same-sex marriage measures in Ohio. In short, counties with higher levels of support for the measures banning same-sex marriage appear to have greater support for Bush in 2004, though not higher turnout, compared with the 2000 election. Support for Issue 1 Consistent with scholarly research showing partisanship at both the individual and aggregate levels to be a strong predictor of vote choice on ballot measures (Smith and Tolbert 2001; Branton 2003), Model A reveals that the anti–gay marriage measure received greater support from counties with greater percentages of Republicans (as measured by the percent vote for Bush in 2000). In Ohio, support for Issue 1 rises by a third of a percentage point (.31) for every one percentage point increase in a county’s support for Bush in 2000. The results also indicate that in Ohio the density of evangelicals is positively related to a county’s vote in support of the gay marriage bans. Stated differently, support for Issue 1 in counties with denser populations of evangelical Protestants is more than the support the measures received in other counties, ceteris paribus. In Ohio, support for Issue 1 rises by almost a quarter of a percentage point for every one point increase in a county’s percent of evangelicals. Furthermore, consistent with research on gay rights ballot issues showing increasing tolerance with increasing education (Button, Rienzo, and Wald 1997; Donovan and Bowler 1998), as educational attainment increases, the vote for gay marriage ban decreases. Finally, controlling for other factors, the model shows counties experiencing poor economic conditions exhibited less support for Issue 1, relative to other counties. In addition to the independent variables that help to explain a county’s level of support for the gay marriage bans in Ohio, there are a handful of other significant independent variables. The model indicates that the density of African Americans in a county is significant; controlling for other factors, as the percentage of African Americans residing in a county increases, support for Issue 1 decreases. This finding appears to run counter to individual-level survey data, which show that nonwhites in Ohio were more likely to support Issue 1 (Donovan, Tolbert, and Smith 2005). More rural counties and those with older populations were significantly more likely to

WAS ROVE RIGHT?

31

support the ban, and those experiencing higher levels of unemployment were less likely to support the ban, all of which were expected. Finally, the percentage of other conservative denominations in a county had no significant impact on a county’s level of support for the measures. Voter Turnout in 2004 The outcome variable in Model B is a county’s percent turnout in 2004. Controlling for several factors, most notably county-level turnout in 2000, the model shows that support for the gay marriage ban has no statistically significant impact on a county’s turnout in 2004. Furthermore, the percent of evangelicals in each county is not related to increased levels of turnout in 2004. Counties with older populations have higher turnout in 2004, controlling for other factors. These findings suggest that the higher county-level turnout in 2004 was not due to the support for the anti–gay measures or to those counties with dense evangelical populations having heavier turnout. Rather, turnout in 2004 was up across all counties, irrespective of the aggregate religious persuasion of a county or the median voter’s support of the anti–gay marriage amendment in a county. Vote Share for Bush in 2004 Model C provides considerable evidence that support for Issue 1 led to increased support for Bush, and that the boost from counties with dense evangelical populations contributed to Bush’s reelection. Consistent with survey-level research showing that the gay marriage bans had a priming effect for Bush in 2004 (Donovan, Tolbert, and Smith 2005), county-level support for Issue 1 appears to have given Bush a boost. Controlling for the level of support the president received in 2000 and other factors, the vote for Bush in 2004 is positively related to support for the ban on gay marriage as well as the density of a county’s evangelical population. In a state where Bush ended up winning by only 118,000 votes in 2004, the model reveals that for every percentage point increase in the vote for Issue 1, support for Bush increases by two-tenths of a percentage point; support for Bush increases an additional one-tenth of a percentage point for every one percentage point increase in a county’s density of evangelicals. So, while there is no evidence that counties in Ohio with denser evangelical populations and support for Issue 1 had higher turnout (Model B), they did provide more support for Bush in 2004, relative to other counties (Model C).7

32

DANIEL A. SMITH

Conclusion The county-level findings from the 2004 battleground state of Ohio support but also raise questions about the electoral impact of the same-sex marriage measure bans. There is evidence that counties with dense levels of evangelical Protestants voted strongly in favor of the anti–gay ballot measures. More impressively, support for Issue 1 was positively related to the vote for Bush in 2004, even after taking into consideration the strong predictive value of the vote for Bush in 2000. Additionally, the percentage of evangelicals in a county was a positive predictor of the vote for the president. More broadly, the findings comport with survey research, particularly that Protestants were significantly more likely than non-Protestants to support the bans on gay marriage and that those who supported the ballot measures were more likely to vote for Bush (Donovan, Tolbert, and Smith 2005). On the question of voter turnout, county-level support for the measures does not appear to be linked to more citizens coming out to vote. The aggregate-level findings show that turnout in Ohio was not higher in counties with greater support for Issue 1. Counter to some postelection claims, there is no evidence that counties with higher proportions of evangelicals, rural voters, or even supporters of the gay marriage bans had higher turnout levels than other counties. Turnout, then, increased across the board, irrespective of support for Issue 1. The strategy of using polarizing ballot measures as partisan cues may rest in part upon the assumption that potential voters are willing to politicize what are considered to be core issues (Smith 2006; Smith and Tolbert 2004; Nicholson 2005). With evangelical leaders as the driving force behind Issue 1 in Ohio, this seems to have been the case, as counties with dense populations of evangelicals not only supported the ballot measure but also supported the president. As for the fact that there is no evidence that counties with higher levels of support for Issue 1 or those with denser evangelical populations had higher turnout relative to other counties, scholars of religion and politics (Jelen 1994; McCann 1997) have noted there are good reasons why socially conservative issues, such as the anti–gay marriage ballot measures, may not mobilize all evangelicals equally. Future research on the impact of ballot measures on candidate elections should certainly bear in mind these considerations and empirical findings. Notes *I would like to thank Todd Donovan, Larry Kenny, Susan Orr, and Caroline Tolbert for their critical feedback on earlier versions of this draft, as well as the 2005 participants at the University of California Center for the Study of

WAS ROVE RIGHT?

33

Democracy/USC-Caltech Center for the Study of Law and Politics/Initiative and Referendum Institute Conference. Portions of this chapter were published in 2006 under the title “Same-Sex Marriage Ballot Measures and the 2004 Presidential Election,” State and Local Government Review 38 (2): 78–91, with coauthors Matthew DeSantis and Jason Kassel. 1. State legislatures in Georgia (76 percent yes), Kentucky (75 percent yes), Mississippi (86 percent yes), Oklahoma (76 percent yes), and Utah (66 percent yes) placed referendums on the ballot. Initiatives were on the ballot in Arkansas (75 percent yes), Michigan (56 percent yes), Montana (67 percent yes), North Dakota (73 percent yes), Ohio (62 percent yes), and Oregon (57 percent yes). 2. There are two different ways to model the second and third hypotheses. The first (results not shown) considers the dependent variable as a dynamic change value, specifically percent change from 2000 to 2004 in turnout (Model B), and percent change from 2000 to 2004 in vote for Bush (Model C). This type of modeling leads to the problem of regression toward the mean, as the dependent variable contains information that should be on the right side of the equation. Alternatively, both hypotheses can be considered as panel models, as they include a lagged value of the dependent variable. The models seek to isolate the effect on something occurring at T2 of some variable measured at T2 (e.g., samesex marriage measures on the ballot), holding initial conditions (T1) constant. This second approach, used here, offers a much more conservative test of the impact of the anti–gay measures, as the lagged T1 values (2000 turnout in Model B; 2000 Bush vote in Model C are huge forces that drive T2 in both models. 3. Due to heteroskedasticity concerns—as the unit of analysis for these models is state counties—weighted least squares regression is used to account for the wide variation in county size and relative influence on statewide electoral results. The weight variable is the 2000 population of a county. 4. Measurement of evangelical Protestants is a subject ripe for academic debate. Following Steensland et al. (2000), a collapsed denominational typology measuring evangelical Protestants, Catholics, other conservative denominations, and liberal denominations/groups with no religious adherence is used because it is compatible with aggregate-level data. For further debate surrounding religious measurement, see Wilcox (1992), Layman (2001), Jelen (1989); Kellstedt et al. (1996). The proportions of evangelical Protestants, Catholics, other conservative denominations, and liberal denominations/groups with no religious adherence per county in Ohio were calculated using census data from the Glenmary Research Center’s Religious Congregations and Membership in the United States: 2000 (Jones et al. 2002). The methodology and actual codes for the religious denominations are available from the authors. 5. Alternative specifications using the county percent vote for the Republican gubernatorial candidate or the Republican attorney general candidate in the 2002 contests were run, but they had no statistical or substantive differences. 6. Unemployment statistics are from the Ohio Department of Job and Family Services, Bureau of Labor Market Information (2004). The models were run using both a static unemployment variable, which measured the rate of unemployment in a county in 2003, as well as with the dynamic variable measuring

34

DANIEL A. SMITH

the rate of change in unemployment from 2000 to 2003. There is no predictive advantage in using one measure over the other, but there is greater theoretical value in utilizing the dynamic unemployment variable, as it helps differentiate between counties that suffered greater increases in unemployment during the first term of the Bush administration with those with chronically high unemployment rates. 7. In two counties—Coshocton and Van Wert—Issue 1 actually had more total votes cast (for and against) than the total votes cast for all the presidential candidates on the ballot.

3

Electoral Supply and Demand: Direct Democracy Campaigns, Political Interest, and Participation Caroline Tolbert and Daniel Bowen

1. Introduction In comparative terms, one of the most distinctive aspects of American politics involves the use of state and local ballot propositions to resolve controversial issues. The constitutions of 24 states and hundreds of municipalities permit votes on proposed laws that reach the ballot when advocates submit the signatures of a required number of citizens. The 2004 election is remembered in part for the massive public and media attention devoted to ballot measures—both initiatives and legislative referenda—banning gay marriage in 13 states. Over the last three decades, the initiative process has allowed major policy decisions in areas such as gambling, affirmative action, the environment, the minimum wage, and tax and spending limitations. Accordingly, ballot initiatives have attracted an increasing amount of attention from political scientists (Gerber 1999; Matsusaka 2004; Boehmke 2005; Nicholson 2005; Bowler and Donovan 1998; Bowler, Donovan, and Tolbert 1998; Magleby 1984). One important line of research focuses on the effects of initiatives on voter turnout. Some studies detect such effects only in midterm elections (Smith 2001; Lacey 2005), although others find a boost to turnout in both presidential and midterm elections (Smith and Tolbert 2004). Recent research using VEP (voter eligible population) turnout, instead of VAP

36

CAROLINE TOLBERT AND DANIEL BOWEN

(voter age population) turnout, found that each initiative appearing on a state’s ballot yielded about 1 percent increase in turnout in presidential elections and 2 percent increase in midterm elections (Tolbert and Smith 2005). Using experimental methods, Lassen (2005) finds that citizens allowed to vote on referenda in Switzerland show increased levels of participation. A central limitation of this literature lies in its inability to identify the causal mechanisms for how ballot measures trigger participation. We know that citizens learn more about politics in information-rich environments (Lupia 1994; Lupia and McCubbins 1998; Nicholson 2003; Popkin 1991). In order for citizens to select representatives and make political choices consistent with their preferences, there must be “sufficient, relevant data available in the political environment” (Freedman, Franz, and Goldstein 2004, 723). Implicit within existing research studying initiatives and turnout is a process whereby issue campaigns and the mass media provide information, arguments, and appeals to the electorate. In doing so, campaign and media attention can stimulate interest in politics, which in turn leads to increased participation. In this account, the supply side of information (salient ballot propositions and associated political campaigns) and the demand side of the equation (citizen interest) work in tandem to increase participation. The literature on the relationship between initiative use and voter turnout, however, addresses only supply-side factors when moving from theory to empirical research. Smith’s article (2001) and others building on his data (Lacey 2005) measure the campaign environment through the volume of newspaper coverage of initiatives. Another approach measures the number of initiatives appearing on the ballot (Tolbert, Grummel, and Smith 2001; Tolbert and Smith 2005). In all of these articles, the intermediate stage of increasing levels of voter interest in response to the campaign is assumed rather than estimated. To improve our understanding of the mechanisms at work, we directly model through two-stage causal models both the supply of information (exposure to ballot propositions) and the demand side of the equation (citizen interest in politics) as factors that predict voter turnout using survey data. The argument builds on Campbell’s (1966) surge-and-decline theory of voter turnout in American elections. Campbell argues that a group of “core voters” consistently vote in each election. During high-stimulus elections these core voters are joined by an additional set of “peripheral” voters with lower political interest. The latter group becomes sufficiently motivated to go to the polls because of the excitement associated with a high-stimulus election. During such elections, individuals are bombarded by election news and campaign discourse, which dramatically lower the cost of obtaining

ELECTORAL SUPPLY AND DEMAND

37

information. If salient policy issues appear on state ballots, the associated campaigns and media attention may stimulate interest and lead to increased political participation, especially in midterm elections. 2. Do Ballot Propositions Increase Civic Engagement? Existing research suggests an engagement effect from exposure to ballot propositions. Drawing on survey data of the 1992 Senate Election Study merged with state-level measure of newspaper coverage of ballot measures, Smith (2002) finds that citizens exposed to salient initiatives and referenda show an increased capacity to correctly answer factual questions about politics. That is, there is a positive relationship between salient ballot measures and political knowledge. Similarly, scholars find that in Switzerland, citizens are better informed when they reside in cantons with more opportunities for direct political participation (Benz and Stutzer 2004). There are also a few studies linking initiatives and referendums to political interest. Research on a Canadian referendum found that exposure to the ballot measure lead to increased citizen interest in politics as well as knowledge (Mendelsohn and Cutler 2000). Scholars studying the effects of ballot initiatives on civic engagement in the U.S. context over time report mixed, though generally positive, results. Citizens residing in states with frequent exposure to ballot initiatives report higher levels of interest in politics in both the 1996 and 1998 elections, but not in the 2000 presidential elections (Smith and Tolbert 2004). While the authors find that opportunities for direct participation in policy making have a more consistent effect on political interest than political discourse, the result is only a partial confirmation of the hypothesis that ballot propositions increase engagement. Caution should be used when interpreting these results because of the way the American National Election Study (NES) samples respondents within states.1 Existing research also suggests that political efficacy, or confidence in government responsiveness, may be higher when individuals are frequently given opportunities to vote on policy issues (Bowler and Donovan 2002; Hero and Tolbert 2004; Mendelsohn and Cutler 2000). But others have challenged these findings, suggesting that efficacy is not increased by exposure to ballot propositions (Dyke and Baldassare 2006). In that efficacy and civic engagement tend to go together, these are important findings as well.2 This literature suggests, however, that engagement effects of ballot measures are the strongest in low-information elections, consistent with previous research on voting. Nicholson (2003) found that awareness of ballot propositions is higher in midterm election years, suggesting that ballot measures will have a greater engagement effect in midterm rather

38

CAROLINE TOLBERT AND DANIEL BOWEN

than presidential elections. Our use of survey data from 2002 midterm elections allows a closer examination of this engagement hypothesis in a recent low-information election. Despite this growing literature, there has been little attempt to link the findings on civic engagement and political participation. A concern is that endogeneity between civic engagement and political participation may bias the results found in previous studies. Do ballot propositions and their associated campaigns increase voter turnout by stimulating interest in the election? We propose a simple causal model: exposure to statewide ballot propositions and issue campaigns should increase attention to, or general interest in, the election and in turn the probability of voting. Since ballot measures boost turnout more in low-information midterm elections, we draw on national opinion data from a recent midterm election (2002) and employ two-stage causal modeling. As a robustness check and to compare the findings to a high-information presidential election, we replicate the model estimating general political interest using 2004 national opinion data. 3. Supply-Side Factors: Ballot Proposition Campaigns Ballot measure campaigns may provide political information to people, leading to increased opportunities for learning about and being interested in politics (Nicholson 2003, 2005). This argument is bolstered by research showing that both political learning and interest are heavily influenced by the political environment (Nie and Andersen 1974; Nie, Verba, and Petrocik 1979). Luskin (1990) argues that political sophistication stems from three broad factors: a person’s ability to assimilate and organize political information; his or her motive or desire to follow political affairs; and his or her exposure to political information—that is, his or her information environment. Exposure to political information may be a function of the frequency at which such information is made available and media use (Luskin 1987, 1990). The effects of exposure to initiative and referendum campaigns on political interest may resemble the effects of media exposure. The consumption of political information from media sources enhances civic engagement by increasing citizens’ knowledge about politics (Delli Carpini and Keeter 1996). Although more learning occurs from reading about politics in newspapers than watching television (Smith 1989), watching television news has been found to be positively associated with political knowledge in a variety of contexts (Iyengar and Kinder 1987; Weaver and Drew 1993; Chaffee and Kanihan 1997). Research has found that campaign advertising increases levels of interest and knowledge among citizens, especially those

ELECTORAL SUPPLY AND DEMAND

39

with low levels of information (Freedman, Franz, and Goldstein 2004; West 1997). This pattern holds even for new technology; consumption of online news has been shown to increase civic engagement and political knowledge (Shah, Kwak, and Holbert 2001; Jennings and Zeitner 2003; Uslaner 2004). Conceptually similar to media exposure studies, ballot measure campaigns may improve the political information environment available to citizens, potentially reducing the costs of being informed and engaged. Depending on how widespread, coordinated, and high-profile their campaigns, as well as the ability of these campaigns to publicize messages about the issues, ballot measures have even been found to help set the agenda in candidate elections (Nicholson 2005; Smith, DeSantis, and Kassel 2006). The vote on most ballot measures is a simple yes or no, and ballot measure campaigns tend to focus on this bottom line, encouraging the consumer to either support or oppose the issue on election day. But campaign ads encouraging either support of or opposition to a ballot measure tend to be rich in informational content (Smith 2002), and research has shown that advertising conveys information in an efficient, easily digestible way that has positive effects on civic engagement (Freedman, Franz, and Goldstein 2004). Exposure to issue campaigns (including advertising, news media coverage, yard signs, direct contacting, etc.) can help citizens learn about the issues and these campaigns are an efficient means to communicate with potential voters (McCuan et al. 1998). Like the media, political organizations such as political parties and interest groups (the sponsors, promoters, and opponents of ballot measure campaigns) may lower citizens’ information-gathering costs (Bowler and Donovan 1998). Studies suggest that campaign effects and the presence of easily available information shortcuts provide minimally informed voters the capacity to make decisions on ballot measures that deal with complex policy choices (Lupia 1994; Karp 1998; Nicholson 2003). As voters use the partisanship of candidates to draw meaningful inferences about their policy stances, similarly, voters can form attitudes on issues without studying them by taking cues from opinion leaders (Bowler and Donovan 1998). In sum, there are good theoretical reasons based on previous research to expect that exposure to ballot measures and campaigns should enhance both engagement and participation. 4. Demand-Side Factors: Campaigns and Political Interest Research on media effects finds that campaign ads (even negative campaigning) can increase civic engagement, particularly political interest, because campaign ads not only convey information about politics but they

40

CAROLINE TOLBERT AND DANIEL BOWEN

do so in an emotional context (Freedman, Franz, and Goldstein 2004). While often nonpartisan, ballot measure campaigns convey information with an emotional coating. Ballot measure campaigns over moral policy (Mooney 2001), for example—in which at least one side frames the argument in moral terms, such as same-sex marriage (Haider-Markel 2001), abortion (Burden 2005; Gerber 1999), affirmative action (Chavez 1998), the environment, or even animal rights—rarely present information in dispassionate, emotionally neutral terms (Schrag 1998). Ballot measure campaigns are skillfully produced, using professional media consultants, pollsters, and campaign ads to persuade voters, often using emotional appeals (McCuan et al. 1998; Broder 2000). Emotionally rich issue campaigns frequently provoke responses from moral indignation to sympathy, anger to fear, either directly, through advertising, or indirectly, by way of elite endorsements from elected officials and interest groups (Bowler and Donovan 1998). As a result, issue campaigns may facilitate the use of the likeability heuristic, by which people make informational inferences on the basis of their likes and dislikes of groups (Sniderman, Brody, and Tetlock 1991; Wenzel, Donovan, and Bowler 1998). In summary, ballot measure campaigns are rife with both informational and emotional content that may spark public interest in politics, contributing to a more attentive citizenry. We know from a long history of voting studies (Campbell et al. 1960; Wolfinger and Rosenstone 1980) that the low educated tend to be the least likely to vote and be interested in politics. Salient ballot propositions over controversial policies may spark political interest even among the low educated. If so, ballot propositions may be an important way to engage the public in politics. 5. Evidence of Citizen Interest in Ballot Propositions Descriptive survey data from the 2004 presidential elections suggest that ballot measures do trigger widespread political interest among the public. A Pew national opinion poll conducted in October 2004 explored the salience of ballot propositions in the presidential election that year.3 In the 2004 elections 163 measures appeared on 34 statewide ballots. Overall, 42 percent of Americans said initiatives, referenda, or state constitutional amendments were on the ballot in their states, and among those residing in the 34 states where statewide proposals actually appeared on the ballot, more than half (56 percent) of voters were aware of such measures (Pew 2004). The survey found that 86 percent of respondents who were aware of ballot measures in their states were “very” or “somewhat interested” in the policies on their state’s ballot (Pew 2004). That is, 41 percent said they

ELECTORAL SUPPLY AND DEMAND

41

were “very interested” in such ballot issues that year; 45 percent claimed being “fairly interested.” Comparable numbers of people across the ideological spectrum expressed interest in ballot measures. Of the policies sparking the most interest among voters, gambling measures were the most visible, followed by proposed constitutional amendments to ban gay marriage. 6. Research Hypothesis From the existing research we can develop an information/engagement hypothesis about the link between exposure to information and emotional appeals from ballot measure campaigns and engagement. We expect that individuals frequently exposed to ballot propositions and salient-issue campaigns will be more interested in politics and, in turn, be more likely to vote. It is worth noting how this hypothesis differs from Smith and Tolbert (2004), who discuss the educative effects of ballot measures, but tend to suggest that the education of citizens occurs because of opportunities for direct participation in policy making, consistent with Progressive reformers’ expectations and theories of participatory democracy, rather than a richer information environment (Nicholson 2005) or high-stimulus election (Campbell 1966). We also hypothesize that issue campaigns may increase engagement/political interest among the low educated more than those with more education. 7. Data and Methods Midterm Election We draw on a Pew survey done after the 2002 midterm election to test our hypothesis about the impact of ballot initiatives on civic engagement and participation.4 One disadvantage of the Pew survey is that it contains fewer questions about individuals’ political attitudes, for example, efficacy, than NES data. The NES, however, does not include representative sample of states (only geographic regions). The Pew data provide more leverage on the effects of state contextual factors on individual attitudes and behavior by sampling randomly within states. We use geographic identifiers to merge the 2002 opinion data with (1) the actual number of initiatives appearing on 2002 election ballots in each respondent’s state and (2) total expenditures on ballot initiatives in each state. Previous research has measured the salience of initiative and referendum campaigns via front-page newspaper coverage or a count of the

42

CAROLINE TOLBERT AND DANIEL BOWEN

number of initiatives appearing on states ballots. Drawing on this literature, one primary explanatory variable is a count of the number of statewide initiatives appearing on the ballot in the respondent’s state.5 We suggest that a better measure of the information in the political environment is total spending on ballot initiative campaigns in a state measured in millions of dollars. In the 2002 election, total expenditures on initiative campaigns ranged from $18,000 in Mississippi to $51 million in California.6 This is the first research of which we are aware to model exposure to issue campaigns by total spending on ballot initiatives. In a number of states, especially when gambling was on the ballot, total spending on ballot propositions rivaled (or exceeded) expenditures for major candidate races such as the governor and senate races. Since the number of initiatives on the ballot does not necessary lead to more information in the election, we predict that spending on ballot initiatives is a better proxy for the amount of information in the election.7 We employ instrumental variable regression (two-stage estimation procedure) to test the hypothesis that exposure to issue campaigns increases engagement and thereby increases political participation. Specifically, we model the impact of total expenditures on ballot initiative campaigns (x1) on probability of being interested in the election (x2) and in turn on the probability of voting (y). In the first stage we estimate general interest in the election as a function of a critical set of independent variables employed in the second stage. We begin (the first stage) by estimating the reduced-form equations for political interest. The measure of interest that we use as the dependent variable is a 4-point ordinal variable, with higher values indicating more interest in the election. The distribution of responses suggested the wisdom of collapsing the responses into a binary variable of “interested in the election” or “not interested.”8 From the reduced-form estimates we produce predicted values for each respondent in the survey. In the second stage, instrumental variable regression procedure substitutes these predicted values for the endogenous variable on the right-hand side of the equation modeling voting. The models include controls for demographic factors that affect individuals’ interest in the election, including income,9 educational attainment, age,10 gender,11 and place of residence.12 Race is held constant with dummy variables for African American, Asian, and Latino (1 ⫽ yes; 0 otherwise) with non-Latino whites serving as the reference group. Partisanship is measured with binary variables with Democrats or Republicans coded 1 and independents coded 0.13 Media consumption is measured with binary variables asking whether or not respondents had read the newspaper or watched a national TV news program the day prior to being surveyed, coding these variables as 1 if they had and as 0

ELECTORAL SUPPLY AND DEMAND

43

if they had not, or had used online news.14 Previous research has reported that use of online news is important in predicting voter turnout, and is controlled for here as well (Krueger 2002; Bimber 2003; Tolbert and McNeal 2003)15 Because of our coding of the primary explanatory variable (initiative exposure and issue campaigns), we also control for other state environmental factors that may affect an individual’s attention to the election. Following Freedman, Franz, and Goldstein (2004), we include variables measuring the competitiveness of the senate and governor races in a state by the vote margin; higher values indicated more competitive elections (one-vote margin). Individuals residing in states with a competitive senate and governor’s races should be exposed to a richer information environment, and be more attentive to the election. 8. Findings Table 3.1 presents models predicting political interest in the 2002 midterm election; since the dependent variable is binary, logistic regression coefficients are reported. In column 1, these data indicate that citizens exposed to ballot initiatives in the 2002 midterm election expressed greater general interest in the election, holding constant demographic and attitudinal factors and the competitiveness of candidate races in the state. That is, as the number of initiatives on the state ballot increased, potentially enriching the respondent’s information environment or campaign exposure, respondents were more likely to be interested in the election. Column 2 directly measures the effects of a richer information environment and indicates that citizens residing in states with higher total expenditures on initiatives campaigns (a direct measure of salience) reported increased interest in the midterm election. Thus, both exposure to ballot initiatives and their associated media campaigns had the effect of increasing interest in politics, a critical precursor to more active forms of participation, including voting. This finding is consistent with previous research on direct democracy and political knowledge based on 1992 survey data (Smith 2002), and suggests that direct democracy may indeed help foster an engaged citizenry. Ballot measures appear to be particularly important in stimulating engagement in low-information elections, consistent with previous research (Nicholson 2003). Many of the control variables are also statistically significant and in the expected direction. Individuals residing in states with a competitive senate race expressed an increased interest in the election. Consumption of all forms of media (newspapers, TV, and online news) increased interest in the election, again consistent with the media effects literature

44

Table 3.1 Impact of exposure to ballot initiatives and initiative campaigns on political interest, 2002 midterm election Interested in politics 2002

b(s.e.) Number of initiatives on state ballot Total expenditures on initiatives Number of initiatives * education Total expenditures * education Newspaper news consumption Television news consumption Online news consumption Republican Democrat Male Age Hispanic Black Asian Suburban Urban Education Income Competitive election—governor Competitive election—senate Constant N Wald Chi2 Pseudo R2

.10 (.04)

Model 2 p⬎z

b(s.e.)

p⬎z

.009 .01 (.00)

.54 (.13) .64 (.10) 1.69 (.15) .35 (.16) .10 (.16) -.23 (.09) .03 (.00) -.25 (.20) .13 (.15) -1.13 (.27) -.23 (.11) -.19 (.15) .17 (.04) .08 (.03) .14 (.20) .64 (.14) -3.70 (.34) 1856 1269.99 .18

Model 3

.000 .000 .000 .026 .557 .007 .000 .228 .402 .000 .041 .423 .000 .002 .507 .000 .000 .000

.54 (.13) .65 (.11) 1.69 (.16) .34 (.16) .08 (.16) -.23 (.09) .03 (.00) -.25 (.20) .11 (.16) -1.18 (.28) -.22 (.11) -.12 (.14) .18 (.04) .08 (.03) .15 (.21) .67 (.15) -3.65 (.35) 1856 1234.55 .18

b(s.e.)

Model 4

p⬎z

.09 (.11)

.423

.00 (.02)

.953

.54 (.13) .64 (.10) 1.69 (.15) .35 (16) .10 (.16) -.23 (.09) .03 (.00) -.25 (.20) .13 (.15) -1.13 (.27) -.23 (.11) -.12 (.15) .17 (.05) .08 (.03) .14 (.20) .64 (.14) -3.69 (.35) 1856 1262.57 .18

.000 .000 .000 .026 .559 .007 .000 .225 .399 .000 .047 .423 .000 .002 .507 .000 .000

.001

.000 .000 .000 .034 .606 .007 .000 .192 .461 .000 .042 .410 .000 .004 .494 .000 .000 .000

.000

b(s.e.)

p⬎z

.02 (.01)

.001

-.00 (.00) .54 (.13) .65 (.11) 1.69 (.16) .34 (16) .09 (.16) -.24 (.09) .03 (.00) -.26 (.20) .11 (.16) -1.15 (.27) -.23 (.11) -.12 (.14) .19 (.04) .08 (.03) .15 (.21) .67 (.15) -3.72 (.35) 1856 1306.95 .18

.055 .000 .000 .000 .033 .603 .006 .000 .194 .481 .000 .036 .396 .000 .004 .499 .000 .000 .000

Note: Unstandardized logistic regression coefficients, with robust standard errors in parentheses to correct for heteroskedasticity. Reported probabilities based on two-tailed test. Clustering by state used to adjust standard errors for multilevel data. Source: Postelection Pew Internet and American Life Daily Tracking Survey (November 2002).

CAROLINE TOLBERT AND DANIEL BOWEN

Model 1

ELECTORAL SUPPLY AND DEMAND

45

(Delli Carpini and Keeter 1996; Freedman, Franz, and Goldstein 2004). The substantive findings do not change when the media use variables are omitted from the model. Partisans were more engaged in the election than nonpartisans, and those with higher socioeconomic status were also more attentive. But do ballot initiatives and their associated campaigns increase political interest among those least likely to be engaged in politics: the low educated? Column 3 (table 3.1) adds to the equation an interaction term: respondent’s education * number of initiatives on the respondent’s state ballot. Column 4 repeats this modeling but includes a different interaction term: respondent’s education * total expenditures on initiatives in the respondent’s state. These interactive models are used to test whether initiative campaigns have a differential effect on increasing interest among low- versus high-educated individuals. While the interaction term in column 3 is not statistically significant (no differential effect among the low and higher educated when exposed to more initiatives on the ballot), the interaction term in column 4 is statistically significant, and negative. The base term for total expenditures on initiative campaigns is positive and statistically significant, indicating that those with low education are more likely to report an increased interest in politics when exposed to salient-issue elections. This is a new finding in the literature and suggests that ballot propositions may mobilize those least likely to be interested in politics. Because both total spending on initiative campaigns and education are interval/ordinal variables, probability simulations are necessary to understand the direction of this conditional effect and its substantive magnitude. Probability simulations show the substantive impact of our findings in the low-information 2002 midterm election (see table 3.2). Increasing total spending on initiatives by one standard deviation above the mean results in a 7 percent increase in the probability that an individual had thought “quite a lot” about the election among the low educated (respondents who have only a high-school diploma or equivalent), holding all other variables at their mean or modal values. Thus, low-educated individuals residing in states with higher spending on initiative campaigns report more interest in elections. The effect of initiative spending on interest decreases as education levels increase. Increasing initiative spending by one standard deviation above the mean results in a 5 percent increase in the probability of being interested in the election among those respondents with some college education and only a 3 percent increase among those with a bachelor’s degree, all else equal. Thus, ballot proposition campaigns and additional information appear to engage the low educated—those least likely to participate in the election.

46

CAROLINE TOLBERT AND DANIEL BOWEN

Table 3.2 What matters? Predicted effect of exposure to ballot initiative campaign spending on political interest—by education level Individuals with low education (high school diploma) Average initiative expenditures in respondent’s state High expenditures (⫹1 S.D.) Very high expenditures (⫹2 S.D.) Change from average to high Individuals with average education (some college) Average initiative expenditures in respondent’s state High expenditures (⫹1 S.D.) Very high expenditures (⫹2 S.D.) Change from average to high Individuals with high education (bachelor’s degree) Average initiative expenditures in respondent’s state High expenditures (⫹1 S.D.) Very high expenditures (⫹2 S.D.) Change from average to high

2002

2004

48% (.04) 55% (.05) 61% (.06) ⫹7%

68% (.03) 74% (.03) 79% (.03) ⫹6%

56% (.04) 61% (.05) 65% (.05) ⫹5%

77% (.02) 77% (.03) 78% (.03) 0%

60% (.04) 63% (.04) 66% (.05) ⫹3%

80% (.02) 79% (.03) 77% (.04) ⫺1%

Note: Predicted probabilities estimated with Clarify from models in tables 3.1 and 3.3. Numbers in parentheses are standard errors. Age, income, education, television, newspaper, and online news consumption, and presidential, senatorial, and gubernatorial election exposure set at mean values. Gender set at female and race/ethnicity at white. The respondent is assumed to reside in a suburban region. Simulation estimated for nonpartisans (independents). Sources: Postelection Pew Internet and American Life Daily Tracking Survey (November 2002) and Preelection Pew Research Center for the People and the Press (October 20, 2004).

Do Issue Campaigns Increase Turnout? The dependent variable in table 3.3 is reported voting in the 2002 midterm election. This second-stage model estimated with two-stage instrumental variable regression method shows that political interest (predicted probability, generated from the model in column 2, table 3.1) is an important

ELECTORAL SUPPLY AND DEMAND

47

Table 3.3 Impact of ballot initiative campaigns on political interest and the probability of voting Voted in 2002 election (Second-stage estimates)

Political interest (Predicted probability) Newspaper news consumption Television news consumption Online news consumption Republican Democrat Male Age Hispanic Black Asian Education Income Constant N F R2

b(s.e.)

p⬎z

.50 (.21) .08 (.04) .04 (.03) ⫺.04 (.07) .06 (.03) .02 (.03) ⫺.01 (.02) .00 (.00) ⫺.04 (.03) .02 (.04) ⫺.20 (.04) .03 (.01) .02 (.01) ⫺.22 (.09) 1414 67.83 .31

.023 .063 .222 .565 .032 .525 .555 .007 .208 .565 .000 .060 .027 .014 .000

Note: Unstandardized instrumental variable regression coefficients, with robust standard errors in parentheses to correct for heteroskedasticity. Reported probabilities based on two-tailed test. Second-stage estimates. First-stage estimate reported in column 2 of table 3.1. The instrumental variable in the first stage is total expenditures on ballot initiatives in the respondent’s state. Source: Postelection Pew Internet and American Life Daily Tracking Survey (November 2002).

predictor of voting in the 2002 election. Initiative campaigns (measured by total expenditures on ballot initiatives) appear to generate political interest, and the increased attention to the election significantly increases the probability of voting, as expected. Again, the demographic control variables are in the expected direction, as the older, wealthier, and more educated were significantly more likely to vote, as were those who regularly consumed newsprint media. Two-stage causal models suggest that ballot measures increase voter turnout by stimulating interest in elections. The findings are consistent with previous research (Smith 2001; Lacey 2005; Smith and Tolbert 2004) that finds that states with salient initiatives or referenda on the ballot have higher voter turnout over time; our survey analysis helps explain why we see a turnout effect. Ballot propositions may help engage citizens in low-information elections, thus creating a positive effect on turnout.

48

CAROLINE TOLBERT AND DANIEL BOWEN

9. Presidential Election As a check on the robustness of the findings and to compare differences between midterm and presidential elections, we draw on the unique 2004 preelection Pew survey data discussed earlier to predict general interest in election.16 The dependent variable is identical to the one used for the 2002 election data analysis: a binary variable measuring how much thought the individual has given to the upcoming presidential election.17 Again, two identical explanatory variables are used: (1) the number of initiatives appearing on the respondent’s statewide ballot in the 2004 election and (2) the total expenditures on ballot initiatives in the respondent’s states. Values on the number of initiatives ranged from 0 in noninitiative states to a high of 8 in California. Total expenditures on initiatives ranged from approximately $10,000 in North Dakota to $202 million in California . State contextual controls measure the competitiveness of senate, governor, and presidential races in 2004 by the vote margin (one-vote margin), with higher values indicating more competitive races. We expect citizens residing in states with more competitive candidate races to be more interested in the election. The model also includes standard ideological and demographic control variables, chosen to mirror those used in the 2002 survey analysis. Democratic identifiers and Republican identifiers are coded 1, and independents and others coded 0. The respondent’s level of educational attainment,18 age (estimated in years), gender (with a binary variable coded 1 for males), race (with a binary variable coded 1 for African Americans, Latinos, and Asians, with whites as the reference group), and income are accounted for.19 Consistent with previous research showing a more limited effect of direct democracy in presidential elections, exposure to ballot initiatives does not seem to increase civic engagement in high-information presidential elections (see table 3.4, column 1).20 This may explain the previous finding that ballot initiatives in presidential elections have about half the turnout effect that they do in midterm elections (Tolbert and Smith 2005). The covariate for the number of initiatives on an individual’s statewide ballot in 2004 is not a statistically significant predictor of general political interest. Similarly, higher total expenditures on ballot initiatives are not associated with increased political interest in 2004 (column 2, table 3.4). Even the more subtle and direct measure of issue-campaign activity does not increase interest in the election for the overall population. However, even in high-information presidential elections, ballot initiatives and their associated campaigns appear to boost engagement among the low educated. The same interaction models used in table 3.1 are reported in table 3.4: respondent’s education * number of initiatives on the ballot (column 3)

Table 3.4 Impact of exposure to ballot initiatives and initiative campaigns on political interest, 2004 presidential election Interested in politics 2004 Model 1 b(s.e.) p⬎z .02 (.02)

b(s.e.)

.500 .00 (.00)

.62 (.30) .30 (.22) 1.16 (.44) .52 (.20) .52 (.17) .03 (.15) .01 (.00) ⫺.60 (.33) .37 (.26) ⫺.45 (.65) .30 (.05) .07 (.03) .05 (.05) .86 (.98) .04 (.20) .07 (.21) ⫺2.68 (1.01) 1284 172.28 .09

p⬎z

.038 .173 .008 .008 .002 .826 .002 .069 .155 .486 .000 .032 .274 .380 .841 .749 .008 .000

.62 (.30) .30 (.22) 1.16 (.44) .52 (.20) .52 (.17) .03 (.15) .013 (.00) ⫺.61 (.33) .37 (.26) ⫺.46 (.64) .30 (.05) .07 (.03) .05 (.05) .91 (.94) .06 (.21) .06 (.20) ⫺2.72 (.98) 1284 168.74 .09

.257

.037 .171 .008 .008 .002 .824 .002 .067 .157 .471 .000 .032 .275 .331 .790 .744 .006 .000

Model 3 b(s.e.)

p⬎z

.14 (.07)

.046

⫺.03 (.015)

.057

.63 (.30) .30 (.22) 1.19 (.44) .52 (.20) .52 (.17) .04 (.15) .01 (.00) ⫺.62 (34) .38 (.26) ⫺.42 (.65) .34 (.05) .07 (.03) .05 (.05) .81 (.98) .04 (.21) .07 (.21) ⫺2.82 (.99) 1284 187.29 .10

.035 .170 .007 .008 .002 .798 .002 .068 .154 .517 .000 .037 .264 .413 .852 .733 .004 .000

Model 4 b(s.e.) p⬎z .01 (.00)

.000

⫺.00 (.00)

.000 .035 .167 .006 .008 .002 .796 .002 .066 .16 .517 .000 .000 .257 .360 .787 .737 .003

.63 (.30) .31 (.22) 1.20 (.44) .52 (.20) .53 (.17) .04 (.15) .01 (.00) ⫺.62 (.34) .37 (.26) ⫺.42 (.64) .34 (.05) .07 (.03) .05 (.05) .86 (.94) .06 (21) .07 (.20) ⫺2.87 (.96) 1284 447.53 .10

.000

49

Note: Unstandardized logistic regression coefficients, with robust standard errors in parentheses to correct for heteroskedasticty. Reported probabilities based on two-tailed test. Clustering by state used to adjust standard errors for multilevel data. Source: Preelection Pew Research Center for the People and the Press (October 20, 2004).

