Donate
Independent, objective, nonpartisan research

R 302ZHR

Authors

R 302ZHR

Tagged with:

Publication PDFs

Database

This is the content currently stored in the post and postmeta tables.

View live version

object(Timber\Post)#3711 (44) { ["ImageClass"]=> string(12) "Timber\Image" ["PostClass"]=> string(11) "Timber\Post" ["TermClass"]=> string(11) "Timber\Term" ["object_type"]=> string(4) "post" ["custom"]=> array(5) { ["_wp_attached_file"]=> string(12) "R_302ZHR.pdf" ["wpmf_size"]=> string(6) "619543" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(197162) "Municipal Elections in California: Turnout, Timing, and Competition ••• Zoltan L. Hajnal Paul G. Lewis Hugh Louch 2002 PUBLIC POLICY INSTITUTE OF CALIFORNIA Library of Congress Cataloging-in-Publication Data Hajnal, Zoltan, 1968– Municipal elections in California : turnout, timing, and competition / Zoltan L. Hajnal, Paul G. Lewis, Hugh Louch. p. cm. Includes bibliographical references. ISBN 1-58213-041-8 1. Local elections—California. I. Lewis, Paul George, 1966– II. Louch, Hugh, 1973– III. Title. JS451.C25 H35 2002 324.9794'053—dc21 2002017332 Copyright © 2002 by Public Policy Institute of California All rights reserved San Francisco, CA Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source and the above copyright notice is included. Research publications reflect the views of the authors and do not necessarily reflect the views of the staff, officers, or Board of Directors of the Public Policy Institute of California. Foreword The PPIC Statewide Survey has shown time and again that California voters distrust their government. The roots of that distrust are deep and complex, but its policy effects are often quite clear. For example, California voters rely more heavily now on the initiative process to guide public policy, in part because they do not believe that their elected representatives will address the problems they think are important. One can argue at length over the advantages and disadvantages of specific initiatives, and one can do the same for the initiative process generally. But it is more difficult to argue that the distrust that has made the initiative process so important is itself salutary. This distrust becomes even more difficult to celebrate when it manifests itself as political apathy or even disgust. These reactions have kept large numbers of citizens from registering to vote in the first place. When we combine these unregistered citizens with the noncitizen population living in California, and then add to this group the registered voters who do not go to the polls, we find that, in many cases, an alarmingly small percentage of California residents decides a local election. This pattern is especially pronounced in off-cycle local elections, when turnout rates typically lag those of local elections that coincide with statewide and national contests. In a political system based on an informed and active citizenry, low and declining participation rates are a great concern. Indeed, increasing those rates may be the most important policy challenge of the early 21st century. But what are the most practical ways to meet that challenge? In this report, Zoltan Hajnal, Paul Lewis, and Hugh Louch offer a detailed description of local turnout patterns and analyze the factors associated with high and low participation rates. Their results indicate that about half the difference in voter turnout across California cities can be traced to a single factor—election timing. Noting that Progressive Era reformers instituted off-cycle local elections, and that one-third of iii California’s cities continue to hold them, the authors calculate how many more citizens would vote in local elections that coincided with state and national contests. The policy solution in this case seems clear enough. If low turnout for local contests is the problem, concurrent elections are a big step in the right direction. This simple change is unlikely to banish political apathy, but it will significantly increase the likelihood that citizens will make their voices heard on local issues. David W. Lyon President and CEO Public Policy Institute of California iv Summary The 2001 race for mayor of Los Angeles was one of the closest and most exciting in decades. With incumbent mayor Richard Riordan leaving office because of term limits, several strong candidates from various political and ethnic backgrounds ran for the office, and surveys indicated that the race was too close to call. It therefore came as a surprise to some that only 33.5 percent of the city’s registered voters came to the polls to cast votes in the April primary nominating election and only 36.2 percent voted in the June runoff election. Given the widely perceived interest in the contest, why did nearly two of three registered voters stay home? Are Los Angeles voters unusual in their degree of civic disinterest? This report finds that voter behavior in Los Angeles is far from atypical. By documenting levels of voter involvement in municipal elections, it investigates factors that help account for differences in turnout across California cities. It also examines the degree of competition for municipal offices as a related barometer of the vibrancy of the local political environment. Most of the key evidence is drawn from a questionnaire mailed to all city clerks in the state in 2000 asking about the most recent elections in the community. Usable data were collected on 350 city council elections, or 74 percent of the 474 municipalities then existing in the state. Analogous information was available for 130 mayoral contests, since only one-third of cities provide for direct election of mayors. City clerks also provided a wealth of information on local electoral procedures and institutional arrangements as well as the context of the particular election. We supplemented this information with published data from the Census and state sources. v Local Voter Turnout Rates: Generally Low, But Varied Observers of politics and community life have raised concerns about low, and in many cases declining, levels of civic engagement in America. Exercising the right to vote is one of the most fundamental and cherished forms of civic participation. Citizen participation in elections is important for several reasons. Low turnout levels may compromise the basic legitimacy of a democratic system, and nonparticipation is related to a host of negative attitudes about government and politics. Voting also serves as an important educational tool because voters learn more about policy issues and government by actively participating in the electoral arena. Finally, low turnout may be linked to an unrepresentative electorate. Racial and ethnic minorities, the poor, and younger Californians vote much less regularly than others in the state. If these nonvoters have systematically different political preferences or needs than voters, then the “voice of the people” will be distorted. Although municipal governments can affect citizens in profound and immediate ways, many Californians and other Americans ignore local elections. Information provided by the city clerks indicates that the average turnout rate in city council elections in California was 48 percent of registered voters and only 44 percent in mayoral races for cities that directly elect the mayor. (Aggregate turnout of registered voters was even lower—43 percent for council races and 39 percent for mayoral contests—because cities with larger populations tended to have lower participation rates.) Measured as a percentage of the voting-age population, turnout was lower still: 32 percent in the average council election and 28 percent in the average mayoral elections. These overall levels of turnout are important, but they conceal important variations in voting rates across communities (Figure S.1). Turnout of registered voters ranges from a low of 10 percent to a high of 89 percent. It is therefore important to consider which factors might account for these major differences in voter participation. vi Number of cities 80 70 60 50 40 30 20 10 0 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70–79 80–89 90–100 Percentage of registered voters casting ballots Figure S.1—Distribution of Voter Turnout Rates in City Council Elections Why Do Local Turnout Rates Differ? To understand variation in voter turnout across cities, we examined three sets of factors: election timing, the electoral and institutional structure of local government, and the specific context surrounding any given election. Although we also control for the demographic and socioeconomic characteristics that are known to affect voter turnout, we focus attention on these broad city-level factors for two reasons. First, anecdotal evidence and findings from other studies suggest that election timing and local institutions can significantly affect voter participation. Second, these city-level characteristics lend themselves more readily to practical policy interventions. Although it is difficult or impossible for city policymakers to affect the demographic or socioeconomic characteristics of their localities, they often can and do alter their electoral institutions and governmental structure. Our analysis indicates that about half of the differences in turnout among California cities can be explained by one simple factor—the vii timing of the local election. By scheduling local elections to occur on the dates of statewide general or primary elections (so called concurrent or oncycle elections), localities make it easier for voters participating in the statewide election to vote in local contests as well. Controlling for a host of other factors, presidential elections are associated with turnouts of registered voters in city elections that are 36 percent higher than off-cycle elections (which are “local-only” elections typically held in the spring); gubernatorial elections and presidential primaries are associated with municipal turnouts of 21 to 26 percent more registered voters (Figure S.2). In short, participation in local elections depends critically on the timing of those elections. At least one local government institutional arrangement is also related to turnout in municipal elections. Cities that provide more services with their own staff (as opposed to contracting out to firms or making service arrangements with other local governments) tend to draw a larger share of voters to the polls. This higher turnout may occur because city governments that provide services directly have more control over some of the basic issues that affect city residents’ quality of life, or 70 60 Registered voters Adult residents 50 Median percentage turnout 40 30 20 10 0 Off-cycle Odd-year Gubernatorial Presidential Presidential November primary Figure S.2—Variation in Turnout by Election Timing (controlling for other factors) viii because municipal employees tend to be a well-mobilized segment of the local electorate. By contrast, the distinctions between charter cities and general-law cities, and between cities with and without term limits for council members or mayors, have no direct relationship with local voter turnout. The context of the local election is also related to turnout. Where there are one or more propositions on the municipal ballot, cities tend to draw about 4 percent more registered voters to the polls. Similarly, the degree of competition for the office (measured by the number of candidates) is positively related to turnout. Predictably, uncontested elections draw especially few voters. If we use a city’s adult residents, rather than registered voters, as our reference group, these factors—timing, service provision, ballot measures, and candidate competition—likewise demonstrate important relationships to turnout. The effect of these factors is slightly diminished because demographic factors and citizenship barriers take on additional importance, heavily influencing whether residents can and do register to vote in the first place. Election Timing, Incumbent Success, and Competition for Office These results indicate that a move to concurrent elections has the greatest potential to expand voter participation in California’s local political arena. More than one-third of the cities responding to the city clerk questionnaire held their elections on a date that was not in the electoral “prime time” of a statewide election. Moreover, the timing of local elections is a subject of policy debate in California. More than 40 percent of the city clerks responding indicated that their city has made a change in the timing of municipal elections in recent years. As the state has changed the date of its primary election in an effort to have more effect on the national presidential primary season, numerous communities with nonconcurrent (stand-alone) spring elections have consolidated their elections with statewide contests, or have considered doing so. The reason is often budgetary; city governments pay a smaller ix share of the cost of holding an election if it is held concurrently with other contests. In considering a timing change, however, policymakers may wish to weigh the benefits of increased participation against other, potentially unintended consequences of changes in election timing. For example, critics have argued that some voters in concurrent elections are unaware of and inattentive to local issues and candidates because the focus of media coverage and popular attention is on the higher-profile state or national races. If true, concurrent elections could provide increased protection for incumbent candidates, who have greater name recognition on average than their challengers. An analysis of the reelection success of incumbent city officeholders provides some mild support for this notion—although the vast majority of incumbents win regardless of election timing. Whereas 80 percent of incumbent council members and 86 percent of incumbent mayors standing for reelection succeeded overall in the cities responding to the questionnaire, the reelection rate for incumbents was higher in cities that have concurrent elections. Nonconcurrent city elections tend to place incumbents in a more vulnerable position, perhaps because voters who come to the polls for stand-alone elections are more motivated or aroused by the issues facing the community. Of course, increased incumbent success in on-cycle elections could be a sign of something else altogether—perhaps greater trust or greater satisfaction resulting from more widespread participation. Our examination of competition for council and mayoral offices also reveals that incumbents have significantly lower reelection success rates in cities with citizen initiatives on the ballot—a possible reflection of community controversy. Cities with larger populations tend to have higher incumbent reelection rates—and lower voter turnout rates— indicating that voters are probably less likely to become engaged in civic issues in larger communities. The presence of an incumbent strongly dissuades potential competitors from entering local races. Competition for council or mayoral positions is lower where incumbents are standing for reelection. It is interesting to note that in cities where independent and third-party voters constitute a larger share of registered voters, there tend to be more x competitors for office, even though local elections in California are nonpartisan (that is, party labels do not appear on the ballot). In addition, the number of candidates for mayor is greater where the rewards of the office (full-time salary and length of the term) are greater. Can Participation in Local Elections Be Increased in California? The first and most important step to increase voter participation in city elections would be to shift their date to coincide with statewide or national contests. We estimate that if all cities in the state that hold nonconcurrent elections switched their elections to a presidential election date, approximately 1.7 million more Californians would have cast a ballot in their most recent city council contest. Yet this step should only be considered a limited reform. Despite the significant boost that concurrent elections could give to city turnouts, they would probably not increase turnout beyond the fairly disappointing levels already found in statewide and national contests. And although voter participation is closely correlated with greater trust, efficacy, and satisfaction in government, increasing turnout would not guarantee improved civic engagement in local communities, along such important dimensions as serving in neighborhood organizations or attending community meetings. Still, in the end, the doctrine of “one person, one vote”—a bedrock of democratic theory—probably outweighs any potential negatives. Some might argue that participation in local contests be left to the most engaged or interested voters. However, political equality and enhanced citizen participation in city politics are important goals in an increasingly diverse state with powerful local governments. xi Contents Foreword ..................................... Summary..................................... Figures ...................................... Tables ....................................... Acknowledgments ............................... iii v xv xvii xix 1. POLITICAL PARTICIPATION AND MUNICIPAL ELECTIONS ............................... Local Elections in the Context of Declining Voter Participation ............................ Do Policymakers Have Any Potential Leverage over Local Participation? ............................ Incumbent Reelection and Competition for Office ....... A Need for New Evidence: Examining Voter Turnout in California Cities .......................... 2. VOTER TURNOUT AND LOCAL CONTEXTS: RESULTS FROM A SURVEY OF CITY CLERKS ...... Mail Survey Procedures and Response Rate ............ Ascertaining Voter Participation Rates: The Core of the Survey ................................ Is Low Voter Turnout a Problem in California Cities? ..... Election Timing, Local Institutions, and Context ........ Election Timing: The First Key ................... Governmental Structure and Electoral Laws ............ District Versus At-Large Elections................. Term Limits .............................. City Service Provision ........................ The Form of Government in Cities ................ Local Structure Summarized .................... Electoral Context ............................. Local Direct Democracy ....................... Competition .............................. 1 1 5 10 11 13 13 14 16 17 18 20 20 21 22 23 24 25 25 26 xiii Incumbency ............................... Electoral Competition Across Races or Ethnicities ...... Supplementing the Survey: City Demographic Characteristics ........................... Summary .................................. 3. ACCOUNTING FOR DIFFERENCES IN VOTER TURNOUT ACROSS CITIES ................... Election Timing ............................. Institutional Factors ........................... Electoral Context ............................. City Demographics ........................... Summary .................................. 4. COMPETITION AND INCUMBENCY IN LOCAL ELECTORAL CONTESTS...................... Competition for Office ......................... The Success of Incumbents ...................... Summary .................................. 5. CONCLUSION ............................. 28 28 29 32 33 35 38 40 44 46 49 49 55 60 63 Appendix A. Survey of City Clerks .......................... B. Mail Survey Procedures and Data Issues .............. C. Voter Turnout Regressions....................... D. Candidate Competition Regression ................. 69 81 87 91 References .................................... 95 About the Authors ............................... 101 Other Related PPIC Publications ..................... 103 xiv Figures S.1. Distribution of Voter Turnout Rates in City Council Elections ............................... vii S.2. Variation in Turnout by Election Timing (controlling for other factors) .......................... viii 2.1. Service Provision by Type and Provider ............ 23 3.1. Distribution of Voter Turnout Rates in City Council Elections ............................... 33 3.2. Variation in Turnout by Election Timing (with controls) ................................ 36 3.3. Variation in Turnout by City Service Provision (with controls) ................................ 39 3.4. Variation in Turnout by Number of City Ballot Propositions (with controls) ................... 42 3.5. Variation in Turnout by City Population (with controls) ................................ 45 4.1. Incumbent Success and Election Concurrency........ 57 4.2. Success of Council Incumbents in Varying Electoral Contexts................................ 58 4.3. Incumbent Success and City Population ........... 59 xv Tables 1.1. The Demographic Profiles of California’s Voters and Nonvoters ............................... 1.2. The Different Preferences of Voters and Nonvoters .... 2.1. Timing of California Municipal Elections .......... 2.2. Local Government Structure in California .......... 2.3. Race or Ethnicity of Mayoral Candidates ........... A.1. Number of Valid Responses for Individual Variables .... B.1. Comparing In-Sample and Out-of-Sample Cities ...... B.2. Geographic Distribution of Cities in Sample ......... B.3. Important Missing Values..................... C.1. Turnout in City Elections: Regression Model ........ C.2. Descriptive Statistics of Variables for Observations in the Regression............................... D.1. Determinants of Competition for Office ........... 3 5 19 25 29 70 82 83 83 88 90 92 xvii Acknowledgments We are indebted to the city clerks of California for giving their time and providing much of the data on which this report is based. We would also like to thank Peter Richardson and Patricia Bedrosian for their expert editorial advice. Amy Bridges, Christopher Hoene, Christopher Jepsen, Joyce Peterson, and Fred Silva all provided detailed comments on earlier drafts of the manuscript, and the report is much improved as a result. Mark Baldassare offered a strong guiding hand throughout the project. Angelina Reyes and Max Neiman provided useful feedback on the survey instrument. Although this report reflects the contributions of many people, the authors are solely responsible for its content. xix 1. Political Participation and Municipal Elections Observers of municipal elections in California have often expressed concern that only a small fraction of Californians actively participate in these local contests. Yet no systematic studies of local voter turnout in contemporary California have been undertaken. In this report, we calculate participation rates for municipal elections across the state and address a number of important related questions about the vitality of those elections. How does political participation vary across cities?1 What political structures and electoral laws are associated with low turnout? What changes might promote higher turnout in local elections? Our data, drawn largely from a questionnaire mailed to all of the city clerks in the state in 2000, also allow us to investigate related issues of competition for local office and the role of incumbency in local elections. Local Elections in the Context of Declining Voter Participation Voter participation at every level of government in California is distressingly low and appears to be getting worse. In presidential elections, almost half of all eligible voters in California do not vote (California Secretary of State, 2001). In midterm congressional elections, over half of eligible voters fail to vote. The numbers are even worse for statewide primaries, where turnout now hovers around onethird of eligible voters. If current trends continue, these rates will decline further. Between 1972 and 2000, statewide turnout dropped roughly 15 percentage points in California (California Secretary of State, 2001). ____________ 1As is common in California, we use the term city as a synonym for municipality—in other words, a general-purpose, sub-county local government. 1 Without that decline, over 3 million more voters would cast ballots in statewide contests today. Nowhere is the problem worse than at the local level. Despite the direct effects that local government activities have on residents—for example, in providing public safety and other essential services, building infrastructure, and making land-use decisions—large numbers of Californians and other Americans ignore local elections. Turnout in municipal elections around the country averages half that of national elections (Morlan, 1984), and local voter turnout often falls below onequarter of the voting-age population (Bridges, 1997; Hampton and Tate, 1996).2 Anecdotal evidence from several cities suggests that turnout in California is even lower than in the rest of the country (Bridges, 1997). At the local level, important public policy decisions are made without input from most of California’s residents. This civic disengagement raises serious concerns (Hill, 2000; Bennett and Resnick, 1990). First, it raises questions about the legitimacy of democratic government. Democracy is supposed to operate at the will of the people. If a small minority of the population elects city, state, and national leaders, political institutions and elected officials may lack the broad support and confidence necessary to govern effectively. Second, low participation may reflect distrust of government (Bobo and Gilliam, 1990), a reduced sense of civic duty (Wattenberg, 1998), and decreased political efficacy (Finkel, 1985). In California, evidence suggests that these conditions are already all too common.3 Third, voting serves as an important educational tool. Active participation teaches citizens about the functioning of government and the issues that are currently under debate (Bennett and Resnick, 1990). ____________ 2Moreover, trends over time suggest that voter turnout in local elections is declining just as rapidly as it is in national elections (Verba et al., 1995; Karnig and Walter, 1993). 3As Baldassare (2000, p. 26) summarizes Californians’ opinions about government, “It is clear that Californians don’t care much and are cynical about politics. . . . They see their governments as bloated bureaucracies and believe that their elected officials are in the pockets of special interest groups. They don’t believe what candidates are telling them in television commercials. They think what they read in the newspapers about politics and government is probably slanted or biased.” 2 Without such participation, Californians may be less able to identify their best interests or to know how to protect those interests. Thus, low participation may increase the likelihood that policy decisions reflect an inaccurate and incomplete understanding of the available policy options and may lead to inefficient outcomes that hurt Californians. The very unequal distribution of voters and nonvoters among different segments of the population heightens each of these concerns. California residents who are highly educated, wealthy, old, and white are much more likely to participate than residents who are poor, young, less educated, and nonwhite (Table 1.1). These differences are especially large across educational levels. Fewer than half of eligible, voting-age Californians have a college degree or some college, yet these citizens constitute nearly three-quarters of all voters. A similar story can be told for racial and ethnic differences. Although about half of California’s total population is non-Hispanic white, some 72 percent of registered voters in 2000 were white. At the opposite end of the spectrum, Table 1.1 The Demographic Profiles of California’s Voters and Nonvoters White Latino Asian American African American Percentage of Percentage of Unregistered Voters Residents 72 44 16 40 49 65 Income under $40,000 Income $40,000 or more 49 63 51 37 High school degree or less education Some college or more education 26 56 74 44 Age 18 to 34 Age 35 to 54 Age 55 and older SOURCE: PPIC Statewide Surveys, 1998–2000. 27 42 31 49 39 12 3 Latinos made up only 16 percent of registered voters but accounted for fully 40 percent of the unregistered population.4 One major problem with an unrepresentative voting population is that it may skew the outcomes of the democratic process (Verba et al., 1995; Guinier, 1994; Casel, 1986). If, for example, a largely white electorate determines policy with little input from nonwhite voters, there may be reason to suspect that minority interests will be short-changed. More generally, if the interests of nonvoters diverge in important ways from those of voters, the candidates who are elected and ultimately the types of policies that are enacted may well favor one group over the other. Put simply, only a few will be speaking for the interests of nonvoters. In presidential elections, where turnout is relatively high, the political preferences of voters and nonvoters are not that dissimilar (Teixeira, 1992; Wolfinger and Rosenstone, 1980). However, differences in policy preferences are likely to be much greater in local contests where turnout is substantially lower and often more skewed along the lines of race and socioeconomic status (Wattenberg, 1998; Alford and Lee, 1968). In California, statewide surveys suggest that the views of registrants do differ from those of unregistered Californians in systematic and sometimes important ways (Table 1.2).5 A sample of voting intentions on recent California propositions suggests that Californians who are registered are significantly more conservative than unregistered Californians across an array of issues including labor union contributions to political campaigns (Proposition 226), bilingual education (Proposition 227), and school bond initiatives (Proposition 1A). However, the differences between registered voters and unregistered residents are not found across all issues. On issues such as defining marriage as a union between a man and a woman ____________ 4In this report, “white” refers to persons who identify as white and not Hispanic. The terms Hispanic and Latino are used interchangeably, as are the terms African American and black, reflecting the manner they are listed in the 2000 Census questionnaire. 5Survey respondents often claim to have voted when they did not, making this analysis suggestive rather than definitive. 4 Table 1.2 The Different Preferences of Voters and Nonvoters Proposition 22 (limit on marriages) Proposition 1A (education bond) Proposition 227 (bilingual education) Proposition 226 (political contributions by unions) SOURCE: PPIC Statewide Surveys, 1998–2000. Percentage in Favor Registered Unregistered Voters Residents 61 63 73 85 71 62 61 49 (Proposition 22), differences between the registered and unregistered are less stark. In the end, it is unknown whether the divergent views of those who participate and those who do not participate translate into distorted policy outcomes, but given the large differences found between voters and nonvoters in California, the relatively low turnout in local elections merits attention. For all of these reasons, low turnout in local elections poses a potentially serious problem. In a political arena that touches regularly and deeply on the daily lives of residents, nonparticipation by a large share of eligible voters is worrisome. Failing to vote may be a sign not only of political disengagement but also of distrust, division, and cynicism. Where participation is low, government outcomes may be less equitable or responsive. Thus we seek to understand which factors affect voting levels in California’s municipal elections and what steps might be taken to try to encourage broader participation. Do Policymakers Have Any Potential Leverage over Local Participation? Can anything be done about this low rate of voter turnout in municipal elections? This report examines four sets of variables to help determine which factors are associated with higher or lower turnout: election timing, the structure of electoral and political institutions, the 5 context of the particular elections, and the demographic attributes of the cities in which they take place.6 Most research on political participation has focused on demographic factors related to voting. 7 We focus most of our attention on timing, institutional structure, and context. Although these three features of local elections have been less well studied, they are potentially significant for policymakers interested in increasing local voter participation. Unlike demographic attributes, which are either impossible or extremely difficult for city policymakers to alter, the three features that we focus on can often be manipulated by local governments. California municipalities (particularly charter cities) generally have the ability to alter the way they conduct elections and can in many cases reform their institutional structure, although in some cases state legislative action may be necessary. Thus, if it turns out that voter participation is related to any of these three features of the local electoral context, changes could be enacted to broaden political participation. To fully understand the role that timing, institutional arrangements, and context might play, it is necessary to outline a series of changes enacted in response to the Progressive movement roughly a century ago (Bridges, 1997; Welch and Bledsoe, 1988). These institutional changes were, according to the Progressives, crucial to the functioning of “good government” at the municipal level. Opponents and subsequent critics have viewed the institutional changes pushed by the Progressives as efforts to limit the participation of the working class and of ethnic groups and to usurp power at the local level (Bridges, 1997). Whatever the true motives of the Progressives, the result of their efforts is that many cities in California and across the country fundamentally altered their basic governing structure. New cities that have since formed in growing ____________ 6Mobilization by parties and campaigns and individual attitudes toward the political arena can also have an effect on voter turnout (Rosenstone and Hansen, 1993; Dennis, 1991). However, study of these factors is beyond the scope of this report. 7Our data relate to municipal political systems rather than to individual voters. Therefore, we will refrain from drawing inferences about individual-level attributes that may affect city-level participation, although we do take account of these demographic factors as “controls.” 6 western states such as California have also typically incorporated most or all of the institutions that were advocated by the Progressives. One change supported by the Progressives was a move from on-cycle or concurrent elections—elections that coincide with statewide or national contests—to off-cycle or nonconcurrent elections. Progressives claimed that this change in election timing would mean that local contests would be decided by the most interested and knowledgeable voters. Critics countered that nonconcurrent elections were pursued as a means to limit participation to those segments of the population most likely to support Progressive candidates. Whatever the initial motivation, many cities across the state have set the timing of their elections so that they do not coincide with state or national contests. Election timing is especially important for our purposes because studies at the national level have revealed a strong relationship between on-cycle elections and higher turnout (Espino, 2001; Hampton and Tate, 1996). The second institutional feature of local elections that we focus on is the electoral mechanism used for the city council. In an attempt to purportedly reduce “narrow” neighborhood interests in favor of citywide concerns, Progressives favored at-large elections over district- or wardbased contests.8 Evidence nationally suggests that this change may have reduced voter interest and participation by distancing leaders from their constituencies (Bullock, 1990). One reform sought by Progressives, direct democracy, may have increased voter turnout (Hahn and Kamieniecki, 1987; but see Magleby, 1994). By giving more decisionmaking power to the people in the form of initiatives, referenda, and recall, Progressives hoped to take power away from entrenched local politicians and interest groups who were, in their view, corrupt.9 In the process, these changes may have increased voter participation by increasing citizen interest in local elections. ____________ 8Critics have seen the move to at-large elections simply as a way for the Progressives to usurp power for their own “narrow” interests and reduce the influence of ethnic neighborhoods. 9Critics have countered that direct democracy was a way for Progressives to pursue their own agenda, and that the initiative process is itself often wielded by special-interest groups. 7 These reforms have not been pursued (or in the case of local voter initiatives, used) uniformly across the state. Chapter 2 provides greater detail on just how widespread change has been. For now, the important point to note is that variation in these institutions and the frequency of local initiative use make it possible to analyze the effects of such electoral and governance arrangements on voter turnout. By studying the relationship between cities’ turnout and their institutional structure, we should be able to determine what, if any, institutional changes might be pursued to increase the participation of Californians in local electoral politics. The effects of two other institutional changes pursued by the Progressives—nonpartisan local elections and the council-manager form of government—cannot be evaluated in California because they exist in virtually every city in the state. There is reason to suspect that both have affected turnout.10 Nonpartisan elections—those in which party labels are not included on the official ballot and parties do not have control over nominating procedures—are mandated for local elections statewide and are thus not included as a variable in our analysis. Similarly, almost all cities in California have chosen a council-manager system in which the day-to-day operations of the city are overseen by an unelected professional city manager or administrator, who is appointed by the council. By contrast, the earlier tradition of the mayor-council plan of government gives stronger executive powers to a directly elected mayor who does not sit as part of the city council. In practice, variation within the two forms of government (council-manager and mayor-council) may now be as important as the distinction between the two plans.11 For this reason we focus more on measures of mayoral strength, where available, ____________ 10Nationwide, nonpartisan elections have reduced voter turnout—perhaps by limiting the efforts of parties to mobilize voters (Schaffner, Wright, and Streb, 2001; Espino, 2001). Similarly, national studies have found that the move to the councilmanager system has reduced voter participation—perhaps by providing for a weaker elected executive and a more insulated administration than its predecessor, the mayorcouncil system of government (Karnig and Walter 1983; Espino 2001). 11Cain, Mullin, and Peele (2001) have detected a number of hybrid forms and have illustrated cases in California where, despite the presence of the council-manager system, mayors have extensive managerial and agenda-setting powers. 8 than on the traditional distinction between council-manager and mayorcouncil forms of government. Several facets of local institutions and elections for which the Progressive movement is not responsible are also of interest. For example, one important consideration concerns local service provision. Although the norm in earlier periods of municipal government was for the city to provide the full array of municipal services, a host of alternative service arrangements have evolved in recent decades. These arrangements range from provision of municipal services by special district governments or county departments to efforts to contract out service provision to private firms. In some “contract cities,” particularly prevalent in Southern California, few local public services are actually directly carried out by city employees. Under these circumstances, some measure of direct influence is removed from the hands of city elected officials, potentially reducing interest in municipal politics and depressing turnout. Term limits are a more recent electoral reform that might also influence local voter turnout. Many cities in California have sought to curb the powers of incumbents, encourage more candidates to run for office, and ultimately increase competition for office by limiting the number of terms that mayors or council members may serve. At this point, however, it is unclear just what effects term limits have had on either competition or turnout (Thompson and Moncrief, 1993). Finally, we also examine the distinction between charter cities and general-law cities as an institutional factor potentially affecting local residents’ participation in politics. About one-fifth of California cities are organized under charters, or local constitutions, in which local voters consent to a specific set of organizational arrangements and powers for the municipality. The rest derive their authority from and operate under the provisions of state law relating generally to local government. Historically, charter cities retained more autonomy from the whims of the state than general-law cities, although in recent decades legislative changes and court decisions have eroded the distinction between the two types of municipalities. In general, all California municipalities have a relatively broad grant of home-rule authority. Charter cities are permitted to adopt a “strong mayor” form of government, alter the 9 number of council members (from the otherwise prescribed five), and have somewhat greater leeway regarding purchasing arrangements and personnel.12 Whether charter status reflects these modest additional powers or a more activist tradition of municipal home rule in the city’s past, some might expect that charter cities would experience more voter interest in local politics and thus perhaps higher turnout. In addition to our strong interest regarding the potential effects of local institutions on turnout, we also focus on a number of contextual features of local elections that may influence voter participation. For example, one important factor that has the potential to drive turnout at any level in politics is the degree of competition among candidates for office (Cox and Munger, 1989). Similarly, the well-known electoral advantages that incumbent officeholders have may reduce the chances of potential challengers (Jacobson, 1983), thereby possibly making elections less interesting to voters. Incumbent Reelection and Competition for Office In this report, we also examine two of the potential pitfalls that might be associated with increased turnout and, specifically, with oncycle elections: reduced competition for office and greater protection for incumbents. A number of observers of local politics suggest that concurrent elections deter challenges to incumbents by reducing the prominence of local elections (Johnson, 1994). With popular attention focused mainly on national and state contests (and in California, statewide initiative elections), local challengers might find it hard to gain attention, garner media coverage, and raise enough money to warrant entering the race. A move to concurrent elections might also conceivably help incumbents by incorporating voters with little interest in or knowledge of local affairs, many of whom might thus vote for incumbents by default. Already, some scholars argue that voters seldom have the necessary knowledge to make informed decisions (Campbell et al., 1980). In this ____________ 12Charter cities, unlike general-law cities, are also allowed to adopt zoning codes that are not strictly consistent with the local general plan. On charter status, see Sokolow and Detwiler (2001); Curtin (2000, p. 18). 10 way, higher turnout and concurrent elections might potentially lead to less competitive elections. Analyzing the level of competition in municipal elections and the success rates of incumbents thus provides additional insight into the political vibrancy of cities around the state. A Need for New Evidence: Examining Voter Turnout in California Cities There is at least anecdotal evidence indicating that each potential institutional and contextual factor we have outlined above may affect voter turnout in America’s municipalities. Yet few of these factors have been tested systematically across a wide range of cases, largely because of a lack of suitable data. The last major published study of municipal election turnout appeared more than 30 years ago (Alford and Lee, 1968). This study is intended to fill this gap. Much of the analysis relies on data from a questionnaire administered to city clerks. In Chapter 2 we describe this fact-gathering survey and provide an overview of the institutional structures and electoral laws in place in the state’s municipalities, as disclosed by the survey respondents. Chapter 3 summarizes our statistical analysis of the factors that influence voter turnout, detailing the role of institutional, contextual, and demographic factors. Chapter 4 turns to other issues related to engagement in city electoral politics, examining competition for office and the success of incumbents. The report concludes in Chapter 5 with a discussion of the implications of these results for understanding citizen participation in California’s local elections and the policy implications of the study’s findings. 