Book Search:  


Google full text of our books:


Voting the Agenda:
Candidates, Elections, and Ballot Propositions
Stephen P. Nicholson

Book Description | Reviews | Table of Contents

COPYRIGHT NOTICE: Published by Princeton University Press and copyrighted, © 2005, by Princeton University Press. All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher, except for reading and browsing via the World Wide Web. Users are not permitted to mount this file on any network servers. Follow links for Class Use and other Permissions. For more information, send e-mail to

This file is also available in Adobe Acrobat PDF format

Chapter 1


Pete Wilson was in trouble. With California smarting from an economic recession, the Republican governor's reelection prospects looked dim. More than a year before his 1994 reelection bid, Wilson had the lowest approval rating of any California governor in the history of the California Field Poll and trailed his likely Democratic opponent, Kathleen Brown, by more than twenty percentage points. Although a relative newcomer to California politics, Brown had some impressive qualities. She was the state's treasurer, an adroit fundraiser, and well-known among voters. In short, many believed Wilson's defeat a foregone conclusion.

Wilson wasn't the only one in trouble. Michael Huffington, the Republican nominee for U.S. Senate, was also a long-shot candidate that year but for different reasons. In contrast to Wilson, Huffington was unknown to most California voters. He had thin political credentials limited to a single term in the House and few legislative accomplishments. On the other hand, Huffington's opponent, Diane Feinstein, was a popular, well-recognized figure in California politics. Few pundits thought Feinstein would have any trouble defeating Huffington for the seat that she had easily won just two years prior (in a special election). Indeed, a little more than a year before the election, the California Field Poll showed Huffington trailing Feinstein by twenty-seven percentage points.

In an ending that few anticipated, both Republican candidates surpassed expectations. Wilson defeated Brown by a fourteen-point margin and Huffington, although falling short of defeating Feinstein, lost by a two-point margin. What helped these long-shot Republicans? In a word, the "agenda." In California's 1994 elections, the agenda included hot-button ballot initiatives, most notably Proposition 187, an initiative that sought to deny illegal immigrants public benefits. The initiative was the topic of extensive political debate in the media as well as in the candidate races, resulting in a majority of voters naming it the most important political event of the elections, including the Senate and governor's contests (Lubenow 1995, 124). Support for Proposition 187 was also high; roughly three out of five voters favored it throughout the campaign season. Since the parties had taken clear positions on the initiative--Republicans were for it and Democrats against it--Proposition 187 cast the election in terms favorable to Republican candidates. Thus, illegal immigration helped define the meaning of the election, boosting the electoral fortunes of Republican candidates, especially Wilson, who had made it the cornerstone of his campaign.

The story of Pete Wilson, Michael Huffington, and Proposition 187 is dramatic but not unique insofar as it illustrates how direct legislation may define the agenda. Although the frequency of direct legislation use varies across the states, in the 2002 elections, voters in forty states cast their ballots on 202 statewide ballot measures (Waters 2003, 8).1 The last twenty years have seen the heaviest use of initiatives (Waters 2003, 8), the types of direct legislation most often associated with politically charged issues. In the last few election cycles, for example, controversial issues such as affirmative action, animal rights, abortion, medicinal marijuana, school choice, gun control, same-sex marriage, physician-assisted suicide, taxes, health care reform, environmental reform, and bilingual education were the topics of ballot initiatives. Issues of this kind, the focus of my research, play an important role in shaping the agenda and thus the types of considerations voters make in judging candidates.

Despite the fact that ballot measures take place alongside candidate races, scholars seldom consider them together. The omission is curious given the aforementioned increase in the frequency of initiative and referendum use over the last twenty years, especially in light of how many ballot measures contain controversial issues, strong ideological overtones, and/or expensive campaigns. Of course, I do not expect all ballot measures to play an important role in candidate races since not all of them have strong partisan or ideological overtones. Nevertheless, a good number do fit this description; many interest groups involved in the initiative process turn to ballot measures precisely because legislators are unwilling to handle many controversial issues.

The study of agendas, of course, is broader than the role of direct legislation in candidate races. Although I focus on a single phenomenon, my goal is to illuminate the role of agendas in elections. The conclusions of this book, therefore, should be applicable to not only how ballot measures affect voting in candidate races but also how issues from one candidate race spill over into another. They should also speak to how issue ad campaigns waged by interest groups and political parties affect voting behavior (Magleby 2002).