ELECTORAL SUPPLY AND DEMAND

Number of initiatives on state ballot Total expenditures on initiatives Number of initiatives * education Total expenditures * education Newspaper news consumption Television news consumption Online news consumption Republican Democrat Male Age Hispanic Black Asian Education Income Attend church Competitive election—president Competitive election—governor Competitive election—senate Constant N Wald Chi2 Pseudo R2

Model 2

50

CAROLINE TOLBERT AND DANIEL BOWEN

and respondent’s education * total expenditures on initiatives (column 4). Both interaction terms are statistically significant, and again, the base terms for number of initiatives on the ballot and total expenditures on issue campaigns are positive and statistically significant. This suggests that the low educated who are exposed to ballot initiatives and their associated media campaigns are more likely to be interested in the election. Using probability simulations, we find that increasing initiative spending by one standard deviation above the mean results in a 6 percent increase of interest among low-educated (high-school graduates) respondents, compared with those with mean spending on ballot initiatives, all else equal. To recap, the empirical analysis provides strong evidence that salient ballot initiatives (measured both by the number of measures on a state’s ballot and by total campaign spending) increase interest in politics in lowinformation midterm elections. More importantly, we find evidence in both low- and high-information elections that ballot initiatives increase political interest among the low educated: those least likely to vote in an election. Finally, we find evidence that campaigns associated with ballot initiatives may indirectly increase the probability of voting by generating greater interest in elections—particularly low-information, midterm elections. The findings suggest that Campbell’s (1966) peripheral voters (low educated), who tend to turn out only in high-information presidential elections, are engaged by ballot initiatives in both low- and high-information elections. Additional research is necessary to further explore these important variations across high- and low-information elections and the effects of ballot propositions on varying types of voters. 10. Conclusion In spite of a growing literature on the impact of direct democracy on elections and political behavior, there has been little attempt to link the disparate findings on civic engagement and political participation. Previous research has assumed that individuals are more interested in elections, and therefore more likely to vote, rather than empirically measuring this phenomenon. To test the causal mechanisms at play, we directly model through two-stage models both the supply of information (exposure to ballot propositions) and the demand side of the equation (citizen interest in politics) as factors that predict voter turnout. Using Pew survey data and the initiatives on 2002 statewide ballots, we show that exposure to salient ballot measures and their associated campaigns produces citizens who are more interested in the election and are considerably more likely to vote. Analysis of the Pew survey data suggests that ballot measures increase voter turnout by stimulating interest in the

ELECTORAL SUPPLY AND DEMAND

51

election, especially in low-information electoral contexts, but not in highinformation presidential elections. Yet, in both high- and low-information elections, we find that the low educated reported increased political interest when exposed to ballot propositions and their media campaigns. That is, ballot propositions may mobilize the segment of the population least likely to vote in both presidential and midterm elections. Freedman, Franz, and Goldstein (2004) eloquently pose the following questions: “Whether one puts faith in miracles of aggregation, looks to the discourse of competing elites (Zaller 1992), or falls back on cognitive shortcuts, someone, somewhere, must be in possession of some supply of information. Where does such information come from? What is the source of the raw informational ingredients that make democracy possible?” The answer they provide is campaign ads, even negative advertising. This research provides an additional answer, suggesting that ballot measures provide alternative sources of information about politics that “make democracy possible.” We conceptualize ballot measure campaigns as providing information and emotional content to citizens, similar to the media. This informational content is necessary for civic engagement and participation in a democracy. Proponents and opponents of ballot measures help subsidize the costs of information, and their interest group and political party allies do so as well (Bowler and Donovan 1998), distributing voting guides and offering heuristic cues with politically relevant information in an easily digestible form. The findings on ballot measures and engagement are consistent with the literature on the importance of information in elections. Given the importance of sustained engagement in politics, the effect of ballot measures on civic engagement may be even more important than on turnout. If ballot proposition campaigns create richer political-information environments, stimulating interest in politics and voter turnout, direct democracy may serve to, in part, counteract a three-decade trend of declining engagement (Abramson and Aldrich 1982; Putnam 2000; Verba, Schlozman, and Brady 1995). Notes 1. The NES samples randomly within regions, but not within states. Thus, NES data are not ideal for studying state contextual effects, because all respondents in the NES may reside in one area of a state. Additionally, the NES does not include respondents from some small states. In contrast, the Pew Research Center samples randomly within states and includes respondents from 48 of the 50 states. This makes inferences about state-level effects on individual behavior less biased.

52

CAROLINE TOLBERT AND DANIEL BOWEN

2. Citizens exposed to ballot measures report more confidence in government, or enhanced political efficacy (Bowler and Donovan 2002; Hero and Tolbert 2004; Mendelsohn and Cutler 2000 and are more likely to contribute money to citizen-interest groups (Tolbert, McNeal, and Smith 2003). Scholars have also found that states with more frequent use of ballot initiatives have more representative interest-group systems, measured by more citizen-interest groups (Boehmke 2002, 2005; see also Gerber 1999) and more citizen and nonprofit groups (Smith and Tolbert 2004). 3. The national random sample telephone survey of registered voters was conducted for Pew by the Princeton Research Associates between October 15 and 19 and included 1,307 registered voters representing all 50 states. The report from the Pew Research Center for the People and the Press (October 20, 2004) is available online: http://www.people-press.org. 4. The 2002 postelection Pew Internet and American Life Daily Tracking Survey of 1,884 respondents is a nationwide random telephone survey conducted by the Princeton Research Associates. 5. The National Conference of State Legislatures served as the source of these data (www.ncsl.org). We measure ballot initiatives, rather than referenda referred by the legislature, as proposals initiated by citizen petition in general tend to be more salient with the public (Gerber 1999; Bowler and Donovan 1998). 6. Data on total expenditures on ballot initiatives in each state in millions of dollars for 2002 and 2004 were downloaded from the Web site of the Ballot Initiative Strategy Center. 7. The Pearson R correlation between the number of initiatives on the ballot and total expenditures on ballot initiatives in 2002 was .63 (p⬍.00). 8. Respondents were asked, “How much thought have you given to the upcoming midterm election?” and the responses—”quite a lot,”“some,”“only a little,” and “none”—were coded as a dichotomous variable, with “quite a lot” and “some” coded 1 and “only a little” and “none” coded 0. 9. Income is measured as an ordinal variable ranging from 1 to 8, where “1” indicates that family income ranges from $0 to $10,000and “8” signifies a family income of $100,000 or more. 10. Education is measured on a 7-point ordinal scale ranging from eighth grade education or less to PhD, and age is measured in years. 11. We control for the respondents’ gender by creating a dummy variable (1 if female and 0 if male). 12. Dummy variables are also used to control for geographic region, with suburban and urban respondents coded 1, with rural residents as the reference group (coded 0). 13. Models using strong partisans, rather than Democrat or Republican, yield similar findings and do not change the significance of the primary covariates. 14. Pew queried respondents about their consumption of online election news by asking, “Have you gone online to get news or information about politics or the campaign” (2002). From responses to this question, we create dummy variables

ELECTORAL SUPPLY AND DEMAND

15. 16.

17.

18. 19. 20.

53

indicating whether respondents had gone online to obtain information about each election (1 ⫽ yes; 0 ⫽ no). We are not able to control for political discussions or efficacy with the Pew data, owing to limitations in the survey questionnaire. Since this is a preelection survey conducted in October 2004, we cannot use this instrument to measure turnout in the election. The Pew Research Center for the People and the Press (October 20, 2004) is available online: http://www.people-press.org. How much thought have you given to the upcoming midterm election?” and the responses—”quite a lot,” “some,” “only a little,” and “none”—were coded as a dichotomous variable, with “quite a lot” and “some” coded 1 and “only a little” and “none” coded 0. In the national sample, education is measured on a 7-point scale ranging from 1 (less than eighth grade) to 7 (postgraduate). In the national sample, family income is measured on a 9-point scale ranging from 1 (less than $10,000 per year) to 9 (over $150,000 per year). Even when additional control variables are included to reduce omitted variable bias, such as campaign contact by a political party or interest group, ideology (liberal/conservative), or a squared term for age, the null findings do not change. When political interest is measured by a question measuring whether the respondent regularly follows politics and government, the null findings remain.

This page intentionally left blank

4

Direct Democracy and the Public Agenda: Ballot Initiatives and Public Beliefs about Important Problems Stephen P. Nicholson

cholars and pundits alike have begun to recognize that initiatives and referenda do not simply “keep to themselves.” On the contrary, scholars have found that direct legislation meaningfully shapes the political lives of citizens. Not only do initiatives and referenda influence public policy (Boehmke 2005; Gerber 1999; Glazer and McGann, this volume; Matsusaka 2004) and governance (see Boehmke, Dalton, and Matsusaka, this volume) but they also shape the political behavior and attitudes of citizens (Bowler, Nicholson, and Segura 2006; Nicholson 2005; Smith 2001; Smith and Tolbert 2004). Direct legislation boosts the civic life of states by increasing political interest (Tolbert and Bowen, this volume), participation (Smith 2001; Smith and Tolbert 2004), political knowledge (Pantoja and Segura 2003; Smith 2002), and political efficacy (Bowler and Donovan 2002). Direct legislation also affects the issues on people’s minds—what politics is about. In my book Voting the Agenda: Candidates, Elections, and Ballot Propositions, I demonstrate how direct legislation helps shape the electoral agenda. In particular, I show how ballot propositions prime voters to consider particular issues, while neglecting others, in evaluating candidates. Examining ballot propositions on topics such as abortion, taxes, the environment, the nuclear freeze, illegal immigration, and affirmative action, I found that these issues shaped voters’ choices of candidates even if the issues were not featured in a given contest or were irrelevant to the

S

56

STEPHEN P. NICHOLSON

job responsibilities of an office. By demonstrating the sweeping effect of these issues across multiple offices up and down the ballot, my research shows that ballot propositions dramatically heighten the salience of issues. In this chapter, I advance the study of the agenda-setting effects of direct legislation by examining whether ballot initiatives help shape the public agenda—the issues citizens deem as the most important problems facing the country. Evidence that direct legislation shapes the public’s national priorities would provide further support of how ballot propositions influence the issues people think about. Although it is counterintuitive that state ballot propositions could affect attitudes about national politics, it is consistent with my earlier research on electoral agendas, where I found that direct legislation did not obey distinctions between state and federal issues in shaping the vote across myriad offices (Nicholson 2005). Using survey data from California’s 1998 ballot proposition elections, I find that ballot initiatives can shape opinion about which issues belong on the national public agenda. Specifically, I find that awareness of Proposition 8, an initiative to reform California public schools, substantially increased the probability of voters mentioning education as an important problem facing the country. In the conclusion, I discuss avenues for future research and discuss the implications of my study for democratic governance. The Agenda-Setting Effects of Direct Legislation The dramatic increase in the number of well-known ballot propositions, especially the initiative, has led scholars to note its importance in setting the agenda (Garrett 1999; Magleby 1984, 1995; Nicholson 2005). Why is direct legislation capable of shaping agendas? How does it work? Since voters must either abstain or vote to accept or reject a ballot proposition, a choice is institutionally presented. In presenting this choice, the initiative is an agenda-setting institution that focuses attention. It informs citizens about what is politically relevant or important. In states with the popular initiative or referendum, political actors such as interest groups, politicians, and political parties may define the agenda by placing issues on the ballot (Nicholson 2005, chapter 6). Since voters are confronted with “take it or leave it” choices in direct-legislation elections (Romer and Rosenthal 1978), they must either vote on the agenda setter’s proposal or abstain. By virtue of being presented with the choice, however, direct legislation focuses attention on a matter the decision maker may not have otherwise considered. In so doing, direct legislation increases the salience of certain issues, shaping opinion on a variety of political phenomena.

DIRECT DEMOCRACY AND THE PUBLIC AGENDA

57

Not all direct legislation issues have the same effect on opinion, of course. Despite the fixed-choice nature of direct legislation issues, an issue’s appearance on the ballot is not a necessary or sufficient condition for its inclusion on the public’s agenda. Many ballot propositions reside in obscurity and may only capture a voter’s attention when he or she confronts them in the voting booth. What explains the differences in visibility among ballot propositions? In earlier research (Nicholson 2003), I found that several factors influenced awareness of ballot propositions. Issues involving morality or civil liberties, for example, heightened awareness of ballot propositions. In recent years, topics such as affirmative action, abortion, medicinal marijuana, and same-sex marriage have all had higher levels of voter awareness. In addition, high levels of campaign spending and extensive media coverage, proximity to election day, and midterm elections also boost voter awareness of ballot propositions (Nicholson 2003). Salient issues on the ballot are the type that have the potential to shape candidate evaluations (Nicholson 2005) and partisan identities (Bowler, Nicholson, and Segura 2006). By capturing the attention of voters, directlegislation issues primed voting decisions across a variety of offices. The exemplar of this phenomenon I examine in Voting the Agenda is California’s 1994 gubernatorial and senatorial elections. In these elections, underdog Republican candidates Pete Wilson and Michael Huffington erased the substantial leads of their opponents in the races for governor and U.S. Senate, respectively. Wilson had erased a more than 20-point lead by his opponent, Kathleen Brown, to defeat her by 14 points on election day. Huffington also erased more than a 20-point deficit, but in the end he fell short of defeating his opponent, incumbent Dianne Feinstein, by a 2point margin. What helped explain the success of these long-shot candidates? Proposition 187, an initiative that sought to deny public benefits to illegal immigrants, was the dominant issue of the campaign season and had a substantial effect on the gubernatorial and senatorial races. Indeed, a majority of voters mentioned Proposition 187 as the most important issue or race. It was also popular among voters, winning nearly 60 percent of the vote. In both the gubernatorial and senatorial contests, Proposition 187 had large effects on increasing the probability of a Republican vote. And its effects, I found, were not restricted to the gubernatorial and senatorial contests. The partisan dimensions of the issue—Republicans were for it and Democrats were against it—allowed its effects to travel down the ballot to lower elected offices (e.g., House races). The agenda-setting effects of direct legislation in candidate races, however, are not restricted to California. In Voting the Agenda, I found that ballot

58

STEPHEN P. NICHOLSON

propositions across the 50 states on issues such as abortion, taxes, and the environment affected the criteria voters used in forming evaluations of congressional candidates. Controlling for a variety of factors, including candidates’ campaign issues, I found that voters in states with an abortion, tax, or environmental ballot measure were more likely to mention these issues when asked about which issues mattered most to them in voting for congressional candidates. I also found that the nuclear-freeze ballot measures from the 1982 midterm elections had potent agenda-setting effects. Since the freeze issue included strong partisan dimensions—Democrats favored it and Republicans opposed it—the issue spilled over into multiple elections, boosting support for Democratic candidates. On the ballot in ten states, I found that support for freeze-ballot measures increased the probability of voting for House, Senate, and gubernatorial races, regardless of whether the candidates in these races had campaigned on the freeze. The freeze issue also influenced voting for gubernatorial candidates, despite the fact that it was outside the purview of state responsibilities and none of the gubernatorial candidates had campaigned on the issue. The issue did not have this effect in states without ballot measures on the subject. The gay marriage ballot propositions from the 2004 election are the most recent example of direct legislation shaping the electoral agenda. Scholars have examined whether it influenced voter turnout in the states where gay marriage was on the ballot, but the evidence is mixed, with some studies suggesting a turnout effect (Campbell and Monson 2005; Donovan, Tolbert, and Smith 2005; McDonald 2004) and others suggesting no such effect (Abramowitz 2004). The question of whether gay marriage ballot propositions influenced the criteria voters used in evaluating candidates is also mixed. Hillygus and Shields (2005) find that the question of a constitutional amendment banning gay marriage did not have a greater effect on vote choice in states with gay marriage ballot measures. Burden (2004), using state-level data, finds that support for Bush was not any higher in states with gay marriage ballot measures. However, using different data, Donovan et al. (2005) find that gay marriage ballot measures increased the salience of the issue, shaping both the ingredients of candidate choice and stimulating turnout. Campbell and Monson (2005) also find that the gay marriage ballot measures mattered by increasing support for Bush among Catholic voters. The effects of ballot propositions may also go beyond a single election by shaping voters’ long-standing political predispositions. Bowler, Nicholson, and Segura (2006) found that racially polarizing ballot propositions sponsored by the Republican Party during the early 1990s influenced Californians’ partisan identities. The cumulative effect of these

DIRECT DEMOCRACY AND THE PUBLIC AGENDA

59

ballot propositions reversed the trend among Latino and Anglo voters toward identifying as Republican by shifting partisan attachments toward the Democratic Party. In shaping both short-term forces (the issues voters use to evaluate candidates in a given election) and long-term forces (party identification), direct legislation has substantial agenda-setting capabilities. In the following section, I discuss how salient ballot propositions may affect citizens’ perceptions of national priorities. Direct Legislation and the Public Agenda That issues made salient by direct legislation can influence voting for candidates and partisan identification suggests a wide-ranging agenda-setting effect, one that is not likely limited to electoral phenomena. Indeed, the ability of direct legislation to shape electoral agendas suggests it may also affect opinion in other, issue-related domains. No less important than the electoral agenda is the public agenda. Whereas the former concerns salient issues in candidate elections, the latter concerns salient issues on the national public agenda, often operationalized as the most important issues or problems facing the country. McCombs and Shaw’s (1972) seminal research on the public agenda examined whether extensive media coverage of an issue increased public perceptions of that issue’s importance. Using the 1968 presidential campaign as their setting, the authors found that the mass media had a substantial effect on the public’s perceptions of important issues. Iyengar and Kinder (1987) confirmed this finding in their experimental analysis, demonstrating that even brief exposure to issues in news stories produced considerable effects in participants’ opinions about the importance of those issues. Since these studies, an extensive body of literature has confirmed and extended the relationship between media coverage and the public agenda (see McCombs 2004). Since I examine the agenda-setting effects of direct legislation on the public agenda, my approach differs from the media-effects tradition. As such, voter awareness of an issue on the ballot is the primary focus rather than mass media coverage, as is typical of most research on the public agenda. Although media coverage is an important predictor of ballot proposition awareness, it is by no means the only ingredient (Nicholson 2003). For this reason, I look directly at the effects of ballot proposition awareness on voters’ opinions of important problems. A potential objection to my hypothesis is that state issues do not, or at least are highly unlikely to, shape opinion about national politics. Some scholars, for example, have argued that voters use different criteria when

60

STEPHEN P. NICHOLSON

voting for federal and state candidates since the national and state governments have different responsibilities (Atkeson and Partin 1995; Stein 1990). Following this logic, an issue on a state ballot would not affect voters’ opinions about national problems. For some issues, this is likely the case. In 1998, for example, a proposition to legalize gambling casinos on tribal grounds became the most expensive initiative campaign in California history. It is unlikely that this type of issue, no matter how salient, would make it onto the national public agenda, even for voters in a state with the issue on the ballot. Although relevant in many states, voters are unlikely to see tribal casinos as an issue requiring national attention. Beyond provincial state issues are diffuse issues that embrace both national and state politics. The environment, taxes, health care, education, and abortion are just a few issues that involve national and state responsibilities, at least in the minds of voters. As I found in Voting the Agenda, ballot propositions on abortion, taxes, and the environment helped define congressional election agendas despite the fact that members of congress have little or no jurisdiction over state policy. Since many voters do not have either the sophistication or the desire (or both) to evaluate candidates on the basis of federal and state distinctions, it is unlikely they would use this sort of fine-tuning in forming beliefs about important national problems. In Voting the Agenda, I found that salient issues on the ballot spill over into multiple offices regardless of state or federal issue responsibilities. Similarly, in a study of Latino voters, Pantoja and Segura (2003) found that Proposition 187 increased the probability of Latino voters mentioning racial problems as an important issue facing the nation. Thus, assuming that the issue is not a provincial matter, I expect awareness of direct legislation issues to shape beliefs about which issues belong on the public agenda. Proposition 8 and the Most Important Problem Facing the Country To evaluate the effects of direct legislation on the public agenda, I use data from a 1998 preelection California Field Poll (9807). The poll of registered voters was conducted between October 22 and November 1. Specifically, I examine whether awareness of Proposition 8 increased the probability of voters mentioning education as an important problem facing the country. Although a good number of issues, states, or election years would have been acceptable for this analysis, I chose this particular case because the field poll had included a battery of questions ideally suited to the study of direct legislation and the public agenda.1 Proposition 8 was Governor Wilson’s response to the state legislature’s rejection of his educational reforms. Although Proposition 8 was dubbed “the Permanent Class Size Reduction and Educational Opportunities Act

DIRECT DEMOCRACY AND THE PUBLIC AGENDA

61

of 1998,” it included other proposals, some controversial. Among the controversial provisions were the creation of parent-teacher councils and the post of a state school inspector and the introduction of new rules for teaching credentials and curriculum approval. The California state Republican Party was the primary financial backer of the initiative, which was endorsed by many Republican politicians. Opponents of the initiative included all major statewide education groups, the California state Democratic Party, and some conservative religious groups (Asimov 1998). The initiative failed, winning only 37 percent of the vote. Although defeated, I expect that Proposition 8’s presence, among those voters aware of it, increased the salience of education. The public agenda, my dependent variable, is derived from the openended question, “What do you think are the most important problems facing this country?” The field poll coded up to three issue mentions. The primary dependent variable in my analysis is coded one (1) if a respondent mentioned education among any of three opportunities and zero (0) if he or she did not. Among all issues, education was the most frequently given response, with nearly 34 percent of the sample mentioning education. A potential shortcoming of this variable is that by including any mention of education (e.g., among three opportunities), I might be casting too wide a net by including all mentions of education, some of which might not be as important a priority. For this reason, I test a second dependent variable derived from the question, “Of those [issues] you’ve mentioned, what would you say is the single most-important problem the country faces?” This question allows voters to identify the one issue they deem most important and is coded one (1) if the respondent mentioned education and zero (0) if he or she did not. Using this question, education remains the most-frequently-given response, but the percentage declines to about 17. In the analysis that follows, I report the results of models using both dependent variables. Central to my analysis is voter awareness of Proposition 8. Voters reported moderate awareness of Proposition 8, with about half the sample (49 percent) reporting they had seen or read something about it. Given my argument about the agenda-setting capacity of direct legislation, I expect voters who reported awareness of Proposition 8 to be significantly more likely to mention education as an important problem facing the country, all else equal. This variable is coded one (1) if a respondent reported having seen or heard anything about Proposition 8 and zero (0) if not. I also control for a variety of variables associated with mentioning education as the most important problem facing the country—party identification, ideology, age, education, and income (Iyengar and Kinder 1987; Zhu and Boroson 1997) (see appendix for coding).

62

STEPHEN P. NICHOLSON

Table 4.1 presents the results from a probit analysis examining the effects of Proposition 8 awareness on voters mentioning education as the most important problem facing the country. The first two columns labeled models 1 and 2 present the results for both dependent variables. As discussed earlier, the dependent variable in model 1 included any mention of education among three issue mentions and the dependent variable in model 2 identified the single most-important problem. Regardless of dependent variable, the results show that awareness of Proposition 8 has a positive and statistically significant effect (p ⬍ .05) on voters’ mentions of education as an important problem. Indeed, the coefficients are identical across the two dependent variables and the standard errors nearly so. Since probit coefficients are nonlinear, they are difficult to interpret. To interpret the magnitude of the effects of independent variables, I calculate changes in predicted probabilities, moving from minimum to maximum values of the Table 4.1 Probit analysis of the agenda-setting effect of Proposition 8, 1998 Model 1 MIP (all)

Model 2 MIP (single)

Model 3 MIP (all)

Aware Prop. 3

.44* (.15) .36* (.17) .58* (.22) ⫺.02 (.04) .17 (.15) .14* (.06) —

.44* (.18) .30 (.19) .56* (.23) ⫺.07 (.04) .10 (.17) .17* (.06) —

Aware Prop. 5

—

—

Aware Prop. 9

—

—

Aware Prop. 10

—

—

⫺1.21* (.30) 310 .

⫺1.48* (.34) 310 .10

.44* (.18) .34 (.17) .52 (.22) ⫺.01 (.12) .16 (.16) .14* (.06) ⫺.17 (.17) ⫺.18 (.36) ⫺.07 (.20) .03 (.21) ⫺1.06* (.45) 282

Independent variable Aware Prop. 8 Democrat Liberal Education Female Income

Constant Number of cases Pseudo R2

Model 4 MIP (single) .54* (.21) .27 (.20) .61* (.24) ⫺.06 (.05) .12 (.18) .18* (.07) .04 (.20) .05 (.44) ⫺.12 (.23) ⫺.25 (.23) ⫺1.43* (.53) 282

Note: * p ⬍ .05 (two-tailed test). Source: Data are from the California Field Poll (9807), conducted between October 22 and November 1, 1998.

DIRECT DEMOCRACY AND THE PUBLIC AGENDA

63

Probability of Mentioning Education as the Most Important Problem Facing the Country

0.3

0.25

0.2

0.15

0.1

0.05

0 Aware Prop. 8

Democrat

Liberal

Income

Note: Quantities represent changes in probabilities calculated by looking at a minimum to maximum change in an independent variable, holding all other variables at their mean values. Source: Data are from the California Field Poll (9807), conducted between October 22 and November 1, 1998.

Figure 4.1

The agenda-setting effect of Prop. 8 on the public agenda, 1998

independent variable of interest while holding all other variables constant at their means. Figure 4.1 depicts changes in probability for awareness of Proposition 8 for model 1.2 If a voter is aware of Proposition 8, his or her probability of mentioning education as an important problem is 16 percent greater than a voter who is not aware of the initiative, holding all other variables at their mean. In model 2, this difference is nearly 12 percent. Regardless of dependent variable, awareness of Proposition 8 increases the probability that voters think education belongs on the public agenda. A few of the control variables merit discussion as well. Liberal identification is a significant predictor of mentioning education as an important problem in both models, whereas partisan identification only has a significant effect in model 1, suggesting a less robust relationship. Curiously, the effect of education is not statistically different from zero in either model, a result I further explore in the following section. On the other hand, the effect of income is significant in both models. Turning to figure 4.1, the changes in probability suggest that income exhibits the largest effect on mentioning education as an important problem. Moving from the lowest to the highest level of income increases the probability of mentioning

64

STEPHEN P. NICHOLSON

education as an important problem by 25 percent, all else equal. Furthermore, if a voter is a liberal, his or her probability of mentioning education as an important problem is nearly 23 percent greater than a voter who is conservative or middle-of-the-road, holding all other variables at their mean. These results suggest that long-standing predispositions, both demographic and political, play a role in voters’ beliefs about the importance of education, but that Proposition 8, a ballot initiative on educational reform, also shaped voters’ beliefs about education as an important problem facing the country. A potential criticism of my results is that awareness of Proposition 8, or any ballot proposition for that matter, might be a proxy for political interest, a variable associated with responding to open-ended questions (Geer 1988). Since political interest and ballot proposition awareness are likely to have a strong association, it is possible that awareness of a ballot proposition is tapping into political interest. If so, the effects of Proposition 8 awareness might be attributed to broader political characteristics that have little to do with the initiative. Unfortunately, the field poll did not include a question about political interest to test this possibility directly. To test this possibility another way, I reestimated models 1 and 2 by including variables for awareness of other ballot propositions. If it is the case that awareness of Proposition 8 is a proxy for political interest, awareness of other ballot propositions will also have significant and positive effects on the dependent variable. Fortunately, the field poll included questions about awareness for four other ballot propositions, each worded in the same way as the awareness question for Proposition 8. Each of these initiatives covered distinct issues: Proposition 3 (partisan presidential primaries), Proposition 5 (tribal casinos), Proposition 9 (electric utilities), and Proposition 10 (tobacco tax for early childhood development programs). As shown in models 3 and 4 of table 4.1, this alternative hypothesis is not supported. These additional analyses, built on models 1 and 2, indicate that awareness of other ballot propositions does not have an effect on mentioning education as an important problem significantly different from zero. Furthermore, the effects of Proposition 8 remain robust with the inclusion of these other ballot awareness variables. Who Is Susceptible to the Agenda-Setting Effects of Direct Legislation? Thus far, I have shown how awareness of direct legislation affects voters’ beliefs about important problems. It might be the case, however, that some voters are more susceptible than others to agenda-setting effects of direct legislation. Theories of attitude change posit two steps—reception and

DIRECT DEMOCRACY AND THE PUBLIC AGENDA

65

acceptance (McGuire 1968; Zaller 1992). Before a message can change an attitude, then, it must first be received and subsequently accepted. Iyengar and Kinder (1987) found this two-step process useful in understanding which members of the public were likely to be the agenda-setting “victims” of mass media exposure. In their research, they found that individuals of “limited political resources and skills” were the most susceptible to media agenda-setting effects (60). Other studies of agenda setting, however, have found contradictory results. Mackuen (1984), for example, finds that individuals with high levels of interest and education are more susceptible to the media’s agendasetting influences. Similarly, Miller and Krosnick (2000) find that citizens with high levels of political knowledge are the most susceptible to the media’s agenda setting. Since the field poll does not include questions tapping political interest or political knowledge, I use education as a proxy. Although imprecise, education is a reasonable surrogate for political resources and skills (Converse 1964). In table 4.1, I found that education does not have an independent effect on mentioning education as an important problem facing the country. However, as a proxy for political resources and skills, education is likely to have a moderating effect. If voters with low political resources and skills are the most susceptible to the agenda-setting effects of direct legislation, I expect awareness of Proposition 8 to have a larger effect on mentioning education as an important problem among those with lower levels of education. On the other hand, I expect that awareness of Proposition 8 will have a larger effect on mentioning education as an important problem if voters with ample political resources and skills are more susceptible to agenda-setting effects. To test these competing hypotheses, I reestimate the models from table 4.1 by looking at two groups: the well educated and the less educated. For the purposes of this analysis, the well-educated group had a college degree or advanced degree, whereas the less-educated group did not. As before, the dependent variable in my analysis is coded one (1) if a respondent mentioned education among any of three opportunities and zero (0) if he or she did not. I also ran the models using the dependent variable coded from the question about the single most-important problem, and the results are not substantively different (results not presented). Table 4.2 presents the results of the analysis. Turning to the less educated, awareness of Proposition 8 has a positive and statistically significant effect on mentioning education as the most important problem facing the country. None of the other variables in the model have an effect statistically distinguishable from zero using a two-tailed test of significance.Furthermore, among the less educated, a voter aware of Proposition 8 has a probability of mentioning

66

STEPHEN P. NICHOLSON

Table 4.2 Probit analysis of the agenda-setting effect of Proposition 8 by education, 1998 Independent variable Aware Prop. 8 Democrat Liberal Female Income Aware Prop. 3 Aware Prop. 5 Aware Prop. 9 Aware Prop. 10 Constant Number of cases Pseudo R2

Less educated .62* (.28) .44 (.26) .42 (.39) ⫺.13 (.25) .17 (.25) .48 (.28) ⫺.50 (.52) ⫺.36 (.30) .11 (.31) ⫺.83 (.30) 130 .07

Well educated .41 (.24) .11 (.25) .69* (.29) .43 (.23) .10 (.08) ⫺.03 (.23) ⫺.02 (.53) .23 (.29) ⫺.06 (.29) ⫺1.36* (.65) 152 .08

Note: *p ⬍ .05 (two-tailed test). Source: Data are from the California Field Poll (9807), conducted between October 22 and November 1, 1998.

education as an important problem 23 percent greater than a voter unaware of the initiative, holding all other variables at their mean. In column two, the model of well-educated voters, the effects of awareness of Proposition 8 is positive but not statistically distinguishable from zero in a two-tailed test. Indeed, the only variable statistically significant in a two-tailed test among the well educated is liberal ideology. Among the well educated, liberal ideology increases the probability of mentioning education as an important problem by 27 percent as compared with middle-ofthe-road and conservative voters. In contrast, if a well-educated voter is aware of Proposition 8, his or her probability of mentioning education as an important problem is 15 percent greater than a voter who is not aware of the initiative, holding all other variables at their mean. These results suggest that less-educated voters are more susceptible to agenda-setting effects than well-educated voters. Well-educated liberals exhibited greater concern about education, thus suggesting that education

DIRECT DEMOCRACY AND THE PUBLIC AGENDA

67

might be a long-standing concern among these voters. On the other hand, the less-educated did not appear to have a long-standing concern for education but were more likely to mention it as an important issue if they were aware of Proposition 8. Conclusion Direct legislation has the capacity to focus attention on an issue and make it politically relevant to the citizens of a state. By increasing the salience of some issues, while ignoring others, direct legislation shapes the agenda. In earlier research, I found that issues on the ballot influenced the issues voters deemed important in candidate elections. Furthermore, I found that direct legislation issues primed voting decisions up and down the ballot. As an integral part of the electoral landscape in many states, I argued that it made little sense to ignore ballot propositions. In this effort, I advanced our understanding of the agenda-setting capabilities of direct legislation. I found that issues on the ballot could shape the public agenda—the most important problems facing the country—for voters in a given state. Specifically, I found that awareness of Proposition 8 increased the probability of a voter mentioning education as an important problem facing the country. Furthermore, I found this effect to be the strongest among those with less political expertise. Although it may seem counterintuitive that a state issue could influence public opinion about national priorities, the psychological processes involved likely do this implicitly, increasing the accessibility of the issue, thus increasing the probability a voter mentions the issue when asked about important problems. As mentioned, this should not happen for all state issues, especially those that involve provincial issues unlikely to be problematic for the nation. To be sure, further research in this area remains. Studies examining the agenda-setting effect of direct legislation on the public agenda need to be done across different issues, states, and election years. In particular, a fruitful area of research would involve comparing states with and without ballot propositions on a given issue to see whether the issue was mentioned as an important problem more by voters in those states with this issue on the ballot. I am not as sanguine as other scholars about the role of direct legislation in democratic governance. Although scholars looking at the effects of direct legislation on citizens emphasize its benefits in promoting civic behavior (Bowler and Donovan 2002; Smith and Tolbert 2004), I find its effects on agenda control contrary to democratic governance. Dahl (1989) and other democratic theorists argue that a fully democratic system requires

68

STEPHEN P. NICHOLSON

final agenda control by the people. During elections, voters typically play the role of final arbiter by deciding what is politically important. Indeed, research on political campaigns shows that campaigns matter by activating voters’ latent preferences (Berelson, Lazarsfeld, and McPhee 1954; Gelman and King 1993). Since voters ultimately decide which issues matter, they have final agenda control. Direct legislation, however, may diminish agenda control. Direct legislation removes the question of whether an issue is politically relevant by putting that issue on the ballot and, for some issues, attracting voter attention through media coverage and campaigns. Access to the ballot, unfortunately, is not democratic. With adequate resources, a group can qualify just about any issue for the ballot (Magleby 1984), thus leaving the people to decide an issue they might not otherwise have thought important to consider. Appendix: Coding of Control Variables Democrat is coded one (1) if a respondent self-identified as a Democrat and zero (0) for Republican and Independent. Liberal is coded one (1) if a respondent self-identified as a liberal and zero (0) otherwise. Education is coded from zero (0) (eighth grade or less) to nine (9) (graduate work past master’s) Female is coded one (1) for female and zero (0) for male. Income is coded from zero (0) (less than $15,000) to five (5) ($100,000) or more. Awareness of Propositions 3, 5, 9, and 10 were individually coded one (1) for awareness and zero (0) otherwise. Notes 1. The field poll does not typically ask questions about the national public agenda. 2. I do not present changes in probabilities for model 2 since they do not differ substantively from model 1.

5

Direct Democracy’s Effects on Political Parties Shaun Bowler and Todd Donovan

olitical parties are central actors in representative democracy. Although it is possible to have political parties without democracy, it does not seem possible—at least in the modern era—to have democracy without political parties.1 A major statement of idea is that of the responsible party government model (APSA 1950): strong, coherent, and persistent political parties provide an essential link between voters and the political system. In damaging political parties, then, one also damages the workings of modern representative democracy. One of the more common arguments made by critics of direct democracy is that the direct democracy process weakens political parties, thereby undercutting not just their role in modern democracy but also some of the central pillars of representative democracy itself (Broder 2000, see also Schrag 1998). In this chapter we examine the question of whether direct democracy diminishes the strength of parties in the U.S. states. In the first empirical section we examine a range of evidence using both aggregate-level indicators and responses from a survey of legislators that is consistent with the argument that the initiative is bad for political parties. But it is important not to state this claim too2 strongly. While it is the case that parties are weakened by direct democracy, they are not passive victims and can respond to the challenges posed by direct democracy. Indeed, some critics of the process argue that parties are far from undercut by direct democracy. In fact we see the reverse: the populist ideal of direct democracy is subverted by political parties, which manage to take over ballot proposition campaigns and elections (see, e.g., Hasen 2005; Smith and Tolbert 2001; Smith, this volume). The second empirical section to this chapter

P

70

SHAUN BOWLER AND TODD DONOVAN

examines this competing claim. As with the first set of claims, however, we need to be careful not to overstate this line of argument. In general the balance of the evidence suggests that direct democracy is associated with the undercutting of political parties rather than vice versa. It is important, however, not to overstate the effects. The patterns we see here are consistent with those found by Russell J. Dalton in chapter 9. Before moving to empirical evidence we outline in more detail the argument that direct democracy is bad for political parties and, therefore, is bad for representative democracy, too. 1. How Direct Democracy May Undermine Political Parties Direct democracy has a great deal of potential to weaken and disrupt political parties and it may do so in at least two ways. First, the existence of direct democracy means that voters may have little need for organizations such as parties that mediate between them and the political system. Rather, voters can interact with the system itself, bypassing institutions of representative democracy, such as legislatures and political parties. Second, and this is more directly threatening, political parties may themselves be the object of direct legislation. Indeed, there is a great deal of anecdotal and historical support for the idea that direct democracy not only weakens political parties but was in fact designed to do so. At least part of the motivation for adopting direct democracy was disaffection between the people on the one hand and the existing political parties and political institutions on the other (see, e.g., Smith 2005; Hofstadter 1955; Barnett 1915). As one early writer put it, the “American people despise legislatures, not because they are averse to representative government, but because legislatures are in fact despicable” (Ford 1912, 72). Such popular contempt was not confined to legislatures but also embraced the political parties that controlled the legislatures. Among the earliest proposals introduced by the process were a string of measures aimed directly at the position of political parties, including attempts to strip state legislators (party in government) of their power to appoint U.S. Senators (in California, Montana, and Oklahoma); remove control of nominations from party organizations via introduction of the direct primary (in Arkansas, California, Maine, and Montana); establish recall procedures (Arkansas and Colorado); limit legislative sessions or abolish a legislative chamber (Arizona, Arkansas, and Colorado); amend election laws (Arkansas and Colorado); expand the franchise (Arizona and California); redistrict (Arizona and Oklahoma); establish civil service (Colorado); enact corrupt practices acts (Montana); eliminate straight party ballots and “irregular” ballots (Colorado and Maine); and limit the

71

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

Table 5.1 Number and pass rates of governance issues, 1911–2000

Examples of proposals All types of propositions on all issues, of which: Administrative reform

Apportionment

Election reform

Term limits

Campaign finance

1908 OR 19: Creation of Hood River County 1990 OK 627: Creation of five-member ethics commission 1991 ME: Reauthorization of widening of the Maine Turnpike 1982 OK 556: Creation of new congressional districts 1981 OH 2: General assembly and congressional redistricting 1904 OR 2: Direct primary nominating convention law 1982 AZ 202: Voter registration by driver’s license 1994 UT A: Term limits on state legislature 8 years. 1996 ND 5: Informed voter law and term limits on state legislature 8/8. 1994 OR 6: Restriction on candidates to only use contributions from district residents 1996 ME 3: Campaign finance reform; including voluntary spending limits and public funding.

N

First order correlation Pass rate % over time *

2,033

41

326

41

⫺.77*

39

39

.07

79

56

⫺.70*

62

79

.52*

46

70

.52*

* Period is divided into 11 decades, and pass rate by decade is correlated with a time counter 1, 2, 3, etc. Source: Initiative and Referendum Institute Web site.

legislatures’ ability to protect legislation with emergency clauses (Nebraska) (Piott 2004, appendix). Anecdotally, then, by both intention and early example, direct democracy does seem to have been aimed at weakening political parties. More systematic evidence seems to bear out this early reading. Table 5.1 and figure 5.1 display the frequency of various election reform measures: some of these measures are relatively innocuous, but many others are more serious. One of the striking features of table 5.1 is simply how many of these proposals were made—they were especially common in the decades immediately after the adoption of direct democracy. In the first few years

72

SHAUN BOWLER AND TODD DONOVAN 25

Percentage

20

15

10

5

00's

80's

60's

40's

20's

00's

0

Time apportionment

election reform

campaign finance

term limits

Source: Initiative and Referendum Institute Web site.

Figure 5.1 Percentage of all initiatives that concern governance issues by decade

that direct democracy was in use, roughly one-third of ballot initiatives dealt with election reform or campaign finance. Right from the outset, one of the main “targets” of direct democracy were the established political parties and passage of reforms or changes that directly shaped fundamental features of party life, such as the ability to control the nomination of candidates. While numbers of reform attempts have reduced from the early enthusiastic levels, governance reform measures that impact political parties have continued to appear on state ballots at a fairly steady rate across time, although the object of change has moved. In the early period, primaries and control of nomination were the main target, while term limits, campaign finance measures, and redistricting measures are all examples of governance reform that are now common (Tolbert 1998). Furthermore, at least in recent years, governance reform measures has tended to pass at higher rates than other initiatives (see table 5.1). All are examples of

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

73

additional ways in which voters have sought to tie the hands of political parties by use of the initiative process. These simple examples illustrate that the initiative process does seem to imply a much more constrained, or simply a less trusting, delegation of power to elected representatives than one familiar from most systems of government. By design the initiative is the citizen’s “gun behind the door” or the “bit between the teeth” (Johnson 1944) that can be used to signal or direct legislators. If legislators are unwilling or unable to pass the appropriate policy, then voters may pass it for them. If legislators try to renege on any deal—for example by overturning policies passed by voters—then it is possible for voters to further constrain politicians by proposing and supporting procedural or governance constraints such as direct primaries, term limits, or tax and expenditure limitations (see, e.g., Tolbert 1998; Matsusaka 2004). In principle these procedural restrictions are more consequential than policy changes because they are institutional changes: they may limit the scope for action of parties and representatives not just over one particular policy or for one period but over a whole range of issues over the long haul. This is, in fact, one reason why institutions matter in a broader sense: individual policies come and go, but institutions affect ranges of policy outcomes for long periods of time. Altering an institution, rather than a specific policy, may be more difficult, but the payoffs will be greater when multiplied over time and over many areas. But do these kinds of changes have a real and demonstrable impact on political parties? While it is possible to point to a series of reform proposals or even reforms themselves, this does not necessarily mean that there is any real effect on the role of political parties. Of course, the answer to whether there is a real impact rests in part on our definition of political party. One way in which we can see the impact that direct democracy and these reforms may have on parties is to examine variation in several statelevel measures of “partyness” to test how this variation correlates with measures of direct democracy. In what follows we use the standard threecomponent definition of political parties to look at the political party as an organization, the political party in the electorate, and party in government (Sorauf 1976; Dalton and Wattenberg 2001). Starting from the assumption that direct democracy weakens parties, our guiding hypothesis is the following: measures of direct democracy should be associated with weakness in measures of partyness. The hypothesis is a straightforward one. What is less straightforward is operationalizing partyness—even using the standard tripartite component—and (to a much lesser extent) how we should operationalize “direct democracy.” The approach we take here is to use a range of different measures for both

74

SHAUN BOWLER AND TODD DONOVAN

independent and dependent variables. We return to the broader question of measurement in the conclusion. Effects of Direct Democracy on Parties as Organizations Table 5.2 reports results from bivariate regressions in which the dependent variables measure several aspects of “parties as organizations” and “parties in the electorate.” In looking at measures of this kind, the robustness of statistical results is always at issue. Particular findings could turn on one specification or definition, and this is especially the case given the date we examine. With an N of 50 or fewer, and with broad measures of statewide properties (such as party organizational strength), results from models of this kind are often viewed suspiciously because they may have been produced by one —very fragile—specification. Therefore, rather than simply rely on one aspects or one measure of partyness, or indeed of direct democracy, we present results from several different measures and specifications. A key starting point is that we need to employ some measure of party organization. The standard measure of the strength of party organization is David Mayhew’s score representing the persistence of traditional party organizations (TPOs) in a state (1986). States with the highest scores have party organizations that were autonomous, long lasting, patronage based, hierarchically structured, and influential in nominating candidates. All of these are ways in which party organizations may be said to be stronger. Measures of a state’s initiative process should be associated with weaker organizations. We should also examine whether or not there is a more or less restrictive legal framework for parties across the states, using a measure of legal constraints placed on parties that was developed by Appleton and Ward (1997). It is all very well listing a series of measures such as term limits and restrictions on party finance, but these measures may have little to do with the process of direct democracy and a lot more to do with the presence of reformist movements or parties in the electorate. If direct democracy does weaken political parties, then we should therefore expect measures of a state’s initiative process to be associated with more restrictive legal frameworks as the power to self-regulate is taken out of the hands of the parties themselves. We also examine the effect of direct democracy on corruption in government. This is a somewhat more controversial measure of partyness. Following the traditional argument of Populist critics that unregulated, autonomous party organizations correspond with corrupt government, we expect that the initiative process should be associated with cleaner government because initiative states may adopt more regulations on parties, and that these regulations act to limit corruption. State levels of public corruption are measured by data from the Boylan and Long (2001) survey.