11 2. Voter Turnout and Local Contexts: Results from a Survey of City Clerks To gather the data necessary for our central research questions, we devised and distributed a mail questionnaire to each California municipality. This chapter describes the survey, the resulting sample, and the information gathered regarding voter turnout patterns, institutional structure, and the context of city elections. The most important finding from the simple survey results is that turnout is quite low in California municipal elections, averaging well below half of registered voters. Mail Survey Procedures and Response Rate The questionnaire was mailed to every city clerk in the state in late 2000. City clerks are designated as the chief municipal elections officials, although the actual task of administering the election and tallying votes rests with the county government. City clerks, furthermore, often have unique, first-hand knowledge about the political life of their communities.1 Of the 474 cities in existence at the time of the survey, a total of 397 clerks (84 percent) returned surveys with at least some of the necessary responses; however, complete and usable data for calculating voter turnout rates themselves were available for a smaller number—350 cities. The 350 responses allow us to report turnout data for 79 percent of the cities in the state, a solid rate of response. Our sample of cities is generally representative of all cities in the state. Comparing cities that responded to those that did not revealed few significant differences. Cities in and out of the sample were similar in ____________ 1For a study that relies on city clerks as informants regarding local politics see Schneider and Teske (1995). 13 terms of regional location (Southern California, San Francisco Bay Area, Central Valley, or other) and in terms of the percentage that were central cities, suburbs, or rural communities. Our sample was also representative in terms of racial demographics, percentage of residents unemployed, median household income, and home ownership rates. There were marginal differences between cities in and out of the sample in the average size of the population, the average household size, and the poverty rate. Compared to other city clerks, those in larger communities were more likely to respond. A random sample of responses was validated using municipal web pages and a variety of other sources. Errors were minimal, and any errors that were found were corrected. Appendix B provides a more detailed discussion of the survey procedure, the representativeness of the sample, and response validation. We asked each city clerk a series of questions about his or her community’s most recent city council election and mayoral election. Every city holds council elections, but only one-third of cities that responded directly elect their mayor, yielding a total of 130 mayoral elections. In subsequent chapters, we analyze a dataset that combines both council and mayoral elections. When we examined each type of contest separately, the results were generally quite similar to those presented in this report. Ascertaining Voter Participation Rates: The Core of the Survey The key variable in this report is voter turnout. For each election, we asked city clerks to report the total number of residents who cast ballots, as well as the total number of registered voters at that time in the city (or in the relevant districts if only certain city council district seats were up for election). In addition, we estimated the size of the votingage population for each city at the time of the election. The Census Bureau publishes the voting-age population of every city but only for decennial Census years. Thus, the measure we use is an interpolation 14 (for the particular year of the election) between the voting-age population in the city in 1990 and in 2000.2 In the analysis that follows, we focus on two dependent variables: the percentage of registered voters who cast ballots for local office in a given election and the percentage of voting-age residents who cast ballots. We focus heavily on turnout of registered voters because we believe that it is likely to be the measure most directly affected by changes in the local institutional and electoral context. Registered voters have already cleared the major entry barrier to political participation (registration) and thus whether or not they vote is likely to be closely related to conditions at the time of the election. Also, our data on registered voters are more accurate than our interpolated estimate of voting-age residents. At the same time, if we want to know how likely the population at large is to participate in local elections, it is important to examine turnout of the voting-age population as well. By focusing on the votingage population and not the registered population, we can take account of the intermediate step of voter registration, which has at times in the past been used to exclude certain segments of the electorate (Parker, 1990; Davidson and Grofman, 1994).3 Although registration procedures are set by state and federal laws, they are typically administered by county officials, who historically have had a fair amount of latitude in carrying out these procedures. Although the average city in our sample had an (estimated) registration rate of 63.8 percent, there was a wide degree of variation in this rate across communities.4 Thus, differences in registration rates across cities could have a number of different causes other than simply the level of interest ____________ 2Obviously, this measure is more precise for those cities that reported on elections held in 2000. 3One could also specifically examine the process of registration. For example, Mitchell (1992) demonstrated how voter rolls have at times been purged, seemingly for arbitrary reasons, such that certain segments of the population were disenfranchised. However, despite the fact that the percentage of residents registered, turnout of votingage population, and turnout of registered voters measure somewhat different aspects of participation, they are often fairly highly correlated (Hampton and Tate, 1996). 4The standard deviation for the percentage of adults registered was 16.2 percent. 15 in elections. As a result, higher turnout of registered voters may not necessarily indicate broader participation but might instead be a sign that barriers to registration at the local level have reduced the number of registered voters. For these reasons, we undertake all subsequent analyses for both turnout of registered voters and turnout of the (estimated) voting-age population. In practice, turnout of registered voters and turnout of the voting-age population are very highly correlated among California cities (r = 0.85), and the factors that lead to increased turnout of registered voters also tend to increase turnout of the adult population. Is Low Voter Turnout a Problem in California Cities? Compared to voter turnout in statewide and national contests, local voter turnout is low. In the last presidential election, 71 percent of California’s registered voters participated. Recent statewide contests for governor have attracted roughly 60 percent of registered voters (California Secretary of State, 2001). Our data indicate that voter turnout in municipal elections falls well below these levels. Of the 130 cities reporting directly elected mayors, only 44 percent of registered voters showed up at the polls and cast a vote in the average contest. In city council elections, where we have complete turnout data for 350 municipalities, the mean turnout was similar—48 percent. Measured as a percentage of the voting-age population, turnout looks even worse. Turnout in mayoral contests averaged only 28 percent of the voting-age population. In council elections, the average was 32 percent of the voting-age population. It is also possible to calculate the overall participation rate among these cities—the number of voters who cast ballots in all of the city elections for which we have data, divided by the total number of registered voters in those cities at election time. We found that the overall participation rate in council elections was 43 percent of registered voters; only 3.7 million voters out of 8.6 million registered in these communities cast ballots in their most recent city council contest. The aggregate turnout rate for mayoral elections was even lower: 39 percent. 16 These low levels of turnout are disturbing. Important policy decisions are being made by local officials chosen by a relatively small group of citizens who may differ in important ways from the nonvoting population. Election Timing, Local Institutions, and Context Why are so many Californians not voting in local contests? What is it about California’s municipal elections that leads to low turnout? To help answer these questions, we endeavored to ask city clerks about as many potentially relevant factors as possible. In particular, we asked several questions about three sets of factors that one might expect to have an effect on voter turnout rates: the timing of mayoral and council elections, the set of political and electoral institutions found in each city, and the political context of the specific election of interest. In the sections that follow, we discuss the motivations behind these questions and describe the city clerks’ responses. Most of our discussion of possible solutions to the problem of low voter turnout focuses on election timing and local institutions, for three reasons. First, election timing and institutional structure have been highlighted as potentially important contributors to low voter turnout across the country (Espino, 2001; Hampton and Tate, 1996).5 Second, it is relatively easy to alter these institutions. In some cases, only a local ordinance would need to be changed; other changes would require alterations to city charters or state law. Third, changes in these institutions and electoral laws are not unprecedented. Throughout the last century, numerous alterations were made to the basic governing structure of cities in the United States and changes are still routinely considered for a variety of reasons. Although our primary purpose in conducting the mail survey was to try to understand voter turnout in municipal elections, we believe that the data we present on the electoral procedures and institutional structures of California’s cities are important in and of themselves. ____________ 5Other electoral laws considered relevant to voter turnout, but not examined here, are the types of registration requirements and the ease of registration (Powell, 1986), along with nonpartisanship (Schaffner, Wright, and Streb, 2001). 17 Given the fluid nature of the electoral procedures and institutional structure of California’s cities, it is important to provide some basic facts on what California’s city governments look like. To date, few systematic data have been collected on the distribution of electoral institutions and governing structures across California’s cities. The following sections should begin to fill in these gaps. Election Timing: The First Key Among the structural factors related to voter turnout, we are most interested in the timing of elections. Scheduling local elections on the same date as statewide primaries or general elections may increase local turnout to levels almost on par with voting in national elections. Having the dates of local elections coincide with the dates of statewide primary or general elections (often referred to as “on-cycle,” “consolidated,” or “concurrent” elections) makes participation in local elections easier. Voters need only check off names further down the ballot to participate in the local political contest. Given the significantly higher turnout rates for national and state elections, gains in turnout could be substantial. Moreover, cities have other incentives besides increased voter participation to switch to concurrent elections. In fact, the primary motivation for this move has usually been cost savings. Because municipalities generally pay the entire cost of holding nonconcurrent elections and only a fraction of the costs of concurrent elections, local officials tend to view this move as a way to cut the costs of administering elections. For example, the city of Concord recently estimated that its cost for running a stand-alone election would be $58,000—more than twice as much as the $25,000 estimate for running a consolidated election (Simerman, 1998). However, local elected officials could have motivations for opposing a move to concurrent elections. By staggering local and statewide contests, local officials often have the opportunity to run for positions in state government without giving up their local positions. On-cycle local elections would limit that ability. Numerous local governments have already moved the dates of their elections to coincide with statewide elections, and many others have considered the change (Rohrs, 2000; Simerman, 1998). In Contra Costa County, for example, 55 cities, school districts, and special districts (out 18 of 73 total local governments) were running consolidated elections as of 2000—up from just 36 in 1996 (Rohrs, 2000). Our survey data indicate that at least 134 cities in California have changed the timing of local elections in recent years, with the vast majority of those switching from stand-alone elections to elections concurrent with statewide contests.6 We asked survey respondents to record the date of their most recent council and mayoral elections (if applicable).7 These election dates can be classified as falling into one of the following five categories: presidential election, presidential primary, gubernatorial election, oddyear November election, or completely off-cycle election (Table 2.1). A large fraction of the cities that responded to our questionnaire (20 percent) hold off-cycle city council elections, usually in the spring. Of the cities that directly elect their mayor, a similar percentage (17 percent) hold off-cycle elections. Roughly one in six cities hold mayoral or council elections in November of odd-years. In total then, about one- Table 2.1 Timing of California Municipal Elections Election Timing Presidential Presidential primary Gubernatorial Odd-year November Off-cycle Percentage Percentage of Council of Mayoral Elections Elections (N=397) (N=134) 12.6 8.2 5.8 6.7 44.6 52.2 16.6 15.7 20.4 17.2 NOTE: Data refer to the most recent election, or in some cases to the most recent contested election. ____________ 6Specifically, of the 308 clerks answering the relevant survey question, 94 (30.5 percent) indicated a change from nonconcurrent to concurrent elections, 3 (1 percent) switched from concurrent to nonconcurrent dates, 37 (12 percent) indicated a change from one nonconcurrent date to another, and 137 (56.5 percent) indicated no change. This calculation excludes 94 cities, or about one-quarter of the total number of respondents, who did not answer this question. Most likely these cities did not change the timing of their elections. 7Several cities whose last election was uncontested—that is, where the number of candidates equaled the number of seats—chose, helpfully, to report instead on their last contested election. 19 third of all the local elections that city clerks identified were nonconcurrent in one form or another (odd-year November or off-cycle). Thus, if election timing does matter and nonconcurrent elections are associated with particularly low turnout, there is clearly room for improvement. Another important question about election timing is whether mayoral and council elections are held simultaneously with other local contests, including those for other citywide offices (such as city attorney or treasurer) or for other governing bodies (such as school boards or county boards of supervisors). The presence of these other local elections might also serve to spur turnout. With this in mind, we asked city clerks whether a number of other local offices were being voted upon on that same day. In just over half (53 percent) of the cities responding to the survey, voters were casting ballots for at least one other local office. The offices most commonly elected on the same day as city offices were school board (31 percent) or county supervisor (17 percent). In response to a related question, 96 percent of the cities that directly elect their mayors held council elections at the same time; 25 percent of the council contests reported by the clerks had a mayoral election on the same day. Governmental Structure and Electoral Laws In addition to election timing, four factors identified by past research have the potential to affect turnout and are to some extent controlled by policymakers or voters. They are the method of electing council members, the use of term limits, the provision of services by the city, and the general form of government. District Versus At-Large Elections As part of their institutional reform agenda a century ago, Progressives sought to create at-large or citywide elections to choose members of the city council. Reformers of the time maintained that replacing district elections with at-large elections would weaken neighborhood and partisan interests in favor of less parochial citywide interests. Recent critics of the move, however, have suggested that atlarge districts not only decrease participation by distancing leaders from 20 their local constituencies (Bullock, 1990) but also may hurt the interests of geographically clustered racial and ethnic minorities who may be overwhelmed by a white voting plurality (Engstrom and McDonald, 1982; Welch, 1990; but see Bullock and MacManus, 1990). Our survey indicates that district elections are not widespread in California. Only 21 cities (5 percent) used the district form at the time of our survey. Another nine cities used some hybrid form that typically involved each district voting to nominate two candidates, with a citywide runoff between each pair.8 The vast majority of California cities (93 percent) use a simple at-large system of electing council members. Term Limits Term limits are a more recent electoral reform that may influence local voter turnout. One motive for term-limit proponents is to increase the competitiveness of elections. Term limits are seen as a tool to level the playing field, encourage more candidates to run for office, and ultimately make elections more dynamic (Copeland, 1997). However, an unintended consequence of term limits may be to increase voter confusion and disinterest by introducing a greater number of “unknown” nonincumbents into electoral contests (Rosales, 2000). At this point, it is unclear what effect term limits have had on either competition or turnout (Thompson and Moncrief, 1993). Californians favor term limits for many of their elective offices, at least at the statewide level.9 Term limits at the local level are far less common, however. Nationwide, about 10 percent of cities place limits on the number of times city council members or mayors can run for reelection (ICMA, 1996). In California, these proportions are somewhat higher. Our data indicate that nearly one in five city councils faces term ____________ 8Since these mixed systems have strong elements of district-level selection, we classify them with the district cities in the analysis in the chapters below. 9Californians voted in a bare majority (52 percent) in 1990 to support Proposition 140, which established term limits for state offices including the state assembly and senate. State voters were much more responsive to a state amendment to require congressional term limits—Proposition 163 passed with 64 percent of the vote in 1994— although this law was later ruled unconstitutional. 21 limits. Of mayors who are directly elected, 27 percent (36 cities) are term-limited. City Service Provision One of the most controversial issues in municipal governance is the question of whether city services should be directly provided by municipal employees or by some other entity. In an effort to reduce costs and provide more efficient services, numerous cities in California and elsewhere have chosen alternative arrangements for providing local services such as police protection or garbage collection. Often these services are performed by private firms, special districts, the county government, or other nearby local governments. Whether such arrangements ultimately reduce costs and improve service quality has been widely debated. However, it is likely that local elected officials in cities that do not provide such services with in-house staff have somewhat less direct influence; if nothing else, they have fewer city jobs to control. The reduced role for city staff and elected officials may have the unintended consequences of reducing interest in municipal politics and depressing voter turnout. To investigate this issue, we asked city clerks to indicate their cities’ arrangements for five common municipal services: police, fire, library, sewerage, and garbage collection. Clerks were asked to indicate whether each service was “mainly carried out by city government personnel, by county personnel, by a special district government, or by a private company under contract with the city.” Cities varied widely in terms of the number and type of services that were provided by other governments or private companies. The mean number of city-provided services was 2.4—roughly half of the five we asked about—indicating that contracting and alternative service arrangements are quite common among the state’s municipalities. However, 13 percent of cities indicated that city staff provided none of the services in question and 25 percent responded that they provided four or five of the services with their own staffs. Police, fire, and sewerage were the functions most likely to be carried out by city government personnel (Figure 2.1). At the other extreme, trash collection was rarely performed by city staff. 22 Percentage of cities 90 City Other government 80 Private Not provided 70 60 50 40 30 20 10 0 Police Fire Library Sewer Garbage Service Figure 2.1—Service Provision by Type and Provider The Form of Government in Cities Historically, California’s cities have been classified into two categories: charter cities and general-law cities. Although the distinction between the two categories has diminished over time (as noted in Chapter 1), charter cities (which tend to be the larger cities in the state) still have somewhat greater autonomy than general-law cities. For example, charter cities can revise certain aspects of their electoral institutions—such as the number of city council members—more easily and have slightly more discretion regarding procurement and personnel arrangements. This greater autonomy suggests that voters in charter cities have more at stake in local elections. Thus, we might expect voter turnout to be slightly greater in charter cities. Although we did not ask specifically about this distinction in the questionnaire, we were able to obtain data on the basic form of government for each city from the California Secretary of State’s office. Most of the cities (79 percent) responding to our questionnaire are general-law municipalities. Some 98 cities (or 21 percent of the sample) are charter cities. 23 We did inquire about the distinction between cities that have a council-manager form of government, where the day-to-day administration of the city is run by an unelected professional city manager, and cities with a mayor-council form of government. More than half of the nation’s cities currently operate under a council-manager form of government (ICMA, 1996). In California, however, nearly every city (97 percent, according to our survey) uses the council-manager form.10 Although nationally studies have found that cities with councilmanager governments tend to have less voter participation (Karnig and Walter, 1983; Espino 2001), there is simply not enough variation in California to test this argument. However, we also asked city clerks to report the extent of two mayoral powers—budgeting authority and the ability to veto council actions. Cain, Mullin, and Peele (2001) suggest that a more refined measure of mayoral power is more telling in California than the simple council-manager versus mayor-council distinction. Of the cities in our survey, only one has direct mayoral control over developing the budget (although in four other communities, the mayor is indicated as jointly participating in this process); 14 mayors had veto power over their councils (about 4 percent of the cities responding to the question). Local Structure Summarized On many of the local government features that we examine in our questionnaire, there is limited variation across California cities (Table 2.2). The vast majority of cities are council-manager governments with council members elected at-large. Mayors are directly elected in about one-third of cities but tend to have few independent executive powers. On other measures, California’s cities appear to be quite mixed. There is a fair degree of variation in the use of term limits and city service arrangements. These more recent municipal reforms (term limits and ____________ 10Considerably more cities directly elect their mayor than use the mayor-council form of government. In most of these cases, mayors are “first among equals” in council matters; the city manager retains control over important functions such as developing the city budget and hiring department heads. 24 Table 2.2 Local Government Structure in California At-large council elections District method Combination Percentage of Cities (Number) 93 (362) 5 (21) 2 (9) Term limits—council members Term limits—mayor (of directly elected mayors) 18 (71) 27 (36) Provide none of five city services with own staff Provide all five General law Charter 14 (54) 8 (30) 79 (309) 21 (83) Council-manager form Mayor-council form 97 (384) 3 (10) Mayor develops (or jointly develops) the budget Mayor has veto power 1 (5) 4 (14) NOTE: Percentages are of the number of cities answering each question. contracting) are less well established, thus far, than the ones initiated by Progressives a century ago. Electoral Context We also asked city clerks about several aspects of the context of specific local elections. Although not as amenable to legislative change as the structures of city government, the political context of city elections might also be expected to play a major role in influencing voter interest and participation. Here we consider various central elements of the local context: the presence of local ballot propositions, the degree of competition for office, the presence of incumbents, and the race or ethnicity of the candidates. Local Direct Democracy Although direct democracy has both proponents and critics, the initiative process does offer the potential to increase turnout (Hahn and 25 Kamieniecki, 1987; but see Magleby, 1994).11 Certain statewide initiatives such as Proposition 187, which sought to eliminate services to illegal immigrants and their children, have sparked greater participation by at least some sectors of the electorate, such as Latinos (Pantoja and Segura, 2000). These gains could also occur at the local level, given that the initiative process can also be employed in California’s cities and counties. Use of the initiative has grown rapidly at the state level, and it is now fairly common to see more than ten propositions on the statewide ballot. We asked cities to identify the number of local propositions put on the ballot, either by the city council or through the popular initiative system, in which citizen activists qualify a measure for the ballot by gathering voter signatures. According to the questionnaire responses, only 43 of the 387 cities (11 percent) had at least one municipal-level citizen initiative on their most recent ballot—indicating that despite the press attention they gather, citizen initiatives occur in relatively few cities each year. By contrast, 98 cities (25 percent) had a question placed on the ballot by the council.12 We also asked city clerks to identify the total number of citizen initiatives placed on their ballots from 1997 through 1999 to get a longer-term perspective on the use of direct democracy in local elections. About 27 percent of cities had citizen initiatives over this period, more than twice as many than in the single election we asked about. Competition One key factor that can influence turnout at any level of government is the degree of competition among candidates for office (Cox and Munger, 1989). Research has generally shown that the more competitive the race for a particular office, the higher voter turnout is likely to be. What is not entirely clear at the local level is whether it is the closeness of ____________ 11For accounts of many of the criticisms of direct democracy see Gerber (1999), Schrag (1998), and Rose (1990). 12Respondents were instructed to report only city-level ballot questions, not, for example, school district or county-level measures. 26 the contest that affects turnout or the sheer number of candidates vying for office. More candidates may be a sign of competing interests and greater divisions that could in turn increase (or conceivably even depress) turnout. Given the necessary brevity of a mail questionnaire, we were able to ask clerks only a limited number of questions about competition. Clerks reported the number of candidates running for office and the presence of incumbents. Because of the variety of methods used in selecting council members, we could ask about the margin of victory only for mayoral elections. The results indicate that elections in California municipalities are not strikingly competitive. Council elections average 2.5 candidates for each seat available; mayoral races attract slightly more but still fewer than three candidates on average. Perhaps more surprisingly, numerous cities in California have uncontested elections—15 council races (or 4 percent of all cities) and 23 mayoral contests (18 percent of all mayoral elections) were reported as having the same number of candidates as seats available. These figures may overstate the actual amount of competition and understate uncontested races by an unknown amount, because a handful of cities reported data for their most recent contested election, omitting a subsequent uncontested election. Indeed, cities with stand-alone elections sometimes choose to cancel elections that are uncontested and simply certify the lone candidate as the winner. For mayoral elections, we asked city clerks to identify the portion of the vote captured by each of the top two finishers.13 Close races indicate a competitive election that might spur turnout, although it is unclear that voters would know in advance that the outcome is likely to be close. As it so happens, mayoral races are rarely very close. In contested races, the average margin of victory was 24 percent. Only 13 percent of contested mayoral contests were decided by a margin of 5 percent or less. However, a few elections might lead one to recall the 2000 presidential ____________ 13For more than half of the mayoral contests, there were fewer than three candidates. 27 contest; three mayoral races, or 2.9 percent of the contested elections, were decided by 1 percent of the vote or less.14 Incumbency Although term limits might influence turnout rates by limiting the presence of incumbents, we also addressed the incumbency issue directly. Some have argued that because incumbents generally have widespread name recognition, can “scare away” quality opponents, and can use the resources of elected office to raise large sums of money, their presence in an election may depress turnout. The survey asked city clerks whether mayoral or council incumbents ran for reelection. For mayoral contests, we used a simple yes-or-no query. For council elections, we asked for the number of council seats up for election and the number of incumbents running. The results show that incumbents are present in most elections. Two-thirds (66 percent) of the council seats available in city elections had an incumbent council member running. Of the mayoral races, 72 percent featured an incumbent mayor defending his or her seat. As might be expected, cities with term limits had lower rates of incumbency, at least for council seats (55 percent). Oddly, the cities with mayoral term limits were slightly more likely to have an incumbent mayor running (77 percent), but this may be an aberration resulting from the low sample size (only 36 cities directly elect mayors and have mayoral term limits). Electoral Competition Across Races or Ethnicities In a diverse state with rapidly changing demographics, one might expect contests for local office to draw more attention or seem more salient if the contestants were from different ethnic or racial ____________ 14 The survey questions queried city clerks about the general mayoral election, not a runoff election. However, runoffs are quite rare, with only eight cities (6 percent) holding runoffs, of the 134 who directly elect the mayor. Even among those cities where the top candidate for mayor received less than 50 percent of the vote (because there were multiple candidates), only a quarter held a runoff. Nearly all of the communities with provisions for runoff elections are large central cities. Given the sizeable number of cities that elect a mayor with less than a majority of the vote, it is perhaps surprising that cities have not considered adopting preference voting schemes, which provide for an “instant runoff.” 28 backgrounds. For this reason, we asked the city clerks to identify the race or ethnicity of the first- and second-place mayoral candidates.15 Although not all respondents were willing or able to identify the candidates in this way, we collected such information for 122 victors and 97 second-place contestants (Table 2.3). As the table indicates, more than three-quarters of winners and runners-up were non-Hispanic whites. In most cases, the top two finishers in a city were of the same race or ethnicity. However, in onethird of the contested elections for which we have race/ethnicity identifiers for both candidates (32 of the 97 elections), the contest was between candidates of different ethnic or racial backgrounds. Most of these cases (21 of 32) involved a race between a Latino candidate and a white candidate.16 Table 2.3 Race or Ethnicity of Mayoral Candidates White, not Hispanic Black, not Hispanic Asian, not Hispanic Hispanic, of any race Other Total Winner 2nd-Place Finisher Number Percent Number Percent 97 79.5 75 77.3 5 4.1 7 7.2 1 0.8 2 2.1 19 15.6 12 12.4 0 0.0 1 1.0 122 100 97 100 Supplementing the Survey: City Demographic Characteristics As we noted earlier, local institutional characteristics and election context are not the only factors that can influence turnout. Individual demographic characteristics can also affect voter participation. Indeed, ____________ 15The large number of council candidates on a given ballot made it impossible to ask city clerks about the race of council candidates on the questionnaire. 16In addition, there were six black/white contests, three Asian/white contests, one Latino/black contest, and one white/other race contest. 29 the bulk of the existing research on voter participation has focused on such topics as socioeconomic status, age, mobility, and race. We therefore supplement our survey data with other data on local demographic characteristics derived mainly from the 2000 Census.17 It is clear from existing research that individuals with higher socioeconomic status participate in politics at significantly higher rates than those with less education or income (Verba et al., 1995; Wolfinger and Rosenstone, 1980). Higher socioeconomic status implies a host of resources such as time and cognitive ability that are important for participation. Several studies have also shown a strong relationship between age, or stage in the lifecycle, and propensity to vote (Rosenstone and Hansen, 1993). In general, younger residents are often not yet politically engaged and tend to be inactive voters. Elderly voters may suffer from physical infirmities, but some studies indicate they are as likely to vote, controlling for other factors (Wolfinger and Rosenstone, 1980), and perhaps to be more involved in community civic affairs than younger generations (Putnam, 2000). Race and ethnicity have also been tied to voter participation. Specifically, Latinos and Asian Americans have participated at lower rates than either whites or African Americans (Hajnal and Baldassare, 2001; Uhlaner et al., 1989). For Latinos, most if not all of this difference is accounted for by socioeconomic status and citizenship factors, but the reasons for lower participation rates among Asian Americans remain something of a mystery (Lien, 1994). Another individual demographic characteristic associated with turnout is residential mobility. Because of the various time and information costs associated with registration and voting, people who have moved recently are much less likely to vote than those who are longtime residents of a community (Squire and Wolfinger, 1987). ____________ 17City population size, however, is measured with state Department of Finance estimates for 1998, rather than 2000 Census data. This is because the largest number of local elections in our dataset (47 percent of council elections) took place in 1998; another 27 percent in 1999, compared to 22 percent who reported on 2000 elections and 4 percent reporting on various pre-1998 elections. In any event, the Department of Finance estimate is very highly correlated with the Census count (r = 0.9998). 30 Along with individual characteristics, some aggregate city-level characteristics can also play a role in voter turnout. Although there is disagreement about the size and the direction of the effect, city population size has been shown to influence turnout. In most cases, a larger population has been associated with less social interaction and lower political participation. Oliver (2000) has argued that as city size grows, citizens are both less interested in participating in local politics and less apt to be mobilized by local campaigns. In our study of city-level voter turnout, we are not specifically interested in any of these demographic factors because they cannot be readily manipulated to increase voter participation. However, to ensure that our analysis of city-level institutions and electoral context is not biased by differences in demographic characteristics across cities, we include controls for each potentially relevant local demographic characteristic. In each of the statistical models investigating voter turnout reported in Chapter 3, we include a specially constructed summary variable representing the socioeconomic status (SES) of the city population,18 along with controls for the percentage of the population age 18 to 24, percentage age 65 or over; percentages of African Americans, Latinos, and Asians in the population; percentage who lived in the same house for five years (residential stability); percentage institutionalized; and the city population size.19 However, because we are examining aggregate turnout rates and aggregate city-level ____________ 18Measuring SES creates difficult statistical modeling issues, because the various measures of status we are interested in (income, educational attainment, home ownership, etc.) are very closely correlated at the city level in California. Thus, we decided to use factor analysis to reduce four important variables that are highly related to one another— median household income, poverty rate, percentage of college graduates among adult residents, and percentage owner-occupancy of housing—into a single summary measure. The resulting factor score is the city’s principal factor loading for these four variables, with higher scores representing higher status. The SES score ranges from a low of –2.0 (city of Huron) to a high of 4.6 (Rolling Hills), with a mean of 0. The SES score is highly correlated with each of its component measures: income (0.98), education (0.84), poverty (–0.79), and home ownership (0.68). 19Because of extremely high collinearity (with percentage Asian and percentage Latino), we could not simultaneously include a measure of percentage immigrant in the model. 31 characteristics, we will not be able to infer any causal relationship about individual behavior from these measures. Summary It is apparent from the questionnaire results that fewer than half of all registered voters bother to come to the polls in municipal elections in California. Measured as a percentage of the voting-age population, turnout looks even more bleak. Just under one-third of voting-age residents turned out in these contests. However, because turnout rates do vary across cities, it is important to determine the timing, institutional, and contextual factors that might shape local participation. We have highlighted several of those factors in this chapter, showing for example that a sizeable number of city elections are held on nonconcurrent or “off-peak” election dates, and that other Progressive Era reforms such as at-large elections are the norm among California cities. Electoral context also varies widely across the city elections we examine. In the next chapter, we analyze the relationships between voter turnout and the structural and contextual factors identified here. 32 3. Accounting for Differences in Voter Turnout Across Cities Voting gives Californians of all races, classes, and beliefs a voice. Through the vote, Californians communicate information about their interests and needs and make important decisions about who should hold office. Yet, as we showed in the last chapter, voter turnout in city elections is low. Well below half of registered voters—possibly a nonrepresentative sample of Californians—are deciding the outcome of political contests at the municipal level. These overall levels of voter turnout are important but they obscure as much as they reveal. Every city is different, and turnout rates vary dramatically across communities (Figure 3.1). 80 70 60 50 40 30 20 10 0 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70–79 80–89 90–100 Percentage of registered voters casting ballots Figure 3.1—Distribution of Voter Turnout Rates in City Council Elections 33 Number of cities Some cities attract many voters to the polls. At one extreme, 89 percent of registered voters in San Marcos turned out for that city’s November 1998 council election. Milpitas was just behind, with 88 percent voter turnout for an election that same day. In sharp contrast, other cities in the state are plagued by extremely low turnout. Lawndale saw only 10 percent of the city’s registered voters participating in its April 2000 election. Even in some major central cities where local politics is much more visible, the story is not much better. Only 20 percent of registered voters in Los Angeles turned out for the April 1999 city council election. What is it about cities like Lawndale and Los Angeles that leads to such low voter participation? Is there anything unique about cities like San Marcos and Milpitas that explains their ability to involve large majorities of the registered population in local elections? If so, can those explanations help increase citizen participation in other local elections? In this chapter, we explore some of the possible systemic factors that underlie this variation and help account for voter turnout rates. We first examine the issue of election timing, which we expect to be of central importance in accounting for variation in voter turnout. Next we examine local governmental institutional arrangements, such as term limits and the degree of city service provision. We then turn to a set of contextual factors, including the presence of propositions on the ballot and the degree of competition for office. Finally, we discuss a series of individual- and city-level demographic characteristics. Throughout the chapter, we analyze a dataset that combines both council and mayoral elections across the state. We combine the two types of elections because of the small number of mayoral contests for which we have sufficient data and because we believe that the dynamics of turnout are nearly identical for mayoral and council contests. Of the 134 respondents indicating that their cities elect mayors directly, we received usable mayoral voter turnout data from 130, but additional missing data (because of other items on the questionnaire being left blank by a few respondents) dropped the analysis down to 122 observations, limiting the statistical power available. When we examined mayoral and council elections separately, results were generally quite similar to those presented here. The same factors that influenced turnout 34 in council elections tended to affect turnout in mayoral elections. However, the small number of mayoral elections did lead to less stable and less robust estimations for these contests. Thus, the discussion focuses on the combined mayoral and council election dataset, for which we have 421 usable observations in the turnout model.1 In each regression, we include a binary variable to estimate differences between mayoral and council contests.2 The regression analysis technique used here allows us to focus on the relationship between various specific factors and voter turnout rates, while holding constant the effects of each of the other variables in the model. We ran two parallel estimations, one geared at explaining the turnout of registered voters and one geared at explaining the turnout of adult residents. The registered-voter model is able to account for 61 percent of the variation in turnout levels across city elections, whereas the model relating to adult-resident turnout accounts for 67 percent of variation. (For complete regression results, as well as summary statistics for all independent variables, see Appendix C.) Election Timing One notable factor in the comparison between San Marcos and Milpitas on the one hand and Lawndale and Los Angeles on the other is that the former pair of cities held elections on a statewide gubernatorial election day, whereas the latter two held nonconcurrent, or off-cycle elections in the spring. As it turns out, our statistical analysis indicates that election timing is, in fact, the most important factor influencing city ____________ 1We were able to calculate council election turnout rates for 350 cities, but some had to be dropped because certain questionnaire items on electoral context that are used in the regression (questions 17 and 18) pertained to the most recent mayoral election, for those cities with directly elected mayors. Fortunately, three-quarters of respondents reported on mayoral and council elections that took place on the same day, and thus their council elections could be retained in the regression analysis. Other cities had to be dropped because of nonresponse to other survey items or because the city was incorporated after 1990 and therefore lacked data for a few variables derived from the 1990 Census. 