I define agenda as the one or more policy issues that matter to voters in a given electoral environment. While actors raise multiple issues within a given electoral environment, typically only a few achieve agenda status. Thus, even though a good number of issues might be well-known, the agenda consists of the most salient issues. My principal argument is that agendas play a profound role in U.S. elections by establishing the criteria voters use in making candidate judgments. Agendas introduce common elements into voters' candidate judgments, and not just for a single office. Thus, agendas structure voting decisions along a common evaluative dimension and do so for offices that are as seemingly different from each other as U.S. Senate, House, and governor. By highlighting decision making at the constituency level, agendas also emphasize the diversity of issues that structure voting decisions across the United States. Taken together, I hope to elucidate how agendas unify voting decisions within a constituency and alter the foundations of candidate evaluation across constituencies.

It is curious that political scientists seldom consider voting for candidates in nonpresidential contests from the perspective of agendas. Instead, voting behavior research typically stresses the uniqueness of voting decisions by distinguishing voting decisions by office or issue on the ballot. The underdog candidates from the beginning of the chapter illustrate the shortcomings of isolating voting decisions. Not recognizing how voting decisions belong to the same family, conventional approaches would consider these contests apart, grouping the Wilson versus Brown race with other gubernatorial contests and the Huffington versus Feinstein race with other Senate contests. In so doing, analysts would likely stress the role of state issues, especially the state economy, in the gubernatorial races (e.g., Chubb 1988; Stein 1990) and national issues in the senatorial contests (e.g., Abramowitz and Segal 1992; Hibbing and Alford 1982). To be sure, these issues play an important role in voting for both types of offices and may even have the greatest overall effect. However, Wilson likely owed much of his reelection to the illegal immigration issue, and although Huffington lost by a slim margin, had illegal immigration not been on the ballot he probably would have lost by a larger margin.

Traditional approaches, therefore, exaggerate the individuality and disconnectedness of voting decisions across issues and candidates on the ballot. Consequently, scholars seldom consider whether direct legislation affects voting in candidate races or the broader question of spillover effects in which "outside" issues influence the criteria voters use to judge candidates. Although political scientists have learned much about elections and voting behavior from the methodological choice of studying elections in isolation from each other, it has privileged some explanatory variables at the expense of others. By using analytical frameworks that stress the distinctiveness of electoral choices, scholars lose sight of how voting decisions--up and down the ballot--belong to a single family by sharing a common understanding of politics centered on the information environment. To be clear, I am not arguing that scholars abandon the study of elections by office type. Rather, I am arguing that voting decisions across offices are not as unique as traditional approaches suggest. In the remainder of this chapter, I critically evaluate conventional approaches and how they elevate considerations that differentiate types of elections while obscuring the electoral information environment, especially the vital role of agendas.


Why study elections apart from each other? Why look at certain types of electoral information and disregard others? More specifically, why ignore issues that share space with candidates on the same ballot? To answer these questions, I begin with the simple insight that elections are busy and confusing marketplaces. In each election, candidates running for different offices try to flood the electorate with campaign messages. Candidates running for president, governor, Congress, and state assembly, to name a few, compete for scarce voter attention not only with their opponents in a given race but also with candidates running for different offices and not all that infrequently the campaigns for and against ballot measures. The electoral process thus generates a cacophony of messages sent by many actors--candidates, mass media, political parties, and interest groups--with the intention of reaching and persuading voters.

To make sense of voting decisions for candidates amid this chaos, analysts isolate different types of elections by office. Studies of electoral behavior treat each office, often implicitly, as self-contained and exempt from external influences.2 In much political science research, critics often direct their energies toward models that leave political processes "inside" the box unexamined, labeling such neglect as "black box" approaches. I take the opposite tack by arguing that students of elections often neglect to look "outside" the box at the broader information environment.

Studying elections and voting decisions by office type is typically done for the sake of unit homogeneity, which requires that cases be identical, or nearly so, for the sake of making causal inferences. The unstated assumption is that the act of voting is largely not comparable across different types of offices, so that organizing the study of voting behavior by office is necessary for comparability. But observe the implication of this assumption: the act of voting varies considerably across elected offices.3 From this vantage point, it is not surprising that voting behavior research orbits around difference rather than similarity.