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

75

Effects of the Initiative Process on Party in the Electorate As Wattenberg (1996) and others have noted, another sign of weak political parties is the erosion of voter attachments to parties—or the decline of what Sorauf called “party in the electorate.” To assess the strength or weakness of party in the electorate, we examine state-level variation in the mass publics’ attitudes about America’s two major parties (as measured with National Election Study Cumulative Data File). If direct democracy weakens party in the electorate, parties should be less well regarded in initiative states. Measuring Direct Democracy If it is important to use different operationalizations of partyness in order to get a better sense of the robustness of any relationship, it is equally important to note that direct democracy may be measured in different ways. Here we use three separate measures of the independent variable: state provisions for the direct citizen initiative and measures of ease of use and of the amendability of ballot proposals. The rationale for using these different measures is that while many states have constitutional provisions that allow the use of direct democracy in some form or other, not all state provisions are the same. Some states, notably Illinois and Mississippi, formally allow the process to be used but in such restricted circumstances and subject to such limitations as to make the process largely unused. By contrast, other states, notably those along the Pacific Coast, have processes that are relatively open to use and where, once a proposal is passed, it is very hard for the legislature to amend that proposal. For this reason, in column 1 of table 5.2, a simple three-category measure of the initiative process is used (0 ⫽ no initiative, 1 ⫽ either constitutional or statutory initiatives, 2 ⫽ both constitutional and statutory initiatives). In column 2 we use an eight-point index where high scores reflect an initiative process that is more difficult to use and 0 means that no initiative process exists. In column 3 we use a ten-point index to measure provisions for direct democracy where higher scores reflect legislatures that are largely immune from the effects of initiatives (see Bowler and Donovan 2004 for discussion of these different indices). The clear impression from the results of table 5.2 is that initiative states have weaker, less autonomous party organizations and generally tougher legal controls on political parties. The electorates in direct democracy states are significantly more likely to have lower levels of public regard for political parties and are also slightly more likely to elect non-major-party governors. This finding of cooler feelings toward parties in direct

76

SHAUN BOWLER AND TODD DONOVAN

Table 5.2 Bivariate regressions on measures of state parties and their environments Use initiative (three Constant categories) Dependent variables: (below) Parties as organizations Party org (TPO) score TPO score adjusted for factionalism Restrictive laws (Appleton and Ward) Public corruption Party in the electorate NES thermometer ranking of parties Election of non-major-party governors (logit)

Qualification Degree difficulty of legislature initiative insulated from initiative

R2

N

2.42 0.57 0.62 2.23 0.67 0.71 4.1 5.7 5.3 3.72 2.54 2.70

⫺0.50 (.24)** —— —— ⫺0.43 (.22) —— —— 0.25 (.26) —— —— ⫺0.33 (.18)** —— ——

—— 0.28 (.10)** —— —— 0.23 (.10)** —— —— ⫺0.25 (.03)** —— —— 0.21 (.08)** ——

—— —— 0.19 (.07)** —— —— 0.16 (.06)** —— —— ⫺0.13 (.07)** —— —— 0.09 (.05)*

.08 .12 .13 .07 .10 .11 .01 .09 .06 .07 .13 .06

50 50 50 50 50 50 50 50 50 47 47 47

54.6 51.0 51.8 ⫺1.68 0.25 0.30

⫺0.85 (.39)** —— —— 0.65 (.37)* —— ——

—— 0.56 (.16)** —— ——⫺0.27 (.17) ——

—— —— 0.30 (.11)** —— —— ⫺0.20 (.11)*

.12 .26 .18 .06 .05 .06

36 36 36 50 50 50

Note: OLS coefficients, unless noted. Standard errors in parentheses. ** ⫽ significant at p ⬍ . 05; * ⫽ significant at p ⬍ .10 (two-tail). See appendix 1 for discussion of robustness. Sources: See appendix 1.

democracy states comes despite the fact that there is some evidence here to suggest that initiative states may have cleaner politics—possibly as a consequence of restrictions on the traditional activities of political parties. These findings are consistent with results we have reported elsewhere demonstrating that liberal provisions for initiative use are associated with tougher term limits and greater regulation of campaign finance (Bowler and Donovan 2004; Pippen, Bowler, and Donovan 2002 2). Those previous results are also consistent with the idea that direct democracy undermines partyness, conceived here as both party as organization and party in the electorate. These findings would seem to support the idea that direct democracy is associated with substantive negative consequences for the health and well-being of political parties across a range of dimensions. The initiative seems to allow for the expression of antiparty sentiment. That is, it is not

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

77

the initiative itself that weakens parties. After all, the initiative is simply a paper process that requires actions and activity on the part of individuals to move into place. But the general thrust of the collection of evidence so far is supportive of the argument that the working of direct democracy undermines the health and well-being of political parties. It is, however, possible to put too much emphasis on the kinds of empirical patterns set out here. As we noted earlier, the concern when using evidence of this kind— that is, statewide measures with a relatively small N—is that any results are likely to be fragile. What of the robustness of the results? Other measures of party organization, for example, demonstrate similar patterns to the ones found in table 5.2, but not in such a direct or straightforward way. For example, when we replicate the analysis in table 5.2 using the measure of local party organization (averaged across both parties for each state) evolved by Cotter et al. (1984), bivariate results (not reported here) are not in the expected direction. Using that measure we find stronger party organizations in noninitiative states, but only after controlling for the level of contestation (party competition) in each state between 1968 and 1995, as measured by Hamm and Moncrief (2004). That is, party organization and competition are—as one might expect—positively correlated, since competition should both induce and reflect organizational capacity. Once we control for this correlation we see an impact for two of our measures of the initiative process that is in the expected direction: the categorical variable and the qualification measure. There is, however, no relationship between the Bibby et al. measure of state party organization and direct democracy.3 Likewise, we found (not reported here) a bivariate correlation in the expected direction with LaRaja’s (2003) measure of GOP party strength and the initiative process, but not with the Democratic Party’s party strength. Similarly, as we noted above, it is possible to find correlations between the strength of campaign finance regulation and direct democracy in the expected direction, with a measure of the frequency of initiative use (states with more frequent use of the initiative see more severe regulation of campaign finance), but not with the alternative measures we see in table 5.1 (see Pippen, Bowler, and Donovan 2002). It is important to note that not all measures show the same relationship in the same degree—although this may perhaps say more about the uncertainty over how to measure state party organizations than it does about any relationship between parties and direct democracy. While some results for some measures do not pass the conventional thresholds of statistical significance, none of the results we examined were in the “wrong” direction; across a range of measures the effect is statistically significant and in the expected direction.

78

SHAUN BOWLER AND TODD DONOVAN

Table 5.3 Effect of cultural variables and frequency of use of direct democracy on weakening of parties Direct democracy (low score means easy to use; higher score means Moralistic Individual Traditional Constant harder to use) -istic -istic

R2

N

.14

48

.24

48

.19

48

.14

48

.22

48

.18

48

Party org (TPO) score (higher score implies stronger organization) .72 ⫺.70 .63

.28** (.11) .31** (.10) .33** (.11)

⫺0.6 (1.3)

⫺.23** (.16) ⫺.23** (.15) ⫺.19** (.15)

0.2 (1.4)

2.1** (.85) ⫺1.2* (.71)

Restrictive legal environment for party 5.6 4.2 45.8

2.5** (.89) ⫺1.7** (.74)

** ⫽Significant at .05 level or better. * ⫽ Significant at .10 level or better.

We return to this question of measurement in the conclusion; for the moment we will simply make the point that, even though we establish correlation, we may not have causation. The introduction of direct democracy predates these measures of organizational and legal climate by many decades, and so it is reasonable to suppose that direct democracy leads to weaker parties rather than both being related to an underlying cultural dimension. One other way to address this is simply to introduce controls for culture. This we do in table 5.3, which reports results from models that look at the impact of direct democracy on two of the measures from table 5.2, this time controlling for Daniel Elazar’s labeling of the cultural identity of states (Sharkansky 1969).4 Historians such as Piott (2004) point to strong-machine politics even in initiative states in the years prior to the adoption of the initiative. But it still could be the case that there is a

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

79

historically persistent and underlying weakness in party organizations. By controlling for the standard measure of state culture, we can see whether direct democracy still has an impact. Table 5.3 shows that the pattern found in the results of table 5.2 holds up even when we introduce statistical controls for at least one version of the cultural argument: ease of use of the initiative is still associated with weaker parties and a more restrictive legal environment. Overall, the weight of the evidence we examined is consistent with the argument that direct democracy is a threat to party organizations and party in the electorate, but the evidence may not be as uniformly strong as we might like. The evidence seems especially muddy when we consider the third of the measures of partyness—party in government. Effects of the Initiative Process on Party in Government In very general terms the third of the dimensions of party —parties in government—would seem not to be terribly fertile ground for a discussion of the impact of direct democracy. Notwithstanding the workings of direct democracy on other features of party and partyness, the past century has seen the two major parties retain their lock on elected office. While there are some examples of minor-party success (and some signs that these are correlated with initiative states), by and large the success of candidates such as Jesse Ventura in Minnesota are rare events, open to over interpretation.5 One preliminary method of examining just how much of an impact direct democracy has upon partisan elites in a more systematic sense is to survey elites themselves. In this way we can begin to gauge the effect that direct democracy has upon the perceptions of those in office. Responses from a mail survey of incumbent legislators and candidates for state legislature in California, Oregon, and Washington allow us to assess how the initiative might impact party in government. In table 5.3 we estimate straightforward models of candidates’ attitudes toward the initiative and other electoral processes. The survey measured how many limits candidates would have placed on initiative use, how many positive and negative perceptions they had of initiatives, and other attitudes about elections and democracy (for a discussion of the survey, see Bowler et al. 2001). Our expectation is that if direct democracy does represent a threat to parties in government, then incumbents and representatives of major parties should exhibit much more mistrust of direct democracy than challengers or minor-party candidates, and a greater willingness to limit the scope of the initiative process. Attitudes about initiatives and elections are estimated as a function of party status and incumbency, controlling for state, age, gender, and education.

80

SHAUN BOWLER AND TODD DONOVAN

Table 5.4 Models of elite opinion toward direct democracy and electoral process

Incumbent Republican Democrat Washington Oregon Age Male Education Constant Cases R-squared

(1)

(2)

(3)

(4)

Number of limits that should be placed on initiative process (0–6)

Number of positive statements about initiative agreed with (0–8)

Number of negative statements about initiative agreed with (0–9)

Initiatives bad thing ? (yes ⫽ 1, no/other ⫽ 0) Probit

0.186 (0.76) 0.30 (1.28) 1.33** (6.15) ⫺0.393* (1.73) ⫺0.27 (1.32) 0.007 (0.94) ⫺0.20 (1.07) 0.003 (0.09) 2.10 (3.36) 286 0.17

⫺0.84** (2.27) ⫺0.14 (0.40) ⫺1.51** (4.71) ⫺0.16 (0.49) ⫺0.04 (0.16) ⫺0.02* (1.93) ⫺0.32 (1.16) ⫺0.06 (0.97) 8.46** (9.24) 278 0.16

0.87** (2.20) 1.09** (2.91) 2.50** (7.09) ⫺0.82** (2.23) 0.004 (0.01) 0.05** (4.36) ⫺0.04 (0.14) 0.06 (0.94) 0.26 (0.25) 257 0.27

0.581** (2.33) 0.628 (1.26) 1.388** (2.93) ⫺0.535* (1.87) ⫺0.186 (0.69) 0.029 (0.39) ⫺0.096 (0.33) ⫺0.242 (0.90) ⫺2.018** (3.72) 232 0.13

Note: Cell entries are OLS coefficients (or, in column 4, logit coefficients). Absolute value of t statistics in parentheses. Pseudo R-squared for column 4. * ⫽ significant at p ⬍ .10, two-tail; ** ⫽ significant at p ⬍ .05, two-tail. 5%. Source: University of California, Riverside / Western Washington University survey of legislators and candidates, 1998–1999.

As we see from table 5.4, incumbent legislators are in fact significantly more critical of the initiative process; they have fewer positive things to say about it than other candidates, and also more negative things to say. We can rephrase these results in another light to state that political outsiders — members of minor parties—are much more supportive of direct democracy (and much more critical of current electoral institutions) than incumbent legislators and major party candidates. With one exception, the results in table 5.3 confirm the general hypothesis that the initiative is the weapon of “outsiders” (or losers) and hostile to the interests of party in government (or winners). The exception is that Democrats remain much more skeptical of the initiative than Republicans. Given that the 1990s were a decade in which right-wing causes were advanced by the initiative, this is, perhaps,

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

81

not so surprising. On balance, then, the evidence would suggest that the major parties in particular are both concerned, and have reason to be, about the impact of direct democracy upon them. 2. The Damaging Consequences of Political Parties for Direct Democracy The preceding discussion took seriously the claim that direct democracy harms political parties. But there are reasons for thinking that direct democracy may not pose as great a threat to political elites and political parties as one might suppose on the basis of those results because parties and partisans may be in a position to resist or limit the effects of direct democracy. First, as we noted above, several states’ direct democracy processes have little impact on policy and institutions and, hence, do not constitute much of a threat to parties or to elected politicians (Bowler and Donovan 2004). Not all direct democracy processes are created equal: West Coast states have a relatively easy to use, Populist initiative process, where legislators have a relatively harder time undercutting citizen-initiated measures, and so the responses of West Coast legislators (table 5.4) may not be typical of inland states. It seems clear that where the process is easiest to use and legislatures are least insulated from direct democracy, then legislators may have something more to be concerned about. This Populist version of direct democracy, however, is not the most common form of the process. Some states that have direct democracy— Illinois and Mississippi, for example—may have them on the books, but the process has no real effect on shaping the political institutions that define conditions political parties operate in. That means that direct democracy per se may not necessarily imply weaker parties, just that certain implementations allow antiparty sentiment to be expressed in law. And, in general, the results of table 5.2 are consistent with this because they suggest that the effects are stronger for the more nuanced measures of direct democracy. Second, parties themselves may be able to take charge of the process. A telling example of the way this may work is that the legislature itself can put proposals on the ballot. In some instances the legislature is compelled to place matters before the voters as, for example, in the case of bond issues or constitutional amendments (this applies even in noninitiative states such as Texas and New Jersey). But the legislature can even put institutional reform proposals on the ballot. Legislators may place measures before voters as their preferred alternative response to an initiative that has qualified (Banducci 1998), and they may enact policies as a preemptive response to the threat of a potential initiative (Gerber 1999).6 The legislatures’ role in affecting the mix of ballot measures that voters face should not be underestimated. In the period from 1998 to 2004,

82

SHAUN BOWLER AND TODD DONOVAN

there were over 760 proposals on the ballot across all states coming from the legislature, and 75 percent of them passed. Between 1884 and 2004 there were 1,230 measures on the California ballot alone: 900 (73 percent of all measures on the ballot) were placed there by the legislature (bond measures, constitutional measures, and so on) and 69 percent (611) passed. Thus, there are large numbers of proposals from legislatures that have been approved (rubber stamped?) by voters, showing that elected representatives can use some components of direct democracy to ratify their decisions.7 It is important to note the legislative authorship of many governance reform proposals, since not all of these measures are initiated outside of the legislature. Figure 5.2 displays data on the number of ballot measures dealing with election reform (initiative and referendum) that were put to voters between 1990 and 2004. Most of these were not initiatives. Figure 5.3 demonstrates that during the same period, voters approved more noninitiative reform proposals than initiative reform proposals. If we assume that proposals from the initiative are hostile to parties and representatives, we would seem to have to assume that proposals placed on

120

100

80

60

40

20

All Proposals

4

2

0

98

96

94

92

90

0

Initiative proposals

Source: National Conference on State Legislatures (NCSL) Web site.

Figure 5.2 Numbers of ballot proposals concerning elections and redistricting, 1990–2004

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

83

45 40 35 30 25 20 15 10 5

Initiated proposals

4

2

0

98

96

94

92

90

0

All proposals

Source: NCSL web site

Figure 5.3 Number of reform proposals concerning elections and redistricting passing 1990–2004

the ballot by the legislature are friendlier. These latter measures have been both numerous and successful. One way of making this point more concrete is to consider term-limit reforms—one of the best examples of the way in which the initiative can be used to constrain parties and legislative elites via a change in institutional design. As noted earlier, our previous work demonstrates that term limits (see appendix 2) rarely exist unless adopted by initiative. Where legislatures have a hand in drafting term limits, the limits are less stringent. Conversely, term limits are more stringent where the initiative process is easier to use and less subject to legislative amendments. Term limits, however, can also be undone altogether. Table 5.5 illustrates the fate of citizen-initiated termlimit measures in six states. Just as nongovernance-related public policies passed by the initiative may be undercut by the actions of public officials (Miller 1999; Gerber et al. 2001), so too can governance policies. Similarly, citizen-initiated attempts to broaden participation in partisan primaries have been blocked by court actions brought by the political parties. California parties eventually succeeded in persuading the United States Supreme Court to overturn that state’s “blanket primary” initiative, Proposition 198 (see California Democratic Party et al. v. Jones et al. 2000).

84

SHAUN BOWLER AND TODD DONOVAN

Table 5.5 Term-limits repeals State Idaho Massachusetts Oregon Utah Washington Wyoming

Year repealed

Year enacted

Who repealed?

2002 1997 2002 2003 1998 2004

1994 1994 1992 1994 1992 1992

Legislature State supreme court State supreme court Legislature State supreme court State supreme court

Note: In six states, term limits have been repealed by the legislature or by court action. Source: NCSL Web site.

That case also gave Washington’s state parties a legal basis to undermine that state’s blanket primary, which had been enacted via an initiative to the legislature in 1934 (see Washington State Democratic Party et al. v. Reed 2002). There are, then, reasons for thinking that the parties are far from being passive victims of the initiative, and that parties can—eventually—undo or at least attenuate the impact of some of the most serious attempts to bind them. We might also note that party organizations may be the occasional beneficiaries of the unintended consequences of institutional reform. One argument, for example, may be that term limits may help strengthen party organizations outside the legislature as a source of advice and funds to relatively inexperienced legislators who are not likely to be able to build up their own machines. Work by Cain and Kousser (2004) on California term limits, for example, shows that party leaders are being given more funds (and hence greater potential for influence) post–termlimits. There are additional examples of party organizations being the beneficiaries of electoral rule changes. Washington’s Initiative 134, authored by a state legislator, changed rules such that contributions to party organizations are now the only (mostly) unregulated contributions relative to contributions to other actors. Since I-134 passed, party organizations have come to supplant interest groups and individuals as the dominant source of funds for legislative candidates (Donovan 2004). All of these points are consistent with the argument that parties can, in fact, use the initiative to promote a particular cause or policy both as an end in itself and, also, as a boost to party fortunes at election time by bringing out partisan supporters to vote for specific issues on the ballot (Smith and Tolbert 2001; Smith, this volume). Indeed, a long-standing part of the literature on voters and the initiative concerns the positive role political parties play in helping voters orient themselves toward proposals (see, especially, Smith and Tolbert 2001; Hasen 2005; Nicholson 2005).

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

85

Cue taking is easier if parties adopt positions: just as in the school science lesson where iron filings are randomly scattered on a sheet of paper until the magnet is placed when they align into patterns, so, too, are voters ignorant and confused until party cues allow them to make sense of proposals. The role of party as cue helps smuggle some elements of the party government model back into the discussion. What this line of argument suggests is that the results of tables 5.2 to 5.4 should be taken with a pinch of salt. Perhaps the complaints of incumbents in table 5.4 are crocodile tears and perhaps the weakening of party organization that is allowed by the initiative process still has not managed to weaken parties enough. Parties may not really be damaged, just a little bruised, and the lower scores for partyness—even across a range of measures—does not correspond to a meaningful qualitative difference in the way parties behave simply, because parties can sidestep or soften many of the real restrictions on their actions. Before we move too quickly to the conclusion that parties are able to manipulate—even dominate—direct democracy, we should note that it is easy to overstate the role parties play in direct democracy contests. We can illustrate this point by reference to campaign-spending patterns. Campaign finance from California indicates a fairly limited direct role for party organizations. For example, between 2000 and 2004—and excluding spending on the recall—the 63 ballot propositions (initiatives, referendums, and legislative proposals combined) saw the central committee of the California Democrats outspend the state central committee of the GOP over four to one in dollars spent on the initiative. Some of that spending by Democrats, as the governing party, was, to either support their legislative proposals that end up on the ballot or to defend themselves against referendums.8 Of course, the argument about the threat of direct democracy to parties would imply that that the parties may spend considerable sums defensively, simply because quite a lot is at stake. As both a major party and a governing party, Democrats potentially have a lot more at stake over this period in some respects than others. Despite this, overall sums of money spent directly by party organizations are very small: the Democrats spent approximately $5.7 million on ballot propositions to the GOP’s $1.3 million over four years. In this period, total spending by the central committees of the Democrats and the GOP was $58.8 million and $23.7 million, respectively. In proportional terms the Democrats spent around 8 percent of central committee money on direct democracy contests, compared with 4.5 percent spent by the GOP central committee. Put differently, both state party organizations spent over 90 percent of their central committee money on candidate races and party-building activities (donations to county-level committees) over

86

SHAUN BOWLER AND TODD DONOVAN

a four-year period. By way of further comparison we can consider how large the party contributions were in terms of total spending on initiatives during this period. The answer again is “very small”: the elections of 2002 alone saw approximately $91 million spent on ballot propositions, and 2000 saw over $280 million.9 Compared with these figures, direct spending by state party central committees is trivial, suggesting that the state parties are not really players in direct democracy campaigns.10 There is, of course, a role for strategy in spending. Presumably, parties consciously choose which campaigns to get involved in and which to ignore. It is also possible that this time period did not contain much of substantive interest to the parties. Yet the overall spending figures suggest that parties are not as deeply involved in the process as some commentators think. It is possible to recover at least some of the argument to the effect that parties dominate ballot proposition elections by reconsidering the importance of party cues in ballot proposition elections. These party cues are both effective and cheap and, hence, pointing to the low financial investment of parties gives a highly misleading picture of low political investment and little consequence. Survey evidence suggests that the pamphlet is very important in helping voters decide which way to vote (see Bowler and Donovan 1998 for a discussion). We might, then, expect to see party figures—possibly the governor, in particular—use the ballot pamphlet to announce their involvement in the campaign. There are, in fact, several celebrated examples of this: Governor Wilson used race/immigration issues as a wedge against the Democrats in the 1990s; prior to that, Governor Reagan supported a forerunner to Proposition 13; both parties objected publicly to Proposition 198 (California’s blanket primary); and, in 2004 , Governor Schwarzenegger was on the ballot pamphlet as endorsing/objecting to several measures. But readily identifiable party elites and party organizations themselves typically take a back seat in endorsements to other players. It is unusual to see party figures take positions on propositions in the California ballot pamphlet. In some instances, that quietness might reflect that parties are not often a cue that the parties wish to deploy (a teacher of the year may be a more meaningful cue than “Democrat” on an education bond). Although published endorsements by state party figures are more common in the voter guides of Washington and Oregon than California, the party label of an partisan endorsee is often entirely absent or is used to signal a “cross-party” or “bipartisan” position—sometimes on both sides of an issue. See table 5.6 for some examples from the ballot pamphlet. What are voters, especially those outside the district , to make of the cue “Marc Boldt, state representative” being against I-685 as a means to help

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

87

Table 5.6 Endorsements in Washington state voter’s guide on selected measures I-872 This measure would allow voters to select among all candidates in a primary. Ballots would indicate candidates’ party preference. The two candidates receiving most votes advance to the general election, regardless of party.

Terry Hunt, President, Washington State Grange; Bill Finkbeiner, State Senator, Republican; Brian Hatfield, State Representative, Democrat; Sam Reed, Secretary of State, Republican; John Stanton, Chairman and CEO, Western Wireless; Darlene Fairley, State Senator, Democrat.

I- 685 Penalties for drug possession, drugrelated violent crime be revised; medical use of Schedule I controlled; substances be permitted; and a drug prevention commission established.

Dr. Rob Killian, Tacoma family physician, Hospice Director, sponsor of I-685; Dr. Rob Thompson, Seattle cardiologist; Rev. Andrew J.W. Mullins, Vice Dean, St. Mark.s Episcopal Cathedral, Seattle. Advisory Committee: Jeffrey T. Haley, Bellevue attorney, Chairman, Citizens for Drug Policy Reform; Ralph Seeley, Tacoma attorney, cancer patient; Dr. Steve Goldman, Seattle surgeon; Dr. G. Alan Marlatt, Director, Addictive Behaviors Research Center, University of Washington; Dr. Warren Gunteroth, Seattle physician.

Judy Golberg, chair President of Washington League of Women Voters; Gary Locke, Governor of the State of Washington, Democrat; Ken Eikenberry, former Washington Attorney General, past State Republican Chair; Jocelyn Langlois, acting chair, Libertarian Party of Washington State; Jody Grage Haug, Membership Chair, Green Party of Washington; Joan Thomas, past President, Seattle LWV, past president Washington LWV. R.W. Seaman, MD, FACS; R. E. Tremblay, MD F.A.S.A.M., addiction medicine; William O. Robertson, MD, medical toxicologist. Advisory Committee: Brad Owen, Lieutenant Governor; Norm Maleng, King County prosecutor; Dr. Terry Bergeson, State Superintendent of public Instruction; Marc Boldt, State Representative.

Source: Secretary of state, Washington.

voters resolve a duel between doctors?11 Perhaps, more simply, on many issues, voters simply do not require or even welcome a cue: gay rights and tax increases would seem to naturally sort voters into distinct groups with very little need for the parties prompting or for overt endorsements. But these simple examples do serve to make the point that, while it is the case that parties can use direct democracy processes by making proposals, there is, often, a very “hands off ” approach to propositions by the central party organizations in California and a slightly different, but nevertheless subdued, partisan approach in Oregon and Washington. These patterns

88

SHAUN BOWLER AND TODD DONOVAN

are somewhat at odds with the idea that parties as organizations are central players in initiative campaigns: it is hard to be so central and so much in the background at the same time. 3. Discussion In the first empirical section of this chapter we outlined the evidence in support of the argument made by critics of the process that direct democracy helps undermine political parties—and hence the party government model. We concluded that section on a note of caution because of a number of measurement issues that meant we needed to be careful not to overstate that evidence. In the second empirical section of the paper, we outlined a very different argument. Indeed, this second line of argument largely contradicts the first. A different set of critics suggests that far from undermining parties, what we see in the actual ballot proposition campaigns and elections is the once proudly Populist process of direct democracy being taken over by the political parties. To be sure, this argument covers a more limited set of features of direct democracy than the first line of criticism, because it tends to be concerned more with the conduct of campaigns and elections than it is with, say, the regulatory framework within which parties may operate. Nevertheless, it is a line of argument that says that rather than direct democracy being bad for political parties, it is political parties that are bad for direct democracy. And there is some evidence in support of that claim. We nevertheless concluded that section, too, with a note of caution because evidence from campaign spending patterns suggests a much more muted role for political parties than many observers suggest. Again, then, we concluded that we need to exercise caution in making the claim that parties predominate. Both of these sections addressed questions of measurement over key terms—not least of which is what we mean by the term “political party”— at some length. Given the kinds of data at hand, it is important to make sure that the empirical demonstrations are robust and do not turn on one particular measure. Studying empirical effects of the initiative process can sometimes have something of the flavour of the story of the blind men and the elephant, where conclusions do depend on which aspect of direct democracy one examines. What makes this worse is that debates over direct democracy can become quickly heated, and motivated commentators (either supporters or opponents of the process) may be more eager to make their case than substantiate or qualify it. Our discussion of a range of measures and their limitations is intended to make the general point

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

89

that assessment of the initiative process does require careful attention to the question of measurement and definition, especially given the heated debates that direct democracy often generates. On balance it would seem that the range of evidence does support the view that direct democracy can be used to undermine aspects of partyness but that the evidence for the argument of party usurpation of the process is a little weaker than the advocates of that argument might like. Term limits, direct primaries, anticorruption laws, and open primary rules are all examples of reforms foisted on parties much against their will and do seem to produce a generally consistent pattern of weaker measures of partyness. But one can go too far in this direction if one does not see how parties can fight back, most notably through the courts, but also over time, by beginning to chip away at reforms produced by direct democracy. Even though the workings of direct democracy may allow lasting, tangible effects on parties, parties may be able—at times—to blunt the impact. Finally, we should note that the phrase “the workings of direct democracy may allow” may seem stilted or hesitant or just downright awkward. One of the points we made in passing and underscore here is that direct democracy in and of itself does nothing to political parties. Although we have talked about the effect of direct democracy, the process itself is inanimate and has neither intent nor agency. What gives life to direct democracy are the wishes and actions of voters, interest groups, and, to an extent, political parties themselves. While critics of direct democracy can point to our findings to note the damage done to parties, it is important to remember that the damage is being done by voters, not the process itself. Critics, scholars, and the leaders of the political parties themselves may well be convinced of the value of political parties: they have yet to persuade voters. Direct democracy allows voters to express their doubts about the value and virtues of political parties, but one should be careful to distinguish the messenger from the message, and direct democracy is the messenger for messages sent by voters. Appendix 1. Definitions for tables 5.2 and 5.3 Independent Variables Direct Democracy: 0 ⫽ no direct democracy, 1 ⫽ Direct democracy for constitutional proposals or statutory proposals, 2 ⫽ direct democracy for both constitutional and statutory Qualification Bowler and Donovan eight-point measure of ease of qualification: 7 ⫽ no direct democracy; 0 ⫽ easy qualification

90

SHAUN BOWLER AND TODD DONOVAN

Insulation Bowler and Donovan ten-point measure of insulation of state legislature from direct democracy: 10 ⫽ no direct democracy; 1 ⫽ not insulated from direct democracy legislation Cultural measures: Dummy measure of Elazar’s state label as found in Sharkansky (1969)

Dependent Variables Traditional party organization (TPO) (from Mayhew 1986): For Mayhew, an organization must claim five traits that distinguish it as a TPO. It must have substantial autonomy, be long lasting, possess a hierarchical command structure, regularly seek to nominate candidates over a wide range of offices, and rely on “material” (patronage) rather than “purposive” (ideological) incentives for inducing activity from supporters (19–20). 5 being a score of high levels of organization on the 5-point scale. He also noted that states with partial factionalism (Indiana, Kentucky, Maryland, Missouri, West Virginia, Louisiana, and Tennessee). We moderate the TPO score by deducting 1 point on the TPO scale for each of these states. Appleton and Ward legal restrictions: AW summary measure based on Appleton and Ward’s “Legal Environment of State Parties,” pp. 374–375. In a series of nine dummy variables, each antiparty measure listed (e.g., state law mandates the composition of party central committees) is scored one. This list is then summed to create a measure of the hostile environment facing political parties. The thermometer score: NES this averages the thermometer scores for political parties by state over the NES’ that asked this question. A minimum state survey sample size of 50 was set as the mark. The pattern reported here is robust until we get down to very low survey responses (basically, single respondents for a couple of states). So, for sample sizes greater than 20, the N increases to 40 and the parameters 54.98 -1.08**Direct Democracy R2 =.12 (.46) 51.4 +0.52** Qualification R2 =.14 (.21) 51.7 + 0.33** Insulation R2 =.13 (.13) Corruption measure, from table 5.3, Boylan and Long Survey (2001) http://www.ripolicyanalysis.org/BoylanLongSurvey.pdf#search=’ stategovernmentcorruptionlevels’

DIRECT DEMOCRACY’S EFFECTS ON POLITICAL PARTIES

91

Appendix 2. Term limits still in effect House State

Year enacted

Maine California Colorado Arkansas Michigan Florida Ohio South Dakota Montana Arizona Missouri Oklahoma Nebraska Louisiana Nevada

1993 1990 1990 1992 1992 1992 1992 1992 1992 1992 1992 1990 2000 1995 1996

Limit 8 6 8 6 6 8 8 8 8 8 8 12 n/a 12 12

Year of impact 1996 1996 1998 1998 1998 2000 2000 2000 2000 2000 2002 2004 n/a 2007 2010

Senate Limit 8 8 8 8 8 8 8 8 8 8 8 12 8 12 12

Year of impact 1996 1998 1998 2000 2002 2000 2000 2000 2000 2000 2002 2004 2006 2007 2010

% Voted yes 67.6 52.2 71 59.9 58.8 76.8 68.4 63.5 67 74.2 75 67.3 56 76 70.4

Notes 1. See Dalton and Wattenberg (2001) for review and evaluation of this argument. 2. Bowler and Donovan (2004) report results using measures here. Pippen, Bowler, and Donovan (2002) show that increasing severity of campaign finance restrictions in the 1990s were correlated with frequency of initiative use. Frequency of use is also associated with weaker party organizations (Mayhew measure) and also the more restrictive legal framework measure (Appleton and Ward). 3. The Cotter et al. state-level measure is, surprisingly, relatively uncorrelated with both their own measure of local measure party organization and also Mayhew’s measure. One of the issues of looking at the impact of direct democracy on party is that the literature has no single “gold standard” measure of party. 4. Not surprisingly, the measure of culture and the assignment of labels such as “moralistic” to whole states based on some assessment of their character a generation or two ago does raise concerns about measurement. However, the Elazar measure is so firmly rooted in the literature on state politics as to have the status of conventional wisdom. 5. If we consider candidates of Democratic Farm Labor and Democratic Populists as minor-party candidates, for example, this allows us to include states such as the Dakotas. But this does seem to load the dice somewhat in our favor. Unfortunately, there seems little alternative to go through minor-party victories on a state-by-state basis. At the end of such an exercise, it still remains the case that the two main parties predominate.

92

SHAUN BOWLER AND TODD DONOVAN

6. Gerber’s work would suggest that such legislative maneuvering is often, if not primarily, defensive. Absent the threat of the initiative, the legislature would behave differently. See also Matsusaka 2004. 7. Popular referendums are a more direct challenge, but even here California and Washington legislators do fairly well. If we count a “yes” vote as supportive of the piece of legislation, then 1911–2004 California legislators had to face a referendum 43 times and won 15 of them. There were not very many challenges out of all the thousands of pieces of legislation, and they won a third of those. Washington does better. The legislators won 70 percent of the referendum challenges between 1970 and 2003. 8. Somewhat surprisingly, some was spent on county-level ballot measures, meaning that the total spent on statewide measures by political parties was even lower than the overall figures suggest. 9. A preliminary look at WA state data suggests a similar pattern. Thirty-six initiatives raised $33.7 million between 1997 and 2004. Of this total, around $160,000 came from identifiable party committees. There is some slippage in handling reports. Another report lists the top 50 contributors in 2004. Bill Gates was fourth with $1.1 million. The Democrat central committee ranked twentysecond with $130,000; 14 of the top 50 were tribes or casinos. Similar patterns seem to obtain in Oregon at least for 2000 http://www.sos.state.or.us/elections/ c&e/summary/G00/toc.pdf. 10. In a recent article, Rick Hasen, one of the more careful students of election issues, notes that candidates can use initiative committees to generate and spend millions of dollars. By his estimate, since 1990, there has been $84 million raised by candidate-controlled ballot-measure committees. One of Hasen’s concerns is that ballot measure funding is a backdoor means to fund candidates. But it is a point that has relevance to our argument here. This emphasis on candidates greatly increases the scope for party involvement if we consider party and candidate involvement. Nevertheless, even adding the party contributions to this $84 million implies an upper figure of party and candidate involvement in a clear minority of spending. (Hasen 2005) 11. Voters can, of course, probably figure out who the party supports from other names on the list and by the topic at hand—Democrats are probably more likely to support gay rights, for example. But this suggests a reversal of the cuetaking argument: using the ballot measure to figure out where the elected official stands rather than the other way round.

6

The Initiative Process and Interest Group Attention to Legislative Activity Frederick J. Boehmke

1. Introduction Hybrid democracy, as embodied in the twenty-four U.S. states with the direct initiative process, changes the available pathways to representation of organized interests. Put simply, the opportunity to propose legislation directly to a state’s voters—as opposed to the sole option of working with elected officials in noninitiative states—presents new alternatives for interest groups. And many groups seize upon this opportunity. But because influencing policy through the initiative process requires tactics different from influencing policy through the legislature, different groups benefit from its presence. Emboldened by their improved stature in state policy making, these groups mobilize in greater numbers and use their ability to propose initiatives to obtain leverage in policy debates. But who are these groups whose fortunes improve under hybrid democracy? Not surprisingly, they are groups who are relatively skilled at proposing and passing initiatives, yet are simultaneously less able to represent their interests and get what they want through the legislative process. Successfully influencing policy through the ballot box typically requires persuading hundreds of thousands of individuals to side with a group; success in the legislature rarely involves persuading more than a few hundred. The kinds of groups that are good at the latter are rarely also good at the former: groups with the monetary resources to maintain a constant, knowledgeable presence in the capitol often lack the broad membership and public support

94

FREDERICK J. BOEHMKE

needed to pass initiatives (Gerber 1999). In short, then, hybrid democracy improves the representation of typically disadvantaged, broad-based citizen groups vis-à-vis more narrow and wealthy economic interests.1 This representational shift is important because groups that dominate legislative lobbying are typically organized around business interests and represent a fairly narrow and wealthy swath of citizens (Schlozman 1984). The initiative process helps overcome these biases in representation by giving citizen groups a better chance at influencing the policy-making process. A number of studies confirm these differences. Comparisons of interest group populations indicate that those in initiative states are larger, include a greater proportion of broad-based citizen or membership groups, and have a greater rate of turnover from year to year (Boehmke 2002, 2004, 2005; Smith and Tolbert 2004). Studies of individual groups reach similar findings, with groups in initiative states having more members and fewer financial resources. These differences translate into different lobbying styles as well, with groups in initiative states relying more on outside lobbying tactics (Boehmke 2005). These findings demonstrate that the initiative process has important consequences for the types and characteristics of interest groups that mobilize. By most interpretations, then, the initiative process increases the representativeness of interest group communities, which are generally portrayed as dominated by business and economic interests (Baumgartner and Leech 1998, 2001; Schattschneider 1960; Schlozman 1984; Walker 1991).2 Representativeness is only the first step along the way to representation, however. The presence of more representative groups does not necessarily translate into better representation in the legislative arena, where most policy is still made. This is more than just a passing concern, as the differences between groups in initiative and noninitiative states suggest that the former may be less experienced at legislative lobbying. First, groups in initiative states tend to be younger (Boehmke 2005), more volatile, and less likely to lobby in more than one state (Boehmke 2004). These differences are particularly great for membership groups. Second, groups in initiative states tend to display a more reactive lobbying style in the sense that their decisions about which legislators to lobby are generally based on legislative characteristics such as constituency, policy positions, and other groups’ contact with that legislator. Third, groups select issues with which to get involved more as a function of what other groups are doing than on the basis of their own desires, placing less importance on items such as legislative connections and more importance on opposition groups’ actions (Boehmke 2005). These findings all suggest that groups influenced by the initiative process may be less stable and less effective in inserting themselves into the

THE INITIATIVE AND ATTENTION TO LEGISLATION

95

legislative process. If this is the case, then the representational gains engendered by shifts in interest group mobilization brought about by the initiative process may be lost in their translation to legislative participation. In this paper I investigate this issue by studying interest group attention to legislative activity. Survey data from over 300 groups in both initiative and noninitiative states allow me to determine whether the initiative process influences the ability of individual groups to participate in the legislative process. My central measure is based on groups’ responses to questions about the number of bills to which they devoted different levels of resources. Using survey data allows me to control for a variety of other factors that might influence groups’ capacity for bill involvement, including financial resources, membership, and experience. My results indicate that groups in initiative states are generally not disadvantaged in the legislative process, despite the presence of initiativemobilized groups—those who organize in part due to the presence of the initiative process. Finally, I also use the survey data to study the reasons that groups prioritize certain bills over others. This allows me to explore whether groups get involved with bills based on their own preferences and objectives or whether they merely respond to shifts in the legislative agenda or the actions of other groups. 2. Interest Groups and Government Activity Most studies of group activities in the legislature tend to focus on lobbying activity in general, whether through campaign contributions to specific legislators, lobby registrations (Gray and Lowery 1996), total lobbying expenditures during that period (Baumgartner and Leech 2001; Boehmke and Witmer 2003), or the number of issues on which the group lobbied (Baumgartner and Leech 2001; Boehmke, Gailmard, and Patty 2006). A couple of recent studies, however, relate interest group lobbying to legislative activity, focusing on whether groups drive the government agenda or whether they merely respond as it changes. At the federal level, Leech et al. (2004) explore this issue by comparing the number of groups in different issue areas with the number of hearings held by Congress, government spending, and the number of organizations in the same issue area over the course of a few years. Their findings indicate that groups respond to changes in government attention, as measured by hearings, and that growth in the interest group universe has been substantially driven by changes in the government agenda. At the state level, Lowery et al. (2004) compare the number of groups registered to lobby with the number of bills in different issue areas. They conclude that changes in group registrations and the government agenda happen concurrently (i.e., during the

96

FREDERICK J. BOEHMKE

same year), rather than changes in legislative activity affecting the number of groups in the previous or subsequent year. Neither of these studies directly links organizations and bills, however.3 So, while they tell us that groups respond to changes in the legislative agenda, they do not tell us how many different items a given group is interested in, or even the average number, since groups are often interested in multiple bills in a given year or session. The most relevant studies use information drawn from interest group lobbying disclosures to determine the number of issues or policy areas in which a group lobbied in a given year (Baumgartner and Leech 2001; Boehmke, Gailmard, and Patty 2006; Boehmke and Witmer 2006). Baumgartner and Leech’s (2001) analysis of federal lobbying data finds that the vast majority of issues have very few groups working on them: the median number of groups per issue was fifteen, and all cases below the median account for only 2.6 percent of group involvement, with the average group mentioning eight different issues. One of the important conclusions based on these findings is that, in many cases, groups are operating on issues with few other groups, and possibly there is no opposition from other groups to check their desires. In particular, this appears to be the case most often for business groups rather than broader citizen groups or nonprofits. The findings can be seen as evidence in favor of niche seeking, by which interest groups seek out relatively narrow areas to lobby in order to avoid conflict and accomplish as much as possible (Browne 1990; Gray and Lowery 1996). These studies do not provide measures of groups’ interest in specific pieces of legislation, rather they tend to focus on the related question of how changes in the government agenda and interest group communities interact. As Baumgartner and Leech (2001) note, however, levels of activity vary greatly across organizations, and measures of representation should account for this. Rather than their use of lobbying reports and issue mentions, I consider the number of actual bills that groups are interested in. 3. Group Indications of Bill Interest To study interest group attention to legislative activity, I use survey data gathered in 1999 from over 300 groups in five states, three with the initiative process and two without. The groups were randomly selected from the lists of groups registered to lobby, which were obtained from the secretaries of state’s offices. A questionnaire sent by mail to 2,050 groups asked them about general characteristics of their group as well as lobbying activities and involvement with the initiative process.4

THE INITIATIVE AND ATTENTION TO LEGISLATION

97

The question of primary importance for understanding group attention to legislative activity asked groups to indicate the number of bills to which they devoted various amounts of resources: Please answer the following questions with respect to legislative bills in your state from last year. ● ●

●

To how many did you consider devoting any resources? Of those, to how many did you actually devote some resources (staff time, money, etc.)? Of those, to how many did you devote a large amount of resources?