2When we ran a model that clustered, or grouped, the observations by city, signs and significance levels did not change. 35 turnout (see Figure 3.2).3 Compared to off-cycle elections, and controlling for a host of other factors, presidential elections are associated with almost 36 percent greater turnout among registered voters and 23 percent greater turnout among adult residents. Presidential primaries and gubernatorial elections are associated with an increased turnout of 21 to 26 percent among the registered and 11 to 15 percent among adults.4 When we examined mayoral and council elections separately, these relationships held true for both types of elections, although the association was somewhat stronger for council contests. Another important aspect of city election timing that could potentially affect turnout is whether city elections are consolidated with elections for other local offices. More than half of all the council and 70 60 Registered voters Adult residents 50 Median percentage turnout 40 30 20 10 0 Off-cycle Odd-year Gubernatorial Presidential Presidential November primary Figure 3.2—Variation in Turnout by Election Timing (with controls) ____________ 3Figure 3.2 and the other bar graphs in this chapter show the relationship between voter turnout rates and the variables of interest, controlling for all other variables in the regression. Specifically, continuous control variables are set at their medians (to represent a typical city), and categorical control variables are set at their modal value. 4Across the two different types of nonconcurrent elections, there is little difference in turnout. That is, with controls, odd-year November elections do not have significantly higher levels of turnout than other off-cycle elections, which are generally held in the spring. 36 mayoral elections we examined were held alongside elections for local officials such as city treasurer, school board members, or county supervisors. Does this have any effect on turnout? It appears that the answer is no. Furthermore, holding the mayoral and council election on the same day does not seem to spur significantly higher turnout. In short, consolidating local elections with statewide elections—as opposed to consolidating various categories of local elections—is the step most likely to yield local turnout gains.5 Taken as a whole, concurrency has a huge influence on whether registered voters will participate in municipal elections. Election timing alone accounts for about half of the variation among cities in turnout of registered voters; for turnout of adult residents, timing accounts for about one-third of the variation across cities.6 In fact, using some simple calculations involving a number of reasonable assumptions, we estimate that if all cities in the state that hold nonconcurrent elections switched their elections to a presidential election date, approximately 1.7 million more Californians would have voted in their most recent city council contest. The estimated turnout increase would be smaller, but still very substantial, if the spring and November off-year elections were moved to the same date as a gubernatorial or presidential primary election.7 ____________ 5Among these timing variables, we have also included a variable indicating whether the election in question was mayoral or council. Notably, mayoral elections tend to draw lower turnouts (with about 3 percent fewer voters casting ballots) than council elections, at least among registered voters. This finding may seem somewhat counterintuitive, since the mayoralty is generally a more visible and important office than individual council seats. However, it does accord with our finding in the last chapter that, overall, council elections in California have higher turnouts. It is conceivable that there is some systematic, unobserved difference between cities that have a separately elected mayor and those that do not—a factor that may be associated with turnout. Given the wide variety of variables that we control for in the regression model, however (notably including the socioeconomic status of the city and the degree of competition for office), it is difficult to imagine what this unaccounted-for factor might be. 6A regression explaining turnout rates and using only the timing dummy variables explains 51 percent of variation for turnout of registrants and 32 percent for turnout of adults. The coefficients of the timing variables are comparable to those in the full model. 7We estimate a turnout increase of about 1.2 million if all nonconcurrent elections were switched to gubernatorial election dates, and about 940,000 if switched to presidential primary dates. We had to make a number of assumptions in these estimations, the most important of which are: that the percentage of city residents who are registered voters is the same in cities that did not respond to our survey as in the cities 37 Nonconcurrent elections clearly are much less inclusive than concurrent elections, on average. Closer examination of the data, however, indicates that there is substantial variation within each electionday category, particularly for nonconcurrent elections and those held on presidential primary days. For example, despite the generally abysmal turnout rates for cities on off-cycle dates, two municipalities with nonconcurrent elections had turnouts of over 70 percent. It is therefore important to move beyond election dates to examine other factors that may influence voter participation rates. Institutional Factors Are the electoral laws and governmental structure of a city related to local voter turnout? According to our estimates, at least one institutional feature of the local government—the degree of city service provision—has an important connection to turnout.8 Many cities in California, particularly in the Los Angeles area, provide only minimal services with their own staff. Instead, they make arrangements with other governments or private firms to provide such functions. Cities that provide these services themselves may find that residents have more at stake in municipal politics. When services are contracted out or provided by other governments, city officials have less direct control over some of the issues that affect city residents’ quality of life the most. The results indicate that a city’s role in service provision influences turnout, if modestly. Each additional service provided by city ________________________________________________________ for which we have data; that the shares of registered voters living in cities having springtime and off-year November elections, respectively, are the same as those shares among the cities in our dataset; and that the factors influencing turnout are the same in cities that are not in our dataset. We estimated that there are just over 12 million registered voters living in municipalities in California, of whom about one-fifth live in cities holding off-cycle spring elections and one-fifth in cities holding off-year November elections. We then calculated how the number of voters casting ballots for city council in these nonconcurrent-election cities would be likely to change based on election timing, using the estimated turnout rates for each type of city shown in Figure 3.2. 8We did not use the institutional variables relating to form of government (mayorcouncil versus council-manager plan), mayoral veto power, or mayoral budgeting authority, because of the lack of variation among California cities on these characteristics. Also, all local California elections are nonpartisan, eliminating another potentially important institutional variable. The district election variable used in this analysis pertains only to council elections, since all directly elected mayors are chosen at-large. 38 staff, of the five services asked about in the survey (fire, police, library, sewerage, and garbage), is associated with approximately 1 percent higher turnout among registered voters (Figure 3.3). This relationship is weaker and statistically insignificant for turnout of adult residents. However, when we substituted a binary variable indicating whether the city provides any of the five services, it was positive and statistically significant for turnout among adult residents as well as registered voters. This finding indicates that there is a particular “turnout penalty” for minimal-service cities—those whose in-house employees provide none of the five services asked about. It is possible that voters find municipal elections in such cities to be less important to them, thus accounting for reduced turnout levels. It is also important to note that municipal employees are often among the most likely residents to participate in local politics, and this pattern may also help account for the lower turnout among cities with fewer in-house staff. A somewhat anomalous relationship is apparent between district council elections and voter turnout, but only in the estimation of turnout among adult residents. We suspected that district elections might draw slightly higher turnout, because district council members might be more 70 Registered voters 60 Adult residents 50 Median percentage turnout 40 30 20 10 0 None Three Five Number of services provided by city staff Figure 3.3—Variation in Turnout by City Service Provision (with controls) 39 closely involved in neighborhood-level issues that are more salient to residents, and voters might thus feel a stronger bond with their representative. However, the estimation indicates a significant negative relationship between district council elections and turnout of adult residents. It appears, however, that this outcome is something of a statistical artifact. The relatively small number of cities in California with district elections have substantially lower rates of voter registration than cities with at-large council elections.9 If we control for the registration rate of the city, district elections are not significantly related to the turnout of adult residents. Thus, we see no strong evidence of a causal connection between the method of council election and voter turnout per se. Two other institutional features of local government had no significant association with voter turnout. Whether a city is chartered or is covered under the state’s general-law provisions for municipal government appears to have no relationship with the share of voters going to the polls. Perhaps surprisingly, we also found no connection between term limits and turnout.10 We hypothesized that term limits would make elections more interesting to residents and therefore increase turnout, but it is also possible that the effect of term limits depends on whether or not an incumbent runs. When an incumbent can no longer run for a seat because of term limits, the election may become more competitive and attract additional interest. We turn to this and other electoral context variables in the next section. Electoral Context Whether voters participate in local elections may be related to the specific context of the elections. How many candidates are running? Is ____________ 9For the 17 cases we observe of a district council election, the estimated registration rate of adults in the city averages 54 percent, compared to 71 percent in the overall sample. This is probably because several of the district cities have very high shares of immigrants in the population, many of whom are ineligible to vote. Examples include Santa Ana, Watsonville, Salinas, and Downey. 10The term-limits variable is specific to the contest examined; that is, the variable measures the influence of council term limits for council-election turnout, and of mayoral term limits for mayoral-election turnout. 40 there an incumbent or is the contest for an open seat? Are there ballot measures to decide? Although it is difficult to influence most of these factors through policy efforts, it is nevertheless important to know what role they play in the local political arena. For example, one of the most controversial elements of California’s electoral system is direct democracy. Despite some criticism, policymaking in California has relied increasingly on initiatives, at least at the statewide level. In some cases, these initiatives have spurred voter interest and participation (Pantoja and Segura, 2000). At the municipal level, as noted in Chapter 2, ballot measures are actually far less prevalent. Moreover, most city propositions reach the ballot through council action, not as the result of citizen initiatives. Although not widespread, local ballot propositions have a significant positive relationship with city turnout rates (Figure 3.4). Cities with at least one citizen initiative or council-placed measure tended to have about a 4 percent higher turnout than those without either. This is true for turnout of both registered voters and adults.11 Another contextual issue to consider is competition for office. The number of candidates competing for a seat is one rough indication of the level of competition and perhaps the level of media and other interest in a given contest. At one extreme, uncontested elections (those with the same number of candidates as seats to be filled) may draw especially low interest. We measure competition with two variables: a binary variable signifying uncontested elections and the number of candidates per seat. For mayoral elections, the latter variable is simply the number of candidates on the ballot; because we do not have information on each specific council contest, we use the number of total candidates running for any council seat divided by the number of seats up for election. ____________ 11Oddly, if entered into the equation separately, only council-placed ballot questions are significantly related to election turnout. The presence of citizen initiatives has a positive relationship to turnout but is statistically insignificant. Despite the extensive attention given to local citizen initiatives, it appears that some measures placed on the ballot by the governing body may be more provocative to voters. 41 70 Median percentage turnout 60 None One or more 50 40 30 20 10 0 Registered voters Adult residents Figure 3.4—Variation in Turnout by Number of City Ballot Propositions (with controls) The results clearly show that uncontested elections are bad news for voter participation in city elections. Uncontested elections draw to the polls about 4 percent fewer adult residents, and about 5 percent fewer registered voters. (Further analysis indicated that this effect was strongest in mayoral elections.) Moreover, for contested elections, the number of candidates appears to be associated with a slightly higher turnout rate. One additional candidate per seat is associated with about 0.7 to 0.9 percent higher turnout. It is possible that each aspirant mobilizes his or her core supporters to come to the polls; therefore, a larger number of combatants would be associated with more voter participation overall. Regardless of the explanation, these results hint that one dimension of local civic vibrancy—competition for office—is connected with another dimension—voter participation. Incumbents at all levels of the political process are extremely likely to win reelection, and one might expect turnout to decrease when incumbents are running. We measure incumbency as the number of incumbents running divided by the number of seats. This variable ranges from zero (no incumbents running) to one (all seats have incumbents); in mayoral elections, of course, the resulting value can only 42 be zero or one since there is only one mayoral seat. Surprisingly, incumbency proved not to be related to turnout at all.12 Also of interest were two other factors that could not be included in the analysis because of high rates of nonresponse: the margin of victory in the election and the issue of whether the top two candidates for office were of different racial or ethnic backgrounds (as discussed in Chapter 2). Although neither variable is included in the final regression models presented here, no significant associations were found between these variables and turnout rates. Unfortunately, these data could feasibly be collected only for mayoral elections, and nonresponse to these questions substantially reduced the number of potential observations. Thus this “nonfinding” cannot be considered definitive. One additional electoral context variable was included in our model of turnout among registered voters as a statistical control: the (estimated) percentage of voting-age residents in the city who are registered to vote.13 This share of potential voters who are registered could be important because it provides some indication of the efforts made to involve residents in electoral politics and helps define the pool of citizens among whom voting will or will not occur. We do not include this variable in the regression for turnout of voting-age residents; in that case, the act of registration is itself a large part of the phenomenon we seek to explain. In cities where a very high share of residents are registered, it is possible that some “low-propensity” voters have been registered, perhaps through especially active outreach efforts by community groups or candidates. Another possibility is simply that the local voter rolls have not been “purged” recently, meaning that there may be a fair number of citizens listed as registered who have died or moved out of the community. In either situation, one might expect that the turnout of registered voters will be lower where the registration rate is higher. This ____________ 12We tried other specifications for the council incumbency variable, such as binary variables for situations where any incumbents were running or where all seats had incumbents, but this did not influence the results. Even when we looked at mayoral elections separately, incumbency had no significant effect on turnout. 13As noted in previous chapters, the variable used is an estimate based on an interpolation (for the particular year of the election) of the voting-age population of the city between 1990 and 2000. 43 is indeed the case, although the relationship is a weak one. Where registration rates are 10 percent higher (all other things equal), turnout is about 1.4 percent lower.14 City Demographics The final set of factors we assess are demographic. Other studies have shown such characteristics to be very important for voter turnout when measured at the individual level. Here we examine their role in California’s municipal elections, viewing them primarily as statistical controls. The model includes several variables of interest to turnout, as described in Chapter 2: city population size, a summary measure of the socioeconomic status of the city, measures of the racial and ethnic makeup of the city, two measures of the age distribution of the local population, residential stability, and (for turnout of adult residents) the proportion of the city population that is institutionalized. One clear-cut finding is that larger cities tended to have lower turnout rates. This generalization is supported by research at the individual level, which finds that residents of more populous jurisdictions generally participate less (Oliver, 2000). Furthermore, the relationship between larger city populations and lower voter turnouts is apparent even when controlling for other factors that may differ between large and small communities. Nevertheless, the magnitude of the relationship is fairly modest (Figure 3.5). Cities with residents of higher socioeconomic status tended to have significantly higher turnout; this result is also consistent with individuallevel research. This pattern is particularly clear for the turnout rates of adult residents, but it is also readily apparent even when the focus is narrowed to those residents who have already been registered to vote. Race and ethnicity appear to be related to turnout in more complex ways. In accordance with past research suggesting that African ____________ 14We also investigated the effects of party registration—the percentage of city residents registered as Democrats, Republicans, or neither—on turnout. This element of electoral context proved to have no influence on turnout. 44 Registered voters 70 Adult residents 60 Median percentage turnout 50 40 30 20 10 0 20,000 50,000 100,000 Population 1,000,000 Figure 3.5—Variation in Turnout by City Population (with controls) Americans participate at rates roughly equivalent to whites, we find no relationship between the proportion of blacks in the city and turnout (Hajnal and Baldassare, 2001). However, the presence of large Asian American or Hispanic populations is associated with lower turnout. Even after controlling for socioeconomic status, the Asian American and Latino share of the population are both tied to lower turnout among adult residents. These results are consistent with the view that citizenship and language barriers make it especially hard for these two immigrant groups to register at rates equivalent to those of whites and African Americans. At the same time, the results indicate that a greater Hispanic share of the population is not associated with lower turnout of registered voters—a finding that seems to support existing research showing that once socioeconomic status and citizenship are accounted for, Latinos participate as much as whites and African Americans (Hajnal and Baldassare, 2001). By contrast, a high proportion of Asian Americans in a city is tied to lower turnout of registered voters. Once again this finding is consistent with individual-level research, which finds that Asian Americans tend to vote significantly less regularly than 45 members of other racial and ethnic groups when controls for citizenship and socioeconomic status are introduced (Lien, 1994). Turnout is also clearly correlated with both the age distribution of the population and household mobility. Cities with higher shares of senior citizens have higher turnout rates. The proportion of young adults in the population, however, appears unrelated to city turnout rates. Household mobility—that is, the proportion of residents who have changed addresses in the past five years—is connected to turnout, with cities of more household stability showing higher turnout among the adult population. Among registered voters, however, mobility levels of the city population are unrelated to turnout. Registration itself seems to be the main hurdle for people who move frequently. Finally, in estimating the turnout of all adults, we draw upon the 2000 Census to measure the proportion of the city population that resides in institutional settings. Our intuition here was that cities with very high institutionalized proportions of the population tend to be the sites of prisons, the inmates of which generally are ineligible to vote. The findings confirm that the institutionalization rate is negatively associated with turnout.15 Summary Although turnout in local elections in California is generally low, there is, in fact, wide variation across cities. Some localities manage to attract the vast majority of registered voters to the polls; others are plagued by extremely low turnout. Our statistical model was designed to ascertain which factors play a significant role in accounting for the substantial variation in voter turnout across California cities. The results show that the largest part of this variation can be explained by timing— the date on which the election is held. Concurrent elections—that is, city elections held on the same day as a presidential or gubernatorial election or a presidential primary—had turnouts that are substantially higher than off-cycle, or nonconcurrent, city elections. The turnout boost for cities holding concurrent elections was 11 to 23 percent among ____________ 15Such individuals would not generally be registered to vote locally, making the issue moot for the turnout of registered voters. 46 all adults, and 21 to 36 percent among registered voters (depending on whether the concurrent election was presidential, gubernatorial, or a statewide primary). Beyond timing, there are a handful of other interesting correlates of turnout. Cities that provide only a minimal number of services with their own staff drew a smaller share of voters than those that provide more services with in-house personnel. Direct democracy plays a role, too; cities with local ballot questions placed before the voters tended to have higher turnout. City elections with a greater number of competitors had slightly higher turnout. All of these factors, however, pale in comparison to the dramatic role played by election timing. Although competition for office and incumbency did not appear to play a decisive role in influencing local turnout, these factors are worthy of additional study in and of themselves. Potential candidates for office are likely to know that turnout is much higher in concurrent elections, and this knowledge may affect their decisions about running for office. And do incumbents fare better in concurrent or nonconcurrent elections? If policymakers consider rescheduling local elections to enhance turnout, they may wish to bear in mind these other potential dynamics of electoral competition. The next chapter addresses these issues. 47 4. Competition and Incumbency in Local Electoral Contests Thus far, our discussion of municipal elections in California has mainly been concerned with voter turnout. However, there are other interesting measures of the vibrancy of local electoral contests. This chapter addresses two such issues: the level of competition for mayoral and council races and the success of incumbents who are seeking reelection. Who runs for city office—and who wins—are different concerns from the level of voter participation, but as we have noted, the issues are linked, since more competition tends to increase turnout. Moreover, as we shall indicate, the question of election scheduling— concurrent as opposed to nonconcurrent elections—may have implications for local candidacies. Competition for Office Local government is the starting point for many political careers— that is, it is the level at which people tend to be recruited into elective politics. As such, the degree of competition for local offices may have broader implications for the quality of democracy. Competition for mayoral and council seats provides a barometer of how much interest and perceived opportunity local politics generates among one type of citizen—those who might consider serving in elective office. Our measure of competition for office is simple: for mayor, the number of candidates whose names appeared on the ballot, and for council, the total number of candidates divided by the number of seats available.1 To estimate a more robust model, we once again merged the ____________ 1Respondents were asked to report on the most recent regular election, not a runoff election or a special election to fill a vacancy. 49 council and mayoral elections, distinguishing between the two where appropriate. The average mayoral contest in California for which we have data had 2.9 candidates running, although just over one-sixth of these elections were uncontested. Among city council elections, each seat attracted an average of 2.5 candidates; less than 4 percent of cities reported that all their council races were uncontested. However, there is wide variation in the number of candidates, with some cities having as many as eight candidates per council seat. It is impossible to capture in a statistical model all the nuances of the local political environment that might motivate individuals to seek city office. Nevertheless, many of the factors we explored above might also influence the vibrancy of electoral competition in cities. Thus, we once again investigate issues of election timing, the institutional arrangements of local government, the electoral context, and the demographic characteristics of the municipality. However, we alter some of the specific variables used in an effort to select measures that are more meaningful in influencing whether city residents will seek to run for office. In estimating the relationship of election timing to competition, we use a simple binary indicator of whether the city election was concurrent with a statewide election day (i.e., a presidential, presidential primary, or gubernatorial election) as opposed to cases where the city held a nonconcurrent election. We suspect that potential candidates will not make fine-grained distinctions about their chances in a gubernatorial versus a presidential election, for example, but they may evaluate their candidacy based on whether the city election will be in electoral “prime time”—and thereby draw a much larger voter turnout, as we have seen. We also use variables indicating whether the mayoral and council elections were held on the same day, and whether other local elections, such as for school board, were held that day. In the case of some of the institutional factors, there may be less of a clear link to competition than to turnout. We do expect term limits to affect competition, in two possible ways. On one hand, term limits may increase competition by providing a clear indication to potential candidates that a seat will be available once an incumbent has been 50 “termed out.” Conversely, the presence of term limits conceivably might reduce the attractiveness of public office and make officeholding a less desirable goal, thereby reducing electoral competition. Along these lines, we very strongly expect the presence of an incumbent to reduce competition, because challengers are well aware of the electoral benefits known to accrue to incumbents, such as name recognition and fundraising potential. The degree of city service provision could affect the salience of city politics for potential candidates, just as it appears to for voters. And two additional features of the mayoralty—whether the position of mayor is considered a full- or part-time job, and the length of the mayor’s term of office—may shape its attractiveness as a goal for potential candidates.2 Charter city status, which provides some nominally increased discretion to city officials, might also potentially affect the number of candidates. Some of the electoral context factors have a less direct, but nevertheless potentially important, link to electoral competition. We include our estimate of the percentage of voting-age residents registered to vote, as this may be an indication of the degree of political engagement in the city, which could also affect competition for office. Simultaneous elections for other local offices may provide alternative outlets for individuals interested in entering local politics, therefore perhaps drawing fewer council or mayoral candidates. The presence of local ballot measures may indicate a degree of community conflict that could motivate more competitors to seek office. Because we are interested in the general level of political interest or rancor within a given community, we turn to broader measures of political engagement. Instead of election-specific indicators of ballot measures, we use a general measure of the number of voter initiatives across a three-year period ____________ 2Regarding the first issue, city clerks were asked, “Is the mayoralty in your city considered a full-time or part-time position? (Generally, the office is considered full-time only if it carries a full-time salary.)” The office was identified as full-time in only 13 cities. In the regression analysis below, the variable used is an interaction variable denoting only mayoral elections in cities with full-time mayors. It is worth bearing in mind that only 2.5 percent of the observations in the regression are of mayoral elections for full-time mayors. Regarding the second issue, about half (49 percent) of directly elected mayors were said to hold a four-year term, whereas the other half held two-year terms of office. 51 (1997–1999) and a subjective response by city clerks indicating the general level of electoral controversy over the past five years. We also include a measure of the percentage of registered voters in the city who are third-party adherents or who decline to state their party affiliations. Even though California city elections are officially nonpartisan (that is, candidates are not identified on the ballot by party affiliation), cities with a more independent electorate may be more critical of the political status quo and have a more nonconformist political style. Moreover, a high percentage of independents may indicate that local party organizations—which often are important in recruiting and screening potential candidates—are weak. Under such circumstances, local parties may be less capable of signaling who the leading candidates are through endorsements or financial contributions. In such a wide-open political arena, more candidates may seek office, believing they have a credible chance at swaying voters and winning. Finally, we explore the relationship between selected local demographic factors and candidate competition. Here our analysis takes us into fairly uncharted waters. Although a city’s population size will almost surely be related to the number of candidates who choose to run, it is not clear from extant theory that the racial or ethnic composition and socioeconomic status of the residents of a city should influence electoral competition—but there is no way to find out for sure without testing. Preliminary analyses confirmed that there were no noticeable effects from socioeconomic status variables, which are therefore omitted from the model; race variables were retained. We also include the measure of residential stability in the city—the proportion of residents who lived in the same dwelling unit from 1985 to 1990—on the thought that more transient communities may present different patterns of candidate recruitment and mobilization from more settled communities. Variables relating to the proportion of senior citizens and young adults in the community were also employed. Retired residents might have more time and interest in local politics, thus increasing the pool of potential candidates. If young adults are especially disengaged politically, cities with many of them may have fewer electoral candidates. We used a regression analysis to estimate which factors are related to competition for office. Detailed results of the model are shown in 52 Appendix D.3 The first finding of note is that mayoral elections draw significantly fewer competitors than council seats, all other things being equal. (Without controls, as we have noted, mayoral races draw slightly more candidates than do council races.) This may reflect the higher cost in time and campaign funds often required for serious mayoral bids. Novice candidates are probably more likely to give politics a try at the council level. The second finding to stress is the lack of a relationship between election timing and competition for office. That is, the number of candidates is not influenced by whether the election is concurrent with a statewide election. Another type of election concurrency does make a difference, however. The presence of other local offices up for election (such as city clerk, school board, or county supervisor) is associated with a significant reduction in the number of aspirants for council seats or mayor. In such circumstances, some potential candidates are possibly being siphoned off into other local contests. An election that deals only with the council and mayor, on the other hand, offers ambitious local campaigners the opportunity to run for those offices and then try again later for a different local position if they lose the council or mayoral race. Certain institutional factors also influence electoral competition in California cities. Neither district elections nor the number of services provided by city government staff produces discernible differences in competition.4 However, a full-time mayoralty is associated with more competition for that office, as is a longer term of office. In short, where the rewards of mayoral office-holding are greater (or where mayoral elections are less frequent), the number of candidates for mayor is greater. In addition, term limits—perhaps by making the office seem less desirable—seem to reduce the level of competition (although the statistical significance is not high). It is likely that term limits also indirectly affect competition by restricting the number of races featuring incumbents—an electoral ____________ 3The dependent variable in the regression (candidates per seat) is expressed in logarithmic form, because of the extreme skewness of the distribution (i.e., many small values and only a few quite large values). 4The district-elections variable applies only to council elections. 53 context variable. Indeed, the presence of an incumbent running for office substantially reduces the number of candidates for mayor and council elections. This finding is consistent with research on congressional elections, which shows that the presence of an incumbent discourages potential opponents from running; candidates for open seats (those with no incumbent) tend to be of “higher quality” (i.e., have more experience in government and politics) than challengers who run against incumbents (Jacobson, 1983, pp. 37–41). Certain other electoral context variables are also related to the level of competition. As one might expect, controversy and a more engaged public appear to be related to greater levels of competition for office. First, general electoral controversy in a city (as judged by the city clerk) has a significant positive association with the number of candidates per seat.5 Second, cities that have had more citizen initiative elections in recent years produce more candidates, perhaps indicating an electorally charged atmosphere with salient issues facing voters and politicians. The percentage of adults registered to vote, another potential indicator of an interested and engaged public, has a positive relationship to competition, but does not attain standard levels of statistical significance. The results also indicate that cities with larger proportions of independent or third-party voters have significantly more competition. (Further analysis showed that this pattern is apparent both for council seats and for the mayoralty.) This finding may reflect the fact that local Democratic and Republican party organizations help winnow the field of candidates in many cities—even in California’s officially nonpartisan electoral system—through their endorsements and campaign support. Major-party organizations generally wish to avoid having their loyalists split their votes among multiple competitors. Where the major parties are weaker (as indicated by a higher proportion declining to state and third-party voters), they may be less effective as “gatekeepers” to office. Potential aspirants to local office also may be encouraged by the presence of a larger share of “free agent” independents among the voting public. ____________ 5It is perhaps unclear whether greater controversy spurs more candidates into action or is itself a function of the clashing views of a large number of candidates. Nevertheless, the questionnaire asked respondents to judge the level of controversy over the past five years, not the controversy level of the specific election for which we have data. 54 Turning to demographic factors, the city characteristic that has the greatest association with competition is the size of its population. Large cities tend to have more candidates, as one would expect.6 Residential stability is associated with fewer council aspirants, perhaps indicating a more consensual, less competitive politics in cities that have a stable, less transient population. Municipalities with higher proportions of blacks and Hispanics in the population have slightly more competition for office. Although we are only speculating, political elites within these two racial/ethnic groups may see a particularly important link between their group interests and the actions of local governments; this link, in turn, may lead more citizens to run for office. There is no such relationship between the size of the Asian population in a city and competition for office. Finally, although the shares of young adults and elderly persons in the population have the expected relationship with electoral competition, neither factor proves statistically significant. The Success of Incumbents Incumbency is a key factor in shaping the degree of competition for office. Indeed, one of the best-documented regularities in electoral politics is the advantage of incumbency. At the federal level, for example, most congressional elections since 1952 have returned over 90 percent of incumbent members of the House of Representatives and about 80 percent of incumbent senators (Squire et al., 1995, pp. 280– 281). At the mayoral level, available evidence suggests that nationwide, 85 percent of incumbents win their reelection bids (Wolman et al., 1990). Whether the power of incumbency extends to the city council level is less well known (but see Krebs, 1998). Also not well known are the factors that lead to incumbent success at the local level. ____________ 6However, the number of candidates does not increase proportionally with the population. That is, there are fewer candidates per thousand city residents in large cities than in smaller communities. As with the lower turnout among large cities noted in Chapter 3, this finding may indicate somewhat lower levels of civic vibrancy in jurisdictions with large populations. (Volunteering and social trust are also behaviors that are less common in large communities, according to Putnam, 2000.) However, lower levels of competition for office in large cities may also be a reflection of the higher financial costs involved in campaigning there, as paid media advertising and direct mail often supplant yard signs and door-to-door campaigning. 55 Our California data indicate that 80 percent of council members and 86 percent of mayors who stood for reelection won their contests. This result underscores the tremendous value of incumbency to political candidates at all levels of politics. Incumbency increases name recognition among voters, deters potential challengers, and very likely enhances perceptions of a candidate’s experience and accomplishment. Nevertheless, some incumbents do lose, and one might ask under what circumstances this outcome is likely. Is incumbent success related to voter turnout? Perhaps more important, is incumbent success linked to the timing of elections? Some local election officials and city residents have expressed concern that concurrent elections might favor incumbents (Simerman, 1998). These critics believe that holding municipal elections on the same day as a statewide contest could create a situation in which some voters who come to the polls are unaware of and inattentive to local issues. Such voters, it is argued, have a tendency to vote for incumbents, who are better known, on average, than their challengers. In contrast, by attracting a higher proportion of purportedly better-informed voters, nonconcurrent elections might make incumbents less secure and ultimately make council members and mayors more responsive to the residents of their city. We cannot directly test the responsiveness of mayors and council members with the available data, but we can analyze the success of incumbents standing for reelection in California. Only contested elections are examined because unopposed incumbents win automatically. We analyze the aggregate number of incumbents who win reelection in various circumstances, rather than engaging in our customary statistical analysis that treats each city as an equally important unit of analysis. Because many cities have multiple incumbents running for reelection and we lack data on the specific attributes of those contests, our remarks are necessarily limited to the city-level factors related to incumbent success.7 Again, we examine the four sets of factors—election ____________ 7The text and charts in this section present the results of simple cross-tabulations of incumbent success and the city characteristics of interest. These relationships were confirmed with a logistic regression analysis of the grouped data. The results of that regression conformed closely to the results presented here. 56 timing, institutional structure, contextual factors, and the demography of the city—for their relationship with incumbent success rates. In line with the views of some proponents of off-cycle elections, concurrent elections are associated with greater success of incumbents in California municipal elections (Figure 4.1). City council incumbents running during gubernatorial, presidential primary, or presidential election dates have about 4 percent greater success rates than incumbents in nonconcurrent contests. Mayoral incumbents win reelection 12 percent more frequently, although this result is not statistically significant, given the limited number of contested mayoral races featuring incumbents.8 We can only speculate whether the greater success of incumbents in concurrent elections indicates that voters “tune out” local issues during statewide elections, since we lack individual-level data on voters. Possibly, the voters who are motivated enough to go to 100 Concurrent elections 90 Nonconcurrent elections Percentage reelected 80 70 60 50 Council members Mayors Figure 4.1—Incumbent Success and Election Concurrency ____________ 8The council-election differences are statistically significant at the p < 0.10 level. For contested mayoral races, the number of observations is only 78, which renders the difference in incumbent success (marginally) statistically insignificant, despite its larger magnitude. 