The notion of unit homogeneity does not require units to be identical, however, but simply that they be comparable. As King, Keohane, and Verba (1994) note, it is never possible to achieve strict unit comparability because units of analysis always differ in some respect. Instead, "two units are homogenous when the expected values of the dependent variables from each unit are the same when our explanatory variable takes on a particular value" (91). Thus, the comparison of voting decisions across different elected offices simply requires that independent variables have a comparable effect on the dependent variable.

Studies of voting behavior provide considerable evidence that independent variables do have common effects across different offices. Few students of voting behavior, for example, dispute that partisan identification is the most important explanatory variable when looking at individual voting behavior across a variety of offices, especially low-information elections. Furthermore, few would disagree that variations in the effects of partisan identification are likely specific to the vagaries of electoral contests rather than the process of voting. Voters in low-information contests rely heavily on partisan cues (Schaffner and Streb 2002) while voters in information-rich contests will be somewhat less reliant on partisan cues given the abundance of other data. The insight here is that the mix of ingredients is the same; what differs is the amount. Gronke (2000) makes much the same point with reference to House and Senate elections when he argues that the "inputs" of voting for these offices are the same--the difference lies in the amount of a given input.

Nevertheless, many studies of voting behavior emphasize how institutional responsibilities (the job requirements of an elected office) and campaign-specific issues (information from a particular race) vary the foundations of candidate evaluation for different offices. If voters draw on institutional responsibilities, they compare the candidates' records against a set of criteria based on the responsibilities associated with a particular office. For example, the local orientation of the House of Representatives directs voters to evaluate House candidates by their ability to handle casework and bring federal projects to the district. The implication of this tendency is that voters use different criteria for evaluating candidates running for different offices.

On the other hand, if voters use campaign-specific issues as evaluative criteria, they focus on policy issues from a given contest.4 According to this approach, voters evaluate candidates in a given race based on the issues the candidates discuss in their campaigns or the issues most prevalent in media coverage of their contests. The implication of this approach is that voters only consider campaign issues (policy issues) relevant to a particular contest and screen out campaign messages from other contests. The picture that emerges from these approaches is one of voters pondering very different criteria as they move down the ballot, considering candidates for governor, senator, or representative. Although many scholars would balk at the notion that voters use subtle distinctions in making candidate judgments, the methodological choice of studying electoral behavior from the vantage point of the office or the flow of campaign information about a particular race nevertheless imposes this type of fine-tuning into the study of the voting decision.

Below, I address a seldom recognized but nonetheless real puzzle in voting research: If most voters pay little or no attention to politics, why do we assume that they correctly link numerous campaign messages with the appropriate candidates or assign certain institutional responsibilities to particular offices and vote accordingly? I argue that, in large part, voters do not make these distinctions and instead rely on the most salient morsels of information available. Often this information takes the form of partisan or incumbency cues, but it might also include prominent policy issues located within electoral environments such as states.

The Methodological Choice of Institutional Responsibilities

As mentioned, the institutional responsibilities approach implies that each office has its duties and requirements, and voters fine-tune their vote choices based on the particular demands of various political offices. Given this logic, the national orientation of the Senate requires voters to evaluate Senate candidates based on national issues such as the economy or presidential approval. Similarly, the Senate's reputation for handling international affairs requires voters to evaluate Senate candidates based on foreign policy. In contrast, because House members have a local orientation, pork barrel politics and constituency service play a large role in voters' decisions. Finally, because voters presumably understand the fine distinctions between federal and state offices, gubernatorial elections should turn on statewide considerations rather than national influences. These perceptions largely guide analysts in deciding what information matters in these races.

But scholars have found little evidence that citizens make meaningful institutional distinctions between voting for House and Senate candidates (Gronke 2000; Krasno 1994). Gronke (2000, chap. 6) found that voter evaluations of the job requirements of House members and senators do not turn on institutional distinctions. Instead of differentiating House members from senators by job requirements, voters ascribe the same duties to both offices, thus debunking the notion that national issues play a greater role in evaluations of senators and local issues play a greater role in evaluations of House members. Similarly, Stein and Bickers (1994) debunk the notion that House incumbents' reelection prospects are intimately tied to their ability to bring home pork barrel projects, a responsibility political scientists strongly associate with House members' institutional roles. Although this may be true for a small segment of voters, the authors find that "[m]ost members of the general public remain indifferent to alterations in the flow of new [pork barrel] rewards" (394). Thus, in many studies, the traditional wisdom that the institutional roles and responsibilities of elected officials shape mass political behavior often is not supported empirically.