Asking groups to indicate the number of bills to which they devoted varying levels of resources is useful for three reasons. First, it provides clearer guidelines for groups to use in responding to the question. Second, it provides more information about the level of group activity. Certain organizations that may have a low level of interest in a variety of topics may focus most of their efforts on one or two narrow areas. Asking groups to prioritize bills in this fashion can help us get a better picture of the breadth of issues and bills that groups are concerned with. Finally, asking groups to prioritize bills in this way then allows me to ask groups how various factors influenced their decisions about which bills to prioritize. Thus, groups were then asked a follow-up question: From the previous question, when choosing from the bills that you considered devoting Medium Attention to, how important were the following in determining which you devoted a large amount of resources to? ● ● ● ● ● ● ●

Total resources available Importance of policy Member support Public support Legislative activities on that issue Lack of other group’s activity Lack of opposition

Groups were asked to rate the importance of these seven factors on a scale ranging from 1 (not very important) to 5 (very important). The bill count variable should provide different information from measures used by other studies, such as the number of issues mentioned in lobbying disclosure reports. Federal reports are filed only if groups spend beyond a certain minimum amount of money (currently $24,500) and may therefore miss groups that spend smaller amounts on lobbying. In addition,

98

FREDERICK J. BOEHMKE

these reports only count expenditures on direct lobbying, which consists of direct interaction with elected officials and certain high-ranking bureaucrats. Crucially, this measure excludes all expenditures on grassroots and outside lobbying tactics. For these reasons, I expect my bill count question to be more comprehensive and comparable across groups lobbying at different levels and using different techniques. On the other hand, this operationalization requires an actual bill to lobby on, which may not be the case for a given issue. Further, there may be variation across issues in the number of bills introduced. In the end, then, the bill count measure has certain advantages and possibly some disadvantages relative to previous measures. Given that the initiative process is a statelevel phenomenon, however, it is the only available measure comparable across states that I can use to investigate the effect of the initiative process on group involvement. 3.1. Group Attention to Legislation Basic summary statistics for these variables and others used in the analysis are contained in table 6.1. In order to increase the representativeness of the survey results, the summary statistics and subsequent analysis are weighted according to organization type; the weights were constructed in order to reflect the true distribution of organization type in the population using information from a short phone survey administered to one hundred groups.5 On average, groups indicated a low level of interest in twenty-three bills, a medium level of interest in twelve bills, and a high level of interest in four bills. These results indicate that groups pay attention to a fairly wide range of legislation, but that they pare that down quite a bit when deciding which to prioritize.6 Only 16 percent of bills that groups were interested in were made a high priority. Further, groups tend to focus a lot of their attention on a fairly small number of bills.7 Not surprisingly, perhaps, there is a large degree of variation across groups in how many bills to which they devote various amounts of resources. Groups reported low attention to between zero and eight hundred bills, though only twenty-nine groups reported interest in at least fifty bills each. For medium attention, groups reported anywhere from zero to five hundred bills, though only thirteen groups surpassed fifty bills. For high attention, groups reported anywhere from zero to two hundred bills, with only three groups reporting interest in more than fifty bills.8 The median number of bills for each category is significantly lower at five, three, and two bills in declining order of interest. In addition to the number of bills groups were interested in, the survey also asked about the reasons for prioritizing certain bills for high levels of

THE INITIATIVE AND ATTENTION TO LEGISLATION

99

Table 6.1 Descriptive statistics Mean Low-attention bills Medium-attention bills High-attention bills

23.23 12.38 4.28

Resources available Policy importance Member support Public support Legislative activities No other groups No opposition

3.45 4.61 4.02 2.75 3.89 2.22 1.77

Initiative state Involved in initiative Supporting initiative Opposing initiative

0.48 0.17 0.12 0.03

Revenue Members Number of groups involved Number of government actors Labor union Government association Trade/professional Other/mixed groups In years active In years on issue

3.42 2.36 3.38 6.83 0.03 0.1 0.39 0.41 0.64 0.38

S.D.

Min.

Max.

Attention to bills 74.26 0 800 39.58 0 500 15.03 0 200 Reasons for prioritizing 1.39 1 5 0.92 1 5 1.24 1 5 1.28 1 5 1.23 1 5 1.32 1 5 1.02 1 5 Initiative involvement variables 0.5 0 1 0.38 0 1 0.32 0 1 0.18 0 1 Control variables 1.79 1 6 2.75 0 9 1.79 1 7 29.38 0 300 0.17 0 1 0.3 0 1 0.49 0 1 0.49 0 1 0.2 0 1 0.23 0 0.92

N 268 272 267 257 267 249 252 256 242 239 306 306 306 306 290 285 249 215 306 306 306 306 291 244

Source: Survey of state interest groups, calculated using weights. See Boehmke (2005) for more information on survey procedure, construction of weights, and survey questionnaire.

attention. As noted previously, groups were asked to rate the importance of each of the seven items on a scale from 1 (not very important) to 5 (very important). Table 6.1 reports summary statistics for these responses as well. Groups’ motivations appear to vary quite a bit. By far the most critical motivation is, not surprisingly, the importance of the policy to the group, which receives an average importance score of 4.6. Only three other items exceeded 3: member support, legislative activities, and resources available. At the other end of the spectrum, three items were not indicated to be very important considerations: public support, lack of other groups, and lack of opposition all received average scores of less than 3, with the last item scoring the lowest average at 1.8.

100

FREDERICK J. BOEHMKE

These results indicate that groups consider both their own objectives and organizational maintenance when deciding which bills to prioritize, but that they also consider their own limited resources and opportunities created by individual legislator’s actions. Public support appears to play a smaller role in this decision, and the activities of other groups seem to play almost no role at all. This suggests that groups do not look for “easy” opportunities by focusing on issues with little opposition and that they may not seek out niches with little activity that they think deserve attention. Thus, these results indicate that groups respond to changes in the political agenda and probably end up paying attention to the same bills as do other groups with similar issue concerns. 4. The Effect of the Initiative on Bill Attention and Priorities In this section, I turn to the question of how the initiative process affects interest group attention to legislation and group motivations for prioritizing certain bills. First, I compare the responses of groups in initiative states with those in noninitiative states. Second, I focus solely on groups in initiative states and compare the responses of groups involved in initiatives with those not involved in initiatives.9 If groups that result from initiative mobilizations are more volatile and less experienced at lobbying, as previous results suggest, then they may be involved with fewer bills and the representational gains brought by the initiative process may be overstated. On the other hand, the initiative process may make it easier for groups to push their issues onto the legislative agenda, thereby increasing their involvement with legislation. 4.1. Comparison of Initiative State and Noninitiative State Groups Table 6.2 presents the first breakdown of the data by whether the group is in a state that permits initiatives. On average, groups in initiative states indicate interest in more bills than groups in noninitiative states: 21.4 and 17.6 received low attention, 12.3 and 8.7 received medium attention, and 4.1 and 2.8 received high attention, respectively. Because of the large standard errors associated with these averages, the differences are never significant at standard levels. Contrary to concerns that might arise over the ability of initiative-mobilized groups to fully participate in the legislative process, these results actually indicate that groups in initiative states are more involved in the process. The bottom panel in the table reports the importance of the seven reasons for prioritizing certain bills. Groups in initiative states gave greater average

101

THE INITIATIVE AND ATTENTION TO LEGISLATION

Table 6.2 Bill involvement and average reasons for choice of high-attention bills by initiative possibility Noninitiative Average

Low attention Medium attention High attention Resources Importance of policy Member support Public support Legislators’ actions No other group No other opposition

Initiative

Standard Error

17.62 8.66 2.83

52.94 19.76 4.98

3.42 4.52 3.83 2.61 4 2.02 1.76

1.45 1.07 1.36 1.27 1.16 1.19 1.01

Average

Standard Error

Bill involvement 21.43 68.93 12.3 41.16 4.1 13.25 Reasons for prioritizing 3.48 1.42 4.6 0.93 3.86 1.45 2.81 1.27 3.8 1.25 2.4 1.35 1.97 1.11

c2

t 0.51 0.96 1.08

3.7 4.07 2.69 2.97 5.41 8.04* 3.25

0.34 0.68 0.15 1.27 1.31* 2.34*** 1.5*

* Test statistic significant at 10\% level, *** significant at 1\% level. Pearson c 2 calculated using response frequencies. One-tailed t-tests performed under assumption of equal variances. Source: Survey of state interest groups, calculated using weights. Average number of responses in noninitiative states: 143. Average number of responses in initiative states: 114.

scores to all categories except legislators’ actions. Further, this category, as well as the lack of other groups or opposition groups, displays significantly different average responses. As noted earlier, then, groups in initiative states appear to be more reactive in their lobbying, choosing bills to prioritize more on the basis of what other groups are doing and less on activities within the legislature. Consistent with groups’ reliance on outside lobbying and the threat of using the initiative process to further their goals, groups in initiatives also indicated that public opinion was more important, with the difference narrowly missing significance at the .10 level with a one-tailed test (p
.103). In order to understand the finding that groups in initiative states report involvement with a greater number of bills, I examine the data in two additional ways. First, to eliminate the possible influence of any outliers and to provide a better feel for the distribution of the data, I report the quartiles for the bill interest variable. Because the data are mostly integer valued, I report the overall proportion of bills in each quartile and the number of bills corresponding to that quartile. The proportions of groups in initiative and noninitiative states reporting bill interest in the same quartiles are calculated to determine how the initiative process influences bill interest. If there are no differences, then the proportions should be 25 percent in both initiative and noninitiative states.

102 FREDERICK J. BOEHMKE

Table 6.3 Tabulations of bill interest variables by initiative possibility Low attention Bills Percentile 25 50 75 100 Weighted N

2 5 10 800

Initiative State

Medium attention Overall

No

Yes

38.73 14.34 20.63 26.31

25.38 32.44 20.20 21.99

33.05 22.03 20.44 24.47

154

114

267

Source: Survey of state interest groups, calculated using weights.

Bills

1 3 8 500

Initiative State

High attention Overall

No

Yes

34.20 20.36 20.90 24.54

17.76 37.35 21.86 23.03

27.06 27.75 21.32 23.88

153

119

272

Bills

0 2 3 200

Initiative State

Overall

No

Yes

36.30 28.14 8.20 27.37

24.35 42.85 15.89 16.91

31.12 34.52 11.53 22.83

151

116

267

THE INITIATIVE AND ATTENTION TO LEGISLATION

103

These tabulations are presented in table 6.3; the values for the different quartiles are much smaller than the averages and suggest that a few very active groups are pushing up the mean number of bills. Over 75 percent of groups indicated low attention to at most five bills; the same proportion indicated medium attention to at most three bills. Further, a full 33 percent of groups indicated no attention at any level to any bills. A fairly consistent pattern emerges with respect to the initiative process. First, groups in initiative states are less likely to report interest in few bills for each category. For example, 24 percent of groups in initiative states indicated a high level of interest in zero bills while the proportion in initiative states is 36 percent. This pattern holds for the lowest quartile across all levels of bill attention. Second, groups in noninitiative states are also more likely to report at the high end of the scale, with greater proportions for the top quartile across all levels of interest. Third, initiative state groups make up for these differences by reporting much more frequently in the second quartile. One-third of groups in initiative states report low interest in three to five bills as opposed to one-seventh in noninitiative states; similar results obtain for higher levels of interest. These results help interpret the finding that initiative state groups report more bills, on average, at each level of attention. Specifically, groups in initiative states are less likely to sit on the sidelines for an extended period of time and ignore legislative activity. These groups shift from the lowest quartile to the second quartile, which increases the average number of bills on its own. This shift could occur for a few reasons, but two possibilities suggest themselves. First, groups in initiative states tend to be more volatile and may therefore fail to register to lobby or disband if they are not currently working on an issue. Second, the presence of the initiative process may make it easier for groups to force their issues onto the legislative agenda, thereby creating at least token legislation for the groups to get involved with. At the same time, there are fewer initiative state groups interested in a large number of bills, and while these differences are not as great as at the lower end, they do suggest that fewer initiative state groups get involved with a wide range of legislation. So, while groups in initiative states are more likely to be interested in at least one bill, most of this shift is for groups interested in a small number of bills. The second, more detailed look at the data is displayed in table 6.4, which disaggregates the effect of the initiative process on bill interest by organizational type. This allows me to determine whether the effect of the initiative process varies across types of organizations, which is important since the initiative process has been found to benefit broad/citizen

104

FREDERICK J. BOEHMKE

Table 6.4 Tabulations of bill interest variables by organization type and initiative possibility Obs.

Levels of attention Low

Trade association Professional association Labor union Business or corporation Government organization Social, charity, research, or foundation Other

Medium

High

No

Yes

No

Yes

No

Yes

No

Yes

24 23 3 19 14 43

30 22 3 10 14 30

6.67 31.30 6.00 17.79 53.00 13.13

22.57 74.59 5.67 10.75 36.75 9.35

4.58 7.61 4.00 9.33 22.75 8.20

13.03 42.07 3.67 7.35 14.79 5.39

2.40 1.33 2.67 2.56 8.15 2.83

3.69 10.14 2.33 2.90 4.43 2.12

20

11

17.90

7.14

9.80

3.36

3.05

1.36

Source: Survey of state interest groups, calculated using weights.

interests more than narrow/economic interests (Boehmke 2005; Gerber 1999; Matsusaka 2004). Thus, its effect on bill interest should vary across these groups as well. Groups are broken down into seven categories based on their response to the survey question about organization type; I combine four of the original categories into one category, owing to the relatively low number of responses.10 These groups are from the social, research, foundation, and charity categories and coincide most directly with broader interests. The results demonstrate that the effect of the initiative process does vary in important ways across different categories of groups. Most clearly, these results indicate that the process has its greatest effect on trade and professional associations, which indicate interest in a much greater number of bills—over twice as many in all categories except one. Labor unions and businesses both have very little difference in the two types of states, though their numbers are slightly lower in initiative states. Government organizations, on the other hand, express themselves in about two-thirds as many bills in initiative states at the three different levels. On the basis of the relatively small sample size, it is too early to speculate on why this might be, but if it turns out to be accurate, it suggests that the initiative process may have important consequences for the political activity of government associations. The combined social category shows more modest decreases in bill interest in initiative states, with about 30 percent fewer bills. Finally, groups in the other category listed 60 percent fewer bills in initiatives states.11 The findings for trade and professional organizations are interesting since they are the type of group most likely to have the combination of members and financial resources necessary to succeed at the ballot box.

105

THE INITIATIVE AND ATTENTION TO LEGISLATION

4.2. Differences by Initiative Involvement In order to determine whether these differences are felt mostly by initiative users or whether they affect all groups in initiative states, I turn my attention to comparisons between these two categories of groups. The average responses for groups in initiative states based on involvement with an initiative is presented in table 6.5. About one-third of groups in initiative states indicated in their survey responses that their current issue involved a potential initiative. The results for bill involvement show that groups involved with initiatives are interested in significantly more bills than groups not involved in initiatives. In fact, the average number of bills is about twice as great for all three categories of attention. Further, the difference is weakly significant (p
.10) for both the medium and high attention levels. These results indicate that groups involved with the initiative process tend to be more involved with current legislation than groups not involved with initiatives. Again, this may result from the ability of groups to pressure the legislature into putting their issues onto the agenda in order to avoid an actual ballot measure on the subject. At the same time, the reason for part of the difference appears to be the activities of groups opposing specific initiatives, of which there are ten in the data: these groups list a Table 6.5 Bill involvement and average reasons for choice of high-attention bills by initiative involvement Not involved Average

Low attention Medium attention High attention Resources Importance of policy Member support Public support Legislators’ actions No other group No other opposition

Standard Error

16.97 8.25 2.78

43.41 17.69 6.77

3.46 4.62 3.97 2.85 3.91 2.01 1.97

1.46 1.06 1.41 1.31 1.28 1.3 1.19

Involved Average

Standard Error

c2

Bill involvement 29.06 98.64 19.03 63.1 6.17 19.46 Reasons for prioritizing 3.5 1.38 6.47 4.58 0.75 13.35** 3.73 1.49 2.91 2.76 1.23 2.09 3.67 1.22 4.27 2.88 1.26 19.9** 1.98 1 8.44*

t 0.91 1.38* 1.35* 0.17 0.2 0.93 0.4 1.06 3.59*** 0.04

* Test statistic significant at 10\% level, ** significant at 5\% level, *** significant at 1\% level. Pearson c 2 calculated using response frequencies. One-tailed t-tests performed under assumption of equal variances. Source: Survey of state interest groups, calculated using weights. Average number of responses from groups not involved with noninitiatives: 75. Average number of responses from groups involved with initiatives: 45.

106

FREDERICK J. BOEHMKE

little over twice as many bills in each category as groups involved with but not opposing an initiative. While groups supporting initiatives are involved with more bills than groups not involved with initiatives, the difference is somewhat inflated by the inclusion of groups that are drawn into initiative battles to defend the status quo. Owing to these different reasons for involvement, I control for support and opposition in the subsequent regression analysis. Comparing the importance of the reasons for prioritizing certain bills shows little difference based on initiative involvement. The one exception is the “lack of other groups” item, which groups in initiative states rated as significantly more important. This may indicate that groups get involved in initiatives on issues where few other groups are working and use the leverage of the initiative process to increase legislative attention to this issue. Note that these groups did not indicate that lack of opposition was more important, though. These results may indicate that the initiative process expands the potential legislative agenda and allows for a broader or different set of issues to be raised. 5. Regression Analysis of Bill Interest Because the characteristics of groups in initiative states are systematically different from those in noninitiative states, it is important to control for those differences when analyzing group attention to legislative bills. In this section, then, I conduct regression analysis of the three levels of bill interest, controlling for group characteristics such as membership, financial resources, and the groups’ age. This allows me to isolate the effect of the initiative process and involvement with specific initiatives. Groups in states with the initiative process are indicated by the variable “Initiative State,” while groups supporting or opposing specific initiatives are indicated with the variables “For Initiative” and “Against Initiative.” Groups can be expected to get involved with legislative bills for a variety of reasons. Perhaps, most importantly, groups need the resources necessary to track legislative activity, whether by hiring a professional lobbyist or by sending a representative to the state capitol. I control for resources using the six-category financial “Revenue” variable. I also control for the number of “Members” in a group; groups with larger and more varied memberships may find occasion to be interested in a broader set of legislative activity. On the other hand, they may also have to focus on a few narrow issues in order to further their organizational identity and maintain their membership. So it is hard to say whether this variable increases or decreases bill attention. Because groups in initiative states have more members, though, it is important to control for it in the analysis.

THE INITIATIVE AND ATTENTION TO LEGISLATION

107

Groups may also get involved when they are more experienced. I control for organizational experience indirectly in two ways. First, I include the natural logarithm of “Group Age”: older groups should have more experience and have more established connections in the legislature to keep them apprised of any activity of interest. Second, I include the natural logarithm of the number of “Years Involved” in their current issue. Groups with more experience on an issue may also be more attuned to current legislative activity. Because this variable is measured with respect to one issue, however, its effect on overall bill involvement may be attenuated. I also control for the activities of other political actors. First, I control for the amount of conflict that a group faces with a variable for the “Number of Groups Involved” in the same issue area. Groups involved in more conflictridden environments may experience a more extensive back-and-forth of legislative activity as both sides are constantly pushing their objectives. Second, I control for the “Number of Government Actors” involved on the issue. Groups involved in areas with attention from officials may be interested in more issues that are on the legislative agenda or in prominent issues that are the subject of multiple bills and therefore have greater opportunity to get involved. Finally, I also control for the type of organization through a series of indicator variables for “Trade/Professional Associations,” “Labor Unions,” “Government Associations,” and a category for all “Other” nonbusiness organizations (with businesses as the omitted category). Because my dependent variable is the number of bills groups are interested in, I employ models for event count data. Event count models such as Poisson and negative binomial regression are superior to linear regression models for these types of data. The difference is particularly important for my analysis since a large number of groups report interest in zero bills; linear regression models have been shown to produce biased coefficients in these situations (King 1989a). While the negative binomial and Poisson models are quite similar, the Poisson model assumes that the dependent variable’s variance is equal to its mean (E[Y]
Var[Y]). The negative binomial relaxes this assumption by allowing for overdispersion, permitting the variance to be greater than the mean. When this is the case, the Poisson parameter estimates are correct, but the standard errors are incorrect (King 1989b).12 To isolate the overall effect of the initiative process as well as the effect of involvement with specific initiatives, I estimate two different models. The first model includes just the initiative state indicator variable, whereas the second model adds indicators for whether a group is involved in support of or in opposition to an initiative on its chosen issue. I separate these two variables because groups using the initiative process to produce policy change tend to be different from those groups that are forced to respond in opposition to initiatives that other groups are putting forth.13

108

FREDERICK J. BOEHMKE

Table 6.6 Negative binomial analysis of number of bills that groups are involved with, by attention level, initiative possibility, and involvement Initiative possibility

Initiative state

Initiative involvement

Low

Med.

High

Low

Med.

High

0.21 (0.18)

0.25 (0.17)

0.33* (0.18)

0.05 (0.11) 0.23**

0.07 (0.11) 0.26**

0.05 (0.08) 0.17**

0.02 (0.22) 0.2 (0.35) 1.3** (0.51) 0.05 (0.11) 0.23**

0.02 (0.19) 0.2 (0.28) 1.48** (0.51) 0.06 (0.11) 0.26**

0.16 (0.2) 0.22 (0.28) 0.95 (0.62) 0.04 (0.08) 0.17**

(0.08) 0.04 (0.03) 1.82** (0.61) 0.29 (0.41) 0.46**

(0.08) 0.05 (0.03) 1.23** (0.48) 0.45 (0.36) 0.36**

(0.06) 0.05 (0.03) 1.38** (0.59) 0.54 (0.44) 0.23

(0.07) 0.04 (0.03) 1.74** (0.57) 0.49 (0.41) 0.43*

(0.07) 0.03 (0.03) 1.19** (0.49) 0.61* (0.36) 0.32*

(0.06) 0.05 (0.03) 1.32** (0.59) 0.72* (0.38) 0.21

(0.22) 0.63 (0.42) 1.88** (0.44) 0.77**

(0.18) 0.68* (0.35) 1.51** (0.39) 0.69**

(0.19) 0.58** (0.29) 0.77** (0.37) 0.85**

(0.22) 0.55 (0.44) 2.27** (0.4) 0.87**

(0.18) 0.55 (0.37) 1.93** (0.36) 0.76**

(0.19) 0.5 (0.3) 1.05** (0.33) 0.89**

(0.36) 0.31 (0.41) 0.29 (0.67) 0.49** (0.09)

(0.32) 0.05 (0.37) 0.81 (0.56) 0.25** (0.1)

(0.32) 0.03 (0.28) 1.6** (0.45) 0.23* (0.12)

(0.39) 0.28 (0.42) 0.25 (0.67) 0.45** (0.09)

(0.33) 0.01 (0.37) 0.74 (0.53) 0.17* (0.1)

(0.33) 0.02 (0.29) 1.56** (0.45) 0.18 (0.12)

For initiative Against initiative Revenue Number of groups involved Members Group age Issue involvement Number of government actors Trade/professional Labor Government association Other groups Constant Overdispersion

N
306. Weighted analysis. Standard errors in parentheses. * Significantly different from zero at the .10 level; ** Significantly different from zero at the .05 level. Source: Survey of state interest groups, calculated using weights. Owing to missing data, results are based on multiple imputation using Amelia (Windows version) by Honaker et al. (2000). See King et al. (2001) for more information.

The results for the first set of models are reported in the “Initiative Possibility” columns in table 6.6. These results confirm the previous findings as the coefficient for the initiative state indicator is positive for all three levels of bill attention; interestingly, the coefficients increase with the level

THE INITIATIVE AND ATTENTION TO LEGISLATION

109

of bill interest. The coefficient of 0.33 in the analysis of high-attention bills is significant at the .10 level and the coefficient for medium-attention bills narrowly misses significance at that level (p
.14). In terms of actual increases in bill involvement, these two coefficients correspond to 28 and 38 percent increases in the number of bills in the corresponding category.14 A variety of other factors also influence bill involvement. Somewhat surprisingly, the coefficient for group resources, while positive, is not significant in any of the models. Group membership also has a positive effect, and the coefficients for medium- and high-attention bills are nearly significant at the .10 level (p
.12). The final two measures of group characteristics correspond to overall experience and experience with the group’s current issue: group age is positive and significant in all three models whereas length of issue involvement is not. These findings indicate weak support for the role of resources in determining how many bills to which groups pay attention, but indicate an important role for processes that accumulate over time, such as learning about an issue area or developing reliable contacts in government. The measures of other actors’ involvement are both associated with more bills. Groups involved with more conflict-ridden issues demonstrate a clear and highly significant pattern of interest in more bills than groups working on issues with fewer groups. Further, groups working in areas associated with attention from more government actors also report activity on more bills, with significant coefficients in the analyses of low and medium levels of attention. These results are not straightforward to interpret, since it may be the case that issue areas characterized by a greater number of bills may attract attention from more groups or government officials. Note that the groups’ responses for the variables are based on their experience with one particular issue area, however, and are not averaged across all of their issue areas of interest. Finally, the organizational type variables indicate that labor groups are involved with fewer bills, and government associations and trade and professional groups are involved with more bills, relative to business groups. The final set of results adds variables for whether the group was involved with and in favor of or in opposition to an initiative on its chosen issue. This analysis can help determine how involvement with the initiative process influences groups’ attention to legislative activity, though one should keep in mind that the involvement variable is also based only on the specific issue chosen by the group in the survey. The results indicate the difference between groups in favor of initiatives and those opposed to them, as the coefficient for the former is the only significant one, with positive and significant coefficients for both low and medium levels of attention and a nearly weakly significant coefficient for high-attention bills (p
.15). The magnitudes of these effects are quite large, corresponding to an over 250 percent increase in the number of bills. At the same time,

110

FREDERICK J. BOEHMKE

groups supporting specific initiatives are not involved with significantly more bills; further, once involvement is accounted for, the effect of being in an initiative state is no longer significant. The coefficients reported represent the difference between groups in initiative states involved with proposals and groups in initiative state not involved with proposals. In order to compare groups in initiative states involved with specific measures with groups in noninitiative states, one must combine the coefficients for initiative possibility and initiative involvement. Because the coefficient for initiative possibility is quite small for low- and medium-attention levels, there is little difference in the results: only groups opposing initiatives are interested in significantly more bills. Yet the net effect for groups supporting initiatives is nearly significant for high-attention bills (p
.15) and corresponds to a 46 percent increase in the number of bills. The findings for most of the other variables are essentially unchanged, but there are a few differences. Years of involvement with the current issue now has a positive and significant effect on the numbers of both mediumand high-attention bills. In addition, the coefficients for the number of government actors involved slip from significance at the .10 level to the .05 level for both low- and medium-attention bills. 6. Conclusion The results from this analysis show that groups in states with the initiative process are not disadvantaged in their ability to participate in the legislative process. This finding is important because previous studies of groups in initiative states have found that they tend to be less experienced with and connected to the legislative process. If that is the case, then groups in initiative states might find it harder to represent their interests in the legislature, potentially undermining apparent representational gains made on the basis of the incentives the initiative offers. The fact that they report interest in just as much, and sometimes more, legislation as groups in noninitiative states alleviates concern about the link from representation through mobilization and representation through lobbying on specific issues. Within these results lies some interesting variation, particularly once involvement with specific initiatives is considered. Groups opposed to initiatives report interest in greater numbers of bills across all levels of interest. I suspect this is due to the fact that these are groups that are well connected in the legislature and have been able to force their opponents to go outside the legislative process by proposing a ballot measure. These groups are not interested in more bills owing to initiative mobilizations; rather they are drawn into initiative battles owing to their interest in

THE INITIATIVE AND ATTENTION TO LEGISLATION

111

multiple issue areas and strong connections in the legislature. Groups supporting specific initiatives also see an increase in the number of bills in which they are interested, but only for those bills accorded a high level of interest. This may be caused by the ability of these groups to use the initiative process to push issues they care about onto the legislative agenda. Of course, these results are only suggestive and do not tell us about how well groups represent their preferences on those bills that they are involved with. All we know is that groups in initiative states devote resources to a greater number of bills; in particular, they are less likely to devote resources to a very small number of bills. Given that groups in initiative states have fewer connections in the legislature, it may be the case that their lobbying efforts are not as effective as those of groups in noninitiative states. Thus, an important question to explore is whether groups in initiative states are as effective at lobbying. Effectiveness has proved to be a tricky concept to measure, of course, meaning that it may be difficult to trace the process from mobilization to policy consequences. There are reasons to think that some of their possibly diminished effectiveness is compensated for by other factors. Groups in initiative states lobby differently because of their different resource mix; thus, their focus on public opinion and constituency preferences when lobbying may be a function of their relative advantage in resources such as members. Further, their lobbying efforts are bolstered by their ability to propose initiatives if the legislature is not sufficiently responsive. This argument helps explain why states with the initiative process tend to adopt policies more in line with voters’ preferences (Matsusaka 2004). It may also result in a shift in the type of issues that are on the legislative agenda toward the kinds of issues that broader-based groups focus on. Groups in initiative states tend to be more interested in some issues and less interested in other issues; variation in these interests may be reflected in the types of bills introduced or the policies debated in the legislature. Finally, addressing the issue of representation in state policy making more generally would require attention to both the legislative and administrative branches. The last century has witnessed the emergence of a vast bureaucratic process that plays on increasingly important role in creating and changing state policy outcomes, with varying degrees of guidance from the legislature. Groups have ever-increasing incentives to pay attention to policy decisions made through the bureaucratic process. Lobbying and participation in the administrative branch is likely to be more difficult for initiative-mobilized groups, however, as it requires detailed, technical information about policy options. Further, the presence of a broad-based membership may matter less for bureaucrats not facing reelection (this depends, of course, on how well legislators are able to influence bureaucratic

112

FREDERICK J. BOEHMKE

behavior). In fact, little is known about how hybrid democracy influences bureaucratic politics in general (however, see Matsusaka’s chapter in this volume for one perspective), not just in the context of interest group participation. Future work is necessary to obtain a broader understanding of the influence of the initiative process and initiative mobilizations on the ability of organized interests to represent their constituents in the policymaking process, in all its forms. Notes 1. A number of empirical studies of initiative sponsorship provide clear evidence that citizen groups disproportionately utilize the initiative process to further their policy goals (Boehmke 2003a; Donovan, Bowler, McCuan, and Fernandez 1998; Ernst 2001; Gerber 1999). 2. However, see Lowery and Gray (2004) for an extended discussion of the difficulty inherent in defining a representative interest group system. 3. As far as I can tell, none of the major surveys of organizations in Washington, D.C. asked groups about the number of bills they were interested in (i.e., Heinz et al. 1993; Schlozman and Tierney 1986; Walker 1991). 4. In addition to the mail questionnaire, I also conducted brief phone interviews with an additional one hundred groups. Given the relatively low response rate to the mail survey, information from the phone survey is used to weight the data by organization type in order to increase its representativeness. For more details on the survey design, the data, and the use of the phone survey data to correct for representativeness, see Boehmke (2003, 2005). 5. For information on the phone survey and the values of the weights used, see Boehmke (2005), specifically appendix C, 172–174. 6. In a few cases, groups listed a range rather than a precise number of bills (i.e., five to ten bills). To include these responses in the subsequent analysis, I took the midpoint of the supplied range. Such cases were not very extensive, however, constituting only 10.4 percent of responses for low attention, 6.6 percent for medium attention, and 1.9 percent for high attention. 7. Recall that the average group listed eight issues in the LDA (Lobbying Disclosure Act) reports analyzed by Baumgartner and Leech (2001). 8. The same group is responsible for all three maxima. While it is impossible to know how comparable their responses are to other groups, the group has been around for quite a long time and has a very large budget and hundreds of paid employees. 9. After responding to questions about general organizational characteristics and lobbying, groups were asked to answer a variety of questions about their lobbying behavior on one issue of their choosing. When asked to choose this issue, the groups were also asked if they had been involved in support of an initiative or potential initiative and use that issue as the focus for their responses. I use this variable to determine which groups are initiative users and which are not.

THE INITIATIVE AND ATTENTION TO LEGISLATION

113

10. Labor unions are not combined despite few responses; it wasn’t clear which category to put them with, and their relatively small averages would have affected whichever category they were combined with. 11. The responses for one group in this category were excluded from this table because the group only provided answers for two of the levels of attention, and their answers were dramatically larger than those from other groups. Using only groups that provided answers for all three categories did not affect the interpretation for the other types of groups. 12. Another possibility is that the variance is less than the mean, known as underdispersion. This case, along with the other two (overdispersion and equidispersion) can simultaneously be estimated using King’s generalized event count model (1989b). My data exhibit overdispersion, so I use negative binomial regression with constant overdispersion. 13. Because many initiative efforts in the survey had not reached the ballot, there are few groups opposed to initiatives in the sample (ten). Thus, results for this variable are probably best considered preliminary. 14. The marginal effect of a continuous variable in a count model is
E[Y|X]; for indicator variables the percent increase is exp(
 )-1.

This page intentionally left blank

7

Direct Democracy and the Executive Branch John G. Matsusaka

1. Introduction The executive branch seems like a bystander in direct democracy. The main instruments of direct democracy—the initiative and the referendum—are legislative in nature, and the ultimate purpose of direct democracy is to break the legislature’s monopoly over law making. Not surprisingly, the study of how direct democracy impacts the legislature has been a staple of the literature for decades. Little, if any, research is available that focuses specifically on how direct democracy impacts the executive branch.1 Yet, in a system where government is fragmented between branches of government and constrained by a system of checks and balances, a reduction in the power of one branch might be expected to have repercussions on the power and functions of the other branches. The executive branch may stand to gain from a vigorous use of direct democracy. Indeed, governors have been the driving force behind adoption of the initiative and the referendum from Hiram Johnson in California in 1911 to Kirk Fordice in Mississippi in 1992. This chapter explores the relation between direct democracy and the executive branch, with an eye toward identifying what changes, if any, are brought about in the executive branch when citizens are given the power to propose and pass laws directly. The chapter is organized around three questions. How does direct democracy change the balance of power between the legislature and the executive branch? What changes does the initiative bring about in the institutional characteristics of the executive branch? And how does direct democracy change the way voters behave when electing governors? I attempt to answer the

116

JOHN G. MATSUSAKA

questions using conventional theories and straightforward empirical approaches. The evidence indicates that direct democracy brings about material changes in the functioning of the executive branch. A counterintuitive implication of the analysis is that even though the initiative and the referendum are targeted at legislative functions, they may end up causing elected executive officials to perform their executive functions more effectively. 2. Checks and Balances This section makes a few observations about how direct democracy affects the balance of power before turning to the more substantive empirical and theoretical analysis in the following sections. The main point is that the initiative and the referendum have an ambiguous effect on the power of the legislative branch and the executive branch relative to each other in a standard spatial model, but unambiguously reduce the power of each branch relative to the voters. Consider a simple model along the lines of Gerber (1996b) and Matsusaka and McCarty (2001). Policy x is a point on the real line, L is the legislature’s ideal point, G is the governor’s ideal point, V is the (median) voter’s ideal point, and all actors have single-peaked preferences. Without loss of generality, suppose L ⬍ G. The equilibrium policy x * is the outcome of a game between the parties. The question of interest is whether x * moves closer to L or G when the initiative becomes available. When the initiative is unavailable, the voter plays no role in the policy decision (as is conventional, this ignores the voter’s role in selecting L and G, a point discussed later in this chapter). For the moment, ignore the possibility that the governor has veto power. Then the legislature simply sets the policy at its ideal point, and the policy with no initiative is x *NI ⫽ L. If the initiative is available, then outside groups can propose alternatives to the legislature’s policy, and the voter chooses between the two options. A realistic case is where the governor is able to propose an alternative. This is not a constitutional right but a de facto power that stems from the governor’s ability to raise funds and marshal interest groups to qualify a measure for the ballot. Arnold Schwarzenegger in California is a recent example. If the governor has the power to make a proposal, the ultimate policy can never be farther from the governor’s ideal point than when the initiative is unavailable (see Gerber 1996), where the governor takes the place of the interest group). A first case to consider, shown in figure 7.1(a), is where V ⬍ L. Here the legislature can set the policy at its ideal point and there is no alternative preferred by the governor that the voter also prefers. Given

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

117

* = xI* xNI

(a)

x V

L

G

* xNI

xI*

L

V

z x

(b) G

* xNI

x

(c) V

L

Q

G

Figure 7.1 Ideal point configurations of legislature (L), governor (G), and voter (V) to illustrate the impact of initiative on balance of power

a fallback of x ⫽ L , the governor does not want any policy to the left of L, and the voter does not want any policy to the right of L. Therefore, the governor will not propose an initiative, and the legislature can set the policy at its ideal point with impunity. The ultimate policy with the initiative is then x*I ⫽ L , the same as without the initiative. The initiative has no effect. The second case, shown in figure 7.1(b), is where L ⬍ V. Here there is always a policy that both the governor and the voter prefer to x ⫽ L , so the legislature cannot set the policy at its ideal point without running the risk of being overridden by an initiative. For example, if the legislature were to choose x ⫽ L , then the governor could propose a policy closer to V, such as x ⫽ V. The voter prefers, and will approve, this alternative, making the voter and the governor both better off. As shown by Romer and Rosenthal (1979), the governor would not be so accommodating to the voter as to choose x ⫽ V, however. Instead the governor would choose a policy that makes the voter indifferent between the legislative option and the alternative; here x ⫽ z if the legislative status quo is x ⫽ L (chosen so that z and L are equidistant from V, measured in the voter’s utility). In anticipation, the legislature will want to preempt the initiative by choosing x ⫽ V at the start (from the legislature’s point of view, x ⫽ V is better than x ⫽ z.) Because

118

JOHN G. MATSUSAKA

the voter will not approve z or any other alternative if the existing policy is x ⫽ V, the legislature’s policy stands: x*I ⫽ V.2 Observe that the final policy is closer to the governor’s position than when the initiative is unavailable. The change is brought about not by an actual initiative but by the legislature’s response to a possible initiative, often called the “indirect effect” or “threat effect” of the initiative. It is fairly easy to see from perturbations of figures 7.1(a) and 7.1(b) that there is no configuration in which policy moves away from the governor’s position when the initiative is introduced. Similarly, it is clear that the voter is never worse off when the initiative is available.3 There is a countervailing effect that can reduce the governor’s power relative to the legislature when the initiative is available. In 43 states, the governor has the power to veto some types of legislation. Initiatives are not subject to this veto. It is easy to think of situations where the loss of veto power causes the governor to lose influence relative to the legislature when the initiative is available. Figure 7.1(c) illustrates one case. Let Q be the status quo policy, that is, the policy that prevails if the governor vetoes the legislature’s proposal. In this example, the legislature and voter would like the policy to move to the left of the status quo, but the governor will veto any proposal in that direction. For simplicity, suppose that the governor’s veto cannot be overridden. Without the initiative, x*NI ⫽ Q will prevail. When the initiative is available, any alternative to the left of Q will be approved by the voters when faced with Q as the reversion point. The exact location of the final policy will depend on the position of the interest group sponsoring the initiative, but will always be to the left of Q: x*I ⬍ x*NI. In short, the ability of outside groups to make proposals that are not subject to the governor’s veto can shift policy in favor of the legislature and against the governor. Such shifts never hurt the voter. To summarize, there is no purely theoretical reason to believe the initiative tips the balance of power in favor of the governor or the legislature. Each case is possible under some scenarios. However, some practical considerations suggest that the governor will usually benefit. The governor is chosen by the electorate at large, and thus is more likely to be representative of the median voter than the legislature, that is, G < V. The legislature is comprised of members elected from geographically distinct areas, each of whom is likely to represent the median voter in his or her district. If the legislature’s position L is the median of its members, the position will seldom correspond to the median of the population, that is, L " V (Gilligan and Matsusaka 2006). A fair amount of evidence suggests that districting does in fact cause a divergence between the position of the legislature and the median voter in the general population.4 If the governor is closer to the median than the legislature, then the configuration most

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

119

likely to hold is the one shown in figure 7.1(b). As discussed, in this case the initiative helps the governor by allowing him to take proposals directly to the voters. The governor loses little from being unable to veto initiatives. The recent history of California provides a good example. The legislature is heavily gerrymandered so that the Democratic Party has a majority in both the upper and lower houses that exceeds its popular vote majority. Nevertheless, after taking office in 2003, Republican Governor Schwarzenegger was able to use the threat of an initiative to force the legislature to repeal legislation granting driver’s licenses to illegal immigrants just months after it was passed. He also used an initiative threat to get the legislature to repeal an unpopular car tax and adopt workers’ compensation legislation that was closer to his ideal point than what the legislature had passed less than a year earlier. Because Schwarzenegger is closer to the median voter than the legislature, direct democracy has been a central part of his governing strategy and has allowed him to implement more of his agenda than he would have been able to if the initiative was unavailable. The initiative only empowers the governor to the extent that the governor is closer than the legislature to the median voter. If the legislature is closer to the median voter, then the tables would be turned. The political actor that always wins (never loses) from having the initiative available is the median voter (in a spatial model, and subject to the caveats discussed earlier and others). In some sense, then, the primary effect of the initiative is power transfer from both branches of government to the median voter. The evidence is fairly strong that the initiative does in fact bring about policies favored by the majority (Lupia and Matsusaka 2004; Matsusaka 2004). 3. Executive Institutions The initiative gives voters the power to change the institutional structure of the executive branch. Voters can rewrite the rules that determine the balance of power, the amount of fragmentation, and so on. It is well known that initiatives often are used to curtail the power of legislatures. Most notably, initiatives or the threat of initiatives recently brought about legislative term limits in 22 of 24 initiative states. Only 2 of 26 noninitiative states have adopted legislative term limits. This section investigates how the initiative changes the institutional structure of the executive branch. Unfortunately, little theory is available to guide the analysis. Most research on institutions has focused on how they affect policy, not on what leads to their adoption. A notable exception is the work of de Figueiredo (2003), which advances the idea that institutional change is driven by incumbent politicians attempting to insulate their policies from revision by future officeholders. Consistent with this idea, de Figueiredo finds that

120

JOHN G. MATSUSAKA

states are more likely to adopt the item veto when the incumbent government’s reelection prospects are bleak. While de Figueiredo’s study represents an important contribution to the issue of institutional change, it does not help much here since it emphasizes political determinants while the question at hand concerns institutional determinants (that is, direct democracy as a causal factor). Absent a compelling theoretical framework, the analysis in this section proceeds at a descriptive level. I look for differences across states in executive institutions and then consider whether those differences can be attributed to the availability of direct democracy. The empirical strategy is to compare the institutions in initiative states with those in noninitiative states, controlling for other determinants of the differences. I will not examine the content of actual initiatives, except in passing. The reason for comparing states with and without the initiative rather than examining the actual measures on the ballot is that initiatives have both a direct effect on policy (when the voters approve a measure) and an indirect effect (when the threat of an initiative causes the legislature to modify its policy choice). If we were to try to measure the effect of initiative by looking only at the measures that actually passes, we would not capture the indirect effect, which turns out to be important. We can capture both direct and indirect effects by comparing initiative and noninitiative states. Intuitively, no matter how the institutional changes come about—directly or indirectly—the effects will show up in the actual institutions. This strategy is conventional in the empirical literature; see Matsusaka (2004, 2005a) for further discussion, examples, and references. Table 7.1 examines difference in executive institutions between initiative and noninitiative states along four dimensions.5 For the purposes of this analysis, an “initiative state” is a state in which voters are allowed to amend the constitution by initiative. States that only allow statutory initiatives are excluded because many of the institutional features under consideration are provided in the state constitution and cannot be reached by statutory initiatives.6 The left panel shows the means for noninitiative and initiative states, and the right panel reports the coefficients from regressions that estimate the difference between initiative and noninitiative states, controlling for region and population. The first row shows that initiative states are 24 percent more likely than noninitiative states to impose term limits on their governors. This may partly be due to regional factors than due to the initiative. The regression in the first row controls for population and whether the state is in the South or not. Southern states are more likely to limit the terms for the governors, all else equal, but there is no measurable difference between large and small states. The difference between initiative and noninitiative states remains after including the controls, and the effect is statistically significant at better

Table 7.1 Comparison of executive branch institutions in initiative and noninitiative states, 2003 Means

Governor is term limited (1 ⫽ yes, 0 ⫽ no) Number of elected officials

.64 5.2

Initiative states .88 6.1

Dummy ⫽ 1 if initiative state

Dummy ⫽ 1 if southern state

Population (log)

Constant

1.75** (0.88)

2.12* (1.18)

⫺0.35 (0.35)

5.46 (5.28)

1.05* (0.59)

0.97 (0.71)

⫺0.13 (0.29)

6.93 (4.31)

Open cabinet meetings required (a) (1 ⫽ yes, 0 ⫽ no)

.15

.43

1.60* (0.80)

0.48 (1.02)

⫺0.33 (0.39)

3.04 (5.69)

Governor has item veto on all bills (1 ⫽ yes, 0 ⫽ no)

.39

.71

1.27* (0.67)

⫺1.46* (0.82)

0.29 (0.32)

⫺4.51 (4.80)

Note: An “initiative state” is a state that allows constitutional initiatives (17 in all). Each row reports the mean values for noninitiative and initiative states and coefficients from a regression in which the dependent variable is in the leftmost column (governor is term limited, etc.) and the independent variables are listed under “Regressions.” Standard errors are in parentheses beneath the coefficient estimates. The coefficient on the intercept is not reported. Logistic regressions are used when the dependent variable is binary. Significance levels are indicated as follows: * ⫽ 10%, ** ⫽ 5%. All regressions use 50 observations except for (a), which uses 41. Source: Institutional variables are taken from the Council of State Governments (2004). Initiative status is taken from Matsusaka (2004). Population is from the Census Bureau (www.census.gov).