57 the polls for off-peak, local-only elections tend to be more attentive to community issues or more critical of incumbent council members.9 Local political institutions appear to have little bearing on incumbent success. Term limits and the degree of city service provision had no relationship with incumbent reelection rates. Cities that use district elections were somewhat more likely to have council members win reelection, although the result did not hold up using more rigorous statistical analyses. Charter cities have higher incumbent reelection rates than general-law cities, but any explanation for this relationship is not readily apparent. The electoral context of a council election, however, appears to play a significant role in incumbent reelection success, and here the findings are very plausible (Figure 4.2). For instance, the existence of citizen initiatives has a substantial negative relationship with incumbent success. 100 No citizen initiatives 90 Citizen initiatives on ballot Percentage reelected 80 70 60 50 1 to 2 3 4 or more Number of council seats being filled Figure 4.2—Success of Council Incumbents in Varying Electoral Contexts ____________ 9We do not include the turnout rate itself as an independent variable regarding incumbent success because the causation could be interpreted as running in the opposite direction—that is, the turnout rate might be affected by the presence of a weak or unpopular incumbent or the expectation of a close race. Moreover, election timing (which is included in the model) is correlated with voter turnout rates at a high level (r = 0.71). 58 Having one or more initiatives on the city ballot is associated with a seven-point reduction in the percentage of incumbents reelected, on average. Elections that include voter initiatives may be marked by more conflict or may draw more critics of local government to the polls. Some initiatives may even be directed against the policies of specific incumbent officeholders. This effect is somewhat contingent on the number of seats being selected, however. When one to three council seats were up for election, the presence of citizen initiatives was associated with a sizeable (13 to 15 point) reduction in the percentage of incumbents winning reelection. By contrast, incumbents running in elections in which numerous council seats were chosen were very likely to win reelection regardless of initiatives on the ballot; better than 90 percent were victorious. It is possible that the high information-gathering demands on voters who have to select multiple council members at the same time render them more likely to take the informational shortcut of relying on incumbency and name recognition. Among demographic variables, city population size is associated with greater success rates for incumbents (Figure 4.3). The name recognition and the fund-raising advantages of incumbency may count for more in 100 Council Mayor 90 Percentage reelected 80 70 60 50 Less than 20,000 20,000–49,000 50,000–99,000 Population category 100,000 or more Figure 4.3—Incumbent Success and City Population 59 large jurisdictions, where constituents are less likely to be intimately familiar with City Hall. Residential stability among the city population also has a positive relationship with the success rate of incumbents. It is possible that more “settled” communities with less population turnover experience more consensus and a quieter, status quo political style. Summary As seen in Chapter 3, the timing of city elections affects turnout. But is timing also related to the competitive dynamics of local contests? Although the level of competition for office is not associated with election timing, incumbents do seem to find some slight additional degree of protection in concurrent elections. This finding brings us full circle to the question of the merits and disadvantages of holding municipal elections simultaneously with elections that attract more attention. The incumbency advantage is a somewhat greater hurdle for challengers to overcome in concurrent elections—whether because local elections during the statewide cycle draw less careful scrutiny and information gathering, or because the much larger number of voters casting ballots in such elections are more inclined to vote for familiar faces. We have raised one possible explanation—that the voters who come to the polls in nonconcurrent elections tend to be more motivated or discontented about local issues, and this spells more trouble for incumbents. Whether this finding provides definitive support for the critics of concurrent scheduling of elections is an issue we will defer to our concluding chapter. Several other interesting patterns emerge from our analysis of competition for office and incumbent success. For example, cities undergoing substantial community controversy (indicated by a large number of voter initiatives) are likely to see more candidates enter the ring and are more likely to send incumbents packing. High-population communities generate more candidates per seat but are nevertheless more likely to reelect incumbents. Towns with a greater proportion of voters who do not identify with one of the two major parties experience more competition for office. Longer mayoral terms and full-time mayoralties also draw more candidates for the job. This finding indicates that 60 institutional choices (such as the powers of the mayor) that were made to address the operations or efficiency of a municipality may have other consequences relating to who runs for office and whether they succeed. 61 5. Conclusion Most Californians do not vote in municipal elections. Even compared to statewide and national contests, participation in local contests is disappointing. Because only a fraction of the adult population is participating in the local electoral arena, relatively few voters are determining the outcome of local elections and presumably affecting policy decisions while the majority of Californians stand on the sidelines. Given that most of the elected officials in the state are elected locally, and given that most of the policies that affect our daily lives are implemented at the local level, this lack of participation is clearly a problem. If the characteristics of those who do vote are different from the makeup of the larger population that does not vote, outcomes may be both biased and inefficient, as not everyone’s preferences will be taken into account. Limited voter participation does not necessarily mean that local democracy is failing but it is surely enough of a red flag to warrant careful examination. Although it is clear that most adult Californians do not vote in the average local contest, it is also clear that there really is no “average” contest. Across cities there is tremendous variation in voter turnout. There are cities where almost all of the registered voters participated in the last city election, there are cities where only 10 percent of all registered voters showed up at the polls, and there is virtually everything in between. Analysis of this variation across cities indicates that one factor has the potential to dramatically alter voter turnout rates in local contests. By far the largest part of this variation can be explained by election timing. City elections held on the same day as statewide contests (that is, on the same day as a presidential or gubernatorial election or a presidential primary) tend to have voter turnout rates almost double those of nonconcurrent city elections. Even if we control for a host of other factors associated with turnout, holding an on-cycle election still 63 increases the turnout of registered voters by an average of 26 to 36 percent above turnout for off-cycle elections. Although timing is clearly the most important factor affecting turnout, this report reveals other important aspects of the local electoral arena that appear to influence voter participation. Whether a city provides its own public services is related to voter turnout. Cities that either do not offer services or contract or arrange with other entities to carry out the bulk of municipal services draw a smaller share of voters than those that provide more services with their own staff. This finding suggests that when the stakes of contest are higher, more voters will participate. Direct democracy also plays a role, as cities with local ballot questions placed before the voters tend to have significantly higher turnout. Although there is much debate about the overall value of direct democracy, placing propositions on the ballot seems to spark greater interest and participation in the municipal political arena. We should also note the factors that do not affect voter participation in local contests. Our analysis suggests that consolidating mayoral and council elections with other local contests (such as school board elections) would have no significant effect on turnout. Similarly, charter cities are no more prone to higher turnout than are general-law cities. Term limits have no direct effect on voter participation. Contrary to expectations, district elections do not generally increase turnout over atlarge elections and may, in fact, marginally decrease turnout (although the unusual characteristics of California’s handful of districted cities complicates this interpretation). Finally, the degree of competition for office has only a limited association with voter turnout rates. What does all of this mean for policymakers? For voter turnout, the implications are clear: If there is a desire to increase voter involvement in municipal electoral contests, the best tool for the job is scheduling concurrent elections. Moving a city’s election from a nonconcurrent date to the date of a presidential election could well mean a doubling of voter turnout. Overall, we estimate that by moving all nonconcurrent elections to coincide with presidential elections, California could have drawn 1.7 million more voters to city council elections statewide (roughly a 31 percent increase in participation in municipal contests). By contrast, altering institutional factors such as city service provision 64 and increasing our reliance on direct democracy at the local level might reasonably be expected to increase turnout rates by about 5 to 10 percent. Unlike any other factor we examine, election timing could dramatically alter the size of the local voter population. Concurrent elections are even more appealing as a policy lever because they would probably produce some modest financial savings for cities. Municipalities often pay the entire cost of stand-alone elections and only a fraction of the cost of concurrent elections.1 The other advantage of concurrent elections is that they are fairly easy to implement. In many cases, city councils need only pass an ordinance. No new structure or agencies would be required. Already the cost savings associated with on-cycle elections and the ease with which election dates can be changed have led a considerable number of municipalities across the state to align their local elections with statewide contests. Encouraging more cities to move in the same direction would presumably not be very difficult—although it is possible that local elected officials would oppose a move to on-cycle elections because it would make it more difficult to retain their local offices while running for positions in state government. As California policymakers reconsider the timing of the statewide primary, in connection with presidential or gubernatorial selection, they might consider whether local governments that still hold stand-alone springtime elections should be encouraged to consolidate these elections with the primary. However, a move to concurrent elections would not fully address the broader dilemma of limited citizen participation in municipal affairs. First, on-cycle local elections would surely lead to greater voter turnout but would not necessarily ensure greater civic engagement. Although greater turnout tends to be coupled with increased political knowledge, trust, and efficacy, it is not yet clear how aligning local elections with statewide elections would affect each of these other measures. Second, ____________ 1Evaluated on the basis of a city’s administrative costs per vote cast, the fiscal advantages of concurrent elections are even clearer. For example, San Francisco election officials recently estimated that the December 2001 runoff for city attorney cost the city $29 per ballot cast. Turnout was only 15.4 percent. This has led city officials to seriously debate the merits of an “instant runoff” voting system so that the November election would be decisive and no runoff would be required (Lelchuk, 2001). 65 although concurrent elections should have a dramatic effect on turnout, in the end local turnout is unlikely to exceed the fairly disappointing levels currently experienced in statewide and national contests. Voter participation is a nationwide problem, and election timing changes for municipal elections would be only an incremental improvement. Ultimately, broader reforms such as extended voting hours, 100 percent mail-in elections, or Internet voting might be more fundamental. Third, a move to concurrent elections raises, for some observers, several concerns about voter attentiveness and knowledge, which may in turn affect the competitiveness and outcomes of city elections. The coupling of local elections with national or statewide contests would lead to longer, more complex ballots that might increase voter confusion. Another worry is that on-cycle elections would draw attention away from local politics. Local contests might simply be overshadowed by more important statewide or national contests. Moving to concurrent elections could also alter the competitive dynamics of elections. As Chapter 4 shows, incumbents win more regularly in cities with concurrent elections. Given that incumbents already win 80 percent or more of their city council and mayoral reelection bids, a widespread move to on-cycle elections might make the outcomes of local contests almost a foregone conclusion. Indeed, some observers have raised concerns that concurrent elections would make it harder for challengers to raise campaign funds, which would mean that municipal incumbents might find it easier to ignore voters and their concerns. Still, the statistical relationship between timing and incumbent success is relatively mild, especially in comparison with more important factors such as city population size. Overall, these concerns are solid arguments for civic education, voter outreach campaigns, higher-quality media coverage of local races, and intensive campaigning by candidates for mayor and council. They are, in our view, not good arguments for scheduling local elections so as to knowingly reduce public participation. Being asked to cast a vote on a given matter may itself spur an educational and information-gathering 66 effort in many citizens.2 Ultimately, which approach California policymakers choose to pursue will depend on how much they value different aspects of the democratic process. There are real reasons to be concerned about low rates of voter participation. Indeed it is difficult to argue with the claim that moving toward full participation of all citizens should be a goal of every democracy. As such, we have tried to identify tools that could encourage voter turnout and expand participation in the local electoral arena. ____________ 2As Lupia (1994) shows, voters need not be experts on a given topic to be able to find informational cues (such as endorsements) that help them determine how they should vote. 67 Appendix A Survey of City Clerks This appendix consists of a copy of the survey administered to city clerks in the fall of 2000 and spring of 2001. For all categorical and “yes or no” questions, we also show the distribution of responses. On several follow-up questionnaires sent to nonrespondents or to respondents with incomplete or unusable responses, we changed the wording of question 11 to read, “Please enter the date of the most recent contested city council election.” The response rate varied widely across questions (Table A.1). Some questions did not apply in some cases or were difficult for city clerks to answer. Only cities with a directly elected mayor answered questions 2 through 10. The maximum number of possible responses for these items is 134 cities. Several questions required an answer only if the previous question was answered in the affirmative. These will naturally have fewer responses. 69 Table A.1 Number of Valid Responses for Individual Variables Survey Item q1 q2 q3 q4 q5 q6 q6a q7a q7b q8 q9(1st) q9(2nd) q10 q11 q12 q12a q12b q13 q14 q15 q15a q16 q17a q17b q17c q17d q17e Number of Cities 391 134 130 133 133 133 96 124 104 134 122 97 134 396 396 389 396 354 393 390 362 301 277 277 274 273 274 Survey Item q18a q18b q19 q20 q21 q21a q22 q23 q24 q25 q26a q26b q27 q28a q28b q28c q28d q29 q30a q30b q30c q30d q30e q31a q31b q32a q32b Number of Cities 387 387 376 353 386 31 378 383 401 385 374 384 393 388 388 388 388 391 395 395 389 383 395 307 125 394 393 70 Voter Turnout in Municipal Elections: A Survey of California City Clerks Thank you for assisting us in our research on voter participation in California municipal elections. Please complete the brief questionnaire below and return it in the enclosed postage-paid envelope. If you are unable to answer certain questions, please answer all of the other questions and return the survey. Your information is still very important to us. The data you provide will be the basis of important research on voter turnout in cities across the state. Please do not hesitate to contact us if you have any questions. Project Directors: Paul Lewis (lewis@ppic.org), Zoltan Hajnal (hajnal@ppic.org) Hugh Louch (louch@ppic.org) Public Policy Institute of California 500 Washington Street, Suite 800 San Francisco, CA 94111 (415) 291-4400 phone (415) 291-4401 fax City: Code: 71 Q-1. How is your mayor (chief elected official) selected? (Check the most appropriate response.) 33.4% Voters elect the mayor directly 54.7% Council selects the mayor from among its members 1.5% Council member receiving most votes in the election becomes the mayor 9.5% Council members rotate into the position of mayor 0.0% Other (please specify) ________________________ If voters DO NOT elect the mayor directly, please skip ahead to question Q-11. Your responses to the other questions are very important to us. Q-2. What was the date of the most recent election for mayor held in your city? (Please enter the date of the general mayoral election, NOT the runoff, if any.) _________ ______ , ________ month day year Q-3. How many voters in your city cast ballots for mayor in that election? _____________ Q-4. How many registered voters were there in your city at the time of that election? _____________ Q-5. How many mayoral candidates were listed on the ballot of that election? _____________ Q-6. Was there an incumbent standing for reelection? 72.2% yes 27.8% no A. If yes, did the incumbent win the election? 86.5% yes 13.5% no 72 Q-7. What percentage of the vote for mayor was captured by the top two candidates in that election? a. 1st Place _______% b. 2nd Place _______% Q-8. Was there a subsequent runoff election? 6.0% yes 94.0 % no Q-9. What is the race/ethnicity of the first and second place finishers in the mayoral election referred to in question Q-2? (Check the most appropriate response.) 1st Place 2nd Place a. White, not Hispanic 79.5% 77.3% b. Black, not Hispanic 4.1 7.2 c. Asian, not Hispanic 0.8 2.1 d. Hispanic, of any race 15.6 12.4 e. Other (please specify) 0.0 1.0 Q-10. Was there also an election for one or more council seats on that day? 95.5% yes 4.5% no All respondents, please answer the following questions. Q-11. Please enter the date of the most recent city council election. (Please use the most recent regular election, not a special election to fill a vacated seat, etc.) _________ ______ , ________ month day year Q-12. How many city council seats were chosen in that election? _____________ A. How many total candidates appeared on the ballot? _____________ B. How many total seats are there on the city council? _____________ 73 Q-13. How many voters in your city cast ballots for city council races in that election? (If council is elected by district please indicate the total number of ballots cast citywide. If council seats are elected at large, please indicate the total number of ballots cast in the council race with the greatest turnout.) ____________ Q-14. How many registered voters were there in your city at the time of that election? _____________ Q-15. How many incumbents ran for reelection to council in that election? _____________ A. How many of those incumbents were reelected? _____________ For the following questions (Q-16 through Q-22), please refer to the most recent mayoral election if the mayor is directly elected by the voters. If not, please refer to the most recent city council election. Q-16. To the best of your knowledge, were there any organized attempts to purge the registered voter roll (removing voters who had died, moved, etc.) in the three years prior to that election? 47.8% yes 52.2% no Q-17. Were voters in your city selecting any other local officials in the election held on that day? (Please check all applicable offices.) 17.7% City attorney or city treasurer 23.8 Other city office(s) 25.5 County supervisor(s) 15.0 County treasurer or sheriff or other county office(s) 45.6 School board member(s) 74 Q-18. How many city-level propositions were on the ballot in that election? (Please indicate the number of each type.) ________ citizen-sponsored initiatives ________ propositions placed by council Q-19. In your judgment, how much popular interest or controversy among voters was generated by the city-level ballot questions in that election? (Check the most appropriate response.) 14.6% More interest/controversy than most elections in this city 10.9 About the same as most other elections in this city 4.3 Less interest/controversy than in most elections in this city 70.2 There were no municipal ballot questions in that election Q-20. How many citizen-sponsored initiatives were on the ballot in the years 1997, 1998, and 1999 (all elections)? Again, please consider only city-level ballot questions. Total number of initiatives 1997–99: ___________ Q-21. Over the years from 1997 through 1999, were any recall initiatives filed against the mayor and/or council members? 8.0% Yes 92.0% No A. If yes, were any successful? 67.7% Yes 32.3% No Q-22. Recognizing how complex a community’s affairs are, please review the following general statements and select the one that best describes elections in your city in the last five years, compared to other municipalities in your region: 35.4% Our elections are low-key affairs with no major controversies 51.1 Elections are generally low-key, but there have been some notable controversies 13.5 Our city often has controversial elections 75 We conclude the survey by asking several basic questions about the structure of government in your city. PLEASE NOTE: In answering these questions, refer to the conditions in your city at the time of the mayoral or city council election you answered questions about above, not the current conditions, if different. Q-23. What is the length of the mayor’s term of office? (Important: If the chief elected official is a member of the council, specify the term for the position of mayor, not of council member.) _______ years Q-24. Is the mayoralty in your city considered a full-time or part-time office? (Generally, the office is considered full-time only if it carries a full-time salary.) 3.2% Full-time 92.8% Part-time 4.0% Don’t know Q-25. Does the mayor in your city have the authority to veto actions of the city council? 3.6% yes 96.4% no Q-26. Does your city limit the number of terms either the mayor or city council members can hold? Mayor*: 27.3% yes 72.7% no City Council: 18.5% yes 81.5% no (*Responses for directly-elected mayoralties only.) Q-27. Which of the following statements best describes the structure of governance in your city? 2.5% MAYOR-COUNCIL Elected council or board serves as the legislative body. The chief elected official is the head of government, generally elected separately from the council, with powers that may range from limited duties to full-scale authority for the daily operation of the government. 76 97.0 COUNCIL-MANAGER Elected council or board is responsible for making policy. A professional administrator or manager appointed by the board or council has full responsibility for the day-to-day operations of the government. 0.5 Other (please specify:___________________________) 0.0 Not sure Q-28. What methods are used to elect council members in your city? Fill in the number of seats elected by each method. a. At-large, elected citywide 92.5% b. Elected from districts 5.4 c. Members represent districts, but 2.1 are elected citywide d. Other method 0.0 (please specify: _________________________________) Q-29. Who has the overall responsibility for developing the budget submitted to the council? (check one) 0.3% Mayor 92.1 Chief Administrative Officer (city manager or equivalent) 1.0 Mayor and CAO are jointly responsible 6.6 Other (please specify:__________________________) Q-30. Regarding each of the following services in your city, are they mainly carried out by city government personnel, by county personnel, by a special district government, or by a private company under contract with the city? (Please check the appropriate response for each service. Leave item blank if the particular service is not provided within your city.) 77 Police Services Fire Services* Library Services Sewer Services Garbage Collection City Staff 72.1% 54.9 30.1 65.0 16.2 County 26.6% 26.8 64.8 10.7 0.8 Special District 0.8% 14.9 4.9 18.0 1.0 Private Firm 0.5% 0.5 0.3 6.3 82.0 Q-31. Has your city switched the timing of mayor/council elections at any time in recent years? 30.6% Yes, switched from “stand-alone” city elections to elections consolidated with statewide primary or general elections 1.0 Yes, switched from consolidated elections to standalone elections 68.4 Other type of change or no change in election timing B. If your city switched, when was the last election using the previous format? _________ ______ , ________ month day year Q-32. A. Has your city made any changes in the structure of city government (mayor-council, council-manager, etc.) in recent years? 4.3% yes 95.7% no B. Has your city made any changes in the way council members are elected (district, at-large, etc.) in recent years? 4.1% yes 95.9% no If the answer to either Q-32A or Q-32B is yes, please describe the change, and list the date it took effect, on the back of this questionnaire. Q-33. Please indicate your job title: _____________________ *An additional 2.8 percent of cities indicated volunteer fire departments. 78 Thank you very much for your participation. Your response is crucial to help inform policymakers and citizens about voter turnout in California cities. Please return the survey in the enclosed postage-paid envelope. You will receive a future mailing that will discuss the results of this survey. We welcome your comments on these topics, as well as comments regarding the questionnaire itself, or clarifications of your responses. You may include any written comments below, or on a separate sheet. The Public Policy Institute of California is a private, nonprofit organization dedicated to independent, nonpartisan research on economic, social, and political issues that affect the lives of Californians. For further information, including the full text of PPIC research reports, please visit our World Wide Web site at www.ppic.org. 79 Appendix B Mail Survey Procedures and Data Issues The analyses presented in this report are based on a survey of California city clerks conducted from October 2000 through January 2001. We purchased a mailing list of city clerks for the 474 cities then in existence from the League of California Cities. Where the contact information was incomplete, we used city web pages for guidance. A questionnaire was sent to each city clerk on the list, accompanied by an explanatory cover letter and a postage-paid return envelope. Follow-up contacts were made to maximize the response rate. Ultimately, respondents from 401 cities returned their surveys. Four cities’ responses were completely unusable because they were either unwilling or unable to answer the survey. One of these had not had a contested election in over a decade. Another 28 cities provided us with information for an uncontested election. We followed up with a second survey for these cities and received new responses from 18 of them, 16 of which were usable. Four cities left significant portions of the survey blank and were contacted with a second survey. Two responded with the additional information. A total of 375 cities returned surveys with enough responses to be useable for some portion of the analyses. Validity of the Sample We compared cities that responded to our survey to those that did not on several characteristics, using census data (Table B.1). There were few statistically significant differences between cities in the sample and those outside of it. Notably, the population of cities in the sample was 20,000 persons larger on average than cities that did not respond, but this difference was not statistically significant. (In part, this is a result of nearly all the state’s largest cities completing the questionnaire.) On most other measures, both the magnitude and the significance of 81 Table B.1 Comparing In-Sample and Out-of-Sample Cities Population Percentage black Percentage Hispanic Percentage in poverty Real median income Percentage unemployed Percentage home owner Average household size In Sample 55,543 3.9 24.4 11.7 $45,501 7.2 58.8 2.8 Not in Sample 30,306 3.7 28.4 13.9 $46,831 7.9 61.0 3.0 P-Value of Difference 0.294 0.812 0.188 0.029 0.662 0.194 0.242 0.047 differences were negligible. Cities in and out of the sample had similar percentages of Hispanics and blacks, unemployed persons, and home owners, as well as similar average incomes. Two factors—poverty and household size—were significantly different, but the magnitude of those differences was minimal. Cities not in the sample had an average poverty rate 2 percent higher than those in the sample. The average household size was also 0.2 persons greater on average. Neither of these numbers raises concerns about the value of the sample. We also compared the distribution of cities across the regions of the state and across city types (central city, suburb, rural), but found no significant differences (Table B.2). Slightly fewer cities from the Central Valley and slightly more cities from outside the three major regions (Los Angeles area, Bay Area, and the Central Valley) responded to our survey, but the differences were quite small. Among the different types of cities, we achieved the best response rate from central cities (95 percent), although they represent the smallest proportion of cities in California (only 10 percent). Cleaning the Data Several changes were made to fix data errors and minimize missing data. In general, we took respondents’ entries at face value and assumed that they answered questions correctly. If we had substantial concrete evidence contradicting a particular response, we used that information. 82 Table B.2 Geographic Distribution of Cities in Sample Region Greater Los Angeles Bay Area Central Valley Other Percentage Responding to Survey 84.5 84.7 83.0 88.2 City type Central city Suburb Rural 95.9 84.3 83.0 NOTE: significance = 0.763 for region, 0.075 for city type. Such evidence included published newspaper reports, vote tallies listed on official election websites, or phone conversations with city clerks or their staff. The major change we undertook was to minimize the extent of missing data. A substantial minority of respondents did not answer one or more crucial survey questions that would have made their responses effectively unusable. Table B.3 lists the number of missing values for questions regarding the number of voters, registered voters, and Table B.3 Important Missing Values Question Description Mayoral Election q3 Number of voters q4 Registered voters q5 Number of candidates Council Election q12a Number of candidates q13 Number of voters q14 Registered voters Missing 9 5 4 25 41 30 83 candidates for mayoral and council elections. These questions provide the core information for establishing turnout and competitiveness of a given election. To fix missing voter registration tallies, we turned to the Report of Registration published by the California Secretary of State. Although not completed each month, a report is available within two months of any election date, providing an accurate assessment of voter registration for every city in California for a given election. To remedy missing data issues for the number of voters and candidates, we conducted a newspaper search of voting results for each city missing any piece of information. Major newspapers in California publish election returns for both concurrent and nonconcurrent elections and can be easily accessed on-line. This search considerably narrowed the number of missing values for voters and candidates listed. For several cities, we were unable to use this information for council elections because of the method of election used in many California cities. Cities in California often elect more than one council member for a given election. When these elections are conducted at-large, voters are typically asked to select multiple candidates. Because many voters cast ballots for fewer candidates than they are permitted, we cannot obtain an accurate assessment of total turnout for these cities. The newspaper check also allowed us to replace missing values for several other variables: • The number and success of incumbents—questions 6 and 15, • Other offices up for election—question 17, • City propositions on a given ballot—question 18, and • Recall elections—question 21. In general we made very few additions to the data for these questions. Typically, respondents left questions blank when there were no other offices, no recall, no city measures on the ballot, or no incumbents up for reelection. For the handful of cases for which this was not the case, we input information from the newspaper sources. 84 Correcting Specific Data Problems Each survey question was checked for impossible values where necessary. In particular, questions 2 and 11 were checked to see that responses conformed to an actual election date. Questions 7a and 7b were checked to be sure that totals did not add up to more than 100 percent and that 7b was not greater than 50 percent. Questions 6a, 15a, and 21a were made to be missing if respondent indicated a “no” answer on Questions 6, 15, or 21, respectively. These second-part questions should not have been answered if the first part was answered in the negative. We assume that if a respondent answered either question 18a or 18b, but not both, that they intended for the other part to be zero. Also, if a respondent left question 19 blank but wrote in a zero for both 18a and 18b we know that 19 should be coded to indicate that there were no propositions on the ballot. If a respondent incorrectly answered question 19 when both 18a and 18b were zero, we assume that this was an honest misreading of the question and alter their answer to indicate that there were no propositions on the ballot. Another major issue is discrepancies between questions 12b and 28. In both cases respondents were asked to indicate the number of council seats. Twenty respondents answered both questions and indicated different numbers of total seats. All 20 of these were cities that directly elected their mayor. For these cities we assume that the lower number is correct and that the mayor sits as part of the city council. For generallaw cities, this is necessarily true, as they are limited in the number of council members they can have and the mayor, if directly elected, serves on the council. For the five charter cities that had inconsistent responses, we examined city websites to make sure this was the most appropriate strategy. In one case, a city had just changed to a directly elected mayor with the current election. Here we use the number of seats at the time the council was elected to appropriately fit with the other responses. Finally, a small number of cities entered the number of seats in the election for question 12b instead of the number of seats on the council. In these cases we changed the response to correspond to question 28. 85 One smaller problem came because cities entered check marks in response to questions 18a or 18b instead of a number. Although we cannot appropriately include these values for the raw versions of these questions, the analyses presented in this report use only binary versions of these variables (i.e., the presence of any city-level ballot propositions), making it possible to include these cases in our analysis. For tabulated responses to question 30, county refers to services provided either by a county or by another city. We also include a separate code for volunteer fire departments for question 30 when this information was volunteered by the respondent. There were 11 cases of this. 86 Appendix C Voter Turnout Regressions This appendix provides information regarding the voter turnout regressions discussed in Chapter 3. Table C.1 shows the results of estimations for two dependent variables: the percentage of registered voters casting ballots and the estimated percentage of all adults casting ballots. As noted in Chapter 3, city council elections and mayoral elections (for those cities that have a separately elected mayor) are merged in this dataset. The “mayoral election” variable is used to denote the mayoral elections in the sample. Finally, for the 421 observations of council or mayoral elections included in the regressions, Table C.2 provides the mean, high value, and low value for each variable. 87 Table C.1 Turnout in City Elections: Regression Model Timing variables Presidential (compared to off-cycle) Presidential primary (compared to off-cycle) Gubernatorial (compared to off-cycle) Odd-year November (compared to off-cycle) Mayor and council election held same day Other local elections held same day Mayoral election (compared to council election) Institutional variables Charter (compared to general-law) city District (compared to at-large) council election Number of services provided by city staff Office has term limits Electoral context variables One or more propositions on city ballot Election uncontested Candidates per seat Incumbents per seat Percentage of voting-age residents registered (est.) Percentage Casting Ballots Registered Adult Voters Residents 35.72*** (14.60) 20.66*** (6.74) 26.02*** (14.70) 1.97 (0.96) 1.43 (0.92) 0.27 (0.22) –3.33** (2.01) 23.39*** (13.07) 11.19*** (5.00) 15.37*** (11.85) –0.82 (0.54) 1.13 (0.99) 0.25 (0.27) –1.79 (1.47) 0.55 (0.33) –1.03 (0.30) 0.98** (2.02) 0.80 (0.47) 0.64 (0.52) –8.44*** (3.64) 0.37 (1.04) 0.64 (0.51) 4.11*** (3.26) –5.48** (2.09) 0.93* (1.86) 0.41 (0.24) –0.14** (1.98) 3.52*** (3.79) –4.28** (2.22) 0.69* (1.87) 0.13 (0.11) n/i 88 Table C.1 (continued) City population characteristics City population (natural log) Socioeconomic status (factor score) Percentage black Percentage Hispanic Percentage Asian Percentage age 18 to 24 Percentage age 65 or older Percentage lived in same house for 5 years Percentage institutionalized Constant Observations Adjusted R-squared Percentage Casting Ballots Registered Adult Voters Residents –3.28*** (5.31) 3.65*** (3.12) –0.04 (0.38) –0.07 (1.43) –0.25*** (3.34) 0.09 (0.41) 0.31** (2.06) 0.02 (0.29) n/i 65.01*** (6.79) 421 0.61 –2.45*** (5.44) 3.94*** (4.79) –0.00 (0.01) –0.20*** (6.66) –0.34*** (6.48) 0.04 (0.23) 0.26** (2.32) 0.12** (1.98) –0.31*** (2.91) 42.14*** (7.40) 421 0.67 NOTES: *p < 0.10, **p < 0.05, ***p < 0.01. n/i indicates variable is not included in regression. Absolute values of t-values are in parentheses. 89 Table C.2 Descriptive Statistics of Variables for Observations in the Regression Dependent variables Turnout of registered voters Turnout of adult residents Mean Value Low Value High Value 47.16 30.04 7.32 88.64 4.46 78.79 Timing variables Presidential Presidential primary Gubernatorial Odd-year November Mayor and council election held same day Other local elections held same day 0.10 0.05 0.49 0.18 0.49 0.54 0 0 0 0 0 0 1 1 1 1 1 1 Institutional variables Charter city District council election Number of services provided by city staff Office has term limits 0.24 0.04 2.57 0.22 0 0 0 0 1 1 5 1 Electoral context variables Any ballot measures Uncontested election Candidates per seat Incumbents per seat Percentage registered, of voting-age residents (est.) 0.34 0.07 2.60 0.69 63.00 0 0 1 0 12.47 1 1 14 1 97.64 Type of election Mayoral 0.30 0 1 City population characteristics City population (natural log) Socioeconomic status (factor score) Percentage black Percentage Hispanic Percentage Asian Percentage age 18 to 24 Percentage age 65 or older Percentage lived in same house for 5 years Percentage institutionalized 10.27 0.01 4.23 30.57 9.09 9.28 11.61 43.70 1.49 5.25 –2.02 0 2.15 0 2.36 3.65 11.35 0 15.13 4.62 46.42 98.27 61.51 33.56 45.06 70.81 39.98 90 Appendix D Candidate Competition Regression Table D.1 displays results for the regression described in Chapter 4. This model attempts to account for the number of candidates running for office (whose names are listed on the ballot). The dependent variable is the natural logarithm of the number of candidates (for mayoral elections) or of the number of total candidates divided by the number of seats (for council elections). Mayoral and council elections are merged in the regression. The “mayoral election” variable is an indicator used to differentiate between mayoral and council elections in the sample. 91 Table D.1 Determinants of Competition for Office Mayoral election (compared to council election) Concurrent with statewide election (compared to nonconcurrent) Mayoral and council election held same day Other local offices elected same day District council elections (compared to at-large) Full-time mayor (compared to part-time) Four-year mayoral term (compared to two-year term) Charter (compared to general-law) city Number of services provided by city staff Office has term limits Percentage voting-age citizens registered (est.) Number of voter initiatives, 1997–1999 Typical degree of controversy of city elections (3-point scale) Incumbents per seat Percentage independent/third party (of registered voters, 1999) City population (natural log) Residential stability Percentage black Percentage Hispanic –0.21*** (3.48) 0.03 (0.69) 0.02 (0.49) –0.08** (2.20) –0.12 (1.07) 0.35** (2.57) 0.27*** (3.68) 0.05 (0.86) –0.00 (0.27) –0.10* (1.77) 0.00 (1.52) 0.03** (2.40) 0.08** (2.45) –0.28*** (5.47) 0.03*** (4.02) 0.12*** (6.43) –0.01** (2.25) 0.01** (2.22) 0.00*** (3.32) 92 Table D.1 (continued) Percentage Asian Percentage age 18 to 24 Percentage age 65 or older Constant Observations Adjusted R-squared –0.00 (0.13) –0.01 (1.45) 0.01 (1.21) –0.90** (2.54) 400 0.40 NOTE: *p < 0.1, **p < 0.05, ***p < 0.01. Absolute values of t-values are in parentheses. Dependent variable is the log of the number of candidates per available council seat (for council elections) and the log of the number of candidates for mayor (for mayoral elections). A potential statistical issue with the regression in Table D.1 is that it treats mayoral and council elections from the same city as independent observations. However, running the model with the observations grouped, or clustered, by city yields the same results and significance levels, with one exception: The percentage of young adults is negative and significant at the 0.05 significance level. We also tried running the competition model with uncontested elections omitted, resulting in 370 observations rather than 400, and an adjusted R-squared of 0.39. Results were again basically the same, with the following exceptions: District council elections had a slightly significant, negative relationship to council competition, as did the percentage of young adults in the city. More interesting, the difference in competition levels between mayoral and council elections vanished. This indicates that the lower level of competition for mayoralties is primarily attributable to the substantial number of uncontested mayoral elections. As noted in Chapter 4, 17 percent of mayoral elections in our data were uncontested, compared to 4 percent of council elections. 