Further, citizens' disregard of institutional distinctions between House and Senate extends beyond the electoral arena. For some time, scholars generally believed that House and Senate members have different representational roles because of the institutional differences between the two chambers, and that citizens appreciated the distinction.5 Scholars typically assumed that House members, given their role as the institution "where the people speak," have closer representational relationships with their constituents than do senators. Questioning this wisdom, Oppenheimer (1996) found that the distinction between House and Senate (the institutional setting) is much less important to constituent-representative relationships than state population size. Indeed, he found that senators from small population states have closer relationships with constituents than do House members.

Some scholars explicitly argue that voters use institutional criteria by differently evaluating senatorial and gubernatorial candidates according to federal-state responsibilities (Atkeson and Partin 1995, 1998; Stein 1990). According to Atkeson and Partin (1995, 1998), voters evaluate candidates running for Senate on national political concerns such as the economy and presidential approval while they evaluate candidates running for governor on matters such as the state economy.

On its face, the case for voters making state versus federal distinctions appears stronger than that for distinctions between the House and Senate. Indeed, the institutional distinction between the House and Senate, both federal offices, appears smaller as compared to that separating the federal and state governments. Yet even here the evidence is mixed. For example, although Stein (1990) and Carsey and Wright (1998) find that state economic conditions affect gubernatorial voting, they also find that national conditions play an important role. Moreover, some scholars have found that national conditions are the primary force behind gubernatorial elections and that state-specific factors play a secondary role (Chubb 1988; Kenney 1983; Peltzman 1987).

The case for voters making institutional distinctions is especially unconvincing examined in light of citizens' lack of political knowledge. The amount of political knowledge, not to mention willingness, required for voters to use institutional distinctions as criteria in voting decisions is contrary to research documenting citizens' minimal knowledge about American politics (e.g., Converse 1964; Delli Carpini and Keeter 1996).6 Delli Carpini and Keeter (1996) show that citizens lack knowledge about basic institutional features of the American political system, many seemingly easier than reasoning about institutional responsibilities for different elected offices. Many citizens, for example, show great ignorance of the president's institutional powers, believing he "has the power to adjourn Congress at his will (29 percent), to suspend the Constitution (49 percent), and to appoint judges to the federal court without the approval of the Senate (60 percent)" (99). Similarly, despite arguments to the contrary, citizens do not have any better grasp of federal versus state powers. The enterprise of distinguishing between federal and state responsibilities, of course, is inherently problematic given the extensive mixing of policy responsibilities. Given the ambiguity of federal-state relations, citizens do not fare well in assigning responsibilities on this basis (Delli Carpini and Keeter 1996, 99). While there is a tendency to assign some policy areas to the states or the national government, partisanship and ideology, explicitly political factors, largely account for these attributions (Schneider and Jacoby 2003). But even if citizens do know a lot about federal and state policy responsibilities, it is unlikely that they would apply this knowledge in evaluating candidates. Put differently, it is doubtful that institutional distinctions are chronically accessible considerations that guide voters' evaluations of candidates (see Arceneaux 2003).

Although the lack of importance citizens and voters ascribe to institutional distinctions are likely surprising to some scholars, they probably would not be surprising to most citizens. Few voters have the knowledge, or desire, to adjust their voting decisions in keeping with institutional responsibilities. When voters do appreciate institutional distinctions, it is likely more the product of the electoral information environment than their knowledge of American political institutions. Therefore, if institutional distinctions matter, they are likely dependent on candidates in high-intensity races, running campaigns that stress the job requirements of their offices. For example, if senatorial and gubernatorial candidates in hotly contested races have stressed issues consistent with institutional responsibilities, voters may draw on such distinctions. The "institution" effect, if it exists, is indirect rather than direct--information environments structure evaluative criteria; institutionally savvy voters do not impose different criteria for different offices. Rudolph (2003) reached the same conclusion in his study of citizen responsibility attributions and institutional context. Although Rudolph found that state institutional context (e.g., the amount of budgetary power given governors) had a significant effect on evaluations of governors, he did not attribute the effect to well-informed, institutionally savvy citizens. Rather, he believes this process likely operates through the flow of information about budgetary politics in states, thus making citizens act as though they were knowledgeable about state institutional structures (210).