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

Noninitiative states

Regressions

121

122

JOHN G. MATSUSAKA

than the 5 percent level. The second row shows that initiative states fill more constitutional offices by election than noninitiative states, and the difference is statistically significant after controlling for region and population. The third row shows that initiative states are almost three times as likely to require open cabinet meetings as noninitiative states, an effect that is statistically significant after controlling for region and population. These results suggest that citizens require higher levels of government accountability in initiative states than in noninitiative states. The question arises whether the initiative caused the differences or whether direct democracy and accountability were themselves caused by an underlying “progressive” sentiment among voters. One reason to suspect that the initiative was the forcing factor is that most states adopted the initiative in the early twentieth century while in many cases gubernatorial term limits were not adopted until relatively recently (for example, Arkansas, California, Colorado, Florida, Maine, Michigan, Montana, Nebraska, Ohio, Utah, Washington, and Wyoming have all adopted gubernatorial term limits measures since 1990).7 Erikson, Wright, and McIver (1993) show that there is virtually no correlation in state ideology before and after World War II. Moreover, survey and other evidence on ideology summarized in Matsusaka (2004, chapter 3) reveal no meaningful difference in the ideology of citizens in initiative and noninitiative states. While the first three comparisons suggest that direct democracy constrains the executive branch, the last column points in the other direction: initiative states are almost twice as likely to give the governor a line item veto on all bills. This difference is significant at better than the 10 percent level. In this case, it is unlikely that the initiative caused the difference: most states adopted some form of line item veto many years before they adopted the initiative (de Figueiredo 2003). Moreover, I could not find instances of actual initiatives that were used to expand or curtail the governor’s veto power. Table 7.2 examines a different characteristic of the executive branch: salaries. Initiatives can and have been used to set or constrain the salaries of government officials. This table compares the salary of five important state officials in initiative and noninitiative states.8 Initiative states consistently pay less than noninitiative states, with the average difference ranging from $9,000 per year for the secretary of state to $3,000 per year for the treasurer. Salaries are affected by a variety of factors in addition to the initiative. The regressions control for region, state population, and state income. As expected, populous and wealthy states pay higher salaries, possibly because the jobs are more complex and require personnel with higher opportunity costs. Even after controlling for these factors, initiative states pay less than noninitiative states on average. None of the differences

Table 7.2 Comparison of executive office salaries between initiative and noninitiative states, 2003 Means

Regressions Population (log)

⫺3 (6)

⫺14** (7)

16*** (3)

76

⫺8 (7)

⫺22** (9)

95

86

⫺4 (5)

105

97

Treasurer

93

All offices (c)

...

Noninitiative states

Initiative states

116

112

Lieutenant governor (a)

83

Secretary of state (b)

Governor

Attorney general

Dummy ⫽ 1 if initiative state

Constant

R2

1.5*** (0.6)

⫺163*** (40)

0.54

16*** (4)

1.3 (0.8)

⫺199*** (53)

0.46

12* (7)

9*** (3)

2.0*** (0.5)

⫺114*** (37)

0.50

⫺7 (5)

2 (7)

12*** (3)

1.1** (0.5)

⫺112*** (39)

0.44

90

⫺2 (6)

2 (8)

12*** (3)

1.3** (0.6)

⫺133*** (42)

0.41

...

⫺5* (3)

⫺4 (3)

13*** (1)

1.4*** (0.3)

...

0.97

GSP per capita

Note: An “initiative state” is a state that allows constitutional initiatives (17 in all). Each row reports the mean values for initiative and noninitiative states and coefficients from a regression in which the dependent variable is the salary of the office in the leftmost column and the independent variables are listed under “Regressions.” GSP is gross state product. Standard errors are in parentheses beneath the coefficient estimates. Significance levels are indicated as follows: * ⫽ 10%, ** ⫽ 5%, *** ⫽ 1%. All regressions use 50 observations except for (a), which uses 46; (b), which uses 49; and (c), which uses 246. The last row pools the data from all five offices into a regression with office-fixed effects. Source: Salary data are from the Council of State Governments (2004). Initiative status is from Matsusaka (2004). Population and GSP are from the Census Bureau (www.census.gov).

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

Dummy ⫽ 1 if southern state

Annual salary ($thousands)

123

124

JOHN G. MATSUSAKA

are statistically distinguishable from zero at conventional levels of significance. To see if this is just an issue of degrees of freedom, the last row reports a regression that pools the observations for all five offices and includes a dummy variable for each office to adjust for differences in means. On average, initiative states pay their officeholders $5,000 per year less than noninitiative states, and this number can be distinguished from zero at better than the 10 percent level. This evidence mirrors the conclusion of Di Tella and Fisman (2004) based on a more extensive panel data set that governors are paid less in initiative states. It also fits with evidence in Matsusaka (2007) that public employees earn less in initiative cities than in noninitiative cities. Given the frequent appearance of actual initiatives that attempt to regulate salaries, there is some reason to believe that the salary differences between initiative and noninitiative states are caused by availability of the initiative. 4. Elections and Accountability Direct democracy changes the nature of the executive branch by allowing policy to be made without the consent of the legislature or the governor (section 2) and by altering the institutional structure of the executive branch (section 3). Because the governor’s job is different in initiative and noninitiative states, there is reason to believe that voters will evaluate the governor’s performance differently in the two types of states. This section develops a simple model that illustrates how voters behave differently in candidate elections when direct democracy is available and provides some empirical evidence from gubernatorial elections. The main conclusion is that direct democracy may lead to better performance by the executive, even on purely administrative tasks. A. A Model of Elections as Incentive Schemes The model adopts the view that voters use elections to provide incentives for their representatives to perform well. It therefore follows the path broken by Barro (1973) and Ferejohn (1986) in focusing on the moral hazard problem of the principal-agent relationship. The model preserves the basic features of previous work but extends the analysis to consider multiple issues and the impact of direct democracy on incentive schemes. The incumbent governor chooses an action gi + [l,h] for i ⫽ 1,...,I different issues. The issues may concern legislation such as whether or not to support a minimum wage increase, or may be purely administrative. The governor derives benefits from each policy of Bi f (gi), where f is increasing

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

125

and concave. There is also a fixed benefit R from being reelected that could represent wages, perquisites, the consumption value of being in the news, and so on; the only critical feature is that R is not a choice variable for the voters. The probability of being reelected, which the voters will choose, is p.9 If the governor does not seek reelection, say because of term limits, p represents the probability that the governor’s party will win reelection and R is a payoff provided by the party to the incumbent, for example, appointment to a cushy commission, as a reward for helping to elect his successor. The governor is risk neutral and maximizes his expected utility I

u ⫽ Bi f (gi) ⫹ pR. i=1

Incentives are only necessary when the incumbent’s preferences differ from the voter’s preferences. To study an extreme case, I assume that voters and politicians have exactly opposite views about the desirability of g for all issues. All voters are identical and suffer a “cost” of Cik(gi) for each issue, where k is increasing and convex.10 I abstract away from the issue of voter heterogeneity and distributional politics to be able to focus on incentive issues. From the governor’s perspective, the optimal choice is gi ⫽ h for all i, while from the voter’s perspective, the optimal choice is gi ⫽ l . Elections are the mechanism by which voters try to control the governor. The voter chooses a probability function p(g1,..., gI) that induces the governor to choose gi’s that minimize his or her cost. Good performance is rewarded with a high probability of reelection, while poor performance earns a low probability of reelection. In principle, candidates for governor could differ in their Bi’s. If so, elections would have a selection function as well as providing incentives: candidates whose campaign promises suggested low Bi’s would be elected, and incumbents whose actions suggested high Bi’s would be turned out of office. In order to draw out the implications of the incentive side of elections, I will put this issue aside by assuming that all candidates for office have identical preferences.11 Elections are thus referendums on the performance of the incumbent, not a means to select candidates of different abilities or with different platforms. There is in fact a large empirical literature showing that voters treat elections as a referendum on the incumbent to a significant degree (see Peltzman 1998 for an overview and references.) In some respects, this problem has the appearance of a standard principal-agent problem where pR plays the role of wages. The critical difference from a standard problem is that p is bounded from above and below (p + [0,1]). This limits the voter’s ability to incentivize the governor

126

JOHN G. MATSUSAKA

and, as Barro (1973) demonstrated, means that voters generally cannot induce the governor to adopt their ideal positions. Intuitively, there may not be “enough p” to reward the governor for making the right choice. The model is essentially a version of Barro (1973) that incorporates multitasking a la Holmstrom and Milgrom (1991). A general property of this type of model is that voters adopt a cutoff rule to determine the reelection probability. In the present model with no uncertainty, the cutoff can be expressed as a maximum total cost imposed on the voter or as issue-by-issue thresholds, each of which incurs a reelection penalty if it is exceeded. The issue-by-issue approach is more useful _ for the subsequent empirical research. Therefore, let g i be the threshold for issue i and let
i be the penalty in reelection probability if the threshold is _ exceeded, that is, if governor chooses gi ⬎ g i , the voters reduce his reelection probability by
i , all else equal. Given such an incentive scheme, the governor’s choice boils down to _ gi ⫽ g i or gi ⫽ h (if he is going to exceed the threshold, he might as well go _ _ all the way to h). He chooses g i if Bi f ( g i ) ⬎ Bi f (h) ⫺
iR. To induce any _ _ given g i , then, the voter sets
i ⫽ Bi( f (h) ⫺ f( g i ))/R. The limits of incen_ tives are apparent: because
i ⱕ 1, there is only so much g i that can be _ induced. Note that
i is decreasing in g i , all else equal. Incentives are also weakened by a multitasking constraint. For
i to be the true marginal probability, it must be the case that Ii⫽1
i ⱕ 1. If the sum exceeds 1, then the probabilities will be ineffective at the margin and the governor will exceed all of the thresholds. It is the adding up constraint that makes the initiative valuable in controlling the governor. Given these basic features of the incentive scheme, we can state the voter’s problem as follows: I

_

min _ _ Ck (g i )

g 1,..., g i i =1

subject to _

(1)


i ⫽ B( f(y) ⫺ f (gi ))/R

(2)


i ⱕ 1.

I

i =1

Constraint (2) is always binding at an optimum. Substitute (1) for (2) and let q be the Lagrange multiplier on the constraint. Assuming an interior solution, the first order condition for an optimum is (3)

_

_

Ci k⬘(gi ) ⫺ qBi f ⬘(gi )/R ⫽ 0 for all i.

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

127

The voter chooses the thresholds so that marginal gain in cost reduction is equal to the marginal cost of inducing the governor to choose the threshold. The multiplier q scales the shadow price of the scarce resource “probability.” Across issues, (3) implies _

_

Cjk⬘(gj) Ci k⬘(gi ) _ _ ⫽ᎏ ᎏ B Bi f ⬘(gi ) j f ⬘(g i ) _

_

Since k⬘/f ⬘ is increasing in g, it is easy to show that gi ⬎ gj if Ci /Bi ⬍ CJ /Bj , and therefore that
i is lower for issues with high Ci /Bi ratios. When the governor derives a large benefit from a particular policy or the voter’s cost of indulging the governor along that dimension is low, the threshold is set high, and the sanction for that activity is low. Intuitively, the voter must allocate his scarce p in an economical way, and it is inefficient to deter activities that the governor really likes or that are not particularly costly for the voter. So far there has been no discussion of direct democracy. The key question is what happens to this scheme when direct democracy is introduced. When the initiative and the referendum are introduced, the voters override the governor and make their own policy decisions along some dimensions. This “frees up” some of the p to be used to provide incentives along other dimensions. Formally, we assume that the governor no longer has the power to choose gi for some i. This has the effect of reducing the shadow price of p, that is, reducing q. Comparative statics on the optimization condition (3) give _

dg Bi f ⬘ ᎏi ⫽ ᎏᎏ ⬎ 0, RCi k⬙ ⫺ qBi f ⬙ dq which implies that d
i /dq ⬍ 0. Thus, when direct democracy is introduced, the voters lower the acceptable thresholds for the issues they do not decide directly or, put differently, they impose larger penalties on the governor when he ignores their wishes. The main empirical implication here is that voters will punish governors more for their transgressions in initiative than noninitiative states. Because stronger incentives are available in initiative than noninitiative states, governors in initiative states will be better agents than governors in noninitiative states. It is often argued that direct democracy will lead to better policy outcomes because voters can override their representatives, or the threat to override will itself bring about better policy choices from the electorate’s perspective. The analysis here implies that there is another more subtle (and I believe previously unrecognized) benefit of direct

128

JOHN G. MATSUSAKA

democracy, that it improves the performance of elected officials on tasks that are not subject to initiatives or initiative threats. The set of issues under consideration could include purely administrative functions that are unreachable by initiatives. The analysis implies that direct democracy could bring about better performance on those tasks by allowing voters to concentrate votes there. Besley and Coate (2003) develop a related model in which direct democracy makes the electorate better off by unbundling issues. Their approach is different in that elections do not provide incentives but are used to select a candidate with the best preferences. Initiatives help by simplifying party competition and increasing the probability that a candidate with the “right” preferences is elected. In contrast to the model developed here, Besley and Coate assume that candidates faithfully implement their promised platforms although there is no mechanism forcing them to do so. Because elections are essentially forward looking in the analysis, their model does not generate empirical implications for how voters would respond to actions of the incumbent while in office. B. Evidence from Gubernatorial Elections Citizens can sanction representatives who do a poor job by denying them votes when they stand for reelection. The model in the previous subsection shows that the threat of being ousted may not be severe enough to guarantee the incumbent’s full compliance with the voter’s wishes. In such a situation, the voter is better off saving the strongest sanctions for a subset of issues he considers most important or easiest to control. In direct democracy states, where some issues are taken out of the hands of elected officials, voters are able to apply stronger sanctions on the remaining issues when the incumbent’s performance falls short. To get a sense of the empirical relevance of the model, this section reports evidence on the strength of electoral sanctions from gubernatorial elections over the period 1950–1988. The main question is whether voters reward and sanction incumbents more when direct democracy is available. The basic empirical setup follows Peltzman (1992).12 Voters are assumed to reward or punish the candidate representing the incumbent’s party on the basis of the performance of the incumbent. Formally, I estimate the following model: (4)

Vit ⫽ a ⫹ b ⭈ Vi,t ⫺ 1 ⫹ c ⭈ ECONit ⫹ d ⭈ SPENDINGit ⫹ e ⭈ INCUMBENTit ⫹ uit ,

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

129

where i indexes a state and t indexes an election year, V is the share of votes received by the candidate of the incumbent party, ECON is an index number describing economic performance during the incumbent’s term (the index is roughly income growth minus inflation), SPENDING is annual spending growth during the incumbent’s term, INCUMBENT is a dummy variable equal to 1 if the incumbent is standing for reelection, a, b, c, d, and e are coefficients to be estimated, and u is an error term.13 Peltzman (1992) documented that voters rewarded strong economic performance (c ⬎ 0) and penalized government spending growth (d ⬍ 0) during the period under investigation. The question here is how the size of the rewards and punishments differ when the initiative is and is not available. Specifically, do voters in initiative states reward good performance and punish bad performance more than voters in noninitiative states?14 Table 7.3 reports the estimates. The results are divided into three panels that differ in the time horizon that voters are assumed to use when assessing the incumbent. Panel A assumes that voters have short memories and only look at the past year, panel B assumes they consider the preceding two years, and panel C assumes they consider all four years of the incumbent’s term. Within each panel, the first two columns present the coefficient estimates and standard errors for regression (4) estimated separately for noninitiative and initiative states, respectively. The last column reports the F-statistic for the hypothesis that the coefficients are the same. Panel A, which assumes that voters evaluate information only from the year preceding the election, shows that voters reward economic performance in both initiative and noninitiative states. Both coefficients on economic performance are positive and statistically different from zero at the 5 percent level. The coefficient in noninitiative states is smaller than the coefficient in initiative states, but they cannot be distinguished from each other with statistical confidence. It seems voters apply roughly the same rewards for economic performance in initiative and noninitiative states. This is not inconsistent with the model because there are policy dimensions where sanctions will not change materially when the initiative is introduced. Peltzman (1992) found that voters punished incumbents for spending growth in elections for president, senator, and governor during the period 1950–1988. Apparently, spending was disliked by the marginal voter during this time. Abundant survey evidence also indicates that citizens favored reductions in government spending during this period (Matsusaka 2004). Panel A shows spending growth was punished more in initiative states than in noninitiative states. In initiative states, each percentage point of annual spending growth cost the incumbent party’s candidate 0.348 percent vote

130

JOHN G. MATSUSAKA

Table 7.3 Regressions of incumbent party vote share on economic performance and government spending for gubernatorial elections, 1950–1988 Variable A. Assessment period: Preceding one year Constant Vote share in last election

Noninitiative states 25.0*** (6.7) 0.4*** (0.1)

Economic performance during assessment period (HAPI)

107.8** (38.0)

Annual state spending growth during assessment period

⫺3.8 (10.0)

Initiative states 57.5*** (7.0) ⫺0.1 (0.1) 116.3*** (34.2) ⫺34.8** (11.6)

Dummy ⫽ 1 if incumbent

4.8*** (1.2)

3.8*** (1.3)

B. Assessment period: Preceding two years Constant

23.5*** (6.6)

57.2*** (7.1)

Vote share in last election

0.5*** (0.1)

Economic performance during assessment period (HAPI)

185.3*** (60.7)

Annual state spending growth during assessment period

6.7 (12.3)

Dummy ⫽ 1 if incumbent C. Assessment period: Preceding four years Constant Vote share in last election Economic performance during assessment period (HAPI) Annual state spending growth during assessment period Dummy ⫽ 1 if incumbent

⫺0.1 (0.1) 151.4*** (56.6) ⫺40.1*** (14.9)

F-statistic 11.3*** 10.4*** 0.0 4.11** 0.4 12.6*** 10.9*** 0.2 5.9**

5.0*** (1.2)

4.0*** (1.3)

25.4*** (6.8)

57.4*** (7.0)

10.7***

0.4*** (0.1)

-0.1 (0.1)

10.2***

285.7*** (89.2)

289.6*** (83.6)

⫺8.5 (14.4) 4.7*** (1.2)

⫺48.4** (20.7) 3.9*** (1.3)

0.3

0.0 2.51 0.2

Note. The panels differ in how far back voters look at economic performance and government spending when assessing the incumbent. The main entries are coefficient estimates multiplied (standard errors in parentheses) from a regression in which the dependent variable is the vote share of the candidate representing the incumbent party. Economic performance is the HAPI index, y – i, where y is income growth in excess of its permanent component and i is inflation in excess of its expected value (see Peltzman (1992) for details). Spending growth is as a fraction, not a percentage. The F-statistic is for the hypothesis that the coefficients are the same. Significance levels are indicated as follows: ** ⫽ 5%, *** ⫽ 1%. Source: The data are those used in Peltzman (1992) with the addition of state initiative status.

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

131

share, a number that is significantly different from zero. The effect in noninitiative states, 0.038 percent, is much smaller and not distinguishable from zero. The F-statistic confirms that the effect is larger in initiative than noninitiative states. For this policy it appears that voters employed stronger sanctions in initiative than noninitiative states, as the model predicts. A similar pattern emerges when the estimates allow the voters to incorporate more information. In panel B, votes are conditioned on information over the two years preceding the election. Again we see that voters reward economic performance but the effects cannot be statistically distinguished between initiative and noninitiative states. Voters punish spending growth only in initiative states. Observe that the coefficients on the performance variables are larger in panel B than panel A. This indicates that voters incorporate information from both years, that is, they do not have extremely short-term memory (Peltzman 1990). Panel C incorporates information from the entire four years of the incumbent’s term. There is an argument that not all of this information is relevant since some of what happens in the first year of office is outside the incumbent’s control. This is especially true when it comes to the first-year budget, much of which may have been approved by the previous administration. In any case, the pattern is essentially unchanged. Voters in initiative states continue to punish spending growth more than voters in noninitiative states. The effect is significantly different from zero only in initiative states and the p-value for the difference is 0.11. To put the numbers in perspective, note that the incumbent starts with a 3.9 percent advantage in initiative states (focusing here on panel C). The coefficient on spending of 48.4 implies that spending can grow by about 8 percent per year during the four-year term before the governor dissipates his incumbency advantage. The mean spending growth in the sample was 4 percent with a standard deviation of 3.7 percent, so it is clear that governors can engage in significant spending growth before seriously endangering their chances of reelection. This is consistent with the idea that sanctions are not strongly linked to a single dimension but are spread across multiple issues. The evidence is generally consistent with the idea that voters in initiative states can deliver stronger incentives to their governors than voters in noninitiative states by taking some issues out of the hands of the governor. If voters use an initiative to legalize capital punishment, they do not have to use their votes to prod the governor in the right direction on that policy; they can save their ire for other issues, spending growth apparently being one of them. The stronger incentives in initiative states appear to be effective: during the sample period, state spending was about 12 percent lower in initiative states than in noninitiative states (Matsusaka 1995).

132

JOHN G. MATSUSAKA

A possible alternative explanation for the findings is that voters in initiative states are simply more fiscally conservative to begin with than voters in noninitiative states. If true, the regressions would be revealing not so much a different incentive scheme in initiative and noninitiative states but different voter preferences over policy. This possibility, while plausible, seems unlikely in light of a large amount of opinion data collected in Matsusaka (2004, chapter 3). There I compare the “ideology” of voters in initiative and noninitiative states using six different metrics over roughly the same time period, and find only trivial differences. 5. Discussion The initiative and the referendum are legislative processes, and most research has approached direct democracy from a legislative perspective. Little attention has been paid to the effect of the direct democracy on the executive branch, and very little empirical evidence is available. This chapter attempts to describe the lay of the land, touching on a variety of topics with an eye toward identifying the main issues and empirical relations. The cost is that it does not drill extremely deep on any particular part of the problem. Nevertheless, I believe a number of interesting issues have emerged. I hope there will be enough here to suggest that further investigation of how the initiative affects the executive branch is worthwhile. Two finding are particularly noteworthy. First, a variety of populist and progressive-style changes to the powers and functions of the executive branch seem to follow in the wake of the constitutional initiative. Voters use the process to bring about term limits, require more offices to be filled by election, and require open cabinet meetings. Governors in initiative states are also more likely to have a full line item veto than governors in initiative states, but this seems to be coincident with the initiative process, not caused by the initiative. A second observation is that, by taking some issues out of the hands of elected officials and deciding them directly, the initiative seems to enable voters to more effectively sanction their representatives for poor performance on the issues that remain under their control. Some evidence from gubernatorial elections is reported, showing that governors were punished more intensely for spending growth by voters in initiative than noninitiative states. The ability to bring to bear stronger punishments means that the performance of elected officials will be more satisfactory to the voters. Thus, the evidence (again, only suggestive at this point) has the interesting implication that direct democracy improves the functioning of elected officials in their role as agents. The analysis is couched in terms of the governor, but the logic

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

133

is general enough to apply to all elected officials, including the legislature. It has long been argued by direct democracy advocates that the initiative is desirable because it allows voters to correct policy mistakes by their representatives. The evidence here suggests that it might have a second, and previously unrecognized, benefit of improving the incentives of elected officials to act as faithful agents for the issues that remain under their control. The point extends to purely administrative as well as legislative functions, that is, the initiative may improve the performance of the executive in his purely executive functions. Russell J. Dalton’s chapter in this volume finds that initiative states fare worse than noninitiative states in terms of several “good government” indexes. At first glance, this appears to contradict the prediction that initiatives lead better governance. A possible explanation for the discrepancy is that the indexes may be a measure of what good government is in the eyes of the scholars and activists that construct the indexes, and not what good government means to the voters. For example, the educational performance measure gives higher scores to states that spend more on higher education and provide greater subsidies to low-income students, which may not be how voters define good government. The tax performance measure gives higher scores to states with more progressive tax structures and states that allow exemptions that are means tested, again not necessarily the tax structure that voters favor. The government performance index, constructed by public administration scholars and officials, tends to reward states that give government administrators a large amount of flexibility. Initiatives that tie the hands of government officials tend to reduce a state’s score, but voters may support such constraints because they do not trust government officials. In short, Dalton’s evidence shows that direct democracy changes the way government works, but whether this is for better or for worse is an open question. Caroline Tolbert’s chapter in this volume (as well as Smith and Tolbert 2004), more generally) shows that ballot propositions increase civic engagement and the degree to which citizens are informed. More informed voters will be better judges of their representatives, and so this effect would tend to reinforce the performance-improving feature of direct democracy that I have emphasized. The intuition behind the model also suggests that sanctions will become less effective as the scope and complexity of government increases. Since government has expanded over the last 100 years, an implication is that it should be becoming increasingly difficult to control representatives with elections—there are simply too many issues to send a clear message on election day. This could be one of the factors fueling the growing demand for direct democracy worldwide.

134

JOHN G. MATSUSAKA

Notes 1. For example, the appendices of Oberholtzer (1912), Key and Crouch (1939), Magleby (1984), Cronin (1989), Bowler, Donovan, and Tolbert (1998), Gerber (1999), and Matsusaka (2004)—a sample that stretches from the Progressive period to the present—refer to a total of 20 pages for “governor” or “executive” compared with 177 pages for “legislature” or related terms. 2. If it is costly to propose an initiative, the legislature can preempt the initiative threat without moving all the way to the voter’s ideal point, that is, by choosing some x ⬍ V. 3. The finding that availability of the initiative helps the governor if he has the power to propose initiatives is robust to incomplete information about policy effects and the preferences of the voter. See Gerber and Lupia (1995) and Matsusaka and McCarty (2001), respectively. However, with incomplete information, the initiative can make the voter worse off. 4. For example, see the evidence on the “Law of 1/n” in Gilligan and Matsusaka (1995, 2001), Bradbury and Crain (2001), and Baqir (2002). 5. Institutional data were drawn from the following tables in The Book of the States (The Council of State Governments 2004). Term limits: any limit on the number of terms or consecutive terms, taken from table 4.1. Number of officials elected from 11 top offices—governor, lieutenant governor, secretary of state, attorney general, treasurer, auditor, comptroller, agriculture, education, labor, and insurance (table 4.9). Open cabinet meetings required (table 4.6). Item veto available on all bills (table 4.4). 6. The 17 constitutional initiative states are Arizona, Arkansas, California, Colorado, Florida, Massachusetts, Michigan, Mississippi, Missouri, Montana, Nebraska, Nevada, North Dakota, Ohio, Oklahoma, Oregon, and South Dakota. Illinois was counted as not having a constitutional initiative because its constitutional initiative cannot be used to modify provisions pertaining to the executive. Alaska, Idaho, Maine, Utah, Washington, and Wyoming allow statutory but not constitutional initiatives. I estimated the effects including statutory initiatives (not reported) and the results typically are statistically insignificant. Classifications were taken from Matsusaka (2004, appendix A1.1). 7. Washington’s measure was struck down by the state Supreme Court in 2002 on the grounds that statutory initiatives cannot be used to alter structures set down in the constitution. 8. Data are taken from table 4.11 in The Book of the States (The Council of State Governments 2004). 9. Alternatively, it can be assumed that the governor cares about his vote share not just winning (Stigler 1972), in which case p is the vote share and R is the value he assigns to each point of support. 10. Nothing hinges on the cost interpretation. The voter’s payoff could be interpreted as benefits that are declining in g. 11. Banks and Sundaram (1998) study an abstract principal-agent model with both moral hazard and adverse selection problems, and provide references to related work. See also Fearon (1999).

DIRECT DEMOCRACY AND THE EXECUTIVE BRANCH

135

12. The data are those used in Peltzman (1992) with the addition of state initiative status. 13. I omitted other control variables that Peltzman used in his study to conserve space: federal spending growth, a party dummy, a “coattails” variable equal to 1 if the incumbent governor and president were from the same party and -1 otherwise, and 31 state/region fixed effects. The first two of these variables invariably fell short of statistical significance, and inclusion of them changes none of the substantive findings. The main features of the data are described in the text and notes to the table. Peltzman made some corrections to the raw spending data to adjust for the Korean War, described in his article. 14. The empirical approach is not entirely consistent with the theoretical model as written. In the model, the incentive scheme is perfectly calibrated so the governor never exceeds the thresholds and therefore the sanctions are never observed in equilibrium. However, sanctions would be observed in equilibrium with slight changes in the model to allow for uncertainty and incomplete information. For example, suppose (1) there is a small exogenous probability that g ⫽ h even if the governor attempts to set the policy at the cutoff level, and (2) voters cannot observe whether the governor deliberately violated the threshold or it was done against his will. This setup might describe a situation in which there are minor bureaucrats involved who, with some probability, disobey orders. In this extended model, the optimal punishment schemes would be the same as before, but the governor would occasionally be sanctioned by the voters when he loses control of the policy.

This page intentionally left blank

8

Direct Democracy and the Stability of State Policy Amihai Glazer and Anthony McGann

overnments change policies. One reason is that the preferences of voters or of leaders change. But policy may also change because democratic institutions make for instability: even with unchanged preferences, different winning coalitions will form, causing instability of policy. Intuitively, it would seem that the initiative, which lets the electorate bypass their elected representatives, would increase responsiveness to public opinion and thus reduce stability. Some social choice theoretic reasons (to be outlined later in the chapter), predict, however, that the initiatives would increase stability. That is, initiatives may act as “veto points,” disrupting logrolling coalitions, and thus leading to stability of policy. This chapter empirically tests these two competing hypotheses, using expenditure data from U.S. states from 1980 through 2002.

G

How the Initiative May Generate Instability The initiative allows the public to bypass unresponsive politicians. If the initiative increases responsiveness to changes in public opinion, then we would expect its presence to reduce stability of policy. For example, in Italy in the 1970s, divorce and abortion laws were liberalized by substantial majorities in initiative, at a time when these measures would have certainly been filibustered to death by Christian Democratic representatives in the regular parliamentary process (or even kept off the agenda altogether). The initiative may overcome another source of stability—the desire of incumbents to maintain status quo. A special characteristic of an incumbent

138

AMIHAI GLAZER AND ANTHONY MCGANN

is that he won. The conditions that got him elected in the previous elections are therefore also likely to help him in the next election. A change in policy, say one that redistributes income or that changes the major concerns of the public, may therefore lead to a loss of political support. Voters on an initiative need not share the incumbent’s preference for the status quo. The initiative may especially reduce stability when the legislature has features that make it prone to gridlock. Among these factors are supermajoritarian decision rules; rules that allow filibusters; the regulation of behavior by party leaders (Kiewiet and McCubbins 1991); the executive veto and the presence of two legislative houses, which requires agreement among all to change policy (Hammond and Miller 1987); and expectations by the current majority that by maintaining current policy it could induce a future majority to maintain policy (Bernholz 1978). How the Initiative May Generate Stability Some features of politics may, however, lead to instability when policy is set by the legislature. In particular, theoretical work shows that in the presence of multiple issues, an agenda setter can package a set of policies that a majority of legislators will prefer over any given status quo. The generality of the problem was discovered by McKelvey (1976) and named “chaos.” Other important works include Black (1948), Black and Newing (1951), Plott (1967), McKelvey (1976), and Schofield (1978), which show that, except under extremely restrictive conditions, whatever is the status quo, a politician can find another bundle of policies that the majority prefers to the current bundle. Riker (1982) goes so far as to argue that politicians strategically exploiting this instability drive politics. Though Riker exaggerates the degree of instability inherent in majority rule (see Miller 1980, among others), it is clear that change in public opinion is not a necessary condition for policy instability. The problem does not, however, generally arise if the alternatives that voters consider lie along a single dimension. Direct democracy often requires voters to vote on a proposal that would change the status quo on only one issue, and may therefore lead to greater stability of policy than when legislatures alone determine policy: initiatives that force each policy dimension to be treated as a separate decision pull policy to the median voter’s position on each dimension, frustrating the manipulative ambitions of political elites. Thus, Tsebelis (2002) argues that initiatives form an extra veto point and therefore increase policy stability. This result holds, of course, if initiatives and referenda address only one issue. Some jurisdictions, such as California, impose such a constitutional requirement. But the restriction to one issue can arise for additional reasons.

DIRECT DEMOCRACY AND POLICY STABILITY

139

Voters may get confused by a complicated referendum, fearing that what they do not understand will hurt them. So, for a referendum to succeed at the polls, it may need to focus on one issue. Moreover, it may be more difficult for organizers of a referendum to logroll and to adopt a complicated policy than it is for legislators (who do see one another all the time). Furthermore, the preference of a majority of the population may change infrequently, compared with more frequent change in the preferences of a majority of the legislature: in the legislature, the majority can change if the majority of voters in one district changes, while the majority in the state does not. Direct democracy will then generate stability of policy. We note that if different voters or representatives put different weights on different issue dimensions, then the initiative can have strongly countermajoritarian effects (Anscombe 1976; Lacy and Niou 2000): the potential for using the initiative can disrupt the construction of logrolling coalitions. Various groups may be able to trade their support on an issue they care little about for support on the issue that is most important to them; the resulting trades can produce an outcome they all prefer. The ability to use the initiative, however, undermines such deals, as an initiative will pull the outcome back to the position of the median voter on each issue. The initiative then frustrates the ability to construct multidimensional coalitions to change the status quo, and so increases stability. Linder (1998) argues that this cause of stability holds in Switzerland. Indeed, he explains the long (until 1972) Swiss denial of suffrage to women in terms of direct democracy: in other countries, female suffrage could be introduced as part of a broader political deal, whereas in Switzerland it needed to win approval from a majority of (male) voters on its own. We thus have two plausible stories about the effect of direct democracy on political stability, producing opposite predictions. After reviewing literature, we turn to testing these theories. Literature An important implication of the chaos theorems discussed earlier is that some institutions can force political issues into a one-dimensional space (Riker 1982; Shepsle 1979; Shepsle and Weingast 1981). Thus, in the U.S. Congress, germaneness rules reduce the policy space to a single dimension, thereby limiting postcommittee cycling between different policy packages and generating a structure-induced stable equilibrium. Several papers argue that policy will be more stable under proportional representation than under plurality voting (where a small shift in the preferences in a few districts can drastically change the majority in the legislature); see, in particular, Morelli and Tertilt (2000). Houser and Freeman (1988)

140

AMIHAI GLAZER AND ANTHONY MCGANN

indeed find that the variability of labor taxes is higher in the United Kingdom (which has plurality voting) than in Germany (which has proportional voting). Grilli, Masciandaro, and Tabellini (1991) find that the countries with the most stable governments use proportional representation. A different approach to studying the stability of policy considers veto players: individuals or institutions whose agreement is required for a change of the status quo. The basic prediction of the veto-player theory is that when the number of veto players increases and their ideological distances increase, policy stability also increases (Tsebelis 1995, 1999, 2002). Though scholars have studied policy stability and studied government budgets, few papers discuss them jointly, as we do. Bawn (1999) studies specific items in the budget of the Federal Republic of Germany from 1961 to 1989, finding that budget allocations varied with the identity of the parties in the coalition government. Konig and Troger (2001) essentially replicate Bawn’s findings over a longer period and use the estimated preferences of the different parties. Our chapter builds on Tsebelis and Chang (2004). They consider the change in budget allocations in 19 OECD (Organisation for Economic Cooperation and Development) countries in the years 1973–1995, finding support for two hypotheses: that parties located farther from each other in a multidimensional space cannot modify the status quo as significantly as a coalition with less-diversified parties, and that the farther is the status quo from the preferences of the veto players, the greater the possible departure from the status quo. Method Our study will examine budget allocations across different states in the United States. Like Tsebelis and Chang (2004), we view a budget allocation as a vector in an n-dimensional Euclidean issue space. (Unlike Tsebelis and Chang we use the absolute value of spending on each item as a proportion of state personal income, rather than the proportion of state spending allocated to each item.) The vector consists of a set of expenditures allocated to different budget categories (a1, a2, . . . ,an). Each year has a different budget allocation, so the above sequence should be indexed by the time it was selected. The difference between two budgets can be represented by the distance between the points that represent them in the n-dimensional Euclidean space. The dependent variable, the change in the budget allocations, is [(a1,t⫺a1,t⫹1)2 ⫹(a2,t⫺a2,t⫹1)2 ⫹ . . . ]1/2. We also calculated the mean squared deviation of the spending vector, [(a1,t⫺
1)2 ⫹(a2,t⫺
2)2 ⫹ . . . ]1/2, where
i is the mean spending by the

DIRECT DEMOCRACY AND POLICY STABILITY

141

state on policy category i. This deviation summarizes the typical variation of the vector from a central tendency, as opposed to year-by-year variation. For budget allocations we look at the fifty states over the years 1980–2002. The budget items we consider are those listed in the Statistical Abstract of the United States, namely, spending on education, highways, public welfare, and health (including hospitals), and other spending. We divided all values by state personal income in that year and state. We regress our measure of policy instability against variables measuring institutional features of state government as well as other explanatory variables. We use a dummy variable for the presence of a popular initiative (whether statutory or constitutional). We also include several dummy variables to capture the effects of other institutional features that may affect stability; these are drawn from Grooters and Eckl (1998). One institutional feature is a supermajoritarian legislative process, which may increase stability by making it harder to pass legislation. The line-item veto would also be expected to increase stability, by increasing the governor’s veto power and allowing the governor to undermine logrolling deals. Lastly, a biennial budget would be expected to increase stability. Patterns of state spending, however, do not simply result from processes within the states. The states are, after all, part of the United States of America, and their spending is likely to be conditioned by the federal government. We therefore include the per capita intergovernmental income of each state in 2002 as an explanatory variable, and we include state population. We would expect small states to show more policy instability, because small demographic changes may change the median policy preferences. Our independent variables are: Fed. transfers: Per capita intergovernmental revenue of state in 2002 (in thousands of dollars) Population: State population in 2002 Supermajority: Dummy variable equal to 1 (and 0 otherwise) for states with a supermajority requirement for passing the budget (Alabama, Arkansas, California, and Rhode Island, plus Alaska, Connecticut, Hawaii, Maine, Massachusetts, and Nebraska, which have supermajoritarian requirement in some circumstances) Line-item veto: Dummy variable for a state that grants the governor a line-item veto Biennial budget: Dummy variable for a state that adopts a two-year budget Initiative: Dummy variable for a state that allows a popular initiative

142

AMIHAI GLAZER AND ANTHONY MCGANN

Results Table 8.1 lists the results of the regression of Euclidean distance of the annual change in spending. We see that the only highly significant variable is federal transfers: states that receive more federal money show greater volatility in their spending patterns than do states that receive little aid. This result is consistent with the hypothesis that much of the instability we observe arises not from processes within the state, but from national policy. Population is significant at the 7 percent level: small states show higher volatility than do larger states. None of the institutional factors, such as the initiative, has any statistically significant effect. We note, following John G. Matsusaka in this volume, however, that initiative states are more likely to impose term limits on their governors, thereby reducing stability. And initiative states are almost three times as likely to give the governor a line-item veto on all bills. Our results on the initiative, therefore, isolate the effects of the initiative alone and do not, therefore, compare states with the initiative with states without it. But our results do incorporate indirect effects. For example, Daniel A. Smith, in this volume, finds that initiatives can affect turnout; different turnout rates by different groups in different elections can generate political instability. The coefficient on initiative we estimate incorporates any such effect. We can see from the raw data that Alaska is an outlier: it receives far more federal money per capita than other states. Since federal grants are so important, Alaska might be driving the results. We therefore repeat the analysis, but without Alaska. Table 8.2 shows the results.