93 References Alford, Robert R., and Eugene C. Lee, “Voting Turnout in American Cities,” American Political Science Review, Vol. 62, 1968, pp. 796– 813. Baldassare, Mark, California in the New Millennium: The Changing Social and Political Landscape, University of California Press, Berkeley, California, 2000. Bennett, Stephen Earl, and David Resnick, “The Implications of Nonvoting for Democracy in the United States,” American Journal of Political Science, Vol. 34, No. 3, 1990, pp. 771–802. Bobo, Lawrence, and Franklin D. Gilliam, Jr., “Race, Sociopolitical Participation, and Black Empowerment,” American Political Science Review, Vol. 84, No. 2, 1990, pp. 377–393. Bridges, Amy, Morning Glories: Municipal Reform in the Southwest, Princeton University Press, Princeton, New Jersey, 1997. Bullock, Charles S., “Turnout in Municipal Elections,” Policy Studies Review, Vol. 9, No. 3, 1990, pp. 539–549. Bullock, Charles S., and Susan A. MacManus, “Structural Features of Municipalities and the Incidence of Hispanic Councilmembers,” Social Science Quarterly, Vol. 71, No. 4, 1990, pp. 665–681. Cain, Bruce E., Megan Mullin, and Gillian Peele, “City Caesars? An Examination of Mayoral Power in California,” paper presented at the Annual Meeting of the American Political Science Association, San Francisco, California, 2001. California Secretary of State, “Registration and Participation, November 7, 2000,” 2001, available at http://www.ss.ca.gov/elections/ elections.htm. 95 Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes, The American Voter: The Unabridged Edition, University of Chicago Press, Chicago, Illinois, 1980 (orig. 1960). Casel, Carol A., “The Nonpartisan Ballot in the United States,” in Bernard Grofman and Arend Lijphart (eds.), Electoral Laws and Their Political Consequences, Agathon Press, New York, 1986, pp. 226–241. Copeland, Gary W., “Term Limits for Urban Settings: One Size Fits All?” Urban News, Vol. 11, No. 1, 1997, pp. 1–10. Cox, Gary, and Michael C. Munger, “Closeness, Expenditures and Turnout in the 1982 House Elections,” American Political Science Review, Vol. 83, 1989, pp. 217–231. Curtin, Daniel J., Jr., Curtin’s California Land Use and Planning Law, 20th ed., Solano Press Books, Point Arena, California, 2000. Davidson, Chandler, and Bernard Grofman (eds.), Quiet Revolution in the South: The Impact of the Voting Rights Act, 1965–1990, Princeton University Press, Princeton, New Jersey, 1994. Dennis, Jack, “Theories of Turnout: An Empirical Comparison of Alienationist and Rationalist Perspectives,” in William Crotty (ed.), Political Participation and American Democracy, Greenwood Press, New York, 1991, pp. 23–65. Engstrom, Richard L., and Michael D. McDonald, “The Underrepresentation of Blacks on City Councils: Comparing the Structural and Socioeconomic Explanations for South/Non-South Differences,” Journal of Politics, Vol. 44, 1982, pp. 1088–1099. Espino, Rodolfo, “The Political Legacy of Municipal Reform,” paper presented at the Annual Meeting of the Midwestern Political Science Association, Chicago, Illinois, 2001. Finkel, Steven E., “Reciprocal Effects of Participation and Political Efficacy: A Panel Analysis,” American Journal of Political Science, Vol. 29, 1985, pp. 891–913. Gerber, Elisabeth R., Prospects for Reforming the Initiative Process, report prepared for the University of Virginia Project on Initiatives, 1999. 96 Guinier, Lani, The Tyranny of the Majority: Fundamental Fairness in Representative Democracy, Free Press, New York, 1994. Hahn, Harlan, and Sheldon Kamieniecki, Referenda Voting: Social Status and Policy Preferences, Greenwood Press, New York, 1987. Hajnal, Zoltan L., and Mark Baldassare, Finding Common Ground: Racial and Ethnic Attitudes in California, Public Policy Institute of California, San Francisco, California, 2001. Hampton, Gloria J., and Katherine Tate, “Political Participation in Urban Elections: The SES Baseline Model Reexamined,” paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, Illinois, 1996. Hill, John, “Voter Turnout Falling in Both U.S. and State,” Sacramento Bee, December 22, 2000. International City/County Management Association (ICMA), The Municipal Year Book, International City/County Management Association, Chicago, Illinois, 1996. Jacobson, Gary C., The Politics of Congressional Elections, Little, Brown, Boston, Massachusetts, 1983. Johnson, Ted, “Cities Delay Elections to Avoid 1996 Primary Conflict,” Los Angeles Times, December 22, 1994. Karnig, Albert K., and B. Oliver Walter, “Decline in Municipal Voter Turnout: A Function of Changing Structure,” American Politics Quarterly, Vol. 11, No. 4, 1983, pp. 491–505. Karnig, Albert K., and B. Oliver Walter, “Municipal Voter Turnout During the 1980s: The Case of Continued Decline,” in Harry A. Bailey and Jay M. Shafritz (eds.), State and Local Government and Politics: Essential Reading, F. E. Peacock Publishers, Itasca, Illinois, 1993. Krebs, Timothy B., “The Determinants of Candidates’ Vote Share and the Advantages of Incumbency in City Council Elections,” American Journal of Political Science, Vol. 42, No. 3, 1998, pp. 921–935. 97 Lelchuk, Irene, “Record-Low Vote Came at High Price,” San Francisco Chronicle, December 13, 2001. Lien, Pei-te, “Ethnicity and Political Participation: A Comparison Between Asian and Mexican Americans,” Political Behavior, Vol. 16, 1994, pp. 237–264. Lupia, Arthur, “Shortcuts Versus Encyclopedias: Information and Voting Behavior in California Insurance Reform Elections,” American Political Science Review, Vol. 88, No. 1, 1994, pp. 63– 76. Magleby, David B., “Direct Legislation in the American States,” in David Butler and Austin Ranney (eds.), Referendums Around the World, AEI Press, Washington, D.C., 1994. Mitchell, Greg, The Campaign of the Century: Upton Sinclair’s Race for Governor of California and the Birth of Media Politics, Random House, New York, 1992. Morlan, Robert L., “Municipal Versus National Election Voter Turnout: Europe and the United States,” Political Science Quarterly, Vol. 99, No. 3, 1984, pp. 457–470. Oliver, J. Eric, “City Size and Civic Involvement in Metropolitan America,” American Political Science Review, Vol. 94, No. 2, 2000, pp. 361–372. Pantoja, Adrian D., and Gary M. Segura, “Citizens by Choice, Voters by Necessity: Patterns in Political Mobilization by Naturalized Latinos,” paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, Illinois, 2000. Parker, Frank R., Black Votes Count: Political Empowerment in Mississippi After 1965, University of North Carolina Press, Chapel Hill, North Carolina, 1990. Powell, G. Bingham, “American Voter Turnout in Comparative Perspective,” American Political Science Review, Vol. 81, 1986, pp. 45–65. Putnam, Robert D., Bowling Alone: The Collapse and Revival of American Community, Simon & Schuster, New York, 2000. 98 Rohrs, Sarah, “Growing Numbers Voting Through Mail Instead of Booth,” San Jose Mercury News, October 20, 2000. Rosales, Rodolfo, The Illusion of Inclusion: The Untold Political Story of San Antonio, University of Texas Press, Austin, Texas, 2000. Rose, Roger (ed.), The California Initiative Process: Current Controversies and Prescriptions, California Policy Seminar, University of California, Berkeley, California, 1990. Rosenstone, Steven J., and John Mark Hansen, Mobilization, Participation, and Democracy in America, Macmillan Publishing Company, New York, 1993. Schaffner, Brian F., Gerald Wright, and Matthew Streb, “Teams Without Uniforms: The Nonpartisan Ballot in State and Local Elections,” Political Research Quarterly, Vol. 54, 2001, pp. 7–30. Schneider, Mark, and Paul Teske, Public Entrepreneurs: Agents for Change in American Government, Princeton University Press, Princeton, New Jersey, 1995. Schrag, Peter, Paradise Lost: California’s Experience, America’s Future. New Press, New York, 1998. Simerman, John, “Debate Surrounds Plan to Move Local Elections,” Contra Costa Times, June 8, 1998. Sokolow, Alvin D., and Peter M. Detwiler, “California,” in Dale Krane, Platon Rigos, and Melvin Hill (eds.), Home Rule in America: A Fifty-State Handbook, CQ Press, Washington, D.C., 2001, pp. 58–68. Squire, Peverill, et al., Dynamics of Democracy, Brown & Benchmark, Madison, Wisconsin, 1995. Squire, Peverill, and Raymond E. Wolfinger, “Residential Mobility and Voter Turnout,” American Political Science Review, Vol. 81, 1987, pp. 45–65. Teixeira, Ruy, The Disappearing American Voter, Brookings Institution Press, Washington, D.C., 1992. 99 Thompson, Joel A., and Gary F. Moncrief, “The Implications of Term Limits for Women and Minorities: Some Evidence from the States,” Social Science Quarterly, Vol. 74, 1993, pp. 300–309. Uhlaner, Carole Jean, Bruce E. Cain, and D. Roderick Kiewiet, “Political Participation of Ethnic Minorities in the 1980s,” Political Behavior, Vol. 11, 1989, pp. 195–221. Verba, Sidney, Kay Lehman Schlozman, and Henry E. Brady, Voice and Equality: Civic Voluntarism in American Politics, Harvard University Press, Cambridge, Massachusetts, 1995. Wattenberg, Martin P., Turnout Decline in the U.S. and Other Advanced Industrial Democracies, Center for the Study of Democracy, University of California, Irvine, California, 1998. Welch, Susan, “The Impact of At-Large Elections on the Representation of Blacks and Hispanics,” Journal of Politics, Vol. 52, 1990, pp. 1050–1076. Welch, Susan, and Timothy Bledsoe, Urban Reform and Its Consequences: A Study in Representation, University of Chicago Press, Chicago, Illinois, 1988. Wolfinger, Raymond E., and Steven J. Rosenstone, Who Votes? Yale University Press, New Haven, Connecticut, 1980. Wolman, Harold, Edward Page, and Martha Reauley, “Mayors and Mayoral Careers,” Urban Affairs Quarterly, Vol. 25, No. 3, 1990, pp. 500–513. 100 About the Authors ZOLTAN L. HAJNAL Zoltan L. Hajnal is an assistant professor of political science at the University of California, San Diego. Before joining UCSD, he was a research fellow at the Public Policy Institute of California and a Mellon Postdoctoral Fellow at Brandeis University. His research interests include racial and ethnic relations, urban politics, and direct democracy. He holds a B.S. in geophysics and a B.A. in political science from Yale University and an M.A. and Ph.D. in political science from the University of Chicago. PAUL G. LEWIS Paul G. Lewis is a research fellow at the Public Policy Institute of California, where he studies local government and urban policy issues. He has authored or co-authored publications on local government structure, local sales taxes, urban development, transportation policy, and regional governance, and has made presentations on such topics to a variety of governmental bodies and civic organizations. He holds an A.B. in political science from Indiana University and an M.A. and Ph.D. in politics from Princeton. HUGH LOUCH Hugh Louch is an analyst at Cambridge Systematics and a former research associate at the Public Policy Institute of California. His research interests include urban transportation, housing, and planning. He received a B.A. in sociology from the University of California, Santa Barbara, and an M.A. in sociology from Princeton University. 101 Other Related PPIC Publications The Effect of Minority Districts and Minority Representation on Political Participation in California Claudine Gay Interest Group Influence in the California Initiative Process Elisabeth R. Gerber Are There Winners and Losers? Race, Ethnicity, and California’s Initiative Process Zoltan Hajnal, Hugh Louch PPIC publications may be ordered by phone or from our website (800) 232-5343 [mainland U.S.] (415) 291-4400 [Canada, Hawaii, overseas] www.ppic.org 103" } ["___content":protected]=> string(102) "

R 302ZHR

" ["_permalink":protected]=> string(107) "https://www.ppic.org/publication/municipal-elections-in-california-turnout-timing-and-competition/r_302zhr/" ["_next":protected]=> array(0) { } ["_prev":protected]=> array(0) { } ["_css_class":protected]=> NULL ["id"]=> int(8163) ["ID"]=> int(8163) ["post_author"]=> string(1) "1" ["post_content"]=> string(0) "" ["post_date"]=> string(19) "2017-05-20 02:35:24" ["post_excerpt"]=> string(0) "" ["post_parent"]=> int(3293) ["post_status"]=> string(7) "inherit" ["post_title"]=> string(8) "R 302ZHR" ["post_type"]=> string(10) "attachment" ["slug"]=> string(8) "r_302zhr" ["__type":protected]=> NULL ["_wp_attached_file"]=> string(12) "R_302ZHR.pdf" ["wpmf_size"]=> string(6) "619543" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(197162) "Municipal Elections in California: Turnout, Timing, and Competition ••• Zoltan L. Hajnal Paul G. Lewis Hugh Louch 2002 PUBLIC POLICY INSTITUTE OF CALIFORNIA Library of Congress Cataloging-in-Publication Data Hajnal, Zoltan, 1968– Municipal elections in California : turnout, timing, and competition / Zoltan L. Hajnal, Paul G. Lewis, Hugh Louch. p. cm. Includes bibliographical references. ISBN 1-58213-041-8 1. Local elections—California. I. Lewis, Paul George, 1966– II. Louch, Hugh, 1973– III. Title. JS451.C25 H35 2002 324.9794'053—dc21 2002017332 Copyright © 2002 by Public Policy Institute of California All rights reserved San Francisco, CA Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source and the above copyright notice is included. Research publications reflect the views of the authors and do not necessarily reflect the views of the staff, officers, or Board of Directors of the Public Policy Institute of California. Foreword The PPIC Statewide Survey has shown time and again that California voters distrust their government. The roots of that distrust are deep and complex, but its policy effects are often quite clear. For example, California voters rely more heavily now on the initiative process to guide public policy, in part because they do not believe that their elected representatives will address the problems they think are important. One can argue at length over the advantages and disadvantages of specific initiatives, and one can do the same for the initiative process generally. But it is more difficult to argue that the distrust that has made the initiative process so important is itself salutary. This distrust becomes even more difficult to celebrate when it manifests itself as political apathy or even disgust. These reactions have kept large numbers of citizens from registering to vote in the first place. When we combine these unregistered citizens with the noncitizen population living in California, and then add to this group the registered voters who do not go to the polls, we find that, in many cases, an alarmingly small percentage of California residents decides a local election. This pattern is especially pronounced in off-cycle local elections, when turnout rates typically lag those of local elections that coincide with statewide and national contests. In a political system based on an informed and active citizenry, low and declining participation rates are a great concern. Indeed, increasing those rates may be the most important policy challenge of the early 21st century. But what are the most practical ways to meet that challenge? In this report, Zoltan Hajnal, Paul Lewis, and Hugh Louch offer a detailed description of local turnout patterns and analyze the factors associated with high and low participation rates. Their results indicate that about half the difference in voter turnout across California cities can be traced to a single factor—election timing. Noting that Progressive Era reformers instituted off-cycle local elections, and that one-third of iii California’s cities continue to hold them, the authors calculate how many more citizens would vote in local elections that coincided with state and national contests. The policy solution in this case seems clear enough. If low turnout for local contests is the problem, concurrent elections are a big step in the right direction. This simple change is unlikely to banish political apathy, but it will significantly increase the likelihood that citizens will make their voices heard on local issues. David W. Lyon President and CEO Public Policy Institute of California iv Summary The 2001 race for mayor of Los Angeles was one of the closest and most exciting in decades. With incumbent mayor Richard Riordan leaving office because of term limits, several strong candidates from various political and ethnic backgrounds ran for the office, and surveys indicated that the race was too close to call. It therefore came as a surprise to some that only 33.5 percent of the city’s registered voters came to the polls to cast votes in the April primary nominating election and only 36.2 percent voted in the June runoff election. Given the widely perceived interest in the contest, why did nearly two of three registered voters stay home? Are Los Angeles voters unusual in their degree of civic disinterest? This report finds that voter behavior in Los Angeles is far from atypical. By documenting levels of voter involvement in municipal elections, it investigates factors that help account for differences in turnout across California cities. It also examines the degree of competition for municipal offices as a related barometer of the vibrancy of the local political environment. Most of the key evidence is drawn from a questionnaire mailed to all city clerks in the state in 2000 asking about the most recent elections in the community. Usable data were collected on 350 city council elections, or 74 percent of the 474 municipalities then existing in the state. Analogous information was available for 130 mayoral contests, since only one-third of cities provide for direct election of mayors. City clerks also provided a wealth of information on local electoral procedures and institutional arrangements as well as the context of the particular election. We supplemented this information with published data from the Census and state sources. v Local Voter Turnout Rates: Generally Low, But Varied Observers of politics and community life have raised concerns about low, and in many cases declining, levels of civic engagement in America. Exercising the right to vote is one of the most fundamental and cherished forms of civic participation. Citizen participation in elections is important for several reasons. Low turnout levels may compromise the basic legitimacy of a democratic system, and nonparticipation is related to a host of negative attitudes about government and politics. Voting also serves as an important educational tool because voters learn more about policy issues and government by actively participating in the electoral arena. Finally, low turnout may be linked to an unrepresentative electorate. Racial and ethnic minorities, the poor, and younger Californians vote much less regularly than others in the state. If these nonvoters have systematically different political preferences or needs than voters, then the “voice of the people” will be distorted. Although municipal governments can affect citizens in profound and immediate ways, many Californians and other Americans ignore local elections. Information provided by the city clerks indicates that the average turnout rate in city council elections in California was 48 percent of registered voters and only 44 percent in mayoral races for cities that directly elect the mayor. (Aggregate turnout of registered voters was even lower—43 percent for council races and 39 percent for mayoral contests—because cities with larger populations tended to have lower participation rates.) Measured as a percentage of the voting-age population, turnout was lower still: 32 percent in the average council election and 28 percent in the average mayoral elections. These overall levels of turnout are important, but they conceal important variations in voting rates across communities (Figure S.1). Turnout of registered voters ranges from a low of 10 percent to a high of 89 percent. It is therefore important to consider which factors might account for these major differences in voter participation. vi Number of cities 80 70 60 50 40 30 20 10 0 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70–79 80–89 90–100 Percentage of registered voters casting ballots Figure S.1—Distribution of Voter Turnout Rates in City Council Elections Why Do Local Turnout Rates Differ? To understand variation in voter turnout across cities, we examined three sets of factors: election timing, the electoral and institutional structure of local government, and the specific context surrounding any given election. Although we also control for the demographic and socioeconomic characteristics that are known to affect voter turnout, we focus attention on these broad city-level factors for two reasons. First, anecdotal evidence and findings from other studies suggest that election timing and local institutions can significantly affect voter participation. Second, these city-level characteristics lend themselves more readily to practical policy interventions. Although it is difficult or impossible for city policymakers to affect the demographic or socioeconomic characteristics of their localities, they often can and do alter their electoral institutions and governmental structure. Our analysis indicates that about half of the differences in turnout among California cities can be explained by one simple factor—the vii timing of the local election. By scheduling local elections to occur on the dates of statewide general or primary elections (so called concurrent or oncycle elections), localities make it easier for voters participating in the statewide election to vote in local contests as well. Controlling for a host of other factors, presidential elections are associated with turnouts of registered voters in city elections that are 36 percent higher than off-cycle elections (which are “local-only” elections typically held in the spring); gubernatorial elections and presidential primaries are associated with municipal turnouts of 21 to 26 percent more registered voters (Figure S.2). In short, participation in local elections depends critically on the timing of those elections. At least one local government institutional arrangement is also related to turnout in municipal elections. Cities that provide more services with their own staff (as opposed to contracting out to firms or making service arrangements with other local governments) tend to draw a larger share of voters to the polls. This higher turnout may occur because city governments that provide services directly have more control over some of the basic issues that affect city residents’ quality of life, or 70 60 Registered voters Adult residents 50 Median percentage turnout 40 30 20 10 0 Off-cycle Odd-year Gubernatorial Presidential Presidential November primary Figure S.2—Variation in Turnout by Election Timing (controlling for other factors) viii because municipal employees tend to be a well-mobilized segment of the local electorate. By contrast, the distinctions between charter cities and general-law cities, and between cities with and without term limits for council members or mayors, have no direct relationship with local voter turnout. The context of the local election is also related to turnout. Where there are one or more propositions on the municipal ballot, cities tend to draw about 4 percent more registered voters to the polls. Similarly, the degree of competition for the office (measured by the number of candidates) is positively related to turnout. Predictably, uncontested elections draw especially few voters. If we use a city’s adult residents, rather than registered voters, as our reference group, these factors—timing, service provision, ballot measures, and candidate competition—likewise demonstrate important relationships to turnout. The effect of these factors is slightly diminished because demographic factors and citizenship barriers take on additional importance, heavily influencing whether residents can and do register to vote in the first place. Election Timing, Incumbent Success, and Competition for Office These results indicate that a move to concurrent elections has the greatest potential to expand voter participation in California’s local political arena. More than one-third of the cities responding to the city clerk questionnaire held their elections on a date that was not in the electoral “prime time” of a statewide election. Moreover, the timing of local elections is a subject of policy debate in California. More than 40 percent of the city clerks responding indicated that their city has made a change in the timing of municipal elections in recent years. As the state has changed the date of its primary election in an effort to have more effect on the national presidential primary season, numerous communities with nonconcurrent (stand-alone) spring elections have consolidated their elections with statewide contests, or have considered doing so. The reason is often budgetary; city governments pay a smaller ix share of the cost of holding an election if it is held concurrently with other contests. In considering a timing change, however, policymakers may wish to weigh the benefits of increased participation against other, potentially unintended consequences of changes in election timing. For example, critics have argued that some voters in concurrent elections are unaware of and inattentive to local issues and candidates because the focus of media coverage and popular attention is on the higher-profile state or national races. If true, concurrent elections could provide increased protection for incumbent candidates, who have greater name recognition on average than their challengers. An analysis of the reelection success of incumbent city officeholders provides some mild support for this notion—although the vast majority of incumbents win regardless of election timing. Whereas 80 percent of incumbent council members and 86 percent of incumbent mayors standing for reelection succeeded overall in the cities responding to the questionnaire, the reelection rate for incumbents was higher in cities that have concurrent elections. Nonconcurrent city elections tend to place incumbents in a more vulnerable position, perhaps because voters who come to the polls for stand-alone elections are more motivated or aroused by the issues facing the community. Of course, increased incumbent success in on-cycle elections could be a sign of something else altogether—perhaps greater trust or greater satisfaction resulting from more widespread participation. Our examination of competition for council and mayoral offices also reveals that incumbents have significantly lower reelection success rates in cities with citizen initiatives on the ballot—a possible reflection of community controversy. Cities with larger populations tend to have higher incumbent reelection rates—and lower voter turnout rates— indicating that voters are probably less likely to become engaged in civic issues in larger communities. The presence of an incumbent strongly dissuades potential competitors from entering local races. Competition for council or mayoral positions is lower where incumbents are standing for reelection. It is interesting to note that in cities where independent and third-party voters constitute a larger share of registered voters, there tend to be more x competitors for office, even though local elections in California are nonpartisan (that is, party labels do not appear on the ballot). In addition, the number of candidates for mayor is greater where the rewards of the office (full-time salary and length of the term) are greater. Can Participation in Local Elections Be Increased in California? The first and most important step to increase voter participation in city elections would be to shift their date to coincide with statewide or national contests. We estimate that if all cities in the state that hold nonconcurrent elections switched their elections to a presidential election date, approximately 1.7 million more Californians would have cast a ballot in their most recent city council contest. Yet this step should only be considered a limited reform. Despite the significant boost that concurrent elections could give to city turnouts, they would probably not increase turnout beyond the fairly disappointing levels already found in statewide and national contests. And although voter participation is closely correlated with greater trust, efficacy, and satisfaction in government, increasing turnout would not guarantee improved civic engagement in local communities, along such important dimensions as serving in neighborhood organizations or attending community meetings. Still, in the end, the doctrine of “one person, one vote”—a bedrock of democratic theory—probably outweighs any potential negatives. Some might argue that participation in local contests be left to the most engaged or interested voters. However, political equality and enhanced citizen participation in city politics are important goals in an increasingly diverse state with powerful local governments. xi Contents Foreword ..................................... Summary..................................... Figures ...................................... Tables ....................................... Acknowledgments ............................... iii v xv xvii xix 1. POLITICAL PARTICIPATION AND MUNICIPAL ELECTIONS ............................... Local Elections in the Context of Declining Voter Participation ............................ Do Policymakers Have Any Potential Leverage over Local Participation? ............................ Incumbent Reelection and Competition for Office ....... A Need for New Evidence: Examining Voter Turnout in California Cities .......................... 2. VOTER TURNOUT AND LOCAL CONTEXTS: RESULTS FROM A SURVEY OF CITY CLERKS ...... Mail Survey Procedures and Response Rate ............ Ascertaining Voter Participation Rates: The Core of the Survey ................................ Is Low Voter Turnout a Problem in California Cities? ..... Election Timing, Local Institutions, and Context ........ Election Timing: The First Key ................... Governmental Structure and Electoral Laws ............ District Versus At-Large Elections................. Term Limits .............................. City Service Provision ........................ The Form of Government in Cities ................ Local Structure Summarized .................... Electoral Context ............................. Local Direct Democracy ....................... Competition .............................. 1 1 5 10 11 13 13 14 16 17 18 20 20 21 22 23 24 25 25 26 xiii Incumbency ............................... Electoral Competition Across Races or Ethnicities ...... Supplementing the Survey: City Demographic Characteristics ........................... Summary .................................. 3. ACCOUNTING FOR DIFFERENCES IN VOTER TURNOUT ACROSS CITIES ................... Election Timing ............................. Institutional Factors ........................... Electoral Context ............................. City Demographics ........................... Summary .................................. 4. COMPETITION AND INCUMBENCY IN LOCAL ELECTORAL CONTESTS...................... Competition for Office ......................... The Success of Incumbents ...................... Summary .................................. 5. CONCLUSION ............................. 28 28 29 32 33 35 38 40 44 46 49 49 55 60 63 Appendix A. Survey of City Clerks .......................... B. Mail Survey Procedures and Data Issues .............. C. Voter Turnout Regressions....................... D. Candidate Competition Regression ................. 69 81 87 91 References .................................... 95 About the Authors ............................... 101 Other Related PPIC Publications ..................... 103 xiv Figures S.1. Distribution of Voter Turnout Rates in City Council Elections ............................... vii S.2. Variation in Turnout by Election Timing (controlling for other factors) .......................... viii 2.1. Service Provision by Type and Provider ............ 23 3.1. Distribution of Voter Turnout Rates in City Council Elections ............................... 33 3.2. Variation in Turnout by Election Timing (with controls) ................................ 36 3.3. Variation in Turnout by City Service Provision (with controls) ................................ 39 3.4. Variation in Turnout by Number of City Ballot Propositions (with controls) ................... 42 3.5. Variation in Turnout by City Population (with controls) ................................ 45 4.1. Incumbent Success and Election Concurrency........ 57 4.2. Success of Council Incumbents in Varying Electoral Contexts................................ 58 4.3. Incumbent Success and City Population ........... 59 xv Tables 1.1. The Demographic Profiles of California’s Voters and Nonvoters ............................... 1.2. The Different Preferences of Voters and Nonvoters .... 2.1. Timing of California Municipal Elections .......... 2.2. Local Government Structure in California .......... 2.3. Race or Ethnicity of Mayoral Candidates ........... A.1. Number of Valid Responses for Individual Variables .... B.1. Comparing In-Sample and Out-of-Sample Cities ...... B.2. Geographic Distribution of Cities in Sample ......... B.3. Important Missing Values..................... C.1. Turnout in City Elections: Regression Model ........ C.2. Descriptive Statistics of Variables for Observations in the Regression............................... D.1. Determinants of Competition for Office ........... 3 5 19 25 29 70 82 83 83 88 90 92 xvii Acknowledgments We are indebted to the city clerks of California for giving their time and providing much of the data on which this report is based. We would also like to thank Peter Richardson and Patricia Bedrosian for their expert editorial advice. Amy Bridges, Christopher Hoene, Christopher Jepsen, Joyce Peterson, and Fred Silva all provided detailed comments on earlier drafts of the manuscript, and the report is much improved as a result. Mark Baldassare offered a strong guiding hand throughout the project. Angelina Reyes and Max Neiman provided useful feedback on the survey instrument. Although this report reflects the contributions of many people, the authors are solely responsible for its content. xix 1. Political Participation and Municipal Elections Observers of municipal elections in California have often expressed concern that only a small fraction of Californians actively participate in these local contests. Yet no systematic studies of local voter turnout in contemporary California have been undertaken. In this report, we calculate participation rates for municipal elections across the state and address a number of important related questions about the vitality of those elections. How does political participation vary across cities?1 What political structures and electoral laws are associated with low turnout? What changes might promote higher turnout in local elections? Our data, drawn largely from a questionnaire mailed to all of the city clerks in the state in 2000, also allow us to investigate related issues of competition for local office and the role of incumbency in local elections. Local Elections in the Context of Declining Voter Participation Voter participation at every level of government in California is distressingly low and appears to be getting worse. In presidential elections, almost half of all eligible voters in California do not vote (California Secretary of State, 2001). In midterm congressional elections, over half of eligible voters fail to vote. The numbers are even worse for statewide primaries, where turnout now hovers around onethird of eligible voters. If current trends continue, these rates will decline further. Between 1972 and 2000, statewide turnout dropped roughly 15 percentage points in California (California Secretary of State, 2001). ____________ 1As is common in California, we use the term city as a synonym for municipality—in other words, a general-purpose, sub-county local government. 1 Without that decline, over 3 million more voters would cast ballots in statewide contests today. Nowhere is the problem worse than at the local level. Despite the direct effects that local government activities have on residents—for example, in providing public safety and other essential services, building infrastructure, and making land-use decisions—large numbers of Californians and other Americans ignore local elections. Turnout in municipal elections around the country averages half that of national elections (Morlan, 1984), and local voter turnout often falls below onequarter of the voting-age population (Bridges, 1997; Hampton and Tate, 1996).2 Anecdotal evidence from several cities suggests that turnout in California is even lower than in the rest of the country (Bridges, 1997). At the local level, important public policy decisions are made without input from most of California’s residents. This civic disengagement raises serious concerns (Hill, 2000; Bennett and Resnick, 1990). First, it raises questions about the legitimacy of democratic government. Democracy is supposed to operate at the will of the people. If a small minority of the population elects city, state, and national leaders, political institutions and elected officials may lack the broad support and confidence necessary to govern effectively. Second, low participation may reflect distrust of government (Bobo and Gilliam, 1990), a reduced sense of civic duty (Wattenberg, 1998), and decreased political efficacy (Finkel, 1985). In California, evidence suggests that these conditions are already all too common.3 Third, voting serves as an important educational tool. Active participation teaches citizens about the functioning of government and the issues that are currently under debate (Bennett and Resnick, 1990). ____________ 2Moreover, trends over time suggest that voter turnout in local elections is declining just as rapidly as it is in national elections (Verba et al., 1995; Karnig and Walter, 1993). 3As Baldassare (2000, p. 26) summarizes Californians’ opinions about government, “It is clear that Californians don’t care much and are cynical about politics. . . . They see their governments as bloated bureaucracies and believe that their elected officials are in the pockets of special interest groups. They don’t believe what candidates are telling them in television commercials. They think what they read in the newspapers about politics and government is probably slanted or biased.” 2 Without such participation, Californians may be less able to identify their best interests or to know how to protect those interests. Thus, low participation may increase the likelihood that policy decisions reflect an inaccurate and incomplete understanding of the available policy options and may lead to inefficient outcomes that hurt Californians. The very unequal distribution of voters and nonvoters among different segments of the population heightens each of these concerns. California residents who are highly educated, wealthy, old, and white are much more likely to participate than residents who are poor, young, less educated, and nonwhite (Table 1.1). These differences are especially large across educational levels. Fewer than half of eligible, voting-age Californians have a college degree or some college, yet these citizens constitute nearly three-quarters of all voters. A similar story can be told for racial and ethnic differences. Although about half of California’s total population is non-Hispanic white, some 72 percent of registered voters in 2000 were white. At the opposite end of the spectrum, Table 1.1 The Demographic Profiles of California’s Voters and Nonvoters White Latino Asian American African American Percentage of Percentage of Unregistered Voters Residents 72 44 16 40 49 65 Income under $40,000 Income $40,000 or more 49 63 51 37 High school degree or less education Some college or more education 26 56 74 44 Age 18 to 34 Age 35 to 54 Age 55 and older SOURCE: PPIC Statewide Surveys, 1998–2000. 27 42 31 49 39 12 3 Latinos made up only 16 percent of registered voters but accounted for fully 40 percent of the unregistered population.4 One major problem with an unrepresentative voting population is that it may skew the outcomes of the democratic process (Verba et al., 1995; Guinier, 1994; Casel, 1986). If, for example, a largely white electorate determines policy with little input from nonwhite voters, there may be reason to suspect that minority interests will be short-changed. More generally, if the interests of nonvoters diverge in important ways from those of voters, the candidates who are elected and ultimately the types of policies that are enacted may well favor one group over the other. Put simply, only a few will be speaking for the interests of nonvoters. In presidential elections, where turnout is relatively high, the political preferences of voters and nonvoters are not that dissimilar (Teixeira, 1992; Wolfinger and Rosenstone, 1980). However, differences in policy preferences are likely to be much greater in local contests where turnout is substantially lower and often more skewed along the lines of race and socioeconomic status (Wattenberg, 1998; Alford and Lee, 1968). In California, statewide surveys suggest that the views of registrants do differ from those of unregistered Californians in systematic and sometimes important ways (Table 1.2).5 A sample of voting intentions on recent California propositions suggests that Californians who are registered are significantly more conservative than unregistered Californians across an array of issues including labor union contributions to political campaigns (Proposition 226), bilingual education (Proposition 227), and school bond initiatives (Proposition 1A). However, the differences between registered voters and unregistered residents are not found across all issues. On issues such as defining marriage as a union between a man and a woman ____________ 4In this report, “white” refers to persons who identify as white and not Hispanic. The terms Hispanic and Latino are used interchangeably, as are the terms African American and black, reflecting the manner they are listed in the 2000 Census questionnaire. 5Survey respondents often claim to have voted when they did not, making this analysis suggestive rather than definitive. 4 Table 1.2 The Different Preferences of Voters and Nonvoters Proposition 22 (limit on marriages) Proposition 1A (education bond) Proposition 227 (bilingual education) Proposition 226 (political contributions by unions) SOURCE: PPIC Statewide Surveys, 1998–2000. Percentage in Favor Registered Unregistered Voters Residents 61 63 73 85 71 62 61 49 (Proposition 22), differences between the registered and unregistered are less stark. In the end, it is unknown whether the divergent views of those who participate and those who do not participate translate into distorted policy outcomes, but given the large differences found between voters and nonvoters in California, the relatively low turnout in local elections merits attention. For all of these reasons, low turnout in local elections poses a potentially serious problem. In a political arena that touches regularly and deeply on the daily lives of residents, nonparticipation by a large share of eligible voters is worrisome. Failing to vote may be a sign not only of political disengagement but also of distrust, division, and cynicism. Where participation is low, government outcomes may be less equitable or responsive. Thus we seek to understand which factors affect voting levels in California’s municipal elections and what steps might be taken to try to encourage broader participation. Do Policymakers Have Any Potential Leverage over Local Participation? Can anything be done about this low rate of voter turnout in municipal elections? This report examines four sets of variables to help determine which factors are associated with higher or lower turnout: election timing, the structure of electoral and political institutions, the 5 context of the particular elections, and the demographic attributes of the cities in which they take place.6 Most research on political participation has focused on demographic factors related to voting. 