The Methodological Choice of Campaign-Specific Issues

A second implication of studying elections by office type is that voters use a narrow range of campaign-specific issues sent by or communicated about the candidates. To test for campaign effects, analysts match a vote choice for a given election to the relevant issues discussed by a candidate or the media about that election. For example, if candidates discussed taxes and crime in their campaigns, scholars examine whether these issues influenced the voting decisions for those candidates. To be sure, campaigns do influence voting decisions. Under a variety of circumstances, scholars have found that Senate (Druckman 2004; Franklin 1991; Kahn and Kenney 1999; Goldstein and Freedman 2000), gubernatorial (Carsey 2000), and House campaign issues (Abbe et al. 2003) affect voter decision making.

However, for campaigns to "matter," voters must receive and accurately link campaign messages to the appropriate candidates. Thus, the less recognized but equally important key to understanding campaign effects from the vantage point of traditional approaches concerns whether voters correctly match issues to candidates. In presidential elections, where information about campaigns is relatively plentiful, fewer voters encounter difficulty recalling important issues and associating them with the presidential contest. Consequently, many scholars have found that presidential campaigns "matter" (Alvarez 1997; Finkel 1993; Holbrook 1996). Nevertheless, elections are bustling affairs involving many different issues floating around with most voters receiving and processing electoral information in a haphazard manner.

Given research on voter sophistication, including revisionist accounts that depict voters as making informed decisions with low-cost sources of information, it is unrealistic to expect voters to be able to make all, or even most, of the linkages between candidates and the issues found in their campaigns. Sometimes issues from the candidates' campaigns stick, but quite often they do not (Dalager 1996; Kahn and Kenney 1999). Kahn and Kenney (1999), for example, found that only 19 percent of voters asked to identify an issue from the Senate contest in their state could do so correctly. Although the authors found that the ability of voters to correctly identify an issue increases with the intensity of Senate campaigns, the fact that so many voters could not strongly suggests that scholars need to look elsewhere for the origins of voter information in low-information contests. When voters make little effort to become informed about elections, as do most voters, it is unlikely that they keep the issues and candidates carefully associated with each other.7

The ability of voters to correctly match issues and candidates depends on the volume (and perhaps clarity) of information available about a contest. For instance, voters choosing Senate candidates are more likely to be knowledgeable of them and use ideological and issue-based criteria in high-intensity campaigns (Kahn and Kenney 1999; Wright and Berkman 1986). Under certain conditions, especially high-spending campaigns, House candidates may influence voting decisions (Herrnson and Patterson 1998). Moreover, voters are most likely to use the campaign messages supplied by presidential candidates because these contests provide abundant information. As a rule of thumb, the more information available about a given contest, the greater the likelihood voters will use criteria from that contest to evaluate candidates.

Despite the fact that some Senate and gubernatorial races are information rich, few of these candidates dictate the terms of debate. Information about issues from these contests is simply less plentiful, and although the candidates from these races do their best to set the agenda, they may often fail in defining the "big" issues. Indeed, for this reason, I refer to all nonpresidential elections as low-information (while recognizing variation within this category). For down-ballot candidates running for House of Representatives or the state legislature the problem is much worse. Their voices are seldom heard among the louder, better-financed, and higher-profile contests that take place within a state (see Wolak 2004). Thus, without the ability to shape the agenda, candidates in low-information elections regularly contend with issues not of their making.

As mentioned, I am not suggesting that students of elections ignore offices. Organizing elections by office type lends analytic tractability to a cluttered electoral marketplace. It makes studying voting behavior a manageable task. Rather, my argument concerns misplaced emphasis. Elections supply an abundance of electoral information, and the content of this information has important implications for voting, and not just for one office type. By emphasizing institutional differences among offices or candidate campaign issues from a specific contest--a by-product of studying elections by office type--researchers often overlook information germane to voters, namely the broader information environment of which agendas play a leading role.


The influence of agendas on voting is unmistakable. As democratic and social choice theorists alike have emphasized, it is difficult to overstate the importance of agendas. Barber asserts, "He who controls the agenda--if only its wording--controls the outcome" (1984, 181). Similarly, Riker posits, "Agendas foreshadow outcomes: the shape of an agenda influences the choices made from it" (1993, 1). In discussing his theory of conflict displacement, an agenda-based approach to studying democratic processes, Schattschneider declares, "[T]he definition of the alternatives is the supreme instrument of power; the antagonists can rarely agree on what the issues are because power is involved in the definition" (1960, 68; emphasis original).