Table 8.1 Effect on budget allocations, annual change (all states) Dependent variable: Euclidean distance of change in spending

(Intercept) Fed. transfers Population Supermajority Line-item veto Biennial budget Initiative

Estimate

Std. Error

t-value

Pr(⬎|t|)

⫺7.18e⫺4 3.70e⫺3 ⫺8.56e⫺8 ⫺1.49e⫺4 8.16e⫺4 7.63e⫺4 2.84e⫺4

1.29e⫺3 7.91e⫺4 4.54e⫺8 7.10e⫺4 7.91e⫺4 5.80e⫺4 5.73e⫺4

⫺0.56 4.67 ⫺1.88 ⫺0.21 1.03 1.32 0.50

0.58 2.92e⫺5 0.07 0.83 0.31 0.20 0.62

Note: Residual standard error: 0.002 on 43 degrees of freedom; multiple R-squared: 0.44; adjusted R-squared: 0.36; F-statistic: 5.64 on 6 and 43 DF; p-value: 0.0002.

DIRECT DEMOCRACY AND POLICY STABILITY

143

Table 8.2 Effect on budget allocations, annual change (Alaska excluded) Dependent variable: Euclidean distance of change in spending

(Intercept) Fed. transfers PopulationSupermajority Line-item veto Biennial budget Initiative

Estimate

Std. Error

t-value

Pr(⬎|t|)

5.25e⫺4 2.62e⫺3 ⫺7.97e⫺8 ⫺6.76e⫺4 8.48e⫺4 1.00e⫺3 1.17e⫺4

1.26e⫺3 8.11e⫺4 4.18e⫺8 6.76e⫺4 7.27e⫺4 5.38e⫺4 5.29e⫺4

0.42 3.23 ⫺1.91 ⫺1.00 1.17 1.86 0.22

0.68 0.00 0.06 0.32 0.25 0.07 0.83

Note: Residual standard error: 0.002 on 42 degrees of freedom; multiple R-squared: 0.35; adjusted R-squared: 0.26; F-statistic: 3.77 on 6 and 42 DF; p-value: 0.004.

Table 8.3 Effect on budget allocations, mean deviation (all states) Dependent variable: Mean deviation of spending

(Intercept) Fed. transfers Population Supermajority Line-item veto Biennial budget Initiative

Estimate

Std. Error

⫺2.28e⫺4 7.01e⫺3 ⫺9.97e⫺8 ⫺1.41e⫺3 8.01e⫺4 4.33e⫺4 1.21e⫺5

1.51e⫺3 9.27e⫺4 5.33e⫺8 8.33e⫺4 9.27e⫺4 6.80e⫺4 6.71e⫺4

t-value ⫺0.15 7.57 ⫺1.87 1.69 0.86 0.64 0.02

Pr(.|t|) 0.88 1.97e⫺9 0.07 0.10 0.39 0.53 0.99

Note: Residual standard error: 0.002 on 43 degrees of freedom; multiple R-squared: 0.65; adjusted R-squared: 0.60; F-statistic: 13.16 on 6 and 43 DF; p-value: 2.14e-8.

The pattern of results is the same: federal transfers continues to be significant, and small states show more volatility. None of the institutional variables is significant at the 5 percent level, but the data show that, at the 7 percent significance level, states with biennial budgets are more volatile. To check for robustness, we ran the same regression with mean deviation as the dependent variable. The results are similar to those reported above. Here again, federal transfers is the most significant explanatory variable, followed by population (see table 8.3). The supermajority variable is significant at the 10 percent level, but the effect appears to result from the outlying case of Alaska. When Alaska is excluded, the supermajority variable loses significance (see table 8.4).

144

AMIHAI GLAZER AND ANTHONY MCGANN

Table 8.4 Effect on budget allocations, mean deviation (Alaska excluded) Dependent variable: Mean deviation of spending Estimate (Intercept) 2.01e⫺3 Fed. transfers 5.08e⫺3 Population ⫺8.90e⫺8 Supermajority 4.60e⫺4 Line-item veto 8.61e⫺4 Biennial budget 8.63e⫺4 Initiative ⫺2.89e⫺4

Std. Error t-value

Pr(⬎|t|)

1.24e⫺3 1.62 8.01e⫺4 6.34 4.13e⫺8 ⫺2.16 6.68e⫺4 0.69 7.19e⫺4 1.20 5.32e⫺4 1.62 5.23e⫺4 ⫺0.55

0.11 1.30e⫺7 0.04 0.49 0.24 0.11 0.58

Note: Residual standard error: 0.002 on 42 degrees of freedom; multiple R-squared: 0.57; adjusted R-squared: 0.51; F-statistic: 9.45 on 6 and 42 DF; p-value: 1.43e-6.

Thus, we find evidence for neither theory about the effect of the initiative on stability. The initiative appears to little affect policy stability. Instead, policy stability seems to depend far more on intergovernmental transfers from the federal government and, to a lesser degree, on state population. Tax and Spending Levels So far we discussed changes in spending, finding few effects. As a test of robustness, we ask whether our variables can explain any important features of government. We first look at the level of taxes. We are particularly interested to see whether the introduction of federal transfers as a control variable has any effect. Table 8.5 gives the results for all states; table 8.6 gives results with Alaska excluded. Federal transfers little affect tax levels. The initiative and supermajoritarianism show effects of similar size but in opposite directions: the initiative is associated with lower levels of taxation, and supermajoritarianism with higher levels. Both supermajoritarianism and the initiative are significant to the 10 percent level when Alaska is excluded, but supermajoritarianism just fails to make the 10 percent level with the entire sample. We note, however, that the R-squared statistic for both these regressions is low, and the magnitude of the effect of supermajoritarianism and the initiative is modest (about $200 per head). These variables explain little of the variation in tax levels. When we consider the level of state spending, we find supermajoritarianism associated with somewhat greater spending; the initiative is associated with somewhat lower spending. However, federal transfers are by far

DIRECT DEMOCRACY AND POLICY STABILITY

145

Table 8.5 Effect on taxes (all states) Dependent variable: Taxes per capita in 2002

(Intercept) Fed. transfers Population Supermajority Line-item veto Biennial budget Initiative

Estimate Std. Error t-value

Pr(⬎|t|)

1774 78 1.63e⫺3 207 ⫺2.32 101 ⫺201

3.46e⫺9 0.59 0.85 0.12 0.98 0.35 0.07

239 146 0.01 131 147 107 106

7.39 0.54 0.19 1.57 ⫺0.02 0.94 ⫺1.89

Note: Residual standard error: 359.3 on 43 degrees of freedom; multiple R-squared: 0.13; adjusted R-squared: 0.01; F-statistic: 1.11 on 6 and 43 DF; p-value: 0.37.

Table 8.6 Effect on taxes (Alaska excluded) Dependent variable: Taxes per capita in 2002

(Intercept) Fed. transfers Population Supermajority Line-item veto Biennial budget Initiative

Estimate

Std. Error t-value

Pr(⬎|t|)

1.70e⫹3 1.45e⫹2 1.27e⫺3 2.40e⫹2 ⫺4.35e⫹0 8.63e⫹1 ⫺1.91e⫹2

2.55e⫹2 6.66 1.64e⫹2 0.88 8.46e⫺3 0.15 1.37e⫹2 1.75 1.47e⫹2 ⫺0.03 1.09e⫹2 0.79 1.07e⫹2 ⫺1.78

4.41e⫺8 0.38 0.88 0.09 0.98 0.43 0.08

Note: Residual standard error: 360.1 on 42 degrees of freedom; multiple R-squared: 0.15; adjusted R-squared: 0.02; F-statistic: 1.2 on 6 and 42 DF; p-value: 0.32.

the strongest predictor. Tables 8.7 and 8.8 report these results with and without the case of Alaska. With Alaska included, the effect of the initiative is insignificant, whereas supermajoritarianism is significant at the 2 percent level. However, when Alaska is excluded, the initiative is significant at the 10 percent level, whereas supermajoritarianism only makes the 13 percent level. The coefficient on Supermajority is still slightly higher than that of the initiative (supermajoritarianism is associated with an increase in spending per head of $390, whereas the initiative is associated with a decrease of $340). Both effects, however, are much smaller than the effect of federal transfers. Increasing transfers by $1,000 increases spending by $1,450. It is notable that this effect is greater than one-to-one. That is, federal transfers lead states to spend more money from other sources. This effect may result from federal matching funds, which reduce the marginal cost of state spending.

146

AMIHAI GLAZER AND ANTHONY MCGANN

Table 8.7 Effect on total spending (all states) Dependent variable: Spending per capita, 2002

(Intercept) Fed. transfers Population Supermajority Line-item veto Biennial budget Initiative

Estimate

Std. Error

t-value Pr(⬎|t|)

1565 2317 0 816 191 ⫺75 ⫺205

622 318 0 342 381 279 276

2.52 0.01 6.09 2.71e⫺7 ⫺0.11 0.92 2.39 0.02* 0.50 0.62 ⫺0.27 ⫺0.79 ⫺0.74 0.46

Note: Residual standard error: 931.1 on 43 degrees of freedom; multiple R-squared: 0.54; adjusted R-squared: 0.47; F-statistic: 8.29 on 6 and 43 DF; p-value: 5.44e-6.

Table 8.8 Effect on total spending (Alaska excluded) Dependent variable: Spending per capita, 2002

(Intercept) Fed. transfers Population Supermajority Line-item veto Biennial budget Initiative

Estimate

Std. Error

2.57e⫹3 1.45e⫹3 2.47e⫺3 3.90e⫹2 2.17e⫹2 1.18e⫹2 ⫺3.40e⫹2

4.73e⫹2 3.05e⫹2 1.57e⫺2 2.55e⫹2 2.74e⫹2 2.03e⫹2 1.99e⫹2

t-value 5.43 4.74 0.16 1.53 0.79 0.58 ⫺1.71

Pr(⬎|t|) 2.61e⫺6 2.45e⫺5 0.88 0.13 0.43 0.56 0.10

Note: Residual standard error: 669.4 on 42 degrees of freedom; multiple R-squared: 0.41; adjusted R-squared: 0.32; F-statistic: 4.81 on 6 and 42 DF; p-value: 0.0008.

It is interesting that states that require a supermajority to pass a budget have higher taxes and spending. The argument often made for a supermajority rule is that it makes it harder for legislators to raise taxes and spending. At least when it is applied to the entire budget, this effect is absent. The absence is unsurprising. If it is necessary to get a two-thirds vote to pass a budget, a broader logroll is required than with a simple majority vote. A supermajority rule empowers veto players who hold out for concessions, which may often come in the form of costly side-payment projects. Conclusion We considered two alternative theories concerning the effects of the initiative process on policy stability, as measured by state spending patterns. The first theory suggests that states with the initiative will be less stable

DIRECT DEMOCRACY AND POLICY STABILITY

147

than states without, because the initiative makes policy more responsive to public opinion, and thus changes in public opinion get directly translated into changes in policy, bypassing (possibly obstructionist) party elites. The other theory suggests that states with the initiative will show more stability of policy. The initiative works as a form of veto point, forcing policy to the position of the median voter on each dimension, and preventing the construction of logrolling coalitions that can challenge the status quo. We find support for neither of these theories. States with the initiative do not seem to have either more or less policy stability than states without it. Interestingly, other institutional features, such as a supermajoritarian budget process and the line-item veto, also appear to have no effect on policy stability. The variable with the strongest explanatory power is the per capita transfer from the federal government. The importance of federal transfers indicates that state governments are not sovereign nations; their behavior cannot be explained only by factors internal to the state. This finding may generate skepticism about using the states as a natural experiment to assess the effect of institutional variables. Many programs are mandated at the national level. It appears that the effect of the federal government on policy stability is far greater than that of state-level institutions. We do, however, find that state-level institutions do affect the aggregate level of taxes and spending. States with the initiative have lower taxes and spending (replicating Matsusaka 2004); a supermajoritarian budget process leads to higher taxes and spending. These effects, however, are modest, amounting to a couple of hundred dollars per head per year. We find that the strongest predictor of state expenditure is once again the level of federal transfers.

This page intentionally left blank

9

Direct Democracy and Good Governance: Does It Matter? Russell J. Dalton

uring the last third of the twentieth century, the American political system underwent a process of democratic expansion and institutional reform that rivaled the populist era reforms early in the century (Cain, Dalton, and Scarrow 2003). This democratic transformation included reforms of the electoral system: direct primaries for presidential elections, term limits, and campaign finance reforms. Citizens gained new access to government through the introduction of Freedom of Information guarantees, “government in the sunshine” provisions, and the introduction of public hearing requirements. Access to the courts also expanded for public interest groups and individual rights complaints. The foremost example of this democratic transformation, however, is the expanded use of initiatives, referendums, and other instruments of direct democracy. As others have noted in this book, the use of initiatives at the state level expanded tremendously. The Initiative and Referendum Institute calculates that there were 118 statewide initiatives in the U.S. during the 1950s; this figure increased to 378 initiatives in the 1990s. A similar trend apparently occurred at the local level, as citizens and interest groups increasingly turned to direct democracy to advocate their positions. This development created a new “direct democracy debate.” Does the increased use of direct democracy methods improve or worsen democratic governing? Much like the populist reformers who introduced direct democracy early in the twentieth century, the contemporary advocates of direct democracy cite initiatives and referendums as a potential check on government, and a mechanism for the expression of the public’s collective preferences (Bowler and Glazer in this volume; Matsusaka 2005a; Braunstein 2004;

D

150

RUSSELL J. DALTON

Bowler and Donovan 1998; Cronin 1989). In contrast, the critics claim that direct democracy actually erodes democratic governance (Gerber et al. 2001; Broder 2000). Direct democracy is weak on deliberation and adjustment, such as occurs as a piece of legislation normally proceeds through hearings, legislative debate, and amendments before reaching a final vote; and normal legislation is more easily modified. Critics also claim that well-funded special interests can more easily capture direct democracy processes, producing a tyranny by the minority. Typically, the debate is based on case studies or anecdotal evidence. Critics select examples of initiatives they view as problematic, and generalize from these examples. This chapter takes a different approach.1 We are interested in whether direct democracy systematically influences the performance of democratic governance. We can readily accept that some initiatives have improved state governance, while others have created problems. (We can make similar stipulations about the laws enacted by state legislatures.) However, the fundamental question is whether a process of direct democracy systematically affects the governance of the American states. To address this question, we have assembled three sets of “good governance” measures recently compiled for the 50 U.S. states. One battery focuses on the management of state governments; a second battery assesses state education policies; and a third battery measures tax policy performance. We chose these three batteries because they span central aspects of state government—from policy inputs to the governing process, to a central state policy output (education). These were also collected to measure the quality of government performance, which is our primary research focus. Our analyses compare the usage of direct democracy across the 50 states with these performance measures to determine whether direct democracy systematically affects government performance. The chapter proceeds in four steps. First, we present the contrasting positions on the “direct democracy debate” and discuss the rival claims for how direct democracy affects governance. Second, we introduce our three areas of state government performance and correlate initiative usage with these performance measures. Third, we develop multivariate models to control for other influences on performance and thus better isolate the impact of initiative usage. Finally, we discuss our findings for the workings of the democratic process in American state governments. The Direct Democracy Debate Because many of the previous chapters have already discussed the theoretical debate about the potential impact of initiatives on the political process, we will only summarize some of the key points as they relate to

DIRECT DEMOCRACY AND GOOD GOVERNANCE

151

the governance themes in this chapter (also see Lupia and Matsusaka 2004; Bowler and Donovan 1998). The populist argument in support of initiatives utilizes variations of principal-agent theory (Matsusaka 2005a; Gerber 1996). If the goal of democracy is to represent the will of the governed (the principals), then elected representatives (the agents) should follow public preferences. In its mild form, this perspective argues that agents are inevitably influenced by other factors—such as interest groups, party pressures, and their own personal ambitions—that can erode the representativeness of their decisions. Elected officials might simply be unsure what the public wants, or be uncertain how to balance various policy demands. Thus, the initiative process provides a means for the public to more clearly translate their preferences into public policy, or at least act as a corrective to the normal channels of representative democracy. A stronger version of the populist argument asks whether elites develop their own interests, separate from the public at large, and systematically distort the democratic process (Ellis 2002). The distortion might take the form of enacting policies that increase their likelihood of reelection rather than representing the public at large. Politicians who cater to their base, rather than the median voter, provide an example of this pattern. Serious distortions occur when elites cater to powerful interest groups and lobbies that might benefit them personally. Indeed, much of the history of populist reform in American is interwoven with concern about the corruption or capture of elites by special interests. Thus, direct democracy’s main benefit is to ensure the accountability of government and its openness to pubic preferences (Cronin 1989). This perspective offers a Jeffersonian view that democratic governance benefits from more democracy. Critics have challenged these populist claims at several levels. One strand of the literature questions the ability of average citizens to make decisions on complex policy matters (Broder 1998; cf. Bowler and Donovan 1998; Lupia and Johnston 2001). There is abundant evidence on the limited political knowledge and sophistication of the public, and researchers question the bases of voting choice even for high-profile electoral decisions such as presidential voting. How then can these same voters make decisions about environmental policy, stem cell research, solar power, health policy, and the wide variety of issues that make their way into initiatives? Without sophisticated knowledge, critics claim, initiatives open the policy process to demagoguery and the manipulation of public opinion. A second critique questions the process of policy making by initiative. While the initiative process was initially a populist tool, it has opened a new path to influence for interest groups. Since modern political campaigns typically require a funding base to support legal drafting, signature

152

RUSSELL J. DALTON

gathering, and the initiative campaign, this benefits well-funded groups that can support an initiative (Magelby 1984; Gerber 1999; Bali 2003). Moreover, this process circumvents the normal legislative process and may even enhance the power of special interests. There are numerous examples of the power of money in stimulating initiatives, ranging from Native Americans proposing initiatives on casino gambling to industries proposing changes in government regulations of their business. In short, while initiatives were created as a populist tool, critics claim that they have morphed into a device that benefits well-funded special interests rather than the public at large. In addition, as shortcuts to policy making, the initiative process may not produce well-crafted legislation. Many initiatives are drafted by groups with limited legal or legislative experience, which shows in the formal legal quality of initiatives. For instance, about half of the initiatives passed in California, Oregon, Washington, and Colorado between 1960 and 1999 faced legal challenges in the courts, and many had significant portions of their content invalidated by the courts (Ellis 2002, chapter 6). Moreover, initiatives do not involve the extended deliberation and debate that accompany the formulation of normal legislative proposals. Thus, the critics of initiatives maintain that normal legislative procedures generally produce better public policy. Finally, another critique is even more directly related to the general quality of governance that is the focus of this chapter. Many analysts claim that the initiative process distorts and circumscribes the normal processes of good governance. Special interests can use initiatives to earmark policies that benefit themselves over the collective interests. By often enshrining initiative decisions in the state constitution or the requirement of future initiatives to approve any changes, the government’s ability to adapt to changing social conditions becomes limited. For instance, California’s infamous Proposition 13 not only rolled back tax rates but mandated that a two-thirds supermajority was required for future tax increases. Similarly, analysts frequently cite California’s Proposition 98 as an example of the rigidity of the initiative process because it requires that at least 40 percent of the state budget be devoted to K-14 education spending; it also mandated a two-thirds majority of the legislature is required to temporarily waive its provisions. Thus, the critics of direct democracy may accept its democratic ideals but argue that the poor policies formulated by the initiative process, the potential exploitation of the process by special interests, and the distortions of interest aggregation severely limit the governability of states with an active initiative process. Even well-intentioned and well-skilled public officials cannot produce good governance if they are constrained in these ways. This debate between the proponents and critics of direct democracy leads to our efforts to test the impact of initiative use on governance.

DIRECT DEMOCRACY AND GOOD GOVERNANCE

153

The Impact of Direct Democracy Is direct democracy a boon or a curse for the democratic process? The most difficult step in this analysis is the measurement of government performance. Previous direct democracy studies have focused on specific policies, or looked at policy outputs such as state budgets or spending categories (e.g., Matsusaka 2004; Bowler and Donovan 2004; Gerber 1999; Camobreco 1998). However, we sought more fundamental measures of performance that encapsulate the idea of whether state governments are doing a good job. Government Management The public administration initiative of performance-based management has measured the performance of public agencies. We draw upon this research to tap one of the most basic dimensions of government performance— management of the public sector. The Government Performance Project (GPP) at the Maxwell School, Syracuse University (GPP 2002), was a six-year research initiative evaluating the management capacity of federal, state, and local government entities. It began with the premise that good management forms the basis of good government performance: The GPP examines capacity by analyzing whether government can hire the right talent when it is needed, whether it has and can use the right information at the right time, and whether the systems that support both the visions and the strategies of leaders are present. Without them in place, high performance is not likely to occur. With them in place, both performance and its effective measurement become more likely. (Ingraham, cited in GPP 2002, 13)

This concept of performance seems to lie at the heart of the direct democracy debate, by asking about the overall quality of the governing process in a state: does direct democracy improve or limit the ability of state governments to manage the process of governance? The project focused on five key systems of government management: 2 ●

●

Financial management evaluates state governments in three broad areas: (1) use of a multiyear perspective on budgeting, (2) state fiscal health and stability, and (3) financial control and financial reporting. Capital management assesses three areas: (1) capital planning, budgeting, and the analysis of future needs, (2) evaluating projects throughout their implementation, and (3) conducting appropriate maintenance of capital assets.

154 ●

●

●

RUSSELL J. DALTON

Human resources management measures the effective and efficient use of human capital to accomplish a government’s goals in five areas: (1) workforce and strategic planning, (2) timely and quality hiring, (3) sophisticated professional development programs, (4) meaningful reward and evaluation structures, and (5) disciplinary procedures. Information technology (IT) management evaluates several areas: (1) an effective and coherent information technology system and planning for future development, (2) adequate IT training, (3) evaluating the value of IT investments, and (4) the use of technology to communicate with citizens. Managing for results measures whether state governments manage for performance through the creation and distribution of performance information through strategic planning and performance measurement routines.

Using surveys, document analysis, and interviews, researchers evaluated each state government. The GPP did not rank states purely on outcomes— we are not measuring how much is spent or whether a policy was enacted. Rather, it focuses on the quality of performance: how well governments manage available resources and how well they deliver services. Experts rated the states in the subcategories of each performance area, and these were summed into an overall score for each of the five domains. The GPP published two separate surveys of state governments, in 1999 and 2002. As a first step, we examined the overall state rankings averaged across the five domains. Table 9.1 presents the state scores in a familiar A to F grading scheme, with 4.0 equaling an A grade and 0.0 an F grade (plus grades are
.25, and minus grades are .25). Our first impression is that state management performance does not follow a simple pattern dictated by affluence, region, or other typical predictors. Alabama ranks fiftieth in performance, but two other relatively poor southern states (Kentucky and South Carolina) rank in the top twelve. In addition, two of the largest and more affluent states, California and New York, rank in the bottom dozen. Any single measurement of government performance has advantages and disadvantages. These measures aim at the broad level of public administration, based on the logic that the management of government resources and processes validly measures good governance. Certainly, good governance involves more than just management—including the other measures studied in this chapter, and the effects examined in other chapters of this book. However, we agree with the GPP that the management of government is an important indicator of good governance. The five separate dimensions are also strongly interrelated, echoing the evidence from Italy

DIRECT DEMOCRACY AND GOOD GOVERNANCE

155

Table 9.1 State rankings in average governmental performance State

2001

1999

Overall

Utah Washington Michigan Missouri Virginia Delaware Iowa Kentucky Maryland Pennsylvania South Carolina Minnesota Ohio Texas Illinois Kansas North Carolina Nebraska North Dakota Wisconsin Louisiana New Jersey Tennessee Vermont Florida

3.75 3.75 3.75 3.25 3.25 3.25 3.25 3.25 3.25 3.25 3.25 3 3 3 3 3 3 2.75 2.75 2.75 2.75 2.75 2.75 2.75 2.75

3.75 3.75 3.25 3.75 3.75 3 3 3 3 3 3 3 3 3 2.75 2.75 2.75 3 3 3 2.75 2.75 2.75 2.75 2.25

3.75 3.75 3.5 3.5 3.5 3.12 3.12 3.12 3.12 3.12 3.12 3 3 3 2.87 2.87 2.87 2.87 2.87 2.87 2.75 2.75 2.75 2.75 2.5

State

2001

1999

Overall

Georgia Indiana Massachusetts Montana Oregon South Dakota Idaho Maine Colorado Mississippi Arizona Nevada New Hampshire West Virginia California New Mexico New York Alaska Oklahoma Wyoming Arkansas Connecticut Hawaii Rhode Island Alabama

2.75 2.75 2.25 2.25 2.25 2.25 2.75 2.75 2.25 2.25 2.25 2 2 2 2.25 2.25 2.25 2 2 2 2 2 2 2 1.75

2.25 2.25 2.75 2.75 2.75 2.75 2 2 2.25 2.25 2 2.25 2.25 2.25 1.75 1.75 1.75 2 2 2 1.75 1.75 1.75 1.75 1

2.5 2.5 2.5 2.5 2.5 2.5 2.37 2.37 2.25 2.25 2.12 2.12 2.12 2.12 2 2 2 2 2 2 1.87 1.87 1.87 1.87 1.37

Note: Table entries are average state scores on the five measures of performance. The scale ranges from A  4 .00, to F  0.0. Source: GPP 2002.

given by Putnam, Leonardi, and Nanetti (1993) that good government extends across policy domains.3 We also expect that most citizens prefer that their government score highly on each of these five dimensions. Thus, the question is whether the usage of direct democracy encourages state government to higher levels of performance because of the potential for public monitoring. Or, does the intervention of initiatives erode the management of state government by restricting governments and distorting policy processes? Indeed, this provides a fairly direct test of the direct democracy debate. To determine whether government performance is systematically related to direct democracy, we measured the usage of initiatives. We counted the number of popular initiatives that were approved in statewide elections from 1960 to 1998, the year before the first GPP survey (we label this variable “initiatives passed”).4 This initiative count

156

RUSSELL J. DALTON

Table 9.2 Correlation between initiative usage and measures of government performance Year

Fiscal Management

Human resources

Information technology

Capital Management

Managing for results

2001 1999

.15 .14

.14 .06

03 .06

.01 .11

.09 .01

Overall index .16 .02

Note: Table entries are the Pearson correlations between state government performance measures and the number of popular initiatives that were passed between 1960 and 1998. Source: GPP 2002 and Initiative and Referendum Institute.

ranges from zero in twenty-seven states that lack provisions for initiatives (including one initiative state where none passed) to fifty-four approved initiatives ballot measures in California. The frequency of initiatives passed is a more robust measure of the use of direct democracy than the simple dichotomy of initiative/noninitiative states that some other analyses have used. We correlated the five GPP measures, and the overall summary score, with the frequency of initiative passed. Table 9.2 demonstrates only a weak relationship between these measures. In the 2001 survey, for example, initiatives passed displays a modest negative relationship with fiscal management (r  .15) and human resource management (r  .14), but neither is statistically significant. The overall summary measure sums up these weak effects to produce a larger correlation (r  .16) in 2001, but in 1999 there is virtually no relationship (r  .02).5 A better impression of the overall relationship can be garnered from figure 9.1, which charts the relationship between initiatives passed and summary scores on the 2001 GPP measure. There is a weak negative relationship, but of a limited magnitude. For instance, the difference between the states with no initiatives and the highest outlier, California, with fifty-four initiatives, is approximately the difference between a Bgrade for the former and a C
grade for the latter. Moreover, this relationship is partially contingent on the score of California, because it is the outlier on the initiative dimension and thus has disproportionate impact on the regression line. Excluding California produces a small positive relationship for 1999 (r  .09) and a negative correlation for 2001 (r .12). In summary, our initial findings do not suggest that the use of initiatives significantly affects the management of state governments. One can cite specific initiatives that have impeded (or improved) the effectiveness of government. However, one can also see laws emanating from the state legislatures that improve or erode the management of state government.

DIRECT DEMOCRACY AND GOOD GOVERNANCE

157

4

Government Management 2001

r = −.156

3

2

1 0

10

20

30 40 Initiatives Passed

50

60

Note: Figure entries plot the state government performance average for 2001 and the count of initiatives passed in each state between 1960 and 1998. Sources: GPP 2002 and Initiative and Referendum Institute.

Figure 9.1 The relationship between initiatives passed and government management

Neither side of the direct democracy debate can find strong support for their position in the general patterns of state governance. Education as a Policy Output In addition to the management of government, we wanted to compare state policy performance as another indicator of good governance. We focused on postsecondary educational policy, partially because most states play a dominant role in setting educational priorities and in funding postsecondary educational spending. This is one policy domain where the role of the federal government remains secondary. In addition, a project by National Center for Public Policy and Higher Education (NCPP&HE) has assessed state educational performance. The “Measuring Up” project evaluated state performance in providing higher education through state public colleges and universities (NCPP&HE 2002). The study measures performance in education, rather than simple measure of expenditures. State level statistics are collected

158

RUSSELL J. DALTON

from national and state educational offices, as well as other educational databases. These data are used to rank the states on several criteria:6 ●

●

●

●

●

Preparation measures preparation for postsecondary education. It includes statistics on high school completion rates, the content of high school curriculums, and several measures of student achievement. Participation assesses the opportunities for residents of varying ages and income levels to enroll in postsecondary education. The index considers enrollment patterns at both two- and four-year institutions and public and private colleges. Affordability is based on three concepts: students’ capacity to pay for college given the type of institution they attend; the amount of needbased grant assistance students receive to offset expenses; and the associate loan burden. Completion calculates the progress in college studies and the completion of certificates and degrees in a timely manner. Benefits considers the states’ investments in human capital in areas: educational achievement, economic benefits, civic benefits, and adult skill levels.

These measures focus upon higher education, but they also indicate the effectiveness of the secondary educational system.7 A populist hypothesis would suggest that initiatives should improve state educational performance across all five dimensions. Even if few citizens focus explicitly on higher education, the indirect impact of public oversight should still have an effect. The critics of direct democracy would hypothesize a negative relationship if government is misdirected or restricted by the initiative process. One also might expect that the public will be more concerned with maximizing certain dimensions, such as the affordability of education and its benefits. Table 9.3 presents the correlations between initiatives passed and state educational performance. We see a pattern of weak and varied relationships.

Table 9.3 Correlation between initiative usage and measures of educational performance Year

Preparation

Participation

Affordability

Completion

Benefits

2002 2000

.01 .02

.03 .02

.13 .08

.21 .28

.23 .20

Overall index .04 .01

Note: Table entries are the Pearson correlations between state education performance measures and the number of popular initiatives that were passed between 1960 and 1998. Source: NCPP&HE (2002) and Initiative and Referendum Institute.

DIRECT DEMOCRACY AND GOOD GOVERNANCE

159

Using the 2002 survey as an example, the benefits derived from postsecondary education are higher in states with frequent initiative usage, as is the affordability of education. These relationships might reflect the public’s expected priorities for affordable education with the maximum benefits. At the same time, the other three measures display weak or negligible relationships with initiative usage—and none of these correlations is statistically significant. We can also test of one of the underlying assumptions of the populist reformers. The value of direct democracy presumably lays in ensuring that states implement public priorities even if the government is not performing well. This implies an interaction effect. Direct democracy presumably has a weaker influence on policy outcomes in well-managed state governments that already respond to public policy preference. Instead, the influence of direct democracy should be more apparent when government is not well managed, and thus voters express their preferences through initiatives. We can indirectly explore this hypothesis by combining the management of state governments with the educational policy measures. We divided the fifty states in two groups: high-performance and low-performance states based on the 2001 GPP scores from table 9.1. We then correlated initiative passage and educational performance for these two separate groups. For states with high government management scores, the number of initiatives is unrelated to state educational performance (the overall education index correlation is .00). However, in states with poorly managed government, there are very strong relationships between initiative usage and education affordability (r  .43) and benefits (r  .41), which are both significant at the .10 level, and the correlation with the overall education index is .19. In summary, education policy suggests that direct democracy systematically affects the policy outcomes of state governments. Presumed public preferences for affordable higher education and maximal education benefits are more common in states with frequent initiative usage. In addition, there is some evidence that the impact of direct public pressure is greatest in states that are poorly managed, and thus presumably less responsive to public opinion without this stimulus. Yet most relationships are weak, and most lack statistical significance. Thus, as with our measures of good governance, direct democracy appears to have less influence than either its proponents or its opponents presume.

State Tax Policies Tax policy signifies a state’s policy potential in providing resources to fund government activity and incentive structures that influence social

160

RUSSELL J. DALTON

and economic development. Moreover, the public at large is generally sensitive to taxation levels, and this is one area where initiative usage has limited and redirected government taxes. Matsusaka (2004, chapter 3), for example, found that initiative states have lower levels of state revenue and lower levels of state spending. Thus, one might expect the impact of initiatives to appear clearly in the taxing (and spending) levels of state governments. Building on these analyses, one must also recognize that tax needs and opportunities vary from state to state. So, instead of measuring simple revenue or spending statistics, our good governance perspective asks about the quality of tax policy. An extension of the GPP examined the tax policies of the states (Barrett et al. 2003). The project tapped broader aspects of taxation that reflect the capabilities of government. For instance, instead of simply measuring total tax receipts, the project determined whether state policy outputs were consistent with the revenue streams to fund these programs, and whether the tax structures were appropriate. Indeed, one of the criticisms of direct democracy is that popular mandates force government into inefficient and insufficient tax policies—something that can be directly examined with these data. The GPP experts judged the success of state tax policies on several criteria that broadly describe a system of good tax governance: 8 ●

●

●

Adequacy: Does the state have adequate current and future revenues to provide reasonable support for the programs the state has historically supported? Is there a balanced, multitax approach? Is a state dramatically out of line with similar states, inhibiting its competitiveness or tax compliance? Is the state experiencing budget shortfalls attributable to a weakness in tax revenues? Management: Does the state have adequate resources and management capacity to optimize voluntary compliance and get taxes from those who do not voluntarily comply? Does the state understand the implications of its tax decisions and improve management and policies? How accurate are revenue estimates and assessments of tax changes? Does it have good information and data to facilitate understanding of the tax system? Is staffing adequate to accomplish the job? Fairness: Are similar taxpayers taxed similarly, resulting in the broadest possible tax base at the lowest possible rates? Is the system overly regressive? How thoroughly does the state tax services? Does the sales tax on goods have a broad base with a minimum of unnecessary exemptions? Do the state’s taxes avoid excessive exemptions and deductions that are not means tested? Is there anything extremely unfair about the state’s approach to corporate taxes? Has the state

DIRECT DEMOCRACY AND GOOD GOVERNANCE

161

Table 9.4 Correlation between initiative usage and measures of tax performance Year 2002

Adequacy of revenue .398

Fairness to taxpayers

Management of system

.122

.052

Overall index .154

Note: Table entries are the Pearson correlations between state education performance measures and the number of popular initiatives that were passed between 1960 and 1998. Source: Barrett et al. 2003 and Initiative and Referendum Institute.

developed corporate taxes to be more evenhanded and less vulnerable to evasive tax planning? Do other factors affect tax fairness? The project team then rated the states on each of these dimensions, using the project’s detailed coding criteria (Barrett et al. 2003).9 Virtually no one wants higher taxes. However, when operationalized in terms of these three measures, we expect that citizens favor a tax system that is adequate, well managed, and fair. Thus, we hypothesize that citizens prefer positive scores on all three of these measures because this implies an efficient and effective use of taxes, independent of tax rates. Therefore, the question is whether initiative usage encourages positive performance—or provides a mechanism for antitax sentiments to distort public policy. Table 9.4 presents the correlation between initiatives passed and the three tax measures, and a summary score averaging the three dimensions. Matsusaka (2004) found that initiative usage is negatively related to the adequacy of tax revenues, and our analyses reflect the same pattern (r  .398). This potential reflects the impact of the antitax movement that uses the initiative process as a tool. Conversely, there is a slight positive relationship between initiative usage and tax fairness, which again is consistent with a public pressuring government through the initiative process. Taken together, we find only limited evidence that the usage of direct democracy notably affects the performance of state governments in America—in a systematically positive or negative direction. To those who argue that initiatives derail democracy, this seems like hyperbole because the impact on governability appears quite limited. Similarly, to those who claim that direct democracy provides a significant new tool for citizens to influence government policy, the evidence on a broad scale of education and tax policy also seems quite limited. Multivariate Models The bivariate relationships provide an initial basis for judging the impact of direct democracy. However, these relationships may arise because of the

162

RUSSELL J. DALTON

spurious impact of other factors that typically influence state performance, or state characteristics may mask the actual influence of direct democracy. The states that regularly use initiatives also systematically differ from low-usage states on a host of social, political, and economic variables that may affect government performance. Thus, the final step of our analysis examines the relationship between initiative usage and government performance while controlling for other factors that also are routinely tied to government performance (see, e.g., Knack 2002; Ingraham, Joyce, and Donahue 2003). State resources are an obvious predictor of government performance, both in terms of governing processes and for policy inputs and outputs. However, the nature of this relationship is often uncertain. Higher-income and better-educated citizens may be more effective in demanding better government and in providing the financial resources and workforce skills needed for successful public policies. Alternatively, these same resources may allow individuals to substitute private resources for state policies—such as private schools and home security—that reduce expectations and demands on government. We measure state resources by personal income per capita in 1999 for each state.10 Population size is also potentially related to government performance. On the one hand, a large state has more human resources to undertake policy initiatives, and size provides potential efficiencies of scale in developing information technologies, government procedures, and other management systems. On the other hand, larger states tend to have more numerous and diverse interests that seek representation. This diversity may make it more difficult for government to fulfill contrasting needs or even identify the public’s preferences. It is presumably easier to manage a small homogeneous state than a large diverse state such as California. We measure population size as the total adult public over age 18 (including both citizens and noncitizens). The political composition of the state might influence government performance. Typically, we might expect greater support for government to be higher in states possessing a Democratic majority, while Republicans are normally more critical of government. However, Republicans may place greater demands on government to be effective and efficient. To examine this relationship, we measure the difference between the percentage of Democrats and the percentage of Republicans in the state legislature in 1999. In a series of recent studies, Robert Putnam has demonstrated a strong relationship between social capital and the performance of state governments in Italy and various aspects of U.S. state government performance (Putnam 1993, 2000).11 Similarly, Knack (2002) found that social capital had

DIRECT DEMOCRACY AND GOOD GOVERNANCE

163

a positive impact on U.S. state government performance; Lovrich and Pierce (1999) found a strong positive relationship between measures of social capital and the performance of city governments in the United States. According to Putnam’s logic, social capital provides the civic resources, norms, and skills that facilitate good government. We measure social capital with the summary social capital index that Putnam has calculated for the states (2000). Finally, various state demographic characteristics might influence performance. For instance, rural/urban composition might affect governance; presumably more urban states present more challenges for a state government. The degree of economic inequality in a state might also affect the policy issues that state governments confront and the political polarization that might influence policy making.12 Therefore, we included these two measures in our analyses. We combined this set of predictors with initiative usage to predict government performance on all three areas we have examined in this chapter: GPP, educational performance, and tax performance. While the impact of initiative usage was somewhat ambiguous in the bivariate relationships, its independent impact on state government becomes more distinct when controlling for the political and social composition of the state. Table 9.5 indicates that initiative usage has a negative impact across all five measures, and most of these relationships are statistically significant. Indeed, these controls systematically strengthen the negative independent effects of initiative usage. These analyses also demonstrate the impact of several of the control variables on the various dimensions of government performance. Population size increases government performance in terms of the GPP management index and the index of state educational performance. Social capital within a state increases performance on all three summary indices. Bowling together generates good governance in the U.S. states, much as Putnam (2000) and Knack (2002) previously found. Other variables in the model had differential effects across these three dimensions. For instance, economic inequality apparently creates difficulties for state governments and lowers performance on the GPP indices. Income levels have a positive influence on educational performance, but surprisingly do not significantly influence the other two areas of state performance. Finally, neither the party composition of the legislature nor the urban/rural composition of the state displays a significant relationship with any of the performance indices.13 Direct democracy represents an overlapping mix of causes and consequences, and this should temper our conclusions. For instance, Smith and Tolbert (2004) argued that initiative usage increases social capital; thus, controlling for social capital in predicting good governance may oversimplify our

164

GPP 2001

GPP 1999

Education 2002

Education 2000

Taxes

Predictor

b

Beta

b

Beta

B

Beta

b

Beta

b

Beta

Initiatives passed Income

.011 (.006) .0000 (.000) .320 (.092) .383 (.274) .104 (.122) .002 (.006) .136 (.053) 2.084 (1.101)

.27*

.006 (.007) .000 (.000) .273 (.113) .160 (.345) .248 (.153) .002 (.008) .151 (.066) 2.366 (1.384)

.14

.009 (.004) .000 (.000) .138 (.070) .236 (.213) .469 (.095) .010 (.005) .057 (.041) .038 (.885)

.19*

.127 (.048) .0011 (.000) .993 (.783) 2.017 (2.386) 4.739 (1.060) .130 (.056) .967 (.459) 61.887 (9.571)

.25*

.0098 (.005) .0000 (.000) .048 (.077) .459 (.234) .280 (.104) .002 (.005) .027 (.045) 2.049 (.940)

.27*

Adult population (log) Party composition Social capital Urban/ rural Economic inequality Constant Multiple R Rsqr

.11 .59* .22 .12 .05 .40*

.629 .396

.08 .45* .08 .32* .04 .38*

.534 .286

.47* .22* .12 .59* .23 .14

.856 .733

.46* .15 .09 .54* .28 .22

.85 .73

Note: Table entries are coefficients from OLS multiple regression analysis; coefficients marked by an asterisk are significant at the .05 level. Sources: See tables 9.1–9.4.