7 We focus most of our attention on timing, institutional structure, and context. Although these three features of local elections have been less well studied, they are potentially significant for policymakers interested in increasing local voter participation. Unlike demographic attributes, which are either impossible or extremely difficult for city policymakers to alter, the three features that we focus on can often be manipulated by local governments. California municipalities (particularly charter cities) generally have the ability to alter the way they conduct elections and can in many cases reform their institutional structure, although in some cases state legislative action may be necessary. Thus, if it turns out that voter participation is related to any of these three features of the local electoral context, changes could be enacted to broaden political participation. To fully understand the role that timing, institutional arrangements, and context might play, it is necessary to outline a series of changes enacted in response to the Progressive movement roughly a century ago (Bridges, 1997; Welch and Bledsoe, 1988). These institutional changes were, according to the Progressives, crucial to the functioning of “good government” at the municipal level. Opponents and subsequent critics have viewed the institutional changes pushed by the Progressives as efforts to limit the participation of the working class and of ethnic groups and to usurp power at the local level (Bridges, 1997). Whatever the true motives of the Progressives, the result of their efforts is that many cities in California and across the country fundamentally altered their basic governing structure. New cities that have since formed in growing ____________ 6Mobilization by parties and campaigns and individual attitudes toward the political arena can also have an effect on voter turnout (Rosenstone and Hansen, 1993; Dennis, 1991). However, study of these factors is beyond the scope of this report. 7Our data relate to municipal political systems rather than to individual voters. Therefore, we will refrain from drawing inferences about individual-level attributes that may affect city-level participation, although we do take account of these demographic factors as “controls.” 6 western states such as California have also typically incorporated most or all of the institutions that were advocated by the Progressives. One change supported by the Progressives was a move from on-cycle or concurrent elections—elections that coincide with statewide or national contests—to off-cycle or nonconcurrent elections. Progressives claimed that this change in election timing would mean that local contests would be decided by the most interested and knowledgeable voters. Critics countered that nonconcurrent elections were pursued as a means to limit participation to those segments of the population most likely to support Progressive candidates. Whatever the initial motivation, many cities across the state have set the timing of their elections so that they do not coincide with state or national contests. Election timing is especially important for our purposes because studies at the national level have revealed a strong relationship between on-cycle elections and higher turnout (Espino, 2001; Hampton and Tate, 1996). The second institutional feature of local elections that we focus on is the electoral mechanism used for the city council. In an attempt to purportedly reduce “narrow” neighborhood interests in favor of citywide concerns, Progressives favored at-large elections over district- or wardbased contests.8 Evidence nationally suggests that this change may have reduced voter interest and participation by distancing leaders from their constituencies (Bullock, 1990). One reform sought by Progressives, direct democracy, may have increased voter turnout (Hahn and Kamieniecki, 1987; but see Magleby, 1994). By giving more decisionmaking power to the people in the form of initiatives, referenda, and recall, Progressives hoped to take power away from entrenched local politicians and interest groups who were, in their view, corrupt.9 In the process, these changes may have increased voter participation by increasing citizen interest in local elections. ____________ 8Critics have seen the move to at-large elections simply as a way for the Progressives to usurp power for their own “narrow” interests and reduce the influence of ethnic neighborhoods. 9Critics have countered that direct democracy was a way for Progressives to pursue their own agenda, and that the initiative process is itself often wielded by special-interest groups. 7 These reforms have not been pursued (or in the case of local voter initiatives, used) uniformly across the state. Chapter 2 provides greater detail on just how widespread change has been. For now, the important point to note is that variation in these institutions and the frequency of local initiative use make it possible to analyze the effects of such electoral and governance arrangements on voter turnout. By studying the relationship between cities’ turnout and their institutional structure, we should be able to determine what, if any, institutional changes might be pursued to increase the participation of Californians in local electoral politics. The effects of two other institutional changes pursued by the Progressives—nonpartisan local elections and the council-manager form of government—cannot be evaluated in California because they exist in virtually every city in the state. There is reason to suspect that both have affected turnout.10 Nonpartisan elections—those in which party labels are not included on the official ballot and parties do not have control over nominating procedures—are mandated for local elections statewide and are thus not included as a variable in our analysis. Similarly, almost all cities in California have chosen a council-manager system in which the day-to-day operations of the city are overseen by an unelected professional city manager or administrator, who is appointed by the council. By contrast, the earlier tradition of the mayor-council plan of government gives stronger executive powers to a directly elected mayor who does not sit as part of the city council. In practice, variation within the two forms of government (council-manager and mayor-council) may now be as important as the distinction between the two plans.11 For this reason we focus more on measures of mayoral strength, where available, ____________ 10Nationwide, nonpartisan elections have reduced voter turnout—perhaps by limiting the efforts of parties to mobilize voters (Schaffner, Wright, and Streb, 2001; Espino, 2001). Similarly, national studies have found that the move to the councilmanager system has reduced voter participation—perhaps by providing for a weaker elected executive and a more insulated administration than its predecessor, the mayorcouncil system of government (Karnig and Walter 1983; Espino 2001). 11Cain, Mullin, and Peele (2001) have detected a number of hybrid forms and have illustrated cases in California where, despite the presence of the council-manager system, mayors have extensive managerial and agenda-setting powers. 8 than on the traditional distinction between council-manager and mayorcouncil forms of government. Several facets of local institutions and elections for which the Progressive movement is not responsible are also of interest. For example, one important consideration concerns local service provision. Although the norm in earlier periods of municipal government was for the city to provide the full array of municipal services, a host of alternative service arrangements have evolved in recent decades. These arrangements range from provision of municipal services by special district governments or county departments to efforts to contract out service provision to private firms. In some “contract cities,” particularly prevalent in Southern California, few local public services are actually directly carried out by city employees. Under these circumstances, some measure of direct influence is removed from the hands of city elected officials, potentially reducing interest in municipal politics and depressing turnout. Term limits are a more recent electoral reform that might also influence local voter turnout. Many cities in California have sought to curb the powers of incumbents, encourage more candidates to run for office, and ultimately increase competition for office by limiting the number of terms that mayors or council members may serve. At this point, however, it is unclear just what effects term limits have had on either competition or turnout (Thompson and Moncrief, 1993). Finally, we also examine the distinction between charter cities and general-law cities as an institutional factor potentially affecting local residents’ participation in politics. About one-fifth of California cities are organized under charters, or local constitutions, in which local voters consent to a specific set of organizational arrangements and powers for the municipality. The rest derive their authority from and operate under the provisions of state law relating generally to local government. Historically, charter cities retained more autonomy from the whims of the state than general-law cities, although in recent decades legislative changes and court decisions have eroded the distinction between the two types of municipalities. In general, all California municipalities have a relatively broad grant of home-rule authority. Charter cities are permitted to adopt a “strong mayor” form of government, alter the 9 number of council members (from the otherwise prescribed five), and have somewhat greater leeway regarding purchasing arrangements and personnel.12 Whether charter status reflects these modest additional powers or a more activist tradition of municipal home rule in the city’s past, some might expect that charter cities would experience more voter interest in local politics and thus perhaps higher turnout. In addition to our strong interest regarding the potential effects of local institutions on turnout, we also focus on a number of contextual features of local elections that may influence voter participation. For example, one important factor that has the potential to drive turnout at any level in politics is the degree of competition among candidates for office (Cox and Munger, 1989). Similarly, the well-known electoral advantages that incumbent officeholders have may reduce the chances of potential challengers (Jacobson, 1983), thereby possibly making elections less interesting to voters. Incumbent Reelection and Competition for Office In this report, we also examine two of the potential pitfalls that might be associated with increased turnout and, specifically, with oncycle elections: reduced competition for office and greater protection for incumbents. A number of observers of local politics suggest that concurrent elections deter challenges to incumbents by reducing the prominence of local elections (Johnson, 1994). With popular attention focused mainly on national and state contests (and in California, statewide initiative elections), local challengers might find it hard to gain attention, garner media coverage, and raise enough money to warrant entering the race. A move to concurrent elections might also conceivably help incumbents by incorporating voters with little interest in or knowledge of local affairs, many of whom might thus vote for incumbents by default. Already, some scholars argue that voters seldom have the necessary knowledge to make informed decisions (Campbell et al., 1980). In this ____________ 12Charter cities, unlike general-law cities, are also allowed to adopt zoning codes that are not strictly consistent with the local general plan. On charter status, see Sokolow and Detwiler (2001); Curtin (2000, p. 18). 10 way, higher turnout and concurrent elections might potentially lead to less competitive elections. Analyzing the level of competition in municipal elections and the success rates of incumbents thus provides additional insight into the political vibrancy of cities around the state. A Need for New Evidence: Examining Voter Turnout in California Cities There is at least anecdotal evidence indicating that each potential institutional and contextual factor we have outlined above may affect voter turnout in America’s municipalities. Yet few of these factors have been tested systematically across a wide range of cases, largely because of a lack of suitable data. The last major published study of municipal election turnout appeared more than 30 years ago (Alford and Lee, 1968). This study is intended to fill this gap. Much of the analysis relies on data from a questionnaire administered to city clerks. In Chapter 2 we describe this fact-gathering survey and provide an overview of the institutional structures and electoral laws in place in the state’s municipalities, as disclosed by the survey respondents. Chapter 3 summarizes our statistical analysis of the factors that influence voter turnout, detailing the role of institutional, contextual, and demographic factors. Chapter 4 turns to other issues related to engagement in city electoral politics, examining competition for office and the success of incumbents. The report concludes in Chapter 5 with a discussion of the implications of these results for understanding citizen participation in California’s local elections and the policy implications of the study’s findings. 11 2. Voter Turnout and Local Contexts: Results from a Survey of City Clerks To gather the data necessary for our central research questions, we devised and distributed a mail questionnaire to each California municipality. This chapter describes the survey, the resulting sample, and the information gathered regarding voter turnout patterns, institutional structure, and the context of city elections. The most important finding from the simple survey results is that turnout is quite low in California municipal elections, averaging well below half of registered voters. Mail Survey Procedures and Response Rate The questionnaire was mailed to every city clerk in the state in late 2000. City clerks are designated as the chief municipal elections officials, although the actual task of administering the election and tallying votes rests with the county government. City clerks, furthermore, often have unique, first-hand knowledge about the political life of their communities.1 Of the 474 cities in existence at the time of the survey, a total of 397 clerks (84 percent) returned surveys with at least some of the necessary responses; however, complete and usable data for calculating voter turnout rates themselves were available for a smaller number—350 cities. The 350 responses allow us to report turnout data for 79 percent of the cities in the state, a solid rate of response. Our sample of cities is generally representative of all cities in the state. Comparing cities that responded to those that did not revealed few significant differences. Cities in and out of the sample were similar in ____________ 1For a study that relies on city clerks as informants regarding local politics see Schneider and Teske (1995). 13 terms of regional location (Southern California, San Francisco Bay Area, Central Valley, or other) and in terms of the percentage that were central cities, suburbs, or rural communities. Our sample was also representative in terms of racial demographics, percentage of residents unemployed, median household income, and home ownership rates. There were marginal differences between cities in and out of the sample in the average size of the population, the average household size, and the poverty rate. Compared to other city clerks, those in larger communities were more likely to respond. A random sample of responses was validated using municipal web pages and a variety of other sources. Errors were minimal, and any errors that were found were corrected. Appendix B provides a more detailed discussion of the survey procedure, the representativeness of the sample, and response validation. We asked each city clerk a series of questions about his or her community’s most recent city council election and mayoral election. Every city holds council elections, but only one-third of cities that responded directly elect their mayor, yielding a total of 130 mayoral elections. In subsequent chapters, we analyze a dataset that combines both council and mayoral elections. When we examined each type of contest separately, the results were generally quite similar to those presented in this report. Ascertaining Voter Participation Rates: The Core of the Survey The key variable in this report is voter turnout. For each election, we asked city clerks to report the total number of residents who cast ballots, as well as the total number of registered voters at that time in the city (or in the relevant districts if only certain city council district seats were up for election). In addition, we estimated the size of the votingage population for each city at the time of the election. The Census Bureau publishes the voting-age population of every city but only for decennial Census years. Thus, the measure we use is an interpolation 14 (for the particular year of the election) between the voting-age population in the city in 1990 and in 2000.2 In the analysis that follows, we focus on two dependent variables: the percentage of registered voters who cast ballots for local office in a given election and the percentage of voting-age residents who cast ballots. We focus heavily on turnout of registered voters because we believe that it is likely to be the measure most directly affected by changes in the local institutional and electoral context. Registered voters have already cleared the major entry barrier to political participation (registration) and thus whether or not they vote is likely to be closely related to conditions at the time of the election. Also, our data on registered voters are more accurate than our interpolated estimate of voting-age residents. At the same time, if we want to know how likely the population at large is to participate in local elections, it is important to examine turnout of the voting-age population as well. By focusing on the votingage population and not the registered population, we can take account of the intermediate step of voter registration, which has at times in the past been used to exclude certain segments of the electorate (Parker, 1990; Davidson and Grofman, 1994).3 Although registration procedures are set by state and federal laws, they are typically administered by county officials, who historically have had a fair amount of latitude in carrying out these procedures. Although the average city in our sample had an (estimated) registration rate of 63.8 percent, there was a wide degree of variation in this rate across communities.4 Thus, differences in registration rates across cities could have a number of different causes other than simply the level of interest ____________ 2Obviously, this measure is more precise for those cities that reported on elections held in 2000. 3One could also specifically examine the process of registration. For example, Mitchell (1992) demonstrated how voter rolls have at times been purged, seemingly for arbitrary reasons, such that certain segments of the population were disenfranchised. However, despite the fact that the percentage of residents registered, turnout of votingage population, and turnout of registered voters measure somewhat different aspects of participation, they are often fairly highly correlated (Hampton and Tate, 1996). 4The standard deviation for the percentage of adults registered was 16.2 percent. 15 in elections. As a result, higher turnout of registered voters may not necessarily indicate broader participation but might instead be a sign that barriers to registration at the local level have reduced the number of registered voters. For these reasons, we undertake all subsequent analyses for both turnout of registered voters and turnout of the (estimated) voting-age population. In practice, turnout of registered voters and turnout of the voting-age population are very highly correlated among California cities (r = 0.85), and the factors that lead to increased turnout of registered voters also tend to increase turnout of the adult population. Is Low Voter Turnout a Problem in California Cities? Compared to voter turnout in statewide and national contests, local voter turnout is low. In the last presidential election, 71 percent of California’s registered voters participated. Recent statewide contests for governor have attracted roughly 60 percent of registered voters (California Secretary of State, 2001). Our data indicate that voter turnout in municipal elections falls well below these levels. Of the 130 cities reporting directly elected mayors, only 44 percent of registered voters showed up at the polls and cast a vote in the average contest. In city council elections, where we have complete turnout data for 350 municipalities, the mean turnout was similar—48 percent. Measured as a percentage of the voting-age population, turnout looks even worse. Turnout in mayoral contests averaged only 28 percent of the voting-age population. In council elections, the average was 32 percent of the voting-age population. It is also possible to calculate the overall participation rate among these cities—the number of voters who cast ballots in all of the city elections for which we have data, divided by the total number of registered voters in those cities at election time. We found that the overall participation rate in council elections was 43 percent of registered voters; only 3.7 million voters out of 8.6 million registered in these communities cast ballots in their most recent city council contest. The aggregate turnout rate for mayoral elections was even lower: 39 percent. 16 These low levels of turnout are disturbing. Important policy decisions are being made by local officials chosen by a relatively small group of citizens who may differ in important ways from the nonvoting population. Election Timing, Local Institutions, and Context Why are so many Californians not voting in local contests? What is it about California’s municipal elections that leads to low turnout? To help answer these questions, we endeavored to ask city clerks about as many potentially relevant factors as possible. In particular, we asked several questions about three sets of factors that one might expect to have an effect on voter turnout rates: the timing of mayoral and council elections, the set of political and electoral institutions found in each city, and the political context of the specific election of interest. In the sections that follow, we discuss the motivations behind these questions and describe the city clerks’ responses. Most of our discussion of possible solutions to the problem of low voter turnout focuses on election timing and local institutions, for three reasons. First, election timing and institutional structure have been highlighted as potentially important contributors to low voter turnout across the country (Espino, 2001; Hampton and Tate, 1996).5 Second, it is relatively easy to alter these institutions. In some cases, only a local ordinance would need to be changed; other changes would require alterations to city charters or state law. Third, changes in these institutions and electoral laws are not unprecedented. Throughout the last century, numerous alterations were made to the basic governing structure of cities in the United States and changes are still routinely considered for a variety of reasons. Although our primary purpose in conducting the mail survey was to try to understand voter turnout in municipal elections, we believe that the data we present on the electoral procedures and institutional structures of California’s cities are important in and of themselves. ____________ 5Other electoral laws considered relevant to voter turnout, but not examined here, are the types of registration requirements and the ease of registration (Powell, 1986), along with nonpartisanship (Schaffner, Wright, and Streb, 2001). 17 Given the fluid nature of the electoral procedures and institutional structure of California’s cities, it is important to provide some basic facts on what California’s city governments look like. To date, few systematic data have been collected on the distribution of electoral institutions and governing structures across California’s cities. The following sections should begin to fill in these gaps. Election Timing: The First Key Among the structural factors related to voter turnout, we are most interested in the timing of elections. Scheduling local elections on the same date as statewide primaries or general elections may increase local turnout to levels almost on par with voting in national elections. Having the dates of local elections coincide with the dates of statewide primary or general elections (often referred to as “on-cycle,” “consolidated,” or “concurrent” elections) makes participation in local elections easier. Voters need only check off names further down the ballot to participate in the local political contest. Given the significantly higher turnout rates for national and state elections, gains in turnout could be substantial. Moreover, cities have other incentives besides increased voter participation to switch to concurrent elections. In fact, the primary motivation for this move has usually been cost savings. Because municipalities generally pay the entire cost of holding nonconcurrent elections and only a fraction of the costs of concurrent elections, local officials tend to view this move as a way to cut the costs of administering elections. For example, the city of Concord recently estimated that its cost for running a stand-alone election would be $58,000—more than twice as much as the $25,000 estimate for running a consolidated election (Simerman, 1998). However, local elected officials could have motivations for opposing a move to concurrent elections. By staggering local and statewide contests, local officials often have the opportunity to run for positions in state government without giving up their local positions. On-cycle local elections would limit that ability. Numerous local governments have already moved the dates of their elections to coincide with statewide elections, and many others have considered the change (Rohrs, 2000; Simerman, 1998). In Contra Costa County, for example, 55 cities, school districts, and special districts (out 18 of 73 total local governments) were running consolidated elections as of 2000—up from just 36 in 1996 (Rohrs, 2000). Our survey data indicate that at least 134 cities in California have changed the timing of local elections in recent years, with the vast majority of those switching from stand-alone elections to elections concurrent with statewide contests.6 We asked survey respondents to record the date of their most recent council and mayoral elections (if applicable).7 These election dates can be classified as falling into one of the following five categories: presidential election, presidential primary, gubernatorial election, oddyear November election, or completely off-cycle election (Table 2.1). A large fraction of the cities that responded to our questionnaire (20 percent) hold off-cycle city council elections, usually in the spring. Of the cities that directly elect their mayor, a similar percentage (17 percent) hold off-cycle elections. Roughly one in six cities hold mayoral or council elections in November of odd-years. In total then, about one- Table 2.1 Timing of California Municipal Elections Election Timing Presidential Presidential primary Gubernatorial Odd-year November Off-cycle Percentage Percentage of Council of Mayoral Elections Elections (N=397) (N=134) 12.6 8.2 5.8 6.7 44.6 52.2 16.6 15.7 20.4 17.2 NOTE: Data refer to the most recent election, or in some cases to the most recent contested election. ____________ 6Specifically, of the 308 clerks answering the relevant survey question, 94 (30.5 percent) indicated a change from nonconcurrent to concurrent elections, 3 (1 percent) switched from concurrent to nonconcurrent dates, 37 (12 percent) indicated a change from one nonconcurrent date to another, and 137 (56.5 percent) indicated no change. This calculation excludes 94 cities, or about one-quarter of the total number of respondents, who did not answer this question. Most likely these cities did not change the timing of their elections. 7Several cities whose last election was uncontested—that is, where the number of candidates equaled the number of seats—chose, helpfully, to report instead on their last contested election. 19 third of all the local elections that city clerks identified were nonconcurrent in one form or another (odd-year November or off-cycle). Thus, if election timing does matter and nonconcurrent elections are associated with particularly low turnout, there is clearly room for improvement. Another important question about election timing is whether mayoral and council elections are held simultaneously with other local contests, including those for other citywide offices (such as city attorney or treasurer) or for other governing bodies (such as school boards or county boards of supervisors). The presence of these other local elections might also serve to spur turnout. With this in mind, we asked city clerks whether a number of other local offices were being voted upon on that same day. In just over half (53 percent) of the cities responding to the survey, voters were casting ballots for at least one other local office. The offices most commonly elected on the same day as city offices were school board (31 percent) or county supervisor (17 percent). In response to a related question, 96 percent of the cities that directly elect their mayors held council elections at the same time; 25 percent of the council contests reported by the clerks had a mayoral election on the same day. Governmental Structure and Electoral Laws In addition to election timing, four factors identified by past research have the potential to affect turnout and are to some extent controlled by policymakers or voters. They are the method of electing council members, the use of term limits, the provision of services by the city, and the general form of government. District Versus At-Large Elections As part of their institutional reform agenda a century ago, Progressives sought to create at-large or citywide elections to choose members of the city council. Reformers of the time maintained that replacing district elections with at-large elections would weaken neighborhood and partisan interests in favor of less parochial citywide interests. Recent critics of the move, however, have suggested that atlarge districts not only decrease participation by distancing leaders from 20 their local constituencies (Bullock, 1990) but also may hurt the interests of geographically clustered racial and ethnic minorities who may be overwhelmed by a white voting plurality (Engstrom and McDonald, 1982; Welch, 1990; but see Bullock and MacManus, 1990). Our survey indicates that district elections are not widespread in California. Only 21 cities (5 percent) used the district form at the time of our survey. Another nine cities used some hybrid form that typically involved each district voting to nominate two candidates, with a citywide runoff between each pair.8 The vast majority of California cities (93 percent) use a simple at-large system of electing council members. Term Limits Term limits are a more recent electoral reform that may influence local voter turnout. One motive for term-limit proponents is to increase the competitiveness of elections. Term limits are seen as a tool to level the playing field, encourage more candidates to run for office, and ultimately make elections more dynamic (Copeland, 1997). However, an unintended consequence of term limits may be to increase voter confusion and disinterest by introducing a greater number of “unknown” nonincumbents into electoral contests (Rosales, 2000). At this point, it is unclear what effect term limits have had on either competition or turnout (Thompson and Moncrief, 1993). Californians favor term limits for many of their elective offices, at least at the statewide level.9 Term limits at the local level are far less common, however. Nationwide, about 10 percent of cities place limits on the number of times city council members or mayors can run for reelection (ICMA, 1996). In California, these proportions are somewhat higher. Our data indicate that nearly one in five city councils faces term ____________ 8Since these mixed systems have strong elements of district-level selection, we classify them with the district cities in the analysis in the chapters below. 9Californians voted in a bare majority (52 percent) in 1990 to support Proposition 140, which established term limits for state offices including the state assembly and senate. State voters were much more responsive to a state amendment to require congressional term limits—Proposition 163 passed with 64 percent of the vote in 1994— although this law was later ruled unconstitutional. 21 limits. Of mayors who are directly elected, 27 percent (36 cities) are term-limited. City Service Provision One of the most controversial issues in municipal governance is the question of whether city services should be directly provided by municipal employees or by some other entity. In an effort to reduce costs and provide more efficient services, numerous cities in California and elsewhere have chosen alternative arrangements for providing local services such as police protection or garbage collection. Often these services are performed by private firms, special districts, the county government, or other nearby local governments. Whether such arrangements ultimately reduce costs and improve service quality has been widely debated. However, it is likely that local elected officials in cities that do not provide such services with in-house staff have somewhat less direct influence; if nothing else, they have fewer city jobs to control. The reduced role for city staff and elected officials may have the unintended consequences of reducing interest in municipal politics and depressing voter turnout. To investigate this issue, we asked city clerks to indicate their cities’ arrangements for five common municipal services: police, fire, library, sewerage, and garbage collection. Clerks were asked to indicate whether each service was “mainly carried out by city government personnel, by county personnel, by a special district government, or by a private company under contract with the city.” Cities varied widely in terms of the number and type of services that were provided by other governments or private companies. The mean number of city-provided services was 2.4—roughly half of the five we asked about—indicating that contracting and alternative service arrangements are quite common among the state’s municipalities. However, 13 percent of cities indicated that city staff provided none of the services in question and 25 percent responded that they provided four or five of the services with their own staffs. Police, fire, and sewerage were the functions most likely to be carried out by city government personnel (Figure 2.1). At the other extreme, trash collection was rarely performed by city staff. 22 Percentage of cities 90 City Other government 80 Private Not provided 70 60 50 40 30 20 10 0 Police Fire Library Sewer Garbage Service Figure 2.1—Service Provision by Type and Provider The Form of Government in Cities Historically, California’s cities have been classified into two categories: charter cities and general-law cities. Although the distinction between the two categories has diminished over time (as noted in Chapter 1), charter cities (which tend to be the larger cities in the state) still have somewhat greater autonomy than general-law cities. For example, charter cities can revise certain aspects of their electoral institutions—such as the number of city council members—more easily and have slightly more discretion regarding procurement and personnel arrangements. This greater autonomy suggests that voters in charter cities have more at stake in local elections. Thus, we might expect voter turnout to be slightly greater in charter cities. Although we did not ask specifically about this distinction in the questionnaire, we were able to obtain data on the basic form of government for each city from the California Secretary of State’s office. Most of the cities (79 percent) responding to our questionnaire are general-law municipalities. Some 98 cities (or 21 percent of the sample) are charter cities. 23 We did inquire about the distinction between cities that have a council-manager form of government, where the day-to-day administration of the city is run by an unelected professional city manager, and cities with a mayor-council form of government. More than half of the nation’s cities currently operate under a council-manager form of government (ICMA, 1996). In California, however, nearly every city (97 percent, according to our survey) uses the council-manager form.10 Although nationally studies have found that cities with councilmanager governments tend to have less voter participation (Karnig and Walter, 1983; Espino 2001), there is simply not enough variation in California to test this argument. However, we also asked city clerks to report the extent of two mayoral powers—budgeting authority and the ability to veto council actions. Cain, Mullin, and Peele (2001) suggest that a more refined measure of mayoral power is more telling in California than the simple council-manager versus mayor-council distinction. Of the cities in our survey, only one has direct mayoral control over developing the budget (although in four other communities, the mayor is indicated as jointly participating in this process); 14 mayors had veto power over their councils (about 4 percent of the cities responding to the question). Local Structure Summarized On many of the local government features that we examine in our questionnaire, there is limited variation across California cities (Table 2.2). The vast majority of cities are council-manager governments with council members elected at-large. Mayors are directly elected in about one-third of cities but tend to have few independent executive powers. On other measures, California’s cities appear to be quite mixed. There is a fair degree of variation in the use of term limits and city service arrangements. These more recent municipal reforms (term limits and ____________ 10Considerably more cities directly elect their mayor than use the mayor-council form of government. In most of these cases, mayors are “first among equals” in council matters; the city manager retains control over important functions such as developing the city budget and hiring department heads. 24 Table 2.2 Local Government Structure in California At-large council elections District method Combination Percentage of Cities (Number) 93 (362) 5 (21) 2 (9) Term limits—council members Term limits—mayor (of directly elected mayors) 18 (71) 27 (36) Provide none of five city services with own staff Provide all five General law Charter 14 (54) 8 (30) 79 (309) 21 (83) Council-manager form Mayor-council form 97 (384) 3 (10) Mayor develops (or jointly develops) the budget Mayor has veto power 1 (5) 4 (14) NOTE: Percentages are of the number of cities answering each question. contracting) are less well established, thus far, than the ones initiated by Progressives a century ago. Electoral Context We also asked city clerks about several aspects of the context of specific local elections. Although not as amenable to legislative change as the structures of city government, the political context of city elections might also be expected to play a major role in influencing voter interest and participation. Here we consider various central elements of the local context: the presence of local ballot propositions, the degree of competition for office, the presence of incumbents, and the race or ethnicity of the candidates. Local Direct Democracy Although direct democracy has both proponents and critics, the initiative process does offer the potential to increase turnout (Hahn and 25 Kamieniecki, 1987; but see Magleby, 1994).11 Certain statewide initiatives such as Proposition 187, which sought to eliminate services to illegal immigrants and their children, have sparked greater participation by at least some sectors of the electorate, such as Latinos (Pantoja and Segura, 2000). These gains could also occur at the local level, given that the initiative process can also be employed in California’s cities and counties. Use of the initiative has grown rapidly at the state level, and it is now fairly common to see more than ten propositions on the statewide ballot. We asked cities to identify the number of local propositions put on the ballot, either by the city council or through the popular initiative system, in which citizen activists qualify a measure for the ballot by gathering voter signatures. According to the questionnaire responses, only 43 of the 387 cities (11 percent) had at least one municipal-level citizen initiative on their most recent ballot—indicating that despite the press attention they gather, citizen initiatives occur in relatively few cities each year. By contrast, 98 cities (25 percent) had a question placed on the ballot by the council.12 We also asked city clerks to identify the total number of citizen initiatives placed on their ballots from 1997 through 1999 to get a longer-term perspective on the use of direct democracy in local elections. About 27 percent of cities had citizen initiatives over this period, more than twice as many than in the single election we asked about. Competition One key factor that can influence turnout at any level of government is the degree of competition among candidates for office (Cox and Munger, 1989). Research has generally shown that the more competitive the race for a particular office, the higher voter turnout is likely to be. What is not entirely clear at the local level is whether it is the closeness of ____________ 11For accounts of many of the criticisms of direct democracy see Gerber (1999), Schrag (1998), and Rose (1990). 12Respondents were instructed to report only city-level ballot questions, not, for example, school district or county-level measures. 26 the contest that affects turnout or the sheer number of candidates vying for office. More candidates may be a sign of competing interests and greater divisions that could in turn increase (or conceivably even depress) turnout. Given the necessary brevity of a mail questionnaire, we were able to ask clerks only a limited number of questions about competition. Clerks reported the number of candidates running for office and the presence of incumbents. Because of the variety of methods used in selecting council members, we could ask about the margin of victory only for mayoral elections. The results indicate that elections in California municipalities are not strikingly competitive. Council elections average 2.5 candidates for each seat available; mayoral races attract slightly more but still fewer than three candidates on average. Perhaps more surprisingly, numerous cities in California have uncontested elections—15 council races (or 4 percent of all cities) and 23 mayoral contests (18 percent of all mayoral elections) were reported as having the same number of candidates as seats available. These figures may overstate the actual amount of competition and understate uncontested races by an unknown amount, because a handful of cities reported data for their most recent contested election, omitting a subsequent uncontested election. Indeed, cities with stand-alone elections sometimes choose to cancel elections that are uncontested and simply certify the lone candidate as the winner. For mayoral elections, we asked city clerks to identify the portion of the vote captured by each of the top two finishers.13 Close races indicate a competitive election that might spur turnout, although it is unclear that voters would know in advance that the outcome is likely to be close. As it so happens, mayoral races are rarely very close. In contested races, the average margin of victory was 24 percent. Only 13 percent of contested mayoral contests were decided by a margin of 5 percent or less. However, a few elections might lead one to recall the 2000 presidential ____________ 13For more than half of the mayoral contests, there were fewer than three candidates. 27 contest; three mayoral races, or 2.9 percent of the contested elections, were decided by 1 percent of the vote or less.14 Incumbency Although term limits might influence turnout rates by limiting the presence of incumbents, we also addressed the incumbency issue directly. Some have argued that because incumbents generally have widespread name recognition, can “scare away” quality opponents, and can use the resources of elected office to raise large sums of money, their presence in an election may depress turnout. The survey asked city clerks whether mayoral or council incumbents ran for reelection. For mayoral contests, we used a simple yes-or-no query. For council elections, we asked for the number of council seats up for election and the number of incumbents running. The results show that incumbents are present in most elections. Two-thirds (66 percent) of the council seats available in city elections had an incumbent council member running. Of the mayoral races, 72 percent featured an incumbent mayor defending his or her seat. As might be expected, cities with term limits had lower rates of incumbency, at least for council seats (55 percent). Oddly, the cities with mayoral term limits were slightly more likely to have an incumbent mayor running (77 percent), but this may be an aberration resulting from the low sample size (only 36 cities directly elect mayors and have mayoral term limits). Electoral Competition Across Races or Ethnicities In a diverse state with rapidly changing demographics, one might expect contests for local office to draw more attention or seem more salient if the contestants were from different ethnic or racial ____________ 14 The survey questions queried city clerks about the general mayoral election, not a runoff election. However, runoffs are quite rare, with only eight cities (6 percent) holding runoffs, of the 134 who directly elect the mayor. Even among those cities where the top candidate for mayor received less than 50 percent of the vote (because there were multiple candidates), only a quarter held a runoff. Nearly all of the communities with provisions for runoff elections are large central cities. Given the sizeable number of cities that elect a mayor with less than a majority of the vote, it is perhaps surprising that cities have not considered adopting preference voting schemes, which provide for an “instant runoff.” 