In contrast, some of the earliest quantitative research on voting behavior found little support for issue voting among the mass public, a necessary requisite of agenda voting. Campbell et al.'s The American Voter (1960) cast doubt on the ability of voters to make issue-based voting decisions. In their classic study, partisan identification and candidate characteristics exceeded the effects of issues. According to Campbell and colleagues, voters lacked the necessary information to vote for presidential candidates on policy issues. Converse's (1964) research on belief systems in the electorate advanced this line of research when he found that voters did not demonstrate a sophisticated enough understanding of politics to link candidates and issues.

Yet scholars have found evidence of issue voting, mostly in presidential contests. Predating The American Voter, Berelson, Lazarsfeld, and McPhee's (1954) research on the 1948 presidential election demonstrated how changes in the weight voters assigned to issues affected standards of candidate evaluation. Specifically, the authors found that the shift in voter support toward Truman followed changes on the weight voters assigned to issues on which they agreed with him. The research following Campbell and colleagues focused on the electoral information environment and how it may help or hinder the ability of voters to make issue-based voting decisions. Key (1966) was among the first to point out that the capability of voters to make informed decisions (e.g., policy based) is contingent on the information available to them. Subsequent research found that voters were more likely to choose candidates based on issues because the information environment had changed (Nie and Anderson 1974; Nie, Verba, and Petrocik 1976; Pomper 1972). Nie, Verba, and Petrocik (1976) showed that changes in the national agenda during the 1960s were responsible for making voters more politically aware and likely to engage in issue voting than were voters during the 1950s. Vietnam and the civil rights movement shaped the politics of the day in a way that differed greatly from the restrained politics of the Eisenhower era.8 In contrast to the 1950s when Campbell and colleagues did their research, Nie, Verba, and Petrocik argued that later decades provided greater opportunities for voters to make issue-based voting decisions. Thus, the change in political climate hastened a greater awareness of and reliance on policy issues in making voting decisions. Despite methodological problems in the revisionist accounts (e.g., see E. Smith 1989; Sullivan, Piereson, and Marcus 1978 for a discussion), contemporary scholarship on voting and elections finds that the electoral information environment, especially campaigns, shapes voter attitudes in presidential elections (Alvarez 1997; Bartels 1988; Gelman and King 1993; Popkin 1991; Repass 1971).

Even though scholars of presidential elections have been comfortable with the notion that policy issues matter, their counterparts who study low-information elections have been less so. The scholarly consensus had been that policy issues are relatively unimportant in these types of contests. Stokes and Miller (1962) found that issue voting was absent in House elections: "[T]he increment of strength that some candidates, especially incumbents, acquire by being known to their constituents is almost entirely free of policy content" (543). Despite research at the aggregate level of analysis demonstrating the effect of national issues on congressional elections (e.g., Tufte 1975), much of the evidence on congressional elections largely supports Stokes and Miller's finding that many voters in congressional elections do not use policy-based considerations. In contrast to studies of presidential elections, many studies of congressional elections demonstrate that "local forces," as defined by the incumbency advantage (e.g., Cain, Ferejohn, and Fiorina 1987; Mayhew 1974), campaign expenditures (e.g., Jacobson 1980), and candidate quality (e.g., Jacobson and Kernell 1983) make up the major ingredients of voting in congressional elections. Taken together, the dominance of "local forces" in congressional voting left little room for policy-based issue voting. No doubt "local forces" explain most of what happens in House elections, especially those races with an incumbent and an unknown challenger. Yet, studies of issue voting in congressional elections have largely neglected salient issues, the type most likely to be on voters' minds.

Studies of gubernatorial elections also neglect salient issues if they are not economic considerations. The major research question for studies of gubernatorial elections is whether these contests turn on national or state economic variables and to a lesser extent presidential approval (Chubb 1988; Holbrook-Provow 1987; Howell and Vanderleeuw 1990; Leyden and Borrelli 1995; Partin 1995; D. Simon 1989; Stein 1990). Thus, although the scholarly debates that characterize studies of congressional and gubernatorial elections help define the relative effect of subnational versus national influences, these debates have largely obscured how policy-based voting follows the rhythms of issue agendas in the states.