.10 .10 .29* .46* .05 .09

.68 .46

RUSSELL J. DALTON

Table 9.5 Predicting government performance

DIRECT DEMOCRACY AND GOOD GOVERNANCE

165

causal assumptions. Yet, the consistency of the results is striking. Controlling for other factors generally influencing state government performance, initiative usage is systematically related to less effective government. To paraphrase Jefferson, the cure for poor democracy may not be more democracy. Direct Democracy and Governance Beyond any single policy outcome or legal procedures, good governance is the essence of democratic government. As noted in the quote from Ingraham earlier in this chapter, good governance is the wellspring from which other desired public policies and democratic processes flow. The paradox for democracy has been to balance its populist foundations with the needs of efficient and effective governability. Too great a stress on populist decision making has potential limitations, just as an overemphasis on maximizing governability. This research posed the question of whether the expansion of direct democracy is likely to improve or worsen the process of democratic governability. We examined this topic from three perspectives: government processes, policy output, and tax inputs. These projects did not just measure quantitative government policies, but tried to tap the qualitative dimension as well. For instance, these studies did not just measure total tax revenues, but asked whether tax policies were sufficient to meet the policy goals set by the legislature and whether tax policy was fair and efficient. In simple bivariate terms, there are only weak relationships between initiative usage in a state and these three measures of performance. For instance, the difference between the states without initiatives and the case with the greatest use of initiatives (California) produced only a small difference in the overall management of state government. In addition, with California excluded as an outlier, the evidence was even more ambiguous. Similar weak relationships appeared for education and tax policy. In some instances, the data suggested a positive influence of initiative use—for aspects of education policy and in producing lower tax levels that voters apparently favor. Yet, when one controls for differences in the social and political composition of noninitiative and initiativeuse states, a clear pattern emerges. The active implementation of initiatives is related to lower performance in the management of state government, education policy, and tax policy. Direct democracy appears to erode the overall performance of state governments in a clear and systematic way. Our results thus can be viewed as supporting both sides of the direct democracy debate. On the one hand, the bivariate relationships provide little support for the critics of direct democracy, who largely based their own conclusions on the simple comparisons of states. California and other

166

RUSSELL J. DALTON

states that frequently use initiatives are not strikingly and consistently different from nonuse states. On the other hand, the multivariate analyses document a consistent negative impact of direct democracy on governance. All else being equal, the active use of initiatives slightly impedes state government performance. This is not a deciding factor in the evaluation of direct democracy. Other research has demonstrated the positive educational impact of initiatives on the citizenry, increasing political knowledge, engagement, feelings of efficacy, and social capital (Tolbert in this volume; Smith in this volume; Bowler and Donovan 2002). There may be other positive effects on political parties, interest groups, and other institutions that counterbalance the patterns described here. Moreover, these tradeoffs may be endemic to the democratic process, because greater citizen access is essential to the definition of democracy—but also lessens the efficiency of government. Democracy is not designed to be efficient or effective, and direct democracy is no different. Notes 1. A previous version of this chapter was presented at the conference “The Impact of Direct Democracy,” Center for the Study of Democracy, University of California, Irvine. I would like to thank Shaun Bowler, Amihai Glazer, and John Matsusaka for their advice on this research. 2. The following explanations are taken from the GPP report (2002). 3. We conducted a principal components analysis to determine the dimensionality of these items, including the five items from both waves. The first component explains 59 percent of the total variance in these ten items, and all the variables have large positive loadings (ranging from .69 to .84). So, even though states can receive different ranks on each dimension, there is a strong tendency for performance to generalize across dimensions, similar to Putnam’s (1993) findings in Italy. 4. We counted popular initiatives and excluded popular referendums or legislative referendums. In our view, this is the most direct measure of citizen “intervention” in the governing process. We also considered the number of initiatives placed on the ballot. These two indices are highly correlated (r .98). We felt that initiatives that were passed measured initiative effects better than initiatives that were not passed. In either case, the results were largely equivalent because the variables are so highly correlated. The data were derived from the Initiative and Referendum Institute Web site (byyear.exl file). 5. If one uses the dichotomy of referendum/nonreferendum states, the correlations are even weaker (2001 r  .04; 1999 r  .12). 6. Details of these measures are available at: measuringup.highereducation.org/ 2002/technicalguide.htm.

DIRECT DEMOCRACY AND GOOD GOVERNANCE

167

7. The scaling of the two years is not identical in these analyses. The 2002 data were graded on a 1–4 scale, and the 2000 data were coded on a 0–100 scale. We ran a principal components analysis to determine the dimensionality of these ten items. The first principal component accounts for 44 percent of the total variance. Eight of the indicators load heavily on this first dimension, with the two affordability questions loading on the second dimension. 8. These summaries are taken from Barrett et al. (2003). 9. States were scored on a 1–4 scale. It appears that tax management measure represents a dimension separate from the other two items. 10. We also analyzed a measure or the educational composition of the state as an indicator of the resources that citizens possess. This variable was so highly correlated with income level (r .66) that we did not include both in the regression analyses because of potential multicollinearity problems. 11. Smith and Tolbert (2004, 156) found that initiative usage was strongly related to social capital levels across the U.S. states. 12. The economic inequality measure is from the Economic Policy Institute for the 1996–98 period. 13. We also used the variable of Republican/Democratic presidential vote shares in the 1996 election and the liberal/conservative composition of the electorate and obtained comparable results.

This page intentionally left blank

Appendix A. Initiative provisions, state-by-state Signature requirement State

Adopted

Type

Total

Basis

Approval

Alaska

1956

S(I)

10%

E

M*

Arizona

1911

S

10%

G

M

Arkansas

1911 1910

C S

15% 8%

G G

M M

California

1910 1911

C S

10% 5%

G G

M M

Florida Idaho Illinois

1911 1910 1910 1968 1912 1970

C S C C S C

8% 5% 5% 8% 6% 8%

G S S P R G

Maine Massachusetts

1908 1918

S(I) 10% S(I) 3% ⫹ ½%

G G

1918

C(I)

3%

G

Michigan

1913

S(I)

8%

G

M M M M M 60% or 50% B M* M and 30% B* M and 30% B* M*

Mississippi

1908 1992

C C(I)

10% 12%

G G

1908 1908 1906 1972

S C S C

3 1/3%* 5 1/3%* 5% 10%

G G G G

Colorado

Missouri Montana

M M and 40% B* M M M M

Amendment by legislature Amend any time, repeal after 2 years, both with majority vote 3/4 vote each house and only to further purpose of measure ... Any time with 2/3 vote each house ... Prohibited, amendment only by vote of people. ... Standard ... ... Standard ... Standard Standard ... Any time with 3/4 vote each house ... ... Standard ... Standard ... (Continued )

170

APPENDIX

Appendix A. (Continued )

Signature requirement State Nebraska

Nevada

North Dakota

Ohio

Oklahoma Oregon South Dakota Utah Washington

Wyoming

Adopted

Type

Total

Basis

Approval

1912

S

7%

R

1912

C

10%

R

1912

S(I)

10%

E

M and 35% B M and 35% B M*

1912

C

10%

E

1914

S

2%

N

M in 2 consecutive elections M

1914 1912 1912 1912 1907 1907 1902 1902 1898 1972 1900 1900 1912

C S(I) C C(I) S C S C S C S S(I) S

4% 3% ⫹ 3% 10% 10% 8% 15% 6% 8% 5% 10% 10% 5% ⫹ 5% 8%

N G G R H H G G G G G G G

M M* M M* M M M M M M M* M* M*

1912

S(I)

8%

G

M*

1968

S(I)

15%

E

50% B*

Amendment by legislature Standard ... After 3 years with majority vote ... 2/3 vote each house in first 7 years, standard thereafter ... Standard ... ... Standard ... Standard ... Standard ... Standard Standard 2/3 vote each house for 2 years, then standard 2/3 vote each house for 2 years then standard Amend any time, repeal after 2 years, majority vote

Source: Initiative and Referendum Institute, and John G. Matsusaka, For the Many or the Few: The Initiative, Public Policy, and American Democracy (University of Chicago Press, 2004). * See note below for elaboration.

Type: Type of initiative available. S ⫽ statutory initiative, C ⫽ constitutional amendment initiative. Measures go directly to the voters (direct initiative) unless indicated with “(I),” in which case the legislature has the option to approve the measure without sending it to the voters (indirect initiative).

APPENDIX

171

Signature requirement: Number of signatures required to place a measure on the ballot. Total: Total number of signatures required, expressed as percent (of votes cast in previous election, registered voters, etc.; see Basis). For indirect initiatives, “X% ⫹ Y%” means that X% is required to place a measure before the legislature, and an additional Y% is required to place the measure before the voters if the legislature does not approve it. Missouri does not have a total signature requirement; the reported numbers are imputed from the distribution requirement. Basis: Signature percentage calculated in terms of G ⫽ votes cast in the preceding gubernatorial election, E ⫽ ballots cast in preceding general election, P ⫽ votes cast in preceding presidential election, S ⫽ votes cast in preceding secretary of state election, R ⫽ registered voters, N ⫽ population, and H ⫽ votes received by state office receiving the highest number of votes in last general election. Approval: M means approval requires 50%⫹1 of the votes cast on the measure. “X% B” means approval requires votes in excess of X% of all ballots cast (which may include ballots that abstained on the measure). Utah requires 2/3 approval for measures involving wildlife. Washington requires 60% approval for gambling measures. In indirect initiative states, the legislature can adopt the measure after signatures are collected, and the proposal will not go to the voters. In Alaska and Wyoming, the measure does not go the voters if the legislature adopts “substantially the same” measure. In Massachusetts, a measure does not go to the voters unless it receives 1/4 of the votes in the legislature for 2 consecutive sessions or if the legislature fails to vote on it. Amendment by legislature: Indicates when and under what condition the legislature can amend a statute. “Standard” means the legislature can amend or repeal the measure at any time with a majority vote of each house. Otherwise, amendment typically requires a supermajority in both houses of the legislature. Constitutional amendments can only be amended by a vote of the people.

Appendix B. Statewide initiatives, five-year intervals Year 1904–05 1906–10 1911–15 1916–20 1921–25 1926–30 1931–35 1936–40 1941–45 1946–50 1951–55 1956–60

Number

Approved

2 54 165 127 79 95 119 149 42 105 45 69

2 23 55 55 18 22 53 53 15 43 14 31 (Continued )

172

APPENDIX

Appendix B. (Continued )

Year

Number

1961–65 1966–70 1971–75 1976–80 1981–85 1986–90 1991–95 1996–2000 2001–05 2006 Total 1904–2006

52 36 65 137 105 169 149 242 146 78 2,230

Approved 24 13 33 50 46 67 69 108 61 32 909

Source: The Initiative and Referendum Institute, Initiative Use, 1904–2006, available at www.iandrinstitute.org.

Appendix C. Initiative activity by state (1904–2006) On ballot

Alaska Arizona Arkansas California Colorado Florida Idaho Illinois Maine Massachusetts Michigan Mississippi Missouri Montana Nebraska Nevada North Dakota Ohio Oklahoma Oregon South Dakota Utah Washington Wyoming

Approved

Number

Rank

Number

Rank

Percent

Rank

41 165 117 315 196 28 28 1 48 65 66 2 75 73 46 53 175 75 85 341 59 20 151 6

18 5 7 2 3 19 20 24 16 13 12 23 9 11 17 15 4 10 8 1 14 21 6 22

22 75 48 105 70 23 14 1 23 32 23 0 29 40 17 11 80 20 39 119 19 4 76 3

15 5 7 2 6 14 19 23 13 10 12 24 11 8 18 20 3 16 9 1 17 21 4 22

53.7 45.5 41.0 33.3 35.7 82.1 50.0 100.0 47.9 49.2 34.8 0.0 38.7 54.8 37.0 20.8 45.7 26.7 45.9 34.9 32.2 20.0 50.3 50.0

4 12 13 19 16 2 6 1 9 8 18 24 14 3 15 22 11 2 10 17 20 23 5 7

Source: The Initiative and Referendum Institute, Initiative Use, 1904–2006, available at www.iandrinstitute. org.

References

Abramowitz, Alan. 2004. Terrorism, Gay Marriage, and Incumbency: Explaining the Republican Victory in the 2004 Presidential Election. The Forum 2 (4): Article 3. www.bepress.com/forum/vol2/iss4/art3. Abramson, Paul, and John Aldrich. 1982. The Decline of Electoral Participation in America. American Political Science Review 76:502–21. Alverez, Michael, and G. Glasgow. 2000. Two-Stage Estimation of Nonrecursive Choice Models. Political Analysis 8:147–65. Anscombe, Gertrude. 1976. On Frustration of the Majority by Fulfillment of the Majority’s Will. Analysis 36:161–68. Appleton, Andrew, and Daniel Ward, eds. 1997. State Party Profiles: A 50-State Guide to Development Organization and Resources. Washington, D.C.: CQ Press. APSA (American Political Science Association). 1950. Toward a More Responsible Two-Party System. American Political Science Review 44 (3): 488–99. Supplement part 2. Asimov, Nanette. 1998. Complex Prop. 8 Creates Divisions: Teachers, GOP Split over Diverse Items in Proposal. San Francisco Chronicle, October 21, A1. Atkeson, Lonna Rae, and Randall W. Partin. 1995. Economic and Referendum Voting: A Comparison of Gubernatorial and Senatorial Elections. American Political Science Review 89:99–107. Bali, V. 2003. Implementing Popular Initiatives: What Matters for Compliance? Journal of Politics 65:1130–46. Banducci, Susan. 1998. Direct Legislation: When Is It Used and When Does It Pass? In Citizens as Legislators: Direct Democracy in the United States, ed. Shaun Bowler, Todd Donovan, and Caroline Tolbert. Columbus: Ohio State University Press. Banks, Jeffrey S., and Rangarajan Sundaram. 1998. Optimal Retention in Agency Problems. Journal of Economic Theory 82:293–323. Baqir, Reza. 2002. Districting and Government Overspending. Journal of Political Economy 110, no. 6 (December): 1318–54. Barnett, James. 1915. The Operation of the Initiative, Referendum and Recall in Oregon. New York: Macmillan. Barrett, Katherine, Richard Greene, Michele Mariani, and Anya Sostek. 2003. The Way We Tax: A Fifty State Report. Governing, February. www.governing.com/gpp/2003.gp3intro. Barro, Robert J. 1973. The Control of Politicians: An Economic Model. Public Choice 14:19–42.

174

REFERENCES

Baumgartner, Frank R., and Beth L. Leech. 1998. Basic Interest: The Importance of Groups in Politics and Political Science. Princeton, NJ: Princeton University Press. ———. 2001. Issue Niches and Policy Bandwagons: Patterns of Interest Group Involvement in National Politics. Journal of Politics 63 (4): 1191–1213. Bawn, Kathleen. 1999. Money and Majorities in the Federal Republic of Germany: Evidence for a Veto Players Model of Government Spending. American Journal of Political Science 43:707–36. Belli, Robert, Michael W. Traugott, Margaret Young, and Katherine A. McGonagle. 1999. Reducing Vote over Reporting in Surveys: Social Desirability, Memory Failure, and Source Monitoring. Public Opinion Quarterly 63:90–108. Benz, M., and A. Stutzer. 2004. Are Voter Better Informed When They Have a Larger Say in Politics? Public Choice 119:21–59. Berelson, Bernard R., Paul F. Lazarsfeld, and William N. McPhee. 1954. Voting. Chicago: University of Chicago Press. Bernholz, Peter. 1978. On the Stability of Logrolling Outcomes in Stochastic Games. Public Choice 33 (3): 65–82. Besley, Timothy, and Stephen Coate. 2003. Issue Unbundling via Citizens’ Initiatives. Working paper, London School of Economics and Cornell University, New York. Bimber, Bruce. 2003. The Internet and American Democracy. New York: Cambridge University Press. Birch, Anthony. 1984. Overload, Ungovernability, and Delegitimation. British Journal of Political Science 14:135–60. Black, Duncan. 1948. On the Rationale of Group Decision Making. Journal of Political Economy 56:23–34. Black, Duncan, and R. A. Newing. [1951] 1998. Committee Decisions with Complimentary Valuation. In The Theory of Committees and Elections by Duncan Black and Committee Decisions with Complimentary Valuation by Duncan Black and R. A. Newing, rev. 2nd ed., ed. Iain McLean, Alistair McMillan, and Burt Monroe. Boston, MA: Kluwer Academic Press. Boehmke, Frederick J. 2002. The Effect of Direct Democracy on the Size and Diversity of State Interest Group Populations. Journal of Politics 3:827–44. ———. 2003a. Using Auxiliary Data to Estimate Selection Bias Models, with an Application to Interest Groups’ Use of the Direct Initiative Process. Political Analysis 11 (3): 234–54. ———. 2003b. Institutional and Political Sources of Variation in Statewide Use of the Initiative Process. Paper presented at the annual meeting of the American Political Science Association, Philadelphia, Pennsylvania. ———. 2004. The Initiative Process and the Volatility of State Interest Group Populations. Typescript, University of Iowa. ———. 2005. The Indirect Effect of Direct Legislation: How Institutions Influence Interest Groups. Columbus: Ohio State University Press. Boehmke, Frederick J., Sean Gailmard, and John W. Patty. 2006. Lobbying in Multiple Venues: Access and Policy Agendas. Paper presented at the annual meeting of the Southern Political Science Association, New Orleans, LA. Boehmke, Frederick J, and Richard C. Witmer. 2003. Hard Money, Soft Money or Direct Lobbying Expenditures? The Emergence and Growth of American Indian

REFERENCES

175

Lobbying. Paper presented at the annual meeting of the Midwest Political Science Association. ———. 2006. Attention to State Legislation by Indian Nations in California: Is All Tribal Lobbying about Gaming? Paper presented at the annual meeting of the Midwest Political Science Association, Chicago, IL. Bowler, Shaun, and Todd Donovan. 1998. Demanding Choices: Opinion, Voting, and Direct Democracy. Ann Arbor: University of Michigan Press. ———. 2002. Democracy, Institutions and Attitudes about Citizen Influence on Government. British Journal of Political Science 32:371–90. ———. 2004. Measuring the Effects of Direct Democracy on State Policy: Not All Initiatives Are Created Equal. State Politics and Policy Quarterly 4 (3): 345–63. Bowler, Shaun, Todd Donovan, Max Neiman, and Johnny Peel. 2001.Institutional Threat and Partisan Outcomes: Legislative Candidates’ Attitudes toward Direct Democracy. State Politics and Policy Quarterly 1 (4): 364–79. Bowler, Shaun, Todd Donovan, and Caroline Tolbert, eds. 1998. Citizens As Legislators: Direct Democracy in the United States. Columbus: Ohio State University Press. Bowler, Shaun, Stephen P. Nicholson, and Gary M. Segura. 2006. Earthquakes and Aftershocks: Race, Direct Democracy and Partisan Change. American Journal of Political Science 50 (1): 146–59. Boylan, Richard T., and Cheryl X. Long. 2001. A Survey of State House Reporter’s Perception of Public Corruption. Manuscript, School of Business, University of Mississippi. Bradbury, John Charles, and W. Mark Crain. 2001. Legislative Organization and Government Spending: Cross Country Evidence. Journal of Public Economics 82 (December): 309–25. Branton, Regina. 2003. Examining Individual-Level Voting Behavior on State Ballot Propositions. Political Research Quarterly 56:367–77. ———. 2004. Voting in Initiative Elections: Does the Context of Racial and Ethnic Diversity Matter? State Politics and Policy Quarterly 4:294–317. Braunstein, Rich. 2004. Initiative and Referendum Voting. New York: LFG. Broder, David S. 2000. Democracy Derailed: Initiative Campaigns and the Power of Money. Orland, CA: Harcourt Brace. Browne, William P. 1990. Organized Interests and Their Issue Niches: A Search for Pluralism in a Policy Domain. Journal of Politics 52:477–509. Burden, Barry C. 2004. An Alternative Account of the 2004 Presidential Election. The Forum 2 (4): Article 2. www.bepress.com/forum/vol2/iss4/art2. ———. 2005. Institutions and Policy Representation in the States. State Politics and Policy Quarterly 5:373–93. Button, James W., Barbara A. Rienzo, and Kenneth D. Wald. 1997. Private Lives, Public Conflicts: Battles over Gay Rights in American Communities. Washington, D.C.: CQ Press. Cain, Bruce, Russell Dalton, and Susan Scarrow, eds. 2003. Democracy Transformed? The Expanding Political Opportunities in Advanced Industrial Democracies. Oxford: Oxford University Press. Cain, Bruce, and Thad Kousser. 2004. Adapting to Term Limits: Recent Experiences and New Directions. San Francisco, CA: Public Policy Institute of California.

176

REFERENCES

Camobreco, J. 1998. Preferences, Fiscal Policies, and the Initiative Process. Journal of Politics 60:819–29. Campbell, Angus. 1966. Surge and Decline: A Study of Electoral Change. In Elections and the Political Order, ed. A. Campbell, P. E. Converse, W. E. Miller, and D. E. Stokes. New York: Wiley. Campbell, Angus, Phillip Converse, Warren Miller, and David Stokes. 1960. The American Voter. Chicago: University of Chicago Press. Campbell, David E., and J. Quin Monson. 2005. The Case of Bush’s Re-Election: Did Gay Marriage Do It? Paper presented at the annual meeting of the Midwest Political Science Association, Chicago, April 7–10. Carmines, Edward G., and Geoffrey C. Layman. 1997. Value Priorities, Partisanship, and Electoral Choice: The Neglected Case of the United States. Political Behavior 19:283–316. Chaffee, S., and S. Kanihan. 1997. Learning about Politics from the Media. Political Communication 14:421–30. Chavez, Lydia. 1998. The Color Bind: California’s Battle to End Affirmative Action. Berkeley: University of California Press. Converse, Phillip E. 1964. The Nature of Belief Systems in Mass Publics. In Ideology and Discontent, ed. David E. Apter. New York: Free Press. Cooperman, Alan, and Thomas Edsall. 2004. Evangelicals Say They Led Charge for the GOP. Washington Post, November 8, A1. Cotter, Cornelius, James Gibson, John Bibby, and Robert Huckshorn. 1984. Party Organizations in American Politics. Westport, CT: Praeger. Council of State Governments, The. 2004. The Book of the States. Vol. 36. Lexington, KY: The Council of State Governments. Cronin, Thomas. 1989. Direct Democracy. Cambridge, MA: Harvard University Press. Cummings, Jeanne. 2004. How Bush Camp Won Ohio. Wall Street Journal, November 19, A-4. Dahl, Robert A. 1989. Democracy and Its Critics. New Haven, CT: Yale University Press. Dalton, Russell J. 2004. Democratic Challenges, Democratic Choices: The Erosion of Political Support in Advanced Industrial Democracies. Oxford: Oxford University Press. Dalton, Russell J., and Martin Wattenberg, eds. 2001. Parties without Partisans: Political Change in Advanced Industrial Democracies. New York: Oxford University Press. Dao, James. 2004. Flush with Victory, Grass-Roots Crusader against Same-Sex Marriage Thinks Big. New York Times, November 26, A28. Davies, Frank. 2004. Bush’s Clear Vision Aided Reelection Bid, Rove Says. Miami Herald, November 10. de Figueiredo Jr., Rui J. P. 2003. Budget Institutions and Political Insulation: Why States Adopt the Item Veto. Journal of Public Economics 87:2677–2701. Delli Carpini, Michael, and Scott Keeter. 1996. What American Know about Politics and Why It Matters. New Haven, CT: Yale University Press. Di Tella, Rafael, and Raymond Fisman. 2004. Are Politicians Really Paid Like Bureaucrats? Journal of Law and Economics, October, 47 (2): 477–514.

REFERENCES

177

Donovan, Todd. 2004. The Legislature. In Washington State Politics, ed. Clayton, Nicholas P. Lovrich, Cornell W. Clayton, and Lance T. Leloup. Pullman: Washington State University Press. Donovas, Todd, and Shaun Bowler. 1998. An Overview of Direct Democracy in the American States. In Citizens as Legislators, ed. Shaun Bowler, Todd Donovan, and Caroline Tolbert. Columbus: Ohio State University Press. Donovan, Todd, Shaun Bowler, David McCuan, and Ken Fernandez. 1998. Contending Players and Strategies: Opposition Advantages in Initiative Campaigns. In Citizens as Legislators: Direct Democracy in the United States, edited by Shaun Bowler, Todd Donovan, and Caroline J. Tolbert. Columbus: Ohio State University Press. Donovan, Todd, Caroline Tolbert, and Daniel Smith. 2005. Do State-Level Ballot Measures Affect Presidential Elections? Paper presented at the annual meeting of the American Political Science Association, Washington, D.C. Donovan, Todd, Caroline Tolbert, Daniel A. Smith, and Janine Perry. 2005. Did Gay Marriage Elect George W. Bush? Paper presented at the State Politics and Polity Conference, East Lansing, Michigan, May 14–15. Donovan, Todd, James Wenzel, and Shaun Bowler. 2000. Direct Democracy Initiatives after Romer. In The Politics of Gay Marriage, ed. Craig Rimmerman, Ken Wald, and Clyde Wilcox. Chicago: University of Chicago Press. Dyke, Joshua, and Mark Baldassare. 2006. The Limits of Support for Direct Democracy: Process-Oriented Preferences and the 2005 California Special Election. Paper presented at the annual meeting of the American Political Science Association, Philadelphia, Pennsylvania. El Nasser, Haya. 2004. For Political Trends, Think Micropolitan. USA Today, November 23, A-1. Ellis, Richard. 2002. Democratic Delusions: The Initiative Process in America. Lawrence: Kansas University Press. Erikson, Robert S., Gerald C. Wright, and John P. McIver. 1993. Statehouse Democracy: Public Opinion and Policy in the American States. New York: Cambridge University Press. Ernst, Howard R. 2001. The Historical Role of Narrow-Minded Interests. In Dangerous Democracy: The Battle over Ballot Initiative in America, ed. Larry J. Sabato, Howard R. Ernst, and Bruce A. Larson. Boulder, CO: Rowman and Littlefield. Fearon, James D. 1999. Electoral Accountability and the Control of Politicians: Selecting Good Types versus Sanctioning Poor Performance. In Democracy, Accountability, and Representation, ed. Adam Przeworski, Susan C. Stokes, and Bernard Manin. Cambridge: Cambridge University Press. Ferejohn, John. 1986. Incumbent Performance and Electoral Control. Public Choice 50:5–25. Foladare, Irving. 1968. The Effect of Neighborhood on Voting Behavior. Political Science Quarterly 83:516–29. Ford, Henry Jones. 1912. Direct Legislation and the Recall. Annals of the American Academy of Political and Social Science The Initiative, Referendum and Recall 43 (September): 65–77.

178

REFERENCES

Freedman, Paul. 2004. The Gay Marriage Myth. Slate, November 5. www.slate. msn.com/id/2109275. Freedman, P., M. Franz, and K. Goldstein. 2004. Campaign Advertising and Democratic Citizenship. American Journal of Political Science 48:723–41. Garrett, Elizabeth. 1999. Money, Agenda Setting and Direct Democracy. Texas Law Review 77:1845–90. ———. 2005. Hybrid Democracy. George Washington University Law Review 73:1096. ———. 2006. The Promise and Perils of Hybrid Democracy. Henry Lecture, University of Oklahoma Law School, and Working Paper 48, USC Gould School of Law, Los Angeles, California. Geer, John G. 1988. What Do Open-Ended Questions Measure? Public Opinion Quarterly 52 (3): 365–71. Gelman, Andrew, and Gary King. 1993. Why Are American Presidential Election Campaign Polls So Variable When Voters Are So Predictable? British Journal of Political Science 23:409–51. Gerber, Elisabeth R. 1996a. Legislatures, Initiatives and Representation: The Effects of State Legislative Institutions on Policy. Political Research Quarterly 49 (2): 263–86. ———. 1996b. Legislative Response to the Threat of Popular Initiatives. American Journal of Political Science 40 (1): 99–128. ———. 1998. The Populist Paradox: Interest Group Influence and the Promise of Direct Legislation. Princeton, NJ: Princeton University Press. Gerber, Elisabeth R., Zoltan L. Hajnal, and Hugh Louch. 2002. Minorities and Direct Legislation: Evidence from California Ballot Proposition Elections. Journal of Politics 64 (1): 154–77. Gerber, Elisabeth R., and Arthur Lupia. 1995. Campaign Competition and Policy Responsiveness in Direct Legislation Elections. Political Behavior, 17 (2): 287–306. Gerber, Elisabeth R., Arthur Lupia, Mathew McCubbins, and D. Roderick Kiewiet. 2000. Stealing the Initiative: How State Government Responds to Direct Democracy. Upper Saddle River, NJ: Prentice Hall. Giles, Michael. 1977. Percent Black and Racial Hostility: An Old Assumption Reexamined. Social Science Quarterly 58:848–65. Giles, Michael, and Arthur Evans. 1986. The Power Approach to Intergroup Hostility. Journal of Conflict Resolution 30:469–86. Gilligan, Thomas W., and John G. Matsusaka. 1995. Systematic Deviations from Constituent Interests: The Role of Legislative Structure and Political Parties in the States. Economic Inquiry 33 (July): 383–401. ———. 2001. Fiscal Policy, Legislature Size, and Political Parties: Evidence from the First Half of the Twentieth Century. National Tax Journal 54 (March): 57–82. ———. 2006. Public Choice Principles of Redistricting. Public Choice 124 (1–2) (December): 381–98. GPP (Government Performance Project). 2002. Paths to Performance in State and Local Government. Syracuse, NY: Maxwell School. www.maxwell.syr.edu/gpp. Gray, Virginia, and David Lowery. 1996. The Population Ecology of Interest Representation. Ann Arbor: University of Michigan Press.

REFERENCES

179

Green, John C. 1994. The Christian Right and the 1994 Elections: A View from the States. PS: Political Science and Politics 28:5–8. Grilli, Vittorio, Donato Masciandaro, and Guido Tabellini. 1991. Political and Monetary Institutions and Public Financial Policies in the Industrial Countries. Economic Policy 13:342–92. Grooters, Jennifer, and Corina Eckl. 1998. Legislative Budget Procedures: A Guide to Appropriations and Budget Processes in the States, Commonwealths and Territories. Washington, D.C.: National Conference of State Legislatures. Haider-Markel, Donald. 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 1:6–26. Haider-Markel, Donald, and Kenneth Meier. 1996. The Politics of Gay and Lesbian Rights: Expanding the Scope of the Conflict. Journal of Politics 58:332–49. Hamm, Keith, and Gary Moncrief. 2004. Legislative Politics in the States. In Politics in the American States: A Comparative Introduction, 8th ed., ed. Gray Virginia and Russell Hanson. Washington, D.C.: CQ Press. Hammond, Thomas H., and Gary J. Miller. 1987. The Core of the Constitution. American Political Science Review 81 (4): 1155–74. Hasen, Rick. 2005. Rethinking the Unconstitutionality of Contribution and Expenditure Limits in Ballot Measure Campaigns. Southern California Law Review 78:885. Heinz, John P., Robert O. Laumann, Robert L. Nelson, and Robert L. Salisbury. 1993. The Hollow Core. Cambridge, MA: Harvard University Press. Hero, Rodney. 1998. Faces of Inequality: Social Diversity in American Politics. New York: Oxford University Press. Hero, Rodney, and Caroline Tolbert. 1996. A Racial/Ethnic Diversity Interpretation of Politics and Policy in the States of the U.S. American Journal of Political Science 40:851–71. ———. 2004. Minority Voices and Citizen Attitudes about Government Responsiveness in the American States: Do Social and Institutional Context Matter? British Journal of Political Science 34:109–21. Hillygus, Sunshine D., and Todd G. Shields. 2005. Moral Issues and Voter Decision Making in the 2004 Presidential Election. PS: Political Science and Politics 38:201–9. Hofstadter, Richard. 1955. The Age of Reform. New York: Vintage. Holmstrom, Bengt, and Paul Milgrom. 1991. Multitask Principal-Agent Analyses: Incentive Contracts, Asset Ownership, and Job Design. Special issue, Journal of Law, Economics, and Organization 70:24–52. Honaker, James, Anne Joseph, Gary King, Kenneth Scheve, and Naunihal Singh. 2000. Amelia: A Program for Missing Data; Windows Version. Cambridge, MA: Harvard University Press. http://gking.harvard.edu. Houser, Daniel, and John Freeman. 1998. Economic Consequences of Political Approval Management in Comparative Perspective. Working paper, University of Minnesota. Huckfeldt, Robert, Eric Plutzer, and John Sprague. 1993. Alternative Contexts of Political Behavior. Journal of Politics 55:365–81.

180

REFERENCES

Ingraham, Patricia W., Phillip G. Joyce, and Amy K. Donahue. 2003. Government Performance: Why Management Matters. Baltimore, MD: Johns Hopkins University Press. Iyengar, Shanto, and Donald R. Kinder. 1987. News That Matters: Television and American Opinion. Chicago: University of Chicago Press. Jackman, Simon. 2004. Same-Sex Marriage Ballot Initiatives and Conservative Mobilization in the 2004 Election. Paper presented at the Institute for Research in the Social Sciences, Palo Alto, California. Jelen, Ted G., ed. 1989. Religion and Political Behavior in the United States. New York: Praeger. ———. 1991. Politicized Group Identification: The Case of Fundamentalists. Western Political Quarterly 44:209–19. ———. 1994. Protestant Clergy as Political Leaders: Theological Limitations. Review of Religious Research 36:23–44. Jelen, Ted G., and Clyde Wilcox. 1992. The Effects of Religious Self-Identification on Support for the New Christian Right: An Analysis of Political Activists. Social Science Journal 29:199–210. Jennings, M. Kent, and V. Zeitner. 2003. Internet Use and Civic Engagement. Public Opinion Quarterly 67 (2): 311–34. Jones, Dale. 2002. Religious Congregations and Membership in the United States 2000. Nashville, TN: Glenmary Research Center. CD-ROM. Karp, Jeffrey A. 1998. The Influence of Elite Endorsements in Initiative Campaigns. In Citizens as Legislators: Direct Democracy in the United States, ed. Shaun Bowler, Todd Donovan, and Caroline Tolbert. Columbus: Ohio State University Press. Kellstedt, Lyman, John Green, and James Guth. 1996. Grasping the Essential: The Social Embodiment of Religion and Political Behavior. In Religion and the Culture Wars, ed. John Green, James Guth, Corwin Smidt, and Lyman Kellstedt. Lanham, MD: Rowman and Littlefield. Key, V. O., Jr. 1949. Southern Politics in State and Nation. Knoxville, TN: University of Tennessee Press. Key, V. O., Jr., and Winston W. Crouch. 1939. The Initiative and Referendum in California. Berkeley: University of California Press. Kiewiet, D. Roderick, and Mathew D. McCubbins. 1991. The Logic of Delegation: Congressional Parties and the Appropriations Process. Chicago: University of Chicago Press. King, Gary. 1989a. Statistical Models for Political Science Event Counts: Bias in Conventional Procedures and Evidence for the Exponential Poisson Regression Model. American Journal of Political Science 32:838–63. ———. 1989b. Variance Specification in Event Count Models: From Restrictive Assumptions to a Generalized Estimator. American Journal of Political Science 33:762–84. ———. 1997. A Solution to the Ecological Inference Problem: Reconstructing Individual Behavior from Aggregate Data. Princeton, NJ: Princeton University Press. King, Gary, James Honaker, Anne Joseph, Kenneth Scheve, and Naunihal Singh. 2001. Analyzing Incomplete Political Science Data. American Political Science Review 95:49–69.

REFERENCES

181

Knack, Stephen. 2002. Social Capital and the Quality of Government: Evidence from the States. American Journal of Political Science 44:772–85. König, Thomas, and Vera Tröger. 2001. Haushaltspolitik und Vetospieler. Paper presented at the Jahrestagung der Deutschen Vereinigung für Politikwissenschaft, Berlin. Korte, Gregory. 2004. Gay Issue May Energize GOP. Cincinnati Enquirer, September 14, A1. Krueger, Brian. 2002. Assessing the Potential of Internet Political Participation in the United States. American Politics Research 30:476–98. Lacey, Robert. 2005. The Electoral Allure of Direct Democracy: The Effect of Initiative Salience on Voting, 1990–1996. State Politics and Policy Quarterly 5 (2): 168–81 Lacy, Dean, and Emerson Niou. 2000. A Problem with Referendums. Journal of Theoretical Politics 12 (1): 5–32. LaRaja, Raymond. 2003. State Parties and Soft Money: How Much Party Building? In The State of the Parties: The Changing Role of Contemporary American Parties, ed. John Green and Rick Farmer. Lanham, MD: Rowman and Littlefield. Lassen, David. 2005. The Effect of Information on Voter Turnout: Evidence from a Natural Experiment. American Journal of Political Science 49 (1): 103–118. Layman, Geoffrey. 2001. The Great Divide: Religious and Cultural Conflict in American Party Politics. New York: Columbia University Press. Leech, Beth L., Frank R. Baumgartner, Timothy La Pira, and Nicholas A. Semanko. 2004. Drawing Lobbyists to Washington: Government Attention and InterestGroup Mobilization. Political Research Quarterly 58: 19–30. Lewis, Gregory B. 2005. Same-Sex Marriage and the 2004 Presidential Election. PS: Political Science and Politics 38:195–200. Linder, Wolf. 1998. Swiss Democracy: Possible Solutions to Conflict in Multicultural Societies. New York: St. Martin’s. Lochhead, Carolyn. 2004. Gay Marriage: Did Issue Help Re-Elect Bush? San Francisco Chronicle, November 4. Lovrich, Nicholas, and John Pierce. 1999. Social Capital and Government Performance. Research paper, University of Colorado at Colorado Springs. Lowery, David, and Virginia Gray. 2004. Bias in the Heavenly Chorus: Interests in Society and Before Government. Journal of Theoretical Politics 16:5–29. Lowery, David, Virginia Gray, Matthew Fellowes, and Jennifer Anderson. 2004. Living in the Moment: Lags, Leads, and the Link between Legislative Agendas and Interest Advocacy. Social Science Quarterly 85 (2): 463–77. Lupia, Arthur. 1994. Shortcuts versus Encyclopedias: Information and Voting Behavior in California Insurance Reform Elections. American Political Science Review 88:63–76. Lupia, A., and R. Johnston. 2001. Are Voters to Blame? Voter Competence and Elite Maneuvers in Public Referendums. In Referendum Democracy: Citizens, Elites, and Deliberation in Referendum Campaigns, ed. M Mendelsohn, A. Parkin, 191–210. Toronto: MacMillan/St. Martin’s. Lupia, Arthur, and John G. Matsusaka. 2004. Direct Democracy: New Approaches to Old Questions. Annual Review of Political Science 7:463–82.