28 backgrounds. For this reason, we asked the city clerks to identify the race or ethnicity of the first- and second-place mayoral candidates.15 Although not all respondents were willing or able to identify the candidates in this way, we collected such information for 122 victors and 97 second-place contestants (Table 2.3). As the table indicates, more than three-quarters of winners and runners-up were non-Hispanic whites. In most cases, the top two finishers in a city were of the same race or ethnicity. However, in onethird of the contested elections for which we have race/ethnicity identifiers for both candidates (32 of the 97 elections), the contest was between candidates of different ethnic or racial backgrounds. Most of these cases (21 of 32) involved a race between a Latino candidate and a white candidate.16 Table 2.3 Race or Ethnicity of Mayoral Candidates White, not Hispanic Black, not Hispanic Asian, not Hispanic Hispanic, of any race Other Total Winner 2nd-Place Finisher Number Percent Number Percent 97 79.5 75 77.3 5 4.1 7 7.2 1 0.8 2 2.1 19 15.6 12 12.4 0 0.0 1 1.0 122 100 97 100 Supplementing the Survey: City Demographic Characteristics As we noted earlier, local institutional characteristics and election context are not the only factors that can influence turnout. Individual demographic characteristics can also affect voter participation. Indeed, ____________ 15The large number of council candidates on a given ballot made it impossible to ask city clerks about the race of council candidates on the questionnaire. 16In addition, there were six black/white contests, three Asian/white contests, one Latino/black contest, and one white/other race contest. 29 the bulk of the existing research on voter participation has focused on such topics as socioeconomic status, age, mobility, and race. We therefore supplement our survey data with other data on local demographic characteristics derived mainly from the 2000 Census.17 It is clear from existing research that individuals with higher socioeconomic status participate in politics at significantly higher rates than those with less education or income (Verba et al., 1995; Wolfinger and Rosenstone, 1980). Higher socioeconomic status implies a host of resources such as time and cognitive ability that are important for participation. Several studies have also shown a strong relationship between age, or stage in the lifecycle, and propensity to vote (Rosenstone and Hansen, 1993). In general, younger residents are often not yet politically engaged and tend to be inactive voters. Elderly voters may suffer from physical infirmities, but some studies indicate they are as likely to vote, controlling for other factors (Wolfinger and Rosenstone, 1980), and perhaps to be more involved in community civic affairs than younger generations (Putnam, 2000). Race and ethnicity have also been tied to voter participation. Specifically, Latinos and Asian Americans have participated at lower rates than either whites or African Americans (Hajnal and Baldassare, 2001; Uhlaner et al., 1989). For Latinos, most if not all of this difference is accounted for by socioeconomic status and citizenship factors, but the reasons for lower participation rates among Asian Americans remain something of a mystery (Lien, 1994). Another individual demographic characteristic associated with turnout is residential mobility. Because of the various time and information costs associated with registration and voting, people who have moved recently are much less likely to vote than those who are longtime residents of a community (Squire and Wolfinger, 1987). ____________ 17City population size, however, is measured with state Department of Finance estimates for 1998, rather than 2000 Census data. This is because the largest number of local elections in our dataset (47 percent of council elections) took place in 1998; another 27 percent in 1999, compared to 22 percent who reported on 2000 elections and 4 percent reporting on various pre-1998 elections. In any event, the Department of Finance estimate is very highly correlated with the Census count (r = 0.9998). 30 Along with individual characteristics, some aggregate city-level characteristics can also play a role in voter turnout. Although there is disagreement about the size and the direction of the effect, city population size has been shown to influence turnout. In most cases, a larger population has been associated with less social interaction and lower political participation. Oliver (2000) has argued that as city size grows, citizens are both less interested in participating in local politics and less apt to be mobilized by local campaigns. In our study of city-level voter turnout, we are not specifically interested in any of these demographic factors because they cannot be readily manipulated to increase voter participation. However, to ensure that our analysis of city-level institutions and electoral context is not biased by differences in demographic characteristics across cities, we include controls for each potentially relevant local demographic characteristic. In each of the statistical models investigating voter turnout reported in Chapter 3, we include a specially constructed summary variable representing the socioeconomic status (SES) of the city population,18 along with controls for the percentage of the population age 18 to 24, percentage age 65 or over; percentages of African Americans, Latinos, and Asians in the population; percentage who lived in the same house for five years (residential stability); percentage institutionalized; and the city population size.19 However, because we are examining aggregate turnout rates and aggregate city-level ____________ 18Measuring SES creates difficult statistical modeling issues, because the various measures of status we are interested in (income, educational attainment, home ownership, etc.) are very closely correlated at the city level in California. Thus, we decided to use factor analysis to reduce four important variables that are highly related to one another— median household income, poverty rate, percentage of college graduates among adult residents, and percentage owner-occupancy of housing—into a single summary measure. The resulting factor score is the city’s principal factor loading for these four variables, with higher scores representing higher status. The SES score ranges from a low of –2.0 (city of Huron) to a high of 4.6 (Rolling Hills), with a mean of 0. The SES score is highly correlated with each of its component measures: income (0.98), education (0.84), poverty (–0.79), and home ownership (0.68). 19Because of extremely high collinearity (with percentage Asian and percentage Latino), we could not simultaneously include a measure of percentage immigrant in the model. 31 characteristics, we will not be able to infer any causal relationship about individual behavior from these measures. Summary It is apparent from the questionnaire results that fewer than half of all registered voters bother to come to the polls in municipal elections in California. Measured as a percentage of the voting-age population, turnout looks even more bleak. Just under one-third of voting-age residents turned out in these contests. However, because turnout rates do vary across cities, it is important to determine the timing, institutional, and contextual factors that might shape local participation. We have highlighted several of those factors in this chapter, showing for example that a sizeable number of city elections are held on nonconcurrent or “off-peak” election dates, and that other Progressive Era reforms such as at-large elections are the norm among California cities. Electoral context also varies widely across the city elections we examine. In the next chapter, we analyze the relationships between voter turnout and the structural and contextual factors identified here. 32 3. Accounting for Differences in Voter Turnout Across Cities Voting gives Californians of all races, classes, and beliefs a voice. Through the vote, Californians communicate information about their interests and needs and make important decisions about who should hold office. Yet, as we showed in the last chapter, voter turnout in city elections is low. Well below half of registered voters—possibly a nonrepresentative sample of Californians—are deciding the outcome of political contests at the municipal level. These overall levels of voter turnout are important but they obscure as much as they reveal. Every city is different, and turnout rates vary dramatically across communities (Figure 3.1). 80 70 60 50 40 30 20 10 0 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70–79 80–89 90–100 Percentage of registered voters casting ballots Figure 3.1—Distribution of Voter Turnout Rates in City Council Elections 33 Number of cities Some cities attract many voters to the polls. At one extreme, 89 percent of registered voters in San Marcos turned out for that city’s November 1998 council election. Milpitas was just behind, with 88 percent voter turnout for an election that same day. In sharp contrast, other cities in the state are plagued by extremely low turnout. Lawndale saw only 10 percent of the city’s registered voters participating in its April 2000 election. Even in some major central cities where local politics is much more visible, the story is not much better. Only 20 percent of registered voters in Los Angeles turned out for the April 1999 city council election. What is it about cities like Lawndale and Los Angeles that leads to such low voter participation? Is there anything unique about cities like San Marcos and Milpitas that explains their ability to involve large majorities of the registered population in local elections? If so, can those explanations help increase citizen participation in other local elections? In this chapter, we explore some of the possible systemic factors that underlie this variation and help account for voter turnout rates. We first examine the issue of election timing, which we expect to be of central importance in accounting for variation in voter turnout. Next we examine local governmental institutional arrangements, such as term limits and the degree of city service provision. We then turn to a set of contextual factors, including the presence of propositions on the ballot and the degree of competition for office. Finally, we discuss a series of individual- and city-level demographic characteristics. Throughout the chapter, we analyze a dataset that combines both council and mayoral elections across the state. We combine the two types of elections because of the small number of mayoral contests for which we have sufficient data and because we believe that the dynamics of turnout are nearly identical for mayoral and council contests. Of the 134 respondents indicating that their cities elect mayors directly, we received usable mayoral voter turnout data from 130, but additional missing data (because of other items on the questionnaire being left blank by a few respondents) dropped the analysis down to 122 observations, limiting the statistical power available. When we examined mayoral and council elections separately, results were generally quite similar to those presented here. The same factors that influenced turnout 34 in council elections tended to affect turnout in mayoral elections. However, the small number of mayoral elections did lead to less stable and less robust estimations for these contests. Thus, the discussion focuses on the combined mayoral and council election dataset, for which we have 421 usable observations in the turnout model.1 In each regression, we include a binary variable to estimate differences between mayoral and council contests.2 The regression analysis technique used here allows us to focus on the relationship between various specific factors and voter turnout rates, while holding constant the effects of each of the other variables in the model. We ran two parallel estimations, one geared at explaining the turnout of registered voters and one geared at explaining the turnout of adult residents. The registered-voter model is able to account for 61 percent of the variation in turnout levels across city elections, whereas the model relating to adult-resident turnout accounts for 67 percent of variation. (For complete regression results, as well as summary statistics for all independent variables, see Appendix C.) Election Timing One notable factor in the comparison between San Marcos and Milpitas on the one hand and Lawndale and Los Angeles on the other is that the former pair of cities held elections on a statewide gubernatorial election day, whereas the latter two held nonconcurrent, or off-cycle elections in the spring. As it turns out, our statistical analysis indicates that election timing is, in fact, the most important factor influencing city ____________ 1We were able to calculate council election turnout rates for 350 cities, but some had to be dropped because certain questionnaire items on electoral context that are used in the regression (questions 17 and 18) pertained to the most recent mayoral election, for those cities with directly elected mayors. Fortunately, three-quarters of respondents reported on mayoral and council elections that took place on the same day, and thus their council elections could be retained in the regression analysis. Other cities had to be dropped because of nonresponse to other survey items or because the city was incorporated after 1990 and therefore lacked data for a few variables derived from the 1990 Census. 2When we ran a model that clustered, or grouped, the observations by city, signs and significance levels did not change. 35 turnout (see Figure 3.2).3 Compared to off-cycle elections, and controlling for a host of other factors, presidential elections are associated with almost 36 percent greater turnout among registered voters and 23 percent greater turnout among adult residents. Presidential primaries and gubernatorial elections are associated with an increased turnout of 21 to 26 percent among the registered and 11 to 15 percent among adults.4 When we examined mayoral and council elections separately, these relationships held true for both types of elections, although the association was somewhat stronger for council contests. Another important aspect of city election timing that could potentially affect turnout is whether city elections are consolidated with elections for other local offices. More than half of all the council and 70 60 Registered voters Adult residents 50 Median percentage turnout 40 30 20 10 0 Off-cycle Odd-year Gubernatorial Presidential Presidential November primary Figure 3.2—Variation in Turnout by Election Timing (with controls) ____________ 3Figure 3.2 and the other bar graphs in this chapter show the relationship between voter turnout rates and the variables of interest, controlling for all other variables in the regression. Specifically, continuous control variables are set at their medians (to represent a typical city), and categorical control variables are set at their modal value. 4Across the two different types of nonconcurrent elections, there is little difference in turnout. That is, with controls, odd-year November elections do not have significantly higher levels of turnout than other off-cycle elections, which are generally held in the spring. 36 mayoral elections we examined were held alongside elections for local officials such as city treasurer, school board members, or county supervisors. Does this have any effect on turnout? It appears that the answer is no. Furthermore, holding the mayoral and council election on the same day does not seem to spur significantly higher turnout. In short, consolidating local elections with statewide elections—as opposed to consolidating various categories of local elections—is the step most likely to yield local turnout gains.5 Taken as a whole, concurrency has a huge influence on whether registered voters will participate in municipal elections. Election timing alone accounts for about half of the variation among cities in turnout of registered voters; for turnout of adult residents, timing accounts for about one-third of the variation across cities.6 In fact, using some simple calculations involving a number of reasonable assumptions, we estimate that if all cities in the state that hold nonconcurrent elections switched their elections to a presidential election date, approximately 1.7 million more Californians would have voted in their most recent city council contest. The estimated turnout increase would be smaller, but still very substantial, if the spring and November off-year elections were moved to the same date as a gubernatorial or presidential primary election.7 ____________ 5Among these timing variables, we have also included a variable indicating whether the election in question was mayoral or council. Notably, mayoral elections tend to draw lower turnouts (with about 3 percent fewer voters casting ballots) than council elections, at least among registered voters. This finding may seem somewhat counterintuitive, since the mayoralty is generally a more visible and important office than individual council seats. However, it does accord with our finding in the last chapter that, overall, council elections in California have higher turnouts. It is conceivable that there is some systematic, unobserved difference between cities that have a separately elected mayor and those that do not—a factor that may be associated with turnout. Given the wide variety of variables that we control for in the regression model, however (notably including the socioeconomic status of the city and the degree of competition for office), it is difficult to imagine what this unaccounted-for factor might be. 6A regression explaining turnout rates and using only the timing dummy variables explains 51 percent of variation for turnout of registrants and 32 percent for turnout of adults. The coefficients of the timing variables are comparable to those in the full model. 7We estimate a turnout increase of about 1.2 million if all nonconcurrent elections were switched to gubernatorial election dates, and about 940,000 if switched to presidential primary dates. We had to make a number of assumptions in these estimations, the most important of which are: that the percentage of city residents who are registered voters is the same in cities that did not respond to our survey as in the cities 37 Nonconcurrent elections clearly are much less inclusive than concurrent elections, on average. Closer examination of the data, however, indicates that there is substantial variation within each electionday category, particularly for nonconcurrent elections and those held on presidential primary days. For example, despite the generally abysmal turnout rates for cities on off-cycle dates, two municipalities with nonconcurrent elections had turnouts of over 70 percent. It is therefore important to move beyond election dates to examine other factors that may influence voter participation rates. Institutional Factors Are the electoral laws and governmental structure of a city related to local voter turnout? According to our estimates, at least one institutional feature of the local government—the degree of city service provision—has an important connection to turnout.8 Many cities in California, particularly in the Los Angeles area, provide only minimal services with their own staff. Instead, they make arrangements with other governments or private firms to provide such functions. Cities that provide these services themselves may find that residents have more at stake in municipal politics. When services are contracted out or provided by other governments, city officials have less direct control over some of the issues that affect city residents’ quality of life the most. The results indicate that a city’s role in service provision influences turnout, if modestly. Each additional service provided by city ________________________________________________________ for which we have data; that the shares of registered voters living in cities having springtime and off-year November elections, respectively, are the same as those shares among the cities in our dataset; and that the factors influencing turnout are the same in cities that are not in our dataset. We estimated that there are just over 12 million registered voters living in municipalities in California, of whom about one-fifth live in cities holding off-cycle spring elections and one-fifth in cities holding off-year November elections. We then calculated how the number of voters casting ballots for city council in these nonconcurrent-election cities would be likely to change based on election timing, using the estimated turnout rates for each type of city shown in Figure 3.2. 8We did not use the institutional variables relating to form of government (mayorcouncil versus council-manager plan), mayoral veto power, or mayoral budgeting authority, because of the lack of variation among California cities on these characteristics. Also, all local California elections are nonpartisan, eliminating another potentially important institutional variable. The district election variable used in this analysis pertains only to council elections, since all directly elected mayors are chosen at-large. 38 staff, of the five services asked about in the survey (fire, police, library, sewerage, and garbage), is associated with approximately 1 percent higher turnout among registered voters (Figure 3.3). This relationship is weaker and statistically insignificant for turnout of adult residents. However, when we substituted a binary variable indicating whether the city provides any of the five services, it was positive and statistically significant for turnout among adult residents as well as registered voters. This finding indicates that there is a particular “turnout penalty” for minimal-service cities—those whose in-house employees provide none of the five services asked about. It is possible that voters find municipal elections in such cities to be less important to them, thus accounting for reduced turnout levels. It is also important to note that municipal employees are often among the most likely residents to participate in local politics, and this pattern may also help account for the lower turnout among cities with fewer in-house staff. A somewhat anomalous relationship is apparent between district council elections and voter turnout, but only in the estimation of turnout among adult residents. We suspected that district elections might draw slightly higher turnout, because district council members might be more 70 Registered voters 60 Adult residents 50 Median percentage turnout 40 30 20 10 0 None Three Five Number of services provided by city staff Figure 3.3—Variation in Turnout by City Service Provision (with controls) 39 closely involved in neighborhood-level issues that are more salient to residents, and voters might thus feel a stronger bond with their representative. However, the estimation indicates a significant negative relationship between district council elections and turnout of adult residents. It appears, however, that this outcome is something of a statistical artifact. The relatively small number of cities in California with district elections have substantially lower rates of voter registration than cities with at-large council elections.9 If we control for the registration rate of the city, district elections are not significantly related to the turnout of adult residents. Thus, we see no strong evidence of a causal connection between the method of council election and voter turnout per se. Two other institutional features of local government had no significant association with voter turnout. Whether a city is chartered or is covered under the state’s general-law provisions for municipal government appears to have no relationship with the share of voters going to the polls. Perhaps surprisingly, we also found no connection between term limits and turnout.10 We hypothesized that term limits would make elections more interesting to residents and therefore increase turnout, but it is also possible that the effect of term limits depends on whether or not an incumbent runs. When an incumbent can no longer run for a seat because of term limits, the election may become more competitive and attract additional interest. We turn to this and other electoral context variables in the next section. Electoral Context Whether voters participate in local elections may be related to the specific context of the elections. How many candidates are running? Is ____________ 9For the 17 cases we observe of a district council election, the estimated registration rate of adults in the city averages 54 percent, compared to 71 percent in the overall sample. This is probably because several of the district cities have very high shares of immigrants in the population, many of whom are ineligible to vote. Examples include Santa Ana, Watsonville, Salinas, and Downey. 10The term-limits variable is specific to the contest examined; that is, the variable measures the influence of council term limits for council-election turnout, and of mayoral term limits for mayoral-election turnout. 40 there an incumbent or is the contest for an open seat? Are there ballot measures to decide? Although it is difficult to influence most of these factors through policy efforts, it is nevertheless important to know what role they play in the local political arena. For example, one of the most controversial elements of California’s electoral system is direct democracy. Despite some criticism, policymaking in California has relied increasingly on initiatives, at least at the statewide level. In some cases, these initiatives have spurred voter interest and participation (Pantoja and Segura, 2000). At the municipal level, as noted in Chapter 2, ballot measures are actually far less prevalent. Moreover, most city propositions reach the ballot through council action, not as the result of citizen initiatives. Although not widespread, local ballot propositions have a significant positive relationship with city turnout rates (Figure 3.4). Cities with at least one citizen initiative or council-placed measure tended to have about a 4 percent higher turnout than those without either. This is true for turnout of both registered voters and adults.11 Another contextual issue to consider is competition for office. The number of candidates competing for a seat is one rough indication of the level of competition and perhaps the level of media and other interest in a given contest. At one extreme, uncontested elections (those with the same number of candidates as seats to be filled) may draw especially low interest. We measure competition with two variables: a binary variable signifying uncontested elections and the number of candidates per seat. For mayoral elections, the latter variable is simply the number of candidates on the ballot; because we do not have information on each specific council contest, we use the number of total candidates running for any council seat divided by the number of seats up for election. ____________ 11Oddly, if entered into the equation separately, only council-placed ballot questions are significantly related to election turnout. The presence of citizen initiatives has a positive relationship to turnout but is statistically insignificant. Despite the extensive attention given to local citizen initiatives, it appears that some measures placed on the ballot by the governing body may be more provocative to voters. 41 70 Median percentage turnout 60 None One or more 50 40 30 20 10 0 Registered voters Adult residents Figure 3.4—Variation in Turnout by Number of City Ballot Propositions (with controls) The results clearly show that uncontested elections are bad news for voter participation in city elections. Uncontested elections draw to the polls about 4 percent fewer adult residents, and about 5 percent fewer registered voters. (Further analysis indicated that this effect was strongest in mayoral elections.) Moreover, for contested elections, the number of candidates appears to be associated with a slightly higher turnout rate. One additional candidate per seat is associated with about 0.7 to 0.9 percent higher turnout. It is possible that each aspirant mobilizes his or her core supporters to come to the polls; therefore, a larger number of combatants would be associated with more voter participation overall. Regardless of the explanation, these results hint that one dimension of local civic vibrancy—competition for office—is connected with another dimension—voter participation. Incumbents at all levels of the political process are extremely likely to win reelection, and one might expect turnout to decrease when incumbents are running. We measure incumbency as the number of incumbents running divided by the number of seats. This variable ranges from zero (no incumbents running) to one (all seats have incumbents); in mayoral elections, of course, the resulting value can only 42 be zero or one since there is only one mayoral seat. Surprisingly, incumbency proved not to be related to turnout at all.12 Also of interest were two other factors that could not be included in the analysis because of high rates of nonresponse: the margin of victory in the election and the issue of whether the top two candidates for office were of different racial or ethnic backgrounds (as discussed in Chapter 2). Although neither variable is included in the final regression models presented here, no significant associations were found between these variables and turnout rates. Unfortunately, these data could feasibly be collected only for mayoral elections, and nonresponse to these questions substantially reduced the number of potential observations. Thus this “nonfinding” cannot be considered definitive. One additional electoral context variable was included in our model of turnout among registered voters as a statistical control: the (estimated) percentage of voting-age residents in the city who are registered to vote.13 This share of potential voters who are registered could be important because it provides some indication of the efforts made to involve residents in electoral politics and helps define the pool of citizens among whom voting will or will not occur. We do not include this variable in the regression for turnout of voting-age residents; in that case, the act of registration is itself a large part of the phenomenon we seek to explain. In cities where a very high share of residents are registered, it is possible that some “low-propensity” voters have been registered, perhaps through especially active outreach efforts by community groups or candidates. Another possibility is simply that the local voter rolls have not been “purged” recently, meaning that there may be a fair number of citizens listed as registered who have died or moved out of the community. In either situation, one might expect that the turnout of registered voters will be lower where the registration rate is higher. This ____________ 12We tried other specifications for the council incumbency variable, such as binary variables for situations where any incumbents were running or where all seats had incumbents, but this did not influence the results. Even when we looked at mayoral elections separately, incumbency had no significant effect on turnout. 13As noted in previous chapters, the variable used is an estimate based on an interpolation (for the particular year of the election) of the voting-age population of the city between 1990 and 2000. 43 is indeed the case, although the relationship is a weak one. Where registration rates are 10 percent higher (all other things equal), turnout is about 1.4 percent lower.14 City Demographics The final set of factors we assess are demographic. Other studies have shown such characteristics to be very important for voter turnout when measured at the individual level. Here we examine their role in California’s municipal elections, viewing them primarily as statistical controls. The model includes several variables of interest to turnout, as described in Chapter 2: city population size, a summary measure of the socioeconomic status of the city, measures of the racial and ethnic makeup of the city, two measures of the age distribution of the local population, residential stability, and (for turnout of adult residents) the proportion of the city population that is institutionalized. One clear-cut finding is that larger cities tended to have lower turnout rates. This generalization is supported by research at the individual level, which finds that residents of more populous jurisdictions generally participate less (Oliver, 2000). Furthermore, the relationship between larger city populations and lower voter turnouts is apparent even when controlling for other factors that may differ between large and small communities. Nevertheless, the magnitude of the relationship is fairly modest (Figure 3.5). Cities with residents of higher socioeconomic status tended to have significantly higher turnout; this result is also consistent with individuallevel research. This pattern is particularly clear for the turnout rates of adult residents, but it is also readily apparent even when the focus is narrowed to those residents who have already been registered to vote. Race and ethnicity appear to be related to turnout in more complex ways. In accordance with past research suggesting that African ____________ 14We also investigated the effects of party registration—the percentage of city residents registered as Democrats, Republicans, or neither—on turnout. This element of electoral context proved to have no influence on turnout. 44 Registered voters 70 Adult residents 60 Median percentage turnout 50 40 30 20 10 0 20,000 50,000 100,000 Population 1,000,000 Figure 3.5—Variation in Turnout by City Population (with controls) Americans participate at rates roughly equivalent to whites, we find no relationship between the proportion of blacks in the city and turnout (Hajnal and Baldassare, 2001). However, the presence of large Asian American or Hispanic populations is associated with lower turnout. Even after controlling for socioeconomic status, the Asian American and Latino share of the population are both tied to lower turnout among adult residents. These results are consistent with the view that citizenship and language barriers make it especially hard for these two immigrant groups to register at rates equivalent to those of whites and African Americans. At the same time, the results indicate that a greater Hispanic share of the population is not associated with lower turnout of registered voters—a finding that seems to support existing research showing that once socioeconomic status and citizenship are accounted for, Latinos participate as much as whites and African Americans (Hajnal and Baldassare, 2001). By contrast, a high proportion of Asian Americans in a city is tied to lower turnout of registered voters. Once again this finding is consistent with individual-level research, which finds that Asian Americans tend to vote significantly less regularly than 45 members of other racial and ethnic groups when controls for citizenship and socioeconomic status are introduced (Lien, 1994). Turnout is also clearly correlated with both the age distribution of the population and household mobility. Cities with higher shares of senior citizens have higher turnout rates. The proportion of young adults in the population, however, appears unrelated to city turnout rates. Household mobility—that is, the proportion of residents who have changed addresses in the past five years—is connected to turnout, with cities of more household stability showing higher turnout among the adult population. Among registered voters, however, mobility levels of the city population are unrelated to turnout. Registration itself seems to be the main hurdle for people who move frequently. Finally, in estimating the turnout of all adults, we draw upon the 2000 Census to measure the proportion of the city population that resides in institutional settings. Our intuition here was that cities with very high institutionalized proportions of the population tend to be the sites of prisons, the inmates of which generally are ineligible to vote. The findings confirm that the institutionalization rate is negatively associated with turnout.15 Summary Although turnout in local elections in California is generally low, there is, in fact, wide variation across cities. Some localities manage to attract the vast majority of registered voters to the polls; others are plagued by extremely low turnout. Our statistical model was designed to ascertain which factors play a significant role in accounting for the substantial variation in voter turnout across California cities. The results show that the largest part of this variation can be explained by timing— the date on which the election is held. Concurrent elections—that is, city elections held on the same day as a presidential or gubernatorial election or a presidential primary—had turnouts that are substantially higher than off-cycle, or nonconcurrent, city elections. The turnout boost for cities holding concurrent elections was 11 to 23 percent among ____________ 15Such individuals would not generally be registered to vote locally, making the issue moot for the turnout of registered voters. 46 all adults, and 21 to 36 percent among registered voters (depending on whether the concurrent election was presidential, gubernatorial, or a statewide primary). Beyond timing, there are a handful of other interesting correlates of turnout. Cities that provide only a minimal number of services with their own staff drew a smaller share of voters than those that provide more services with in-house personnel. Direct democracy plays a role, too; cities with local ballot questions placed before the voters tended to have higher turnout. City elections with a greater number of competitors had slightly higher turnout. All of these factors, however, pale in comparison to the dramatic role played by election timing. Although competition for office and incumbency did not appear to play a decisive role in influencing local turnout, these factors are worthy of additional study in and of themselves. Potential candidates for office are likely to know that turnout is much higher in concurrent elections, and this knowledge may affect their decisions about running for office. And do incumbents fare better in concurrent or nonconcurrent elections? If policymakers consider rescheduling local elections to enhance turnout, they may wish to bear in mind these other potential dynamics of electoral competition. The next chapter addresses these issues. 47 4. Competition and Incumbency in Local Electoral Contests Thus far, our discussion of municipal elections in California has mainly been concerned with voter turnout. However, there are other interesting measures of the vibrancy of local electoral contests. This chapter addresses two such issues: the level of competition for mayoral and council races and the success of incumbents who are seeking reelection. Who runs for city office—and who wins—are different concerns from the level of voter participation, but as we have noted, the issues are linked, since more competition tends to increase turnout. Moreover, as we shall indicate, the question of election scheduling— concurrent as opposed to nonconcurrent elections—may have implications for local candidacies. Competition for Office Local government is the starting point for many political careers— that is, it is the level at which people tend to be recruited into elective politics. As such, the degree of competition for local offices may have broader implications for the quality of democracy. Competition for mayoral and council seats provides a barometer of how much interest and perceived opportunity local politics generates among one type of citizen—those who might consider serving in elective office. Our measure of competition for office is simple: for mayor, the number of candidates whose names appeared on the ballot, and for council, the total number of candidates divided by the number of seats available.1 To estimate a more robust model, we once again merged the ____________ 1Respondents were asked to report on the most recent regular election, not a runoff election or a special election to fill a vacancy. 49 council and mayoral elections, distinguishing between the two where appropriate. The average mayoral contest in California for which we have data had 2.9 candidates running, although just over one-sixth of these elections were uncontested. Among city council elections, each seat attracted an average of 2.5 candidates; less than 4 percent of cities reported that all their council races were uncontested. However, there is wide variation in the number of candidates, with some cities having as many as eight candidates per council seat. It is impossible to capture in a statistical model all the nuances of the local political environment that might motivate individuals to seek city office. Nevertheless, many of the factors we explored above might also influence the vibrancy of electoral competition in cities. Thus, we once again investigate issues of election timing, the institutional arrangements of local government, the electoral context, and the demographic characteristics of the municipality. However, we alter some of the specific variables used in an effort to select measures that are more meaningful in influencing whether city residents will seek to run for office. In estimating the relationship of election timing to competition, we use a simple binary indicator of whether the city election was concurrent with a statewide election day (i.e., a presidential, presidential primary, or gubernatorial election) as opposed to cases where the city held a nonconcurrent election. We suspect that potential candidates will not make fine-grained distinctions about their chances in a gubernatorial versus a presidential election, for example, but they may evaluate their candidacy based on whether the city election will be in electoral “prime time”—and thereby draw a much larger voter turnout, as we have seen. We also use variables indicating whether the mayoral and council elections were held on the same day, and whether other local elections, such as for school board, were held that day. In the case of some of the institutional factors, there may be less of a clear link to competition than to turnout. We do expect term limits to affect competition, in two possible ways. On one hand, term limits may increase competition by providing a clear indication to potential candidates that a seat will be available once an incumbent has been 50 “termed out.” Conversely, the presence of term limits conceivably might reduce the attractiveness of public office and make officeholding a less desirable goal, thereby reducing electoral competition. Along these lines, we very strongly expect the presence of an incumbent to reduce competition, because challengers are well aware of the electoral benefits known to accrue to incumbents, such as name recognition and fundraising potential. The degree of city service provision could affect the salience of city politics for potential candidates, just as it appears to for voters. And two additional features of the mayoralty—whether the position of mayor is considered a full- or part-time job, and the length of the mayor’s term of office—may shape its attractiveness as a goal for potential candidates.2 Charter city status, which provides some nominally increased discretion to city officials, might also potentially affect the number of candidates. Some of the electoral context factors have a less direct, but nevertheless potentially important, link to electoral competition. We include our estimate of the percentage of voting-age residents registered to vote, as this may be an indication of the degree of political engagement in the city, which could also affect competition for office. Simultaneous elections for other local offices may provide alternative outlets for individuals interested in entering local politics, therefore perhaps drawing fewer council or mayoral candidates. The presence of local ballot measures may indicate a degree of community conflict that could motivate more competitors to seek office. Because we are interested in the general level of political interest or rancor within a given community, we turn to broader measures of political engagement. Instead of election-specific indicators of ballot measures, we use a general measure of the number of voter initiatives across a three-year period ____________ 2Regarding the first issue, city clerks were asked, “Is the mayoralty in your city considered a full-time or part-time position? (Generally, the office is considered full-time only if it carries a full-time salary.)” The office was identified as full-time in only 13 cities. In the regression analysis below, the variable used is an interaction variable denoting only mayoral elections in cities with full-time mayors. It is worth bearing in mind that only 2.5 percent of the observations in the regression are of mayoral elections for full-time mayors. Regarding the second issue, about half (49 percent) of directly elected mayors were said to hold a four-year term, whereas the other half held two-year terms of office. 