Evaluating the effects of the electoral information environment, of which agendas play a central role, is a crucial step toward understanding how voters make decisions in low-information contests. Indeed, if some information environments educate citizens on matters of public policy better than others (Kuklinski et al. 2001; Jerit and Barabas 2003), it is not far-fetched to expect related effects for the less Herculean task of evaluating candidates in nonpresidential elections. Studies of issue voting in these types of contests suggest that salient issues have a significant effect on voting for candidates in gubernatorial (e.g., Carsey 2000; Cook, Jelen, and Wilcox 1994; Hutchings 2003), Senate (e.g., Jacobson and Wolfinger 1989; Hutchings 2003; Kahn and Kenney 1999), and House (e.g., Iyengar and Kinder 1987, chap. 11; Zaller 1992, chap. 10) elections. Few analysts, however, take the crucial step of looking outside a particular contest to examine the broader information environment and how it might prime voters to assign greater weight to some issues. And while the candidates' campaigns may play a major role, especially if an electoral contest is highly visible, I contend that agendas, whether discussed by a pair of candidates in a given race or not, may play an equally important or perhaps even greater role.


I offer a theory of agenda voting in chapter 2. Here, I bring together well-known analytical tools in the study of political behavior and opinion--agenda-setting, priming, spillover effects, and partisan stereotypes. Together, these concepts provide the theoretical framework for how agendas infuse voting decisions with common meaning. Although these concepts are not new to political science, my use of them is somewhat novel for political science. For example, most applications of priming in political science examine how a prime shapes evaluative criteria for a single politician. In contrast, I recognize how priming effects influence multiple voting decisions. Priming effects, therefore, ignore institutional boundaries, showing little regard for institutional arrangements or the content of candidate campaigns. Thus, priming effects travel much further than previously recognized.

In chapter 3, I address questions of research design. Central to my research strategy is the use of direct legislation or ballot measures as a means of assessing the influence of agendas. Using direct legislation has several advantages. With the exception of a few states, most have some kind of process where voters decide matters of public policy at the polls. Institutionally defined as a state-level concern, ballot measures are often well-known and are often the focus of their own campaigns. Furthermore, they are often distinct from candidate campaigns and thus offer an opportunity to examine how "exogenous" issues affect voting decisions.

In contrast to the conventional approaches that characterize policy issues in congressional elections as exclusively national, chapter 4 examines the agenda-setting effects of statewide ballot measures in House and Senate elections. Here, I show that direct legislation has a substantial agenda-setting effect in these contests, controlling for rival explanations such as candidates' campaign messages. Using data from the National Election Study/Senate Election Study, I find evidence that statewide ballot measures on the environment, taxes, and abortion influence the agenda in congressional elections.

In chapter 5, I examine how agendas--via priming--affect voters' choice of candidates running for a variety of offices. Taking advantage of a natural experiment from the 1982 elections wherein some voters were exposed to freeze campaigns through initiatives and referenda and others were not, I examine whether nuclear freeze ballot measures affect the standards of candidate evaluation in congressional and gubernatorial elections. Comparing voters in states with and without freeze ballot measures, I find evidence of substantial priming effects. Voters in states with freeze measures were primed to evaluate candidates using this issue regardless of whether the freeze was a part of the campaign or relevant to a particular office.

Chapter 6 examines the interplay of candidates and ballot measures. The focus of this chapter is the use of ballot propositions as agenda-setting tools of candidates and political parties. Although direct legislation issues are mostly exogenous to candidate races, on occasion, and with increasing frequency, candidates and political parties make them endogenous to their campaigns. In the cases examined in this chapter, I look at how California's Republican Party used the issues of illegal immigration and affirmative action in 1994 and 1996, respectively, to infuse the "wedge issue" of race into elections.

Finally, in chapter 7, I discuss the democratic implications of agendas in light of my research findings and conclude with some thoughts on future research.

Return to Book Description

File created: 8/7/2007

Questions and comments to:
Princeton University Press

New Book E-mails
New In Print
PUP Blog
Princeton APPS
Sample Chapters
Princeton Legacy Library
Exam/Desk Copy
Recent Awards
Princeton Shorts
Freshman Reading
PUP Europe
About Us
Contact Us
PUP Home

Bookmark and Share