182

REFERENCES

Lupia, Arthur, and Matthew McCubbins. 1998. The Democratic Dilemma. Cambridge: Cambridge University Press. Luskin, Robert. 1987. Measuring Political Sophistication. American Journal of Political Science 31:856–99. ———. 1990. Explaining Political Sophistication. Political Behavior 12:331–61. MacKuen, Michael. 1984. Exposure to Information, Belief Integration, and Individual Responsiveness to Agenda Change. American Political Science Review 78:372–91. Magleby, David B. 1984. Direct Legislation: Voting on Ballot Propositions in the United States. Baltimore, MD: Johns Hopkins University Press. ———. 1995. Direct Legislation in the American States. In Referendums around the World, ed. Austin Ranney and David Butler. Washington, D.C.: American Enterprise Institute. Matsusaka, John G. 1995. Fiscal Effects of the Voter Initiative: Evidence from the Last 30 Years. Journal of Political Economy 103 (June): 587–623. ———. 2004. For the Many or the Few: The Initiative, Public Policy, and American Democracy. Chicago: University of Chicago Press. ———. 2005a. Direct Democracy Works. Journal of Economic Perspectives 19 (2): 185–206. ———. 2005b. Direct Democracy and Fiscal Gridlock: Have Voter Initiatives Paralyzed the California Budget? State Politics and Policy Quarterly 5 (3): 248–64. ———. 2007. Direct Democracy and Public Employees. Working paper, USC Marshall School of Business. Matsusaka, John G., and Nolan M. McCarty. 2001. Political Resource Allocation: Benefits and Costs of Voter Initiatives. Journal of Law, Economics, and Organization, 17 (October): 413–48. Mayhew, David. 1986. Placing Parties in American Politics: Organization, Electoral Settings and Government Activity in the 20th Century. Princeton, NJ: Princeton University Press. McCann, James. 1997. Electoral Choices and Core Value Change: The 1992 Presidential Campaign. American Journal of Political Science 41:564–83. McCombs, Maxwell. 2004. Setting the Agenda: The Mass Media and Public Opinion. Malden, MA: Blackwell. McCombs, Maxwell, and Donald L. Shaw. 1972. The Agenda-Setting Function of the Mass Media. Public Opinion Quarterly 36:176–87. McCuan, David, Shaun Bowler, Todd Donovan, and Ken Fernandez. 1998. California’s Political Warriors: Campaign Professionals and the Initiative Process. In Citizens as Legislators, ed. Bowler, Donovan, and Tolbert. Columbus: Ohio State University Press. McDonald, Michael P. 2004. Up, Up and Away! Voter Participation in the 2004 Presidential Election. The Forum 2 (4): Article 4. www.bepress.com/forum/ vol2/iss4/art4. McKelvey, Richard. 1976. Intransitivities in Multidimensional Voting Models and Some Implications for Agenda Control. Journal of Economic Theory 12:472–82.

REFERENCES

183

Mendelsohn, Matthew, and F. Cutler. 2000. The Effect of Referenda on Democratic Citizens: Information, Politicization, Efficacy and Tolerance. British Journal of Political Science 30:669–98. Miller, Joanne M., and Jon A. Krosnick. 2000. News Media Impact on the Ingredients of Presidential Evaluations: Politically Knowledgeable Citizens Are Guided by a Trusted Source. American Journal of Political Science 44 (2): 301–15. Miller, Kenneth P. 1999. The Role of Courts in the Initiative Process: A Search for Standards. Paper presented at the annual meeting of the American Political Science Association, Atlanta, GA, September 2–5. Miller, Nicholas R. 1980. A New Solution Set for Tournaments and Majority Voting: Further Graph-Theoretical Approaches to the Theory of Voting. American Journal of Political Science 24 (1): 68–96. ———. 1983. Pluralism and Social Choice. American Political Science Review 77 (3): 734–47. Mooney, Christopher, ed. 2001. The Public Clash of Private Values: The Politics of Morality Policy. New York: Chatham House. Morelli, Massimo, and Michele Tertilt. 2000. Policy Stability under Different Electoral Systems. Working Paper No. 00–13, Ohio State University. NCPP&HE (National Center for Public Policy and Higher Education). 2002. Measuring Up. http://measuringup.highereducation.org. NCSL (National Conference on State Legislatures). 2001. Initiative and Referendum in the 21st Century: Final Report and Recommendations of the I&R Task Force. Denver, CO: National Conference on State Legislatures. http://www.ncsl.org/ programs/legman/irtaskfc/IandR_report.pdf. Nicholson, Stephen P. 2003. The Political Environment and Ballot Proposition Awareness. American Journal of Political Science 47:403–10. ———. 2005. Voting the Agenda: Candidates, Elections, and Ballot Propositions. Princeton, NJ: Princeton University Press. Nie, Norman, and K. Andersen. 1974. Mass Belief System Revisited: Political Change and Attitude Structure. Journal of Politics 36:541–91. Nie, Norman, Sidney Verba, and J. Petrocik. 1979. The Changing American Voter. Cambridge, MA: Harvard University Press. Norris, Pippa, ed. 1999. Critical Citizens Global Support for Democratic Government. Oxford: Oxford University Press. Oberholtzer, Ellis Paxson. 1912. The Referendum in America, rev. ed. New York: Charles Scribner’s Sons. O’Connell, Sue. 2006. The Money behind the Marriage Amendments. Helena, MT: Institute on Money in State Politics. www.followthemoney.org/press/Reports/ 200601241.pdf. Oliver, Eric, and Tali Mendelberg. 2000. Reconsidering the Environmental Determinants of White Racial Attitudes. American Journal of Political Science 44:574–89. Pantoja, Adrian D., and Gary M. Segura. 2003. Fear and Loathing in California: Contextual Threat and Political Sophistication among Latino Voters. Political Behavior 25 (3): 265–86.

184

REFERENCES

Peltzman, Sam. 1990. How Efficient Is the Voting Market? Journal of Law and Economics 33 (April): 27–63. ———. 1992. Voters as Fiscal Conservatives. Quarterly Journal of Economics, 107 (2): 327–61. ———. 1998. Political Participation and Government Regulation. Chicago: University of Chicago Press. Pippen, John, Shaun Bowler, and Todd Donovan. 2002. Election Reform and Direct Democracy: Campaign Finance Regulations in the American States. American Politics Research 30 (6): 559–82. Plott, Charles. 1967. A Notion of Equilibrium under Majority Rule. American Economic Review 57:787–806. Popkin, Samuel. 1991. The Reasoning Voter. Chicago: University of Chicago Press. Putnam, Robert D. 1966. Political Attitudes and the Local Community. American Political Science Review 60:640–54. ———. 2000. Bowling Alone: The Collapse and Revival of American Community. New York: Simon and Schuster. Putnam, Robert, Robert Leonardi, and Raffaella Nanetti. 1993. Making Democracy Work. Princeton, NJ: Princeton University Press. Reichley, A. James. 1986. Religion and the Future of American Politics. Political Science Quarterly 101:23–47. Riker, William H. 1982 Liberalism against Populism: A Confrontation between the Theory of Democracy and the Theory of Social Choice. San Francisco : W.H. Freeman. Romer, Thomas, and Howard Rosenthal. 1978. Political Resource Allocation, Controlled Agendas, and the Status Quo. Public Choice 33:27–43. ———. 1979. Bureaucrats versus Voters: On the Political Economy of Resource Allocation by Direct Democracy. Quarterly Journal of Economics 93:563–87. Rozell, Mark J., and Clyde Wilcox. 1996. Second Coming: The Strategies of the New Christian Right. Political Science Quarterly 111:271–94. Scarrow, Susan. 2004. Making Elections More Direct? Reducing the Role of Parties in Elections. In Democracy Transformed? Expanding Political Opportunities in Advanced Industrial Democracies ed. Cain Bruce, Russell Dalton, and Susan Scarrow. Oxford: Oxford University Press. Schattschneider, E. E. 1960. The Semi-Sovereign People. New York: Holtz, Rinehart and Winston. Schlozman, Kay Lehman. 1984. What Accent the Heavenly Chorus? Political Equality and the American Pressure System. Journal of Politics 46:1006–32. Schlozman, Kay Lehman, and John T. Tierney. 1986. Organized Interests and American Democracy. New York: Harper and Row. Schofield, Norman. 1978. Instability of Simple Dynamic Games. Review of Economic Studies 45 (3): 557–94. Schrag, Peter. 1998. Paradise Lost: California’s Experience, America’s Future. New York: New Press. ———. 2003. Vote Early, Vote Often—but Vote on Everything? Sacramento Bee, October 15.

REFERENCES

185

Shah, Darron, N. Kwak, and R. Holbert. 2001. “Connecting” and “Disconnecting” with Civic Life: Patterns of Internet Use and the Production of Social Capital. Political Communication 18:141–62. Shapiro, Walter. 2004. Presidential Election May Have Hinged on One Issue: Issue 1. USA Today, November 4, A9. Sharkansky, Ira. 1969. The Utility of Elazar’s Political Culture: A Research Note. Polity 2 (Fall): 66–83. Shepsle, K. 1979. Institutional Arrangements and Equilibrium in Multidimensional Voting Models. American Journal of Political Science 23:27–60. Shepsle, K., and B. Weingast. 1981. Structure-Induced Equilibrium and Legislative Choice. Public Choice 37:503–19. Shibley, Mark A. 1998. Contemporary Evangelicals: Born-Again and World Affirming. Annals of American Academy of Political and Social Science (July): 67–87. Smidt, Corwin. 1988. Evangelicals within Contemporary American Politics: Differentiating between Fundamentalist and Non-Fundamentalist Evangelicals. Western Political Quarterly 41:601–20. Smith, Daniel A. 2006. Initiatives and Referendums: The Effects of Direct Democracy on Candidate Elections. In The Electoral Challenge: Theory Meets Practice, ed. Steven Craig. Washington, D.C.: CQ Press. Smith, Daniel A., Matthew DeSantis, and Jason Kassel. 2006. Same-Sex Marriage Ballot Measures and the 2004 Presidential Election. State and Local Government Review 38 (2). Smith, Daniel A., and Caroline J. Tolbert. 2001. The Initiative to Party: Partisanship and Ballot Initiatives in California. Party Politics 7:781–99. ———. 2004. Educated by Initiative: The Effects of Direct Democracy on Citizens and Political Organizations in the American States. Ann Arbor: University of Michigan Press. Smith, Eric. 1989. The Unchanging American Voter. Berkeley: University of California Press. Smith, Mark A. 2001. The Contingent Effects of Ballot Initiatives and Candidate Races on Turnout. American Journal of Political Science 45:700–6. ———. 2002. Ballot Initiatives and the Democratic Citizen. Journal of Politics 64:892–903. Sniderman, Paul, Richard Brody, and P. Tetlock. 1991. Reasoning and Choice: Explorations in Political Psychology. New York: Cambridge University Press. Sorauf, Frank. 1976. Party Politics in America. New York: Little Brown. Steel, Brent S., and Nicholas Lovrich. 1998. Determinants of Public Support for Tax and Expenditure Initiatives. Social Science Journal 35:213–29. Steensland, Brian, Jerry Z. Park, Mark D. Regnerus, Lynn D. Robinson, W. Bradford Wilcox, and Robert D. Woodberry. 2000. The Measure of American Religion: Toward Improving the State of the Art. Social Forces 79:291–318. Stein, Robert M. 1990. Economic Voting for Governor and U.S. Senator: The Electoral Consequences of Federalism. Journal of Politics 52:29–53. Stigler, George J. 1972. Economic Competition and Political Competition. Public Choice 13 (Fall): 91–106.

186

REFERENCES

Teixeira, Ruy. 2004. Public Opinion Watch. Center for American Progress, November 10. www.americanprogress.org/site/pp.asp?c⫽biJRJ8OVF&b⫽246518. Tolbert, Caroline. 1998. Changing Rules for State Legislatures: Direct Democracy and Governance Policies. In Citizens as Legislators, ed. Bowler, Donovan, and Tolbert. Columbus: Ohio State University Press. Tolbert, Caroline, John Grummel, and Daniel A. Smith. 2001. The Effect of Ballot Initiatives on Voter Turnout in the American States. American Politics Research 29:625–48. Tolbert, Caroline J., and Rodney Hero. 2001. Facing Diversity: Racial/Ethnic Context and Social Policy Change. Political Research Quarterly 54:571–604. Tolbert, Caroline, and Ramona McNeal. 2003. Unraveling the Effects of the Internet on Political Participation. Political Research Quarterly 56:175–85. Tolbert, Caroline J., Ramona McNeal, and Daniel Smith. 2003. Enhancing Civic Engagement: The Effects of Direct Democracy on Political Participation and Knowledge. State Politics and Policy Quarterly 3:23–41. Tolbert, Caroline, and Daniel A. Smith. 2005. The Educative Effects of Ballot Initiatives on Voter Turnout. American Politics Research 33:283–309. Tsebelis, George. 1995. Decision Making in Political Systems: Veto Players in Presidentialism, Parliamentalism, Multicameralism, and Multipartyism. British Journal of Political Science 25:289–326. ———. 1999. Veto Players and Law Production in Parliamentary Democracies: An Empirical Analysis. American Political Science Review 93:591–608. ———. 2002. Veto Players: How Political Institutions Work. Princeton, NJ: Princeton University Press. Tsebelis, George, and Eric C. C. Chang. 2004. Veto Players and the Structure of Budgets in Advanced Industrial Countries. European Journal of Political Research 43:449–76. Uslaner, Eric. 2004. Trust, Civic Engagement, and the Internet. Political Communications 21 (2): 223–42. Verba, Sidney, Kay Schlozman, and Henry Brady. 1995. Voice and Inequality: Civic Voluntarism in American Politics. Cambridge, MA: Harvard University Press. Walker, Jack L., Jr. 1991. Mobilizing Interest Groups in America: Patrons, Professionals and Social Movements. Ann Arbor: University of Michigan Press. Wattenberg, Martin. 1996. The Decline of Political Parties. Cambridge, MA: Harvard University Press. Weaver, David, and D. Drew. 1993. Voter Learning in the 1990 Off-Year Election: Did the Media Matter? Journalism Quarterly 70:356–68. Wenzel, James, Todd Donovan, and Shaun Bowler. 1998. Direct Democracy and Minorities: Changing Attitudes about Minorities Targeted by Initiatives. In Citizens as Legislators, ed. Shaun Bowler, Ted Donovan, and Caroline Tolbert. Columbus: Ohio State University Press. West, Darrell. 1997. Air Wars: Television Advertising in Election Campaigns, 1952–1996. Washington, D.C.: CQ Press. Wilcox, Clyde. 1989. The New Christian Right and the Mobilization of the Evangelicals. In Religion and Political Behavior in the United States, ed. Ted G. Jelen. New York: Praeger.

REFERENCES

187

———. 1992. God’s Warriors. Baltimore, MD: Johns Hopkins University Press. ———. 2000. Onward Christian Soldiers: The Religious Right in American Politics. Washington, D.C.: Westview. Williams, Vanessa. 2004. Ohio Gay Marriage Initiative Roils Skeptics: Some See Conservatives Pushing Bans in Close States to Spur Turnout for Bush. Washington Post, May 10, A5. Witt, Stephanie L., and Suzanne McCorkel. 1997. Anti–Gay Rights: Assessing Voter Initiatives. Westport, CT: Praeger. Wolfinger, Raymond E., and Steven J. Rosenstone. 1980. Who Votes? New Haven, CT: Yale University Press. Zaller, John R. 1992. The Nature and Origins of Mass Opinion. Cambridge: Cambridge University Press. Zhu, Jhian-Hua, and William Boroson. 1997. Susceptibility to Agenda Setting: A Cross-Sectional and Longitudinal Analysis of Individual Differences. In Communication and Democracy: Exploring the Intellectual Frontiers in AgendaSetting Theory, ed. Maxwell McCombs, Donald L. Shaw, and David Weaver, 69–83. Mahwah, NJ: Erlbaum.

This page intentionally left blank

Further Reading

Alvarez, R. Michael, and L. Bedolla. 2004. The Revolution against Affirmative Action in California: Racism, Economics, and Proposition 209. State Politics and Policy Quarterly 4:1–17. Alvarez, R. Michael, and Tara Butterfield. 2000. The Resurgence of Nativism in California? The Case of Proposition 187 and Illegal Immigration. Social Science Quarterly 81:167–79. Baumgartner, Frank R., and Beth L. Leech. 1999. Studying Interest Groups Using Lobby Disclosure Reports. VOX POP Newsletter of the Political Organizations and Parties Section of the APSA 18:1–3. Berman, David. 1997. State and Local Politics. Armonk, NY: ME Sharpe. Bowler, Shaun, Todd Donovan, and Jeffrey Karp. 2002. When Might Institutions Change? Elite Support for Direct Democracy in Three Nations. Political Research Quarterly 55 (4): 731–54. ———. 2004. Preferences over Democratic Institutions: A Politician’s View. Paper presented at the meeting of the Western Political Science Association, Portland, Oregon. Broder, David S. 1970. The Party’s Over. Harper & Row: New York. Budge, Ian. 1996. The New Challenge of Direct Democracy. Cambridge: Polity. Donovan, Todd, and Shaun Bowler. 1997. Direct Democracy and Minority Rights: Opinions on Anti–Gay and Lesbian Ballot Initiatives. In Anti–Gay Rights: Assessing Voter Initiatives, ed. Stephanie Witt and Suzanne McCorkle. Westport, CT: Praeger. Donovan, Todd, and Daniel Smith. 2004. Tuning in and Turning out. Paper presented at the State Politics and Policy Conference, Kent State University, Ohio. Gordon, Tracy M. 2004. The Local Initiative in California. San Francisco, CA: Public Policy Institute of California. Hill, Kim, and Jan Leighley. 1999. Racial Diversity, Voter Turnout, and Mobilizing Institutions in the United States. American Politics Quarterly 27:275–95. Krebhiel, Keith. 1987. Why Are Congressional Committees Powerful? American Political Science Review 81:929–35. ———. 1991. Information and Legislative Organization. Ann Arbor: University of Michigan Press. Matsusaka, John G. 2000. Fiscal Effects of the Voter Initiative in the First Half of the 20th Century. Journal of Law and Economics 43 (October): 619–48. Nownes, Anthony J., and Patricia Freeman. 1998. Interest Groups Activity in the States. Journal of Politics 60:86–112.

190

FURTHER READING

Piott, Steven. 2003. Giving Voters a Voice. Columbia: University of Missouri Press. Qvortrup, Mads. 2001. The Courts v. the People: An Essay on Judicial Review of Initiatives. In The Battle Over Citizen Law Making, ed. M. Dane Waters. Durham, NC: Carolina Academic Press. Rosenstone, S., and J. Hansen. 2003. Mobilization, Participation, and Democracy in America. New York: Longman. Saari, Donald, and Katri Sieberg. 2001. The Sum of the Parts Can Violate the Whole. American Political Science Review 95 (2): 415–34. Sabato, Larry J., Bruce A. Larson, and Howard R. Ernst, eds. 2001. Dangerous Democracy? Boulder, CO: Rowman and Littlefield. Scarrow, Susan. 2004. Making Elections More Direct? Reducing the Role of Parties in Elections. In Democracy Transformed? Expanding Political Opportunities in Advanced Industrial Democracies ed. Cain Bruce, Russell Dalton, and Susan Scarrow. Oxford: Oxford University Press. Shepsle, K., and B. Weingast. 1987. The Institutional Foundations of Committee Power. American Political Science Review 81:85–104. Tolbert, Caroline, Daniel H. Lowenstein, and Todd Donovan. 1998. Election Laws and Rules for Initiatives. In Citizens as Legislators, ed. Shaun Bowler, Todd Donovan, and Caroline Tolbert. Columbus: Ohio State University Press. Wald, Kenneth D., Dennis E. Owen, and Samuel S. Hill, Jr. 1990. Political Cohesion in Churches. Journal of Politics 52:197–215.

Index

abortion, 28, 40, 55, 57–8, 60, 137 Abramowitz, Alan, 22, 58 Abramson, Paul, 51 administrative reform, 71 affirmative action, 40, 55, 57 African-American, 27, 29–30 agenda, electoral, 55–61, 66, 65, 67–8 agenda, public, See agenda, electoral agenda setting, 3, 4, 56–9, 61–7 Alabama, 141 Alaska, 134n6, 141–6 alcohol regulation, 4, 5 Aldrich, John, 51 amendment, constitutional, 3–4, 17, 23–4, 40–1, 58, 81, 120 American National Election Study (NES), 37, 41, 51, 75–6, 90 Anderson, Jennifer, 95 Anderson, K., 38 animal welfare, 4–5 Anscombe, Gertrude, 139 anticorruption, 70, 89 Appleton, Andrew, 74, 91 apportionment, 18, 21, 71–2, 118 APSA, 10 Arizona, 4–5, 70, 91, 134n6 Arizona 202 (voter registration), 71 Arkansas, 4, 33n1, 70, 91, 134n6, 141 Asimov, Nanette, 61 Atkeson, Lonna Rae, 60 Baldassare, Mark, 37 Bali, V., 152

ballot measures, Arizona 202 (voter registration), 71 California 3 (partisan presidential Primaries), 64 California 5 (tribal casinos), 64 California 8 (public school reform), 56, 60–7 California 9 (electric utilities), 64 California 10 (tobacco tax), 64 California 13 (tax/spending limits), 5, 13, 19, 86, 152 California 98 (education spending), 5, 13, 152 California 187 (illegal immigration), 57 California 198 (blanket primary), 83, 86 Maine 3 (campaign finance reform), 71 North Dakota 5 (term limits), 71 Ohio 1 (same-sex marriage ban), 22, 24–32, 34 Ohio 2 (redistricting), 71 Oklahoma 556 (redistricting), 71 Oklahoma 627 (ethics commission), 71 Oregon 2 (direct primary), 71 Oregon 5 (tax/spending limits), 12 Oregon 6 (campaign finance), 71 Oregon 19 (creation of a county), 71 Utah A (term limits), 71 Washington 134 (campaign finance), 84 Washington 601 (tax/spending limits), 12

192

INDEX

ballot measures (Continued) Washington 685 (medical marijuana), 87 Washington 872 (blanket primary), 87 Banducci, Susan, 81 Banks, Jeffrey, S., 134n11 Baqir, Reza, 134n4 Barnett, James, 70 Barrett, Katherine, 160–1, 166n8 Barro, Robert, J., 124, 126 Baumgartner, Frank, 94–6, 112n7 Bawn, Kathleen, 140 Belli, Robert, 24 Benz, M., 37 Berelson, Bernard R., 68 Bernholz, Peter, 138 Besley, Timothy, 128 Bibby, John, 77 Bimber, Bruce, 43 Birch, Anthony, 12 Black, Duncan, 138 Boehmke, Fred, 15, 35, 52n2, 55, 94–6, 112n1, 112n4, 112n5 Boroson, William, 61 Bowler, Shaun, 7–8, 15, 19, 27–8, 30 35, 37, 39–40, 51, 52n2, 52n5, 55, 57–8, 67, 75–7, 81, 91, 112n1, 134n1, 149–51, 153, 166 Boylan, Richard T., 74, 90 Bradbury, John Charles, 134n4 Brady, Henry, 51 Branton, Regina, 23, 25, 27, 30 Braunstein, Rich, 149 Broder, David, 69, 150–1 Brody, Richard, 40 Brown, Kathleen, 57 Browne, William P., 96 Burden, Barry C., 40, 58 Bush, George W., 10, 21–34 Button, James W., 26–7, 30 Cain, Bruce, 84, 149 Cain, W. Mark, 134n4

California, 3–5, 7, 9–10, 12–14, 42, 48, 56–8, 60–3, 70, 79, 82–87, 91, 92n7, 115–16, 119, 122, 134n6, 138, 141, 152, 154–6, 162, 165, 166n1 Brown, Kathleen, 57 California 3 (partisan presidential Primaries), 64 California 5 (tribal casinos), 64 California 8 (public school reform), 56, 60–7 California 9 (electric utilities), 64 California 10 (tobacco tax), 64 California 13 (tax/spending limits), 5, 13, 19, 86, 152 California 98 (education spending), 5, 13, 152 California 187 (illegal immigration), 57 California 198 (blanket primary), 83, 86 Davis, Gray, 3 Feinstein, Dianne, 57 Huffington, Michael, 57 Johnson, Hiram, 115 Reagan, Ronald, 86 Schwarzenegger, Arnold, 3, 10, 86, 116, 119 Wilson, Pete, 15, 57, 86 California Democratic Party et al. v. Jones et al. 2000, See ballot measures, CA 198 California Field Poll, 60–6, 68 Camobreco, J., 153 campaign finance, 71, 84, 95 campaign finance, reform, 71–2, 76–7, 91n2, 149 campaigning, negative, 39, 51 Campbell, Angus, 36, 40, 41, 50 Campbell, David E., 23, 58 Carmines, Edward G., 28 Catholic, 22, 25–6, 29, 33 Center for the Study of Democracy, 14, 32, 166n1 Chaffee, S., 38 Chang, Eric C., 140 Chavez, Lydia, 40

INDEX

Christian Right, 27 coalitions (winning), 137, 139–40, 147 Coate, Stephen, 128 Colorado, 5, 70, 91, 122, 134n6 Connecticut, 141 constitution, state, 35 Converse, Phillip, 40, 65 Cooperman, Alan, 22, 24 Corrupt Practices Act. See anticorruption Cotter, Cornelius, 77 Cronin, Thomas, 134n1, 150–1 Crouch, Winston W., 134n1 cues, 39, 51, 85–7, 92 cultural issues/values. See moral issues/values Culture and Family Institute, 22 Cummings, Jeanne, 22 Cutler, F., 37, 52n2 cycling, 139 Dahl, Robert A., 67 Dalton, Russell J., 7, 10, 16, 70, 73, 91, 149 Dao, James, 22 Davies, Frank, 21 Davis, Gray, 3 de Figueiredo, Jr., Rui J. P., 119–20, 122 Defense of Marriage Act (DOMA), 23–4 Delli Carpini, Michael, 38, 45 democracy, direct constitutional provisions for, 18, 75 impact on campaign strategies, 6, 10, 15 candidates, 10, 115 elections, 124–128 engagement, 43, 48, 50 executive branch, 115–135 expenditures, 129–31 fiscal policy, 141–6, 159–61 governability, 1, 11–12, 18 governance, 149–66 institutions, 2, 8, 17, 21 interest groups, 10 participation, 1, 35–51

193

parties, 10, 21, 69–89 policy bundling, 1, 138 policy stability, 137–47 policymaking, 23 political actors, 23 public discourse, 11, 56–7, 59–60, 67 strength of political parties, 2, 69, 75, 77, 84 tax and spending levels, 144–7 voters, 8 democracy, hybrid, 93–4, 112 democracy, representative, 69 Democratic candidates, 58 Democratic Farm Labor Party, 91 Democratic Party, 22, 27, 59, 61, 77, 83–5, 119 Democratic Populist Party, 91 Democratic vote, 10, 167n13 Democratic voters, 80, 162 DeSantis, Matthew, 39 Di Tella, Rafael, 124 direct and representative democracy, inter-relationship between, 1 Donahue, Amy K., 162 Donovan, Todd, 7–8, 10, 15, 19, 23, 27–8, 30–2, 35, 37, 39–40, 51, 52n2, 52n5, 55, 67, 75–7, 81, 84, 91, 112n1, 134n1, 150–1, 153, 166 Drew, D., 38 Dyke, Joshua, 37 Eckl, Corina, 141 ecological fallacy, 24 Edsall, Thomas, 22, 24 efficacy, 37, 41, 52–3 El Nasser, Haya, 22 elections, gubernatorial, 25, 124, 128–32 elections, midterm, 35–9, 41, 43–6, 48, 50–3 elections, presidential, 15, 24, 35–8, 40, 46, 48–51 elections, primary, 4 blanket, 83–4, 86–7 direct, 70–1, 149 open, 89

194

INDEX

electoral reform, 71, 84 elites, 138, 147, 151 Ellis, Richard, 151 engagement, civic, 15, 23, 37–9, 41–3, 48, 50–1, 133 Erikson, Robert S., 122 Ernst, Howard, 112n1 European Union, 3 evangelical, 22, 24–33 Evans, Arthur, 25 event count model, 107, 113n12 experimental methods, 36, 59 Family Research Council, 22 Fearon, James D., 134n11 Feinstein, Dianne, 57 Fellowes, Matthew, 95 Ferejohn, John, 124 Fernandez, Ken, 39–40, 112n1 Fisman, Raymond, 124 Florida, 91, 122, 134n6 Foladare, Irving, 25 Ford, Henry Jones, 11, 70 Fordice, Kirk, 115 framing, 15 Franz, M., 36, 39–40, 43, 45, 51 Freedman, Paul, 23, 36, 39–40, 43, 45, 51 Freeman, John, 139 Gailmard, Sean, 65–6 gambling, 60 Garrett, Elizabeth, 1, 56 Geer, John G., 64 Gelman, Andrew, 68 Georgia, 33n1 Gerber, Elisabeth, 6, 8–9, 12, 35, 40, 52n2, 52n5, 55, 81, 83, 94, 112n1, 116, 134n1, 134n3, 150–3 Gibson, James, 77 Giles, Michael, 24–5 Gilligan, Thomas W., 118, 134n4 Glazer, Amihai, 16, 55, 149 Goldstein, K., 36, 39–40, 43, 45, 51 GOP, See Republican Party

government performance, measurements of, 153–4 Government Performance Project (GPP), 153 Gray, Virginia, 95–6, 112n2 Green, John, 28–9, 33 Greene, Richard, 160–1, 166n8 Grilli, Vittorio, 140 Grooters, Jennifer, 141 group membership, 106, 108, 111 Grummel, John, 23, 36 “gun behind the door,” 4, 73 Guth, James, 33 Haider-Markel, Donald, 28, 40 Hajnal, Zolton L., 8 Hamm, Keith, 77 Hammond, Thomas H., 138 Hasen, Rick, 69, 84, 92 Hawaii, 141 Heinz, 112n3 Hero, Rodney, 24, 24–5, 37, 52n2 Hillygus, Sunshine D., 23, 58 Hofstadter, Richard, 70 Holbert, R., 39 Holmstrom, Bengt, 126 Honaker, James, 108 Houser, Daniel, 139 Huckfeldt, Robert 25 Huckshorn, Robert, 77 Huffington, Michael, 57 hybrid democracy, 1, 2, 4, 14 Idaho, 84, 134n6 ideal point, 116–17, 119, 134n2 identification, party, 58–9, 61, 63 Illinois, 14, 75, 81, 134n6 immigration, 7, 10, 55, 57, 119 incumbency, 7, 25 inequality, economic, 163 information, political, 38–9, 51 Ingraham, Patricia W., 162 initiative constitutional, 120–1, 123, 132, 134n6, 141

INDEX

constitutional constraints on, 138, 152 legal challenges to, 152 threat effect of, 118 Initiative and Referendum Institute, 14, 19n1, 149, 156–8, 161, 166n4 initiative vs. noninitiative states, 100, 110–12, 120–33, 156, 166n5 interest group, 13, 39–40, 51–3, 56 interview, 112n4, 154 issue area, 95, 107, 109, 111 issue salience, 56, 58, 61, 67 Iyengar, Shanto, 38, 59, 61, 65 Jackman, Simon, 23 Jelen, Ted G., 28–9, 32–3 Jennings, M. Kent, 39 Johnson, Hiram, 115 Johnston, R., 151 Jones, Dale, 33 Joseph, Anne, 108 Joyce, Phillip G., 162 Kanihan, S., 38 Karp, Jeffrey A., 39 Kassel, Jason, 39 Keeter, Scott, 38, 45 Kellstedt, Lyman, 33 Kentucky, 33n1 Kerry, John, 22 Key, V. O., 24, 134n1 Kiewiet, D. Roderik, 83, 138 Kinder, Donald R., 38, 59, 61, 65 King, Gary, 24, 68, 107–8, 113n12 Knack, Stephen, 162–3 Knight, Robert, 22 König, Thomas, 140 Korte, Gregory, 22 Kousser, Thad, 84 Krosnick, Jon A., 65 Krueger, Brian, 43 Kwak, N., 39 La Pira, Timothy, 95 labor unions, 9, 104, 109

195

Lacey, Robert, 35–6, 47 Lacy, Dean, 139 LaRaja, Raymond, 77 Lassen, David, 36 Laumann, Robert O., 112n3 Layman, Geoffrey C., 28, 33 Lazarsfeld, Paul F., 68 Leech, Beth L., 94–6, 112n7 legislative agenda, 95–6, 100, 103, 105–7, 111 Leonardi, Robert, 155 Lewis, Gregory B., 23 Linder, Wolf, 139 lobbying, 15, 94–98, 100–01, 110–11, 112n9 Lochhead, Carolyn, 22 logrolling, 137, 139, 141, 147 Long, Cheryl X., 74, 90 Louch, Hugh, 8 Louisiana, 91 Louvrich, Nicholas, 12, 163 Lowery, David, 95–6, 112n2 Lupia, Arthur, 23, 36, 39, 83, 119, 134n3, 151 Luskin, Robert, 38 MacKuen, Michael, 65 Magleby, David B., 35, 56, 68, 134n1, 152 Maine, 70–1, 91, 122, 134n6, 141 Maine 3 (campaign finance reform), 71 Mariani, Michele, 160–1, 166n8 marijuana, medical, 57, 86–7 marriage, same-sex, 10, 15, 21–30, 32–3, 35, 40–1, 56–7 Masciandaro, Donato, 140 Massachusetts, 84, 134n6, 141 Matsusaka, John, 13, 16, 35, 55, 73, 111, 116, 118–20, 122, 124, 129, 131–2, 134n1, 134n3, 134n4, 142, 147, 149, 151, 153, 160–1 Mayhew, David, 74, 91 McCann, James, 32 McCarty, Nolan M., 116, 134n3

196

INDEX

McCombs, Maxwell, 59 McCorkel, Suzanne, 28 McCuan, David, 39–40, 112n1 McCubbins, Mathew, 36, 83, 138 McDonald, Michael P., 58 McGann, Tony, 16, 55 McGonagle, Katherine A., 24 McIver, John P., 122 McKelvey, Richard, 138 McNeal, Roberta, 43, 52n2 McPhee, William N., 68 Measure Up project, 157–8 Meier, Kenneth, 28 member support, 97, 99, 101, 105 membership group, 94 Mendelberg, Tali, 24 Mendelsohn, Matthew, 37, 52n2 Michigan, 33n1, 91, 122, 134n6 Milgrom, Paul, 126 Miller, Gary J., 138 Miller, Joanne M., 65 Miller, Kenneth, 83 Miller, Nicholas, 8, 138 Miller, Warren, 40 minorities, oppression of, 11 minority rights, 11 Mississippi, 14, 33n1, 42, 75, 81, 115, 134n6 Fordice, Kirk, 115 Missouri, 91 mobilization cognitive, 7 of interest groups, 93–5, 100, 111 of voters, 23–4, 27–8, 32 Moncrief, Gary, 77 Monson, J. Quin, 23, 58 Montana, 33n1, 91, 122, 134n6 Mooney, Christopher, 40 moral issues/values, 22–3, 26, 28 Morelli, Massimo, 139 multidimensional issue space. See policy bundle Nanetti, Raffaella, 155 National Conference of State Legislatures (NCSL), 5, 19n1, 52

Nebraska, 91, 134n6, 141 Nelson, Robert L., 112n3 Nevada, 91 Newing, R.A., 138 Nicholson, Stephen P., 7, 10, 15, 23, 32, 35–9, 41, 43, 55–8, 59, 84 Nie, Norman, 38 Niou, Emerson, 139 Norris, Pippa, 7 North Dakota, 33n1, 48, 91n5, 134n6 North Dakota 5 (term limits), 71 O’Connell, Sue, 24 Oberholtzer, Ellis Paxson, 134n1 OECD, 140 Ohio, 5, 10, 15, 21–33, 91, 122, 134n6 Ohio 1 (same-sex marriage ban), 22, 24–32, 34 Ohio 2 (redistricting), 71 Ohio Campaign to Protect Marriage, 24, 26 Oklahoma, 33n1, 70, 91, 134n6 Oklahoma 556 (redistricting), 71 Oklahoma 627 (ethics commission), 71 Oliver, Eric, 24 omitted variable bias, 53 Oregon, 4–5, 12, 14, 33n1, 92n9, 134n6, 152 Oregon 2 (direct primary), 71 Oregon 5 (tax/spending limits), 12 Oregon 6 (campaign finance), 71 Oregon 19 (creation of a county), 71 organizational type, 103, 106–7 overdispersion, 107–8, 113n12 Pantoja, Adrian D., 55, 60 Park, Jerry Z., 33 participatory democracy, 41 parties. See political parties Partin, Randall W., 60 partisanship, 14, 23, 25, 27, 28, 30, 32, 52n13, 58, 84 party behavior, 19

INDEX

party government, 16 party labels, 10 party organization, 19 Patty, John W., 95–6 Peltzman, Sam., 128–9, 131, 135n12 Perkins, Tony, 21 Perry, Janine, 58 Petrocik, J., 38 Pew Research Center for the People and the Press, 46, 49, 52–3 Pierce, John, 163 Pippen, John, 76–7 Plott, Charles, 138 pluralism, 8–9 plurality voting, 139 Plutzer, Eric, 25 polarization, 163 policy bundle, 138 policymaking, constraints on, 21–22 political participation, 11 political party (or parties), 1, 39, 51, 53, 56 Popkin, Samuel, 36 population, 162–3 Populist, 4, 74, 81, 88, 132, 158 preferences, candidates’, 125, 128 incumbent, 125 interest groups’, 95 party, 140 policy, 12, 141 public, 149, 151, 159, 162 single-peaked, 116, 138 voters’, 11, 36, 68, 111, 125, 132, 134n3, 137, 159 Princeton Research Associates, 52 principal components analysis, 166 principal-agent theory, 124–8, 134n11, 151 probability simulation, 45, 50 progressive movement, 4, 6, 26, 41 Prohibition, 4 proportional representation, 139 public policy, 55 public schools, 56 Putnam, Robert, D., 25, 51, 155, 162–3

197

Reagan, Ronald, 86 Regnerus, Mark D., 33 regression bivariate, 74, 76 instrumental variables, 36, 38, 42, 46–7, 50 logistic, 43–4, 49, 121 multivariate, 120, 124, 129–30, 132, 142–4, 156, 161, 164, 167n10 negative binomial, 107–8, 113n12 weighted least squares (WLS), 25, 33n3 Reichley, A. James, 28 Republican candidates, 22, 33n5, 57 Republican National Convention, 22 Republican Party, 10, 15, 27, 30, 58, 61, 85 Republican vote, 57, 167n13 Republican voters, 25, 30, 44, 58, 162 Rhode Island, 141 Rienzo, Barbara A., 26–7, 30 Riker, William H., 138–9 Robinson, Lynn D., 33 Romer, Thomas, 56 Rosenstone, Steven J., 40 Rosenthal, Howard, 56 Rozell, Mark J., 28–9 Salisbury, Robert L., 112n3 Scarrow, Susan, 149 Schattschneider, E. E., 94 Scheve, Kenneth, 108 Schlozman, Kay Lehman, 51, 94, 112n3 Schofield, Norman, 138 Schrag, Peter, 5, 69 Schwarzenegger, Arnold, 3, 10, 86, 116, 119 Segura, Gary M., 55, 57–8, 60 Semanko, Nicholas A., 95 Shah, Darron, 39 Shapiro, Walter, 22 Sharkansky, Ira, 78, 90 Shaw, Donald L., 59 Shepsle, K., 139 Shibley, Mark A., 29 Shields, Todd G., 23, 58

198

INDEX

Singh, Naunihal, 108 Smidt, Corwin, 28 Smith, Daniel, 8, 10, 15, 23, 27, 30–2, 35–7, 39, 41, 47–8, 52n2, 55, 58, 67, 69, 70, 84, 94, 133, 142, 163, 166, 166n11 Smith, Eric, 38 Smith, Mark A., 23, 35–7, 39, 43, 47, 55 Sniderman, Paul, 40 Sorauf, Frank, 73 Sostek, Anya, 160–1, 166n8 South Dakota, 5, 91, 91n5, 134n6 Sprague, John, 25 Steel, Brent S., 12 Steensland, Brian, 33 Stein, Robert M., 60 Stigler, George J., 134n9 Stokes, David, 40 Stutzer, A., 37 Sundaram, Rangarajan, 134n11 supermajoritarian, 138–9, 141, 144–5, 147, 152 Supreme Court, U.S., 83 survey instrument(s), 129, 154 survey, of interest groups, 96–100 Switzerland, 36–7 Tabellini, Guido, 140 tax and expenditure limit (TEL), 5, 12–3, 35, 73 tax and/or spending limits, 73 tax performance, 16, 133, 150–2, 159–61 tax, car, 119 labor, 140 progressive, 133 tobacco, 64 Teixeira, Ruy, 28 term limits, 5, 12, 18, 21, 71–4, 76, 83–4, 89, 91, 119–20, 122, 125, 132, 134n5, 142, 149 Tertilt, Michele, 139 Tetlock, P., 40

Tierney, John T., 112n3 Tolbert, Caroline, 8, 10, 15, 23–5, 27, 30, 32, 35–7, 41, 43, 47–8, 52n2, 55, 58, 67, 69, 72–3, 84, 94, 133, 134n1, 163, 166, 166n11 traditional party organization (TPO), 74 transfers, federal, 141–7 Traugott, Michael W., 24 Tröger, Vera, 140 Tsebelis, George, 13, 16, 138, 140 turnout, 11, 21–6, 28–33, 35–6, 38, 43, 46–8, 48, 50–1, 53 two-stage (estimation) model. See instrumental variables regression unemployment, 27–9, 31, 33–4 Uslaner, Eric, 39 Utah, 33n1, 84, 122, 134n6 Utah A (term limits), 71 Ventura, Jesse, 79 Verba, Sydney, 38, 51 veto, 3, 116, 118–19, 138, 140 line–item, 120–2, 132, 134n5, 141–7 veto players, 20, 140, 146 veto points, 137–8, 147 voter, median, 31, 116, 118–19, 138, 141, 147, 151 voters conservative, 10, 22, 66 latino, 59 minority, 8 peripheral, 36, 50 registered, 4, 27, 52n3, 60 well-educated, 66 voting behavior, 1, 6, 24 Wald, Kenneth D., 26–7, 30 Walker, Jr., Jack L., 94, 112n3 Ward, Daniel, 74, 91 Washington, 9, 12, 79–80, 84, 86–7, 92n7

INDEX

Washington 134 (campaign finance), 84 Washington 601 (tax/spending limits), 12 Washington 685 (medical marijuana), 87 Washington 872 (blanket primary), 87 Washington State Democratic Party et al. v. Reed 2002, 84 Wattenberg, Martin, 10, 73, 75, 91 Weaver, David, 38 Weingast, B., 139 Wenzel, James, 28, 40 West, Darrell, 39 Wilcox, Clyde, 27–9, 32–3

Wilcox, W. Bradford, 33 Williams, Vanessa, 23 Wilson, Pete, 15, 57, 86 Witmer, Richard C., 95–6 Witt, Stephanie L., 28 Wolfinger, Raymond E., 40 women’s suffrage, 4 Woodberry, Robert D., 33 Wright, Gerald C., 122 Wyoming, 84, 122, 134n6 Young, Margaret, 24 Zaller, John R., 51, 65 Zeitner, V., 39 Zhu, Jhian-Hua, 61

199