51 (1997–1999) and a subjective response by city clerks indicating the general level of electoral controversy over the past five years. We also include a measure of the percentage of registered voters in the city who are third-party adherents or who decline to state their party affiliations. Even though California city elections are officially nonpartisan (that is, candidates are not identified on the ballot by party affiliation), cities with a more independent electorate may be more critical of the political status quo and have a more nonconformist political style. Moreover, a high percentage of independents may indicate that local party organizations—which often are important in recruiting and screening potential candidates—are weak. Under such circumstances, local parties may be less capable of signaling who the leading candidates are through endorsements or financial contributions. In such a wide-open political arena, more candidates may seek office, believing they have a credible chance at swaying voters and winning. Finally, we explore the relationship between selected local demographic factors and candidate competition. Here our analysis takes us into fairly uncharted waters. Although a city’s population size will almost surely be related to the number of candidates who choose to run, it is not clear from extant theory that the racial or ethnic composition and socioeconomic status of the residents of a city should influence electoral competition—but there is no way to find out for sure without testing. Preliminary analyses confirmed that there were no noticeable effects from socioeconomic status variables, which are therefore omitted from the model; race variables were retained. We also include the measure of residential stability in the city—the proportion of residents who lived in the same dwelling unit from 1985 to 1990—on the thought that more transient communities may present different patterns of candidate recruitment and mobilization from more settled communities. Variables relating to the proportion of senior citizens and young adults in the community were also employed. Retired residents might have more time and interest in local politics, thus increasing the pool of potential candidates. If young adults are especially disengaged politically, cities with many of them may have fewer electoral candidates. We used a regression analysis to estimate which factors are related to competition for office. Detailed results of the model are shown in 52 Appendix D.3 The first finding of note is that mayoral elections draw significantly fewer competitors than council seats, all other things being equal. (Without controls, as we have noted, mayoral races draw slightly more candidates than do council races.) This may reflect the higher cost in time and campaign funds often required for serious mayoral bids. Novice candidates are probably more likely to give politics a try at the council level. The second finding to stress is the lack of a relationship between election timing and competition for office. That is, the number of candidates is not influenced by whether the election is concurrent with a statewide election. Another type of election concurrency does make a difference, however. The presence of other local offices up for election (such as city clerk, school board, or county supervisor) is associated with a significant reduction in the number of aspirants for council seats or mayor. In such circumstances, some potential candidates are possibly being siphoned off into other local contests. An election that deals only with the council and mayor, on the other hand, offers ambitious local campaigners the opportunity to run for those offices and then try again later for a different local position if they lose the council or mayoral race. Certain institutional factors also influence electoral competition in California cities. Neither district elections nor the number of services provided by city government staff produces discernible differences in competition.4 However, a full-time mayoralty is associated with more competition for that office, as is a longer term of office. In short, where the rewards of mayoral office-holding are greater (or where mayoral elections are less frequent), the number of candidates for mayor is greater. In addition, term limits—perhaps by making the office seem less desirable—seem to reduce the level of competition (although the statistical significance is not high). It is likely that term limits also indirectly affect competition by restricting the number of races featuring incumbents—an electoral ____________ 3The dependent variable in the regression (candidates per seat) is expressed in logarithmic form, because of the extreme skewness of the distribution (i.e., many small values and only a few quite large values). 4The district-elections variable applies only to council elections. 53 context variable. Indeed, the presence of an incumbent running for office substantially reduces the number of candidates for mayor and council elections. This finding is consistent with research on congressional elections, which shows that the presence of an incumbent discourages potential opponents from running; candidates for open seats (those with no incumbent) tend to be of “higher quality” (i.e., have more experience in government and politics) than challengers who run against incumbents (Jacobson, 1983, pp. 37–41). Certain other electoral context variables are also related to the level of competition. As one might expect, controversy and a more engaged public appear to be related to greater levels of competition for office. First, general electoral controversy in a city (as judged by the city clerk) has a significant positive association with the number of candidates per seat.5 Second, cities that have had more citizen initiative elections in recent years produce more candidates, perhaps indicating an electorally charged atmosphere with salient issues facing voters and politicians. The percentage of adults registered to vote, another potential indicator of an interested and engaged public, has a positive relationship to competition, but does not attain standard levels of statistical significance. The results also indicate that cities with larger proportions of independent or third-party voters have significantly more competition. (Further analysis showed that this pattern is apparent both for council seats and for the mayoralty.) This finding may reflect the fact that local Democratic and Republican party organizations help winnow the field of candidates in many cities—even in California’s officially nonpartisan electoral system—through their endorsements and campaign support. Major-party organizations generally wish to avoid having their loyalists split their votes among multiple competitors. Where the major parties are weaker (as indicated by a higher proportion declining to state and third-party voters), they may be less effective as “gatekeepers” to office. Potential aspirants to local office also may be encouraged by the presence of a larger share of “free agent” independents among the voting public. ____________ 5It is perhaps unclear whether greater controversy spurs more candidates into action or is itself a function of the clashing views of a large number of candidates. Nevertheless, the questionnaire asked respondents to judge the level of controversy over the past five years, not the controversy level of the specific election for which we have data. 54 Turning to demographic factors, the city characteristic that has the greatest association with competition is the size of its population. Large cities tend to have more candidates, as one would expect.6 Residential stability is associated with fewer council aspirants, perhaps indicating a more consensual, less competitive politics in cities that have a stable, less transient population. Municipalities with higher proportions of blacks and Hispanics in the population have slightly more competition for office. Although we are only speculating, political elites within these two racial/ethnic groups may see a particularly important link between their group interests and the actions of local governments; this link, in turn, may lead more citizens to run for office. There is no such relationship between the size of the Asian population in a city and competition for office. Finally, although the shares of young adults and elderly persons in the population have the expected relationship with electoral competition, neither factor proves statistically significant. The Success of Incumbents Incumbency is a key factor in shaping the degree of competition for office. Indeed, one of the best-documented regularities in electoral politics is the advantage of incumbency. At the federal level, for example, most congressional elections since 1952 have returned over 90 percent of incumbent members of the House of Representatives and about 80 percent of incumbent senators (Squire et al., 1995, pp. 280– 281). At the mayoral level, available evidence suggests that nationwide, 85 percent of incumbents win their reelection bids (Wolman et al., 1990). Whether the power of incumbency extends to the city council level is less well known (but see Krebs, 1998). Also not well known are the factors that lead to incumbent success at the local level. ____________ 6However, the number of candidates does not increase proportionally with the population. That is, there are fewer candidates per thousand city residents in large cities than in smaller communities. As with the lower turnout among large cities noted in Chapter 3, this finding may indicate somewhat lower levels of civic vibrancy in jurisdictions with large populations. (Volunteering and social trust are also behaviors that are less common in large communities, according to Putnam, 2000.) However, lower levels of competition for office in large cities may also be a reflection of the higher financial costs involved in campaigning there, as paid media advertising and direct mail often supplant yard signs and door-to-door campaigning. 55 Our California data indicate that 80 percent of council members and 86 percent of mayors who stood for reelection won their contests. This result underscores the tremendous value of incumbency to political candidates at all levels of politics. Incumbency increases name recognition among voters, deters potential challengers, and very likely enhances perceptions of a candidate’s experience and accomplishment. Nevertheless, some incumbents do lose, and one might ask under what circumstances this outcome is likely. Is incumbent success related to voter turnout? Perhaps more important, is incumbent success linked to the timing of elections? Some local election officials and city residents have expressed concern that concurrent elections might favor incumbents (Simerman, 1998). These critics believe that holding municipal elections on the same day as a statewide contest could create a situation in which some voters who come to the polls are unaware of and inattentive to local issues. Such voters, it is argued, have a tendency to vote for incumbents, who are better known, on average, than their challengers. In contrast, by attracting a higher proportion of purportedly better-informed voters, nonconcurrent elections might make incumbents less secure and ultimately make council members and mayors more responsive to the residents of their city. We cannot directly test the responsiveness of mayors and council members with the available data, but we can analyze the success of incumbents standing for reelection in California. Only contested elections are examined because unopposed incumbents win automatically. We analyze the aggregate number of incumbents who win reelection in various circumstances, rather than engaging in our customary statistical analysis that treats each city as an equally important unit of analysis. Because many cities have multiple incumbents running for reelection and we lack data on the specific attributes of those contests, our remarks are necessarily limited to the city-level factors related to incumbent success.7 Again, we examine the four sets of factors—election ____________ 7The text and charts in this section present the results of simple cross-tabulations of incumbent success and the city characteristics of interest. These relationships were confirmed with a logistic regression analysis of the grouped data. The results of that regression conformed closely to the results presented here. 56 timing, institutional structure, contextual factors, and the demography of the city—for their relationship with incumbent success rates. In line with the views of some proponents of off-cycle elections, concurrent elections are associated with greater success of incumbents in California municipal elections (Figure 4.1). City council incumbents running during gubernatorial, presidential primary, or presidential election dates have about 4 percent greater success rates than incumbents in nonconcurrent contests. Mayoral incumbents win reelection 12 percent more frequently, although this result is not statistically significant, given the limited number of contested mayoral races featuring incumbents.8 We can only speculate whether the greater success of incumbents in concurrent elections indicates that voters “tune out” local issues during statewide elections, since we lack individual-level data on voters. Possibly, the voters who are motivated enough to go to 100 Concurrent elections 90 Nonconcurrent elections Percentage reelected 80 70 60 50 Council members Mayors Figure 4.1—Incumbent Success and Election Concurrency ____________ 8The council-election differences are statistically significant at the p < 0.10 level. For contested mayoral races, the number of observations is only 78, which renders the difference in incumbent success (marginally) statistically insignificant, despite its larger magnitude. 57 the polls for off-peak, local-only elections tend to be more attentive to community issues or more critical of incumbent council members.9 Local political institutions appear to have little bearing on incumbent success. Term limits and the degree of city service provision had no relationship with incumbent reelection rates. Cities that use district elections were somewhat more likely to have council members win reelection, although the result did not hold up using more rigorous statistical analyses. Charter cities have higher incumbent reelection rates than general-law cities, but any explanation for this relationship is not readily apparent. The electoral context of a council election, however, appears to play a significant role in incumbent reelection success, and here the findings are very plausible (Figure 4.2). For instance, the existence of citizen initiatives has a substantial negative relationship with incumbent success. 100 No citizen initiatives 90 Citizen initiatives on ballot Percentage reelected 80 70 60 50 1 to 2 3 4 or more Number of council seats being filled Figure 4.2—Success of Council Incumbents in Varying Electoral Contexts ____________ 9We do not include the turnout rate itself as an independent variable regarding incumbent success because the causation could be interpreted as running in the opposite direction—that is, the turnout rate might be affected by the presence of a weak or unpopular incumbent or the expectation of a close race. Moreover, election timing (which is included in the model) is correlated with voter turnout rates at a high level (r = 0.71). 58 Having one or more initiatives on the city ballot is associated with a seven-point reduction in the percentage of incumbents reelected, on average. Elections that include voter initiatives may be marked by more conflict or may draw more critics of local government to the polls. Some initiatives may even be directed against the policies of specific incumbent officeholders. This effect is somewhat contingent on the number of seats being selected, however. When one to three council seats were up for election, the presence of citizen initiatives was associated with a sizeable (13 to 15 point) reduction in the percentage of incumbents winning reelection. By contrast, incumbents running in elections in which numerous council seats were chosen were very likely to win reelection regardless of initiatives on the ballot; better than 90 percent were victorious. It is possible that the high information-gathering demands on voters who have to select multiple council members at the same time render them more likely to take the informational shortcut of relying on incumbency and name recognition. Among demographic variables, city population size is associated with greater success rates for incumbents (Figure 4.3). The name recognition and the fund-raising advantages of incumbency may count for more in 100 Council Mayor 90 Percentage reelected 80 70 60 50 Less than 20,000 20,000–49,000 50,000–99,000 Population category 100,000 or more Figure 4.3—Incumbent Success and City Population 59 large jurisdictions, where constituents are less likely to be intimately familiar with City Hall. Residential stability among the city population also has a positive relationship with the success rate of incumbents. It is possible that more “settled” communities with less population turnover experience more consensus and a quieter, status quo political style. Summary As seen in Chapter 3, the timing of city elections affects turnout. But is timing also related to the competitive dynamics of local contests? Although the level of competition for office is not associated with election timing, incumbents do seem to find some slight additional degree of protection in concurrent elections. This finding brings us full circle to the question of the merits and disadvantages of holding municipal elections simultaneously with elections that attract more attention. The incumbency advantage is a somewhat greater hurdle for challengers to overcome in concurrent elections—whether because local elections during the statewide cycle draw less careful scrutiny and information gathering, or because the much larger number of voters casting ballots in such elections are more inclined to vote for familiar faces. We have raised one possible explanation—that the voters who come to the polls in nonconcurrent elections tend to be more motivated or discontented about local issues, and this spells more trouble for incumbents. Whether this finding provides definitive support for the critics of concurrent scheduling of elections is an issue we will defer to our concluding chapter. Several other interesting patterns emerge from our analysis of competition for office and incumbent success. For example, cities undergoing substantial community controversy (indicated by a large number of voter initiatives) are likely to see more candidates enter the ring and are more likely to send incumbents packing. High-population communities generate more candidates per seat but are nevertheless more likely to reelect incumbents. Towns with a greater proportion of voters who do not identify with one of the two major parties experience more competition for office. Longer mayoral terms and full-time mayoralties also draw more candidates for the job. This finding indicates that 60 institutional choices (such as the powers of the mayor) that were made to address the operations or efficiency of a municipality may have other consequences relating to who runs for office and whether they succeed. 61 5. Conclusion Most Californians do not vote in municipal elections. Even compared to statewide and national contests, participation in local contests is disappointing. Because only a fraction of the adult population is participating in the local electoral arena, relatively few voters are determining the outcome of local elections and presumably affecting policy decisions while the majority of Californians stand on the sidelines. Given that most of the elected officials in the state are elected locally, and given that most of the policies that affect our daily lives are implemented at the local level, this lack of participation is clearly a problem. If the characteristics of those who do vote are different from the makeup of the larger population that does not vote, outcomes may be both biased and inefficient, as not everyone’s preferences will be taken into account. Limited voter participation does not necessarily mean that local democracy is failing but it is surely enough of a red flag to warrant careful examination. Although it is clear that most adult Californians do not vote in the average local contest, it is also clear that there really is no “average” contest. Across cities there is tremendous variation in voter turnout. There are cities where almost all of the registered voters participated in the last city election, there are cities where only 10 percent of all registered voters showed up at the polls, and there is virtually everything in between. Analysis of this variation across cities indicates that one factor has the potential to dramatically alter voter turnout rates in local contests. By far the largest part of this variation can be explained by election timing. City elections held on the same day as statewide contests (that is, on the same day as a presidential or gubernatorial election or a presidential primary) tend to have voter turnout rates almost double those of nonconcurrent city elections. Even if we control for a host of other factors associated with turnout, holding an on-cycle election still 63 increases the turnout of registered voters by an average of 26 to 36 percent above turnout for off-cycle elections. Although timing is clearly the most important factor affecting turnout, this report reveals other important aspects of the local electoral arena that appear to influence voter participation. Whether a city provides its own public services is related to voter turnout. Cities that either do not offer services or contract or arrange with other entities to carry out the bulk of municipal services draw a smaller share of voters than those that provide more services with their own staff. This finding suggests that when the stakes of contest are higher, more voters will participate. Direct democracy also plays a role, as cities with local ballot questions placed before the voters tend to have significantly higher turnout. Although there is much debate about the overall value of direct democracy, placing propositions on the ballot seems to spark greater interest and participation in the municipal political arena. We should also note the factors that do not affect voter participation in local contests. Our analysis suggests that consolidating mayoral and council elections with other local contests (such as school board elections) would have no significant effect on turnout. Similarly, charter cities are no more prone to higher turnout than are general-law cities. Term limits have no direct effect on voter participation. Contrary to expectations, district elections do not generally increase turnout over atlarge elections and may, in fact, marginally decrease turnout (although the unusual characteristics of California’s handful of districted cities complicates this interpretation). Finally, the degree of competition for office has only a limited association with voter turnout rates. What does all of this mean for policymakers? For voter turnout, the implications are clear: If there is a desire to increase voter involvement in municipal electoral contests, the best tool for the job is scheduling concurrent elections. Moving a city’s election from a nonconcurrent date to the date of a presidential election could well mean a doubling of voter turnout. Overall, we estimate that by moving all nonconcurrent elections to coincide with presidential elections, California could have drawn 1.7 million more voters to city council elections statewide (roughly a 31 percent increase in participation in municipal contests). By contrast, altering institutional factors such as city service provision 64 and increasing our reliance on direct democracy at the local level might reasonably be expected to increase turnout rates by about 5 to 10 percent. Unlike any other factor we examine, election timing could dramatically alter the size of the local voter population. Concurrent elections are even more appealing as a policy lever because they would probably produce some modest financial savings for cities. Municipalities often pay the entire cost of stand-alone elections and only a fraction of the cost of concurrent elections.1 The other advantage of concurrent elections is that they are fairly easy to implement. In many cases, city councils need only pass an ordinance. No new structure or agencies would be required. Already the cost savings associated with on-cycle elections and the ease with which election dates can be changed have led a considerable number of municipalities across the state to align their local elections with statewide contests. Encouraging more cities to move in the same direction would presumably not be very difficult—although it is possible that local elected officials would oppose a move to on-cycle elections because it would make it more difficult to retain their local offices while running for positions in state government. As California policymakers reconsider the timing of the statewide primary, in connection with presidential or gubernatorial selection, they might consider whether local governments that still hold stand-alone springtime elections should be encouraged to consolidate these elections with the primary. However, a move to concurrent elections would not fully address the broader dilemma of limited citizen participation in municipal affairs. First, on-cycle local elections would surely lead to greater voter turnout but would not necessarily ensure greater civic engagement. Although greater turnout tends to be coupled with increased political knowledge, trust, and efficacy, it is not yet clear how aligning local elections with statewide elections would affect each of these other measures. Second, ____________ 1Evaluated on the basis of a city’s administrative costs per vote cast, the fiscal advantages of concurrent elections are even clearer. For example, San Francisco election officials recently estimated that the December 2001 runoff for city attorney cost the city $29 per ballot cast. Turnout was only 15.4 percent. This has led city officials to seriously debate the merits of an “instant runoff” voting system so that the November election would be decisive and no runoff would be required (Lelchuk, 2001). 65 although concurrent elections should have a dramatic effect on turnout, in the end local turnout is unlikely to exceed the fairly disappointing levels currently experienced in statewide and national contests. Voter participation is a nationwide problem, and election timing changes for municipal elections would be only an incremental improvement. Ultimately, broader reforms such as extended voting hours, 100 percent mail-in elections, or Internet voting might be more fundamental. Third, a move to concurrent elections raises, for some observers, several concerns about voter attentiveness and knowledge, which may in turn affect the competitiveness and outcomes of city elections. The coupling of local elections with national or statewide contests would lead to longer, more complex ballots that might increase voter confusion. Another worry is that on-cycle elections would draw attention away from local politics. Local contests might simply be overshadowed by more important statewide or national contests. Moving to concurrent elections could also alter the competitive dynamics of elections. As Chapter 4 shows, incumbents win more regularly in cities with concurrent elections. Given that incumbents already win 80 percent or more of their city council and mayoral reelection bids, a widespread move to on-cycle elections might make the outcomes of local contests almost a foregone conclusion. Indeed, some observers have raised concerns that concurrent elections would make it harder for challengers to raise campaign funds, which would mean that municipal incumbents might find it easier to ignore voters and their concerns. Still, the statistical relationship between timing and incumbent success is relatively mild, especially in comparison with more important factors such as city population size. Overall, these concerns are solid arguments for civic education, voter outreach campaigns, higher-quality media coverage of local races, and intensive campaigning by candidates for mayor and council. They are, in our view, not good arguments for scheduling local elections so as to knowingly reduce public participation. Being asked to cast a vote on a given matter may itself spur an educational and information-gathering 66 effort in many citizens.2 Ultimately, which approach California policymakers choose to pursue will depend on how much they value different aspects of the democratic process. There are real reasons to be concerned about low rates of voter participation. Indeed it is difficult to argue with the claim that moving toward full participation of all citizens should be a goal of every democracy. As such, we have tried to identify tools that could encourage voter turnout and expand participation in the local electoral arena. ____________ 2As Lupia (1994) shows, voters need not be experts on a given topic to be able to find informational cues (such as endorsements) that help them determine how they should vote. 67 Appendix A Survey of City Clerks This appendix consists of a copy of the survey administered to city clerks in the fall of 2000 and spring of 2001. For all categorical and “yes or no” questions, we also show the distribution of responses. On several follow-up questionnaires sent to nonrespondents or to respondents with incomplete or unusable responses, we changed the wording of question 11 to read, “Please enter the date of the most recent contested city council election.” The response rate varied widely across questions (Table A.1). Some questions did not apply in some cases or were difficult for city clerks to answer. Only cities with a directly elected mayor answered questions 2 through 10. The maximum number of possible responses for these items is 134 cities. Several questions required an answer only if the previous question was answered in the affirmative. These will naturally have fewer responses. 69 Table A.1 Number of Valid Responses for Individual Variables Survey Item q1 q2 q3 q4 q5 q6 q6a q7a q7b q8 q9(1st) q9(2nd) q10 q11 q12 q12a q12b q13 q14 q15 q15a q16 q17a q17b q17c q17d q17e Number of Cities 391 134 130 133 133 133 96 124 104 134 122 97 134 396 396 389 396 354 393 390 362 301 277 277 274 273 274 Survey Item q18a q18b q19 q20 q21 q21a q22 q23 q24 q25 q26a q26b q27 q28a q28b q28c q28d q29 q30a q30b q30c q30d q30e q31a q31b q32a q32b Number of Cities 387 387 376 353 386 31 378 383 401 385 374 384 393 388 388 388 388 391 395 395 389 383 395 307 125 394 393 70 Voter Turnout in Municipal Elections: A Survey of California City Clerks Thank you for assisting us in our research on voter participation in California municipal elections. Please complete the brief questionnaire below and return it in the enclosed postage-paid envelope. If you are unable to answer certain questions, please answer all of the other questions and return the survey. Your information is still very important to us. The data you provide will be the basis of important research on voter turnout in cities across the state. Please do not hesitate to contact us if you have any questions. Project Directors: Paul Lewis (lewis@ppic.org), Zoltan Hajnal (hajnal@ppic.org) Hugh Louch (louch@ppic.org) Public Policy Institute of California 500 Washington Street, Suite 800 San Francisco, CA 94111 (415) 291-4400 phone (415) 291-4401 fax City: Code: 71 Q-1. How is your mayor (chief elected official) selected? (Check the most appropriate response.) 33.4% Voters elect the mayor directly 54.7% Council selects the mayor from among its members 1.5% Council member receiving most votes in the election becomes the mayor 9.5% Council members rotate into the position of mayor 0.0% Other (please specify) ________________________ If voters DO NOT elect the mayor directly, please skip ahead to question Q-11. Your responses to the other questions are very important to us. Q-2. What was the date of the most recent election for mayor held in your city? (Please enter the date of the general mayoral election, NOT the runoff, if any.) _________ ______ , ________ month day year Q-3. How many voters in your city cast ballots for mayor in that election? _____________ Q-4. How many registered voters were there in your city at the time of that election? _____________ Q-5. How many mayoral candidates were listed on the ballot of that election? _____________ Q-6. Was there an incumbent standing for reelection? 72.2% yes 27.8% no A. If yes, did the incumbent win the election? 86.5% yes 13.5% no 72 Q-7. What percentage of the vote for mayor was captured by the top two candidates in that election? a. 1st Place _______% b. 2nd Place _______% Q-8. Was there a subsequent runoff election? 6.0% yes 94.0 % no Q-9. What is the race/ethnicity of the first and second place finishers in the mayoral election referred to in question Q-2? (Check the most appropriate response.) 1st Place 2nd Place a. White, not Hispanic 79.5% 77.3% b. Black, not Hispanic 4.1 7.2 c. Asian, not Hispanic 0.8 2.1 d. Hispanic, of any race 15.6 12.4 e. Other (please specify) 0.0 1.0 Q-10. Was there also an election for one or more council seats on that day? 95.5% yes 4.5% no All respondents, please answer the following questions. Q-11. Please enter the date of the most recent city council election. (Please use the most recent regular election, not a special election to fill a vacated seat, etc.) _________ ______ , ________ month day year Q-12. How many city council seats were chosen in that election? _____________ A. How many total candidates appeared on the ballot? _____________ B. How many total seats are there on the city council? _____________ 73 Q-13. How many voters in your city cast ballots for city council races in that election? (If council is elected by district please indicate the total number of ballots cast citywide. If council seats are elected at large, please indicate the total number of ballots cast in the council race with the greatest turnout.) ____________ Q-14. How many registered voters were there in your city at the time of that election? _____________ Q-15. How many incumbents ran for reelection to council in that election? _____________ A. How many of those incumbents were reelected? _____________ For the following questions (Q-16 through Q-22), please refer to the most recent mayoral election if the mayor is directly elected by the voters. If not, please refer to the most recent city council election. Q-16. To the best of your knowledge, were there any organized attempts to purge the registered voter roll (removing voters who had died, moved, etc.) in the three years prior to that election? 47.8% yes 52.2% no Q-17. Were voters in your city selecting any other local officials in the election held on that day? (Please check all applicable offices.) 17.7% City attorney or city treasurer 23.8 Other city office(s) 25.5 County supervisor(s) 15.0 County treasurer or sheriff or other county office(s) 45.6 School board member(s) 74 Q-18. How many city-level propositions were on the ballot in that election? (Please indicate the number of each type.) ________ citizen-sponsored initiatives ________ propositions placed by council Q-19. In your judgment, how much popular interest or controversy among voters was generated by the city-level ballot questions in that election? (Check the most appropriate response.) 14.6% More interest/controversy than most elections in this city 10.9 About the same as most other elections in this city 4.3 Less interest/controversy than in most elections in this city 70.2 There were no municipal ballot questions in that election Q-20. How many citizen-sponsored initiatives were on the ballot in the years 1997, 1998, and 1999 (all elections)? Again, please consider only city-level ballot questions. Total number of initiatives 1997–99: ___________ Q-21. Over the years from 1997 through 1999, were any recall initiatives filed against the mayor and/or council members? 8.0% Yes 92.0% No A. If yes, were any successful? 67.7% Yes 32.3% No Q-22. Recognizing how complex a community’s affairs are, please review the following general statements and select the one that best describes elections in your city in the last five years, compared to other municipalities in your region: 35.4% Our elections are low-key affairs with no major controversies 51.1 Elections are generally low-key, but there have been some notable controversies 13.5 Our city often has controversial elections 75 We conclude the survey by asking several basic questions about the structure of government in your city. PLEASE NOTE: In answering these questions, refer to the conditions in your city at the time of the mayoral or city council election you answered questions about above, not the current conditions, if different. Q-23. What is the length of the mayor’s term of office? (Important: If the chief elected official is a member of the council, specify the term for the position of mayor, not of council member.) _______ years Q-24. Is the mayoralty in your city considered a full-time or part-time office? (Generally, the office is considered full-time only if it carries a full-time salary.) 3.2% Full-time 92.8% Part-time 4.0% Don’t know Q-25. Does the mayor in your city have the authority to veto actions of the city council? 3.6% yes 96.4% no Q-26. Does your city limit the number of terms either the mayor or city council members can hold? Mayor*: 27.3% yes 72.7% no City Council: 18.5% yes 81.5% no (*Responses for directly-elected mayoralties only.) Q-27. Which of the following statements best describes the structure of governance in your city? 2.5% MAYOR-COUNCIL Elected council or board serves as the legislative body. The chief elected official is the head of government, generally elected separately from the council, with powers that may range from limited duties to full-scale authority for the daily operation of the government. 76 97.0 COUNCIL-MANAGER Elected council or board is responsible for making policy. A professional administrator or manager appointed by the board or council has full responsibility for the day-to-day operations of the government. 0.5 Other (please specify:___________________________) 0.0 Not sure Q-28. What methods are used to elect council members in your city? Fill in the number of seats elected by each method. a. At-large, elected citywide 92.5% b. Elected from districts 5.4 c. Members represent districts, but 2.1 are elected citywide d. Other method 0.0 (please specify: _________________________________) Q-29. Who has the overall responsibility for developing the budget submitted to the council? (check one) 0.3% Mayor 92.1 Chief Administrative Officer (city manager or equivalent) 1.0 Mayor and CAO are jointly responsible 6.6 Other (please specify:__________________________) Q-30. Regarding each of the following services in your city, are they mainly carried out by city government personnel, by county personnel, by a special district government, or by a private company under contract with the city? (Please check the appropriate response for each service. Leave item blank if the particular service is not provided within your city.) 77 Police Services Fire Services* Library Services Sewer Services Garbage Collection City Staff 72.1% 54.9 30.1 65.0 16.2 County 26.6% 26.8 64.8 10.7 0.8 Special District 0.8% 14.9 4.9 18.0 1.0 Private Firm 0.5% 0.5 0.3 6.3 82.0 Q-31. Has your city switched the timing of mayor/council elections at any time in recent years? 30.6% Yes, switched from “stand-alone” city elections to elections consolidated with statewide primary or general elections 1.0 Yes, switched from consolidated elections to standalone elections 68.4 Other type of change or no change in election timing B. If your city switched, when was the last election using the previous format? _________ ______ , ________ month day year Q-32. A. Has your city made any changes in the structure of city government (mayor-council, council-manager, etc.) in recent years? 4.3% yes 95.7% no B. Has your city made any changes in the way council members are elected (district, at-large, etc.) in recent years? 4.1% yes 95.9% no If the answer to either Q-32A or Q-32B is yes, please describe the change, and list the date it took effect, on the back of this questionnaire. Q-33. Please indicate your job title: _____________________ *An additional 2.8 percent of cities indicated volunteer fire departments. 78 Thank you very much for your participation. Your response is crucial to help inform policymakers and citizens about voter turnout in California cities. Please return the survey in the enclosed postage-paid envelope. You will receive a future mailing that will discuss the results of this survey. We welcome your comments on these topics, as well as comments regarding the questionnaire itself, or clarifications of your responses. You may include any written comments below, or on a separate sheet. The Public Policy Institute of California is a private, nonprofit organization dedicated to independent, nonpartisan research on economic, social, and political issues that affect the lives of Californians. For further information, including the full text of PPIC research reports, please visit our World Wide Web site at www.ppic.org. 79 Appendix B Mail Survey Procedures and Data Issues The analyses presented in this report are based on a survey of California city clerks conducted from October 2000 through January 2001. We purchased a mailing list of city clerks for the 474 cities then in existence from the League of California Cities. Where the contact information was incomplete, we used city web pages for guidance. A questionnaire was sent to each city clerk on the list, accompanied by an explanatory cover letter and a postage-paid return envelope. Follow-up contacts were made to maximize the response rate. Ultimately, respondents from 401 cities returned their surveys. Four cities’ responses were completely unusable because they were either unwilling or unable to answer the survey. One of these had not had a contested election in over a decade. Another 28 cities provided us with information for an uncontested election. We followed up with a second survey for these cities and received new responses from 18 of them, 16 of which were usable. Four cities left significant portions of the survey blank and were contacted with a second survey. Two responded with the additional information. A total of 375 cities returned surveys with enough responses to be useable for some portion of the analyses. Validity of the Sample We compared cities that responded to our survey to those that did not on several characteristics, using census data (Table B.1). There were few statistically significant differences between cities in the sample and those outside of it. Notably, the population of cities in the sample was 20,000 persons larger on average than cities that did not respond, but this difference was not statistically significant. (In part, this is a result of nearly all the state’s largest cities completing the questionnaire.) On most other measures, both the magnitude and the significance of 81 Table B.1 Comparing In-Sample and Out-of-Sample Cities Population Percentage black Percentage Hispanic Percentage in poverty Real median income Percentage unemployed Percentage home owner Average household size In Sample 55,543 3.9 24.4 11.7 $45,501 7.2 58.8 2.8 Not in Sample 30,306 3.7 28.4 13.9 $46,831 7.9 61.0 3.0 P-Value of Difference 0.294 0.812 0.188 0.029 0.662 0.194 0.242 0.047 differences were negligible. Cities in and out of the sample had similar percentages of Hispanics and blacks, unemployed persons, and home owners, as well as similar average incomes. Two factors—poverty and household size—were significantly different, but the magnitude of those differences was minimal. Cities not in the sample had an average poverty rate 2 percent higher than those in the sample. The average household size was also 0.2 persons greater on average. Neither of these numbers raises concerns about the value of the sample. We also compared the distribution of cities across the regions of the state and across city types (central city, suburb, rural), but found no significant differences (Table B.2). Slightly fewer cities from the Central Valley and slightly more cities from outside the three major regions (Los Angeles area, Bay Area, and the Central Valley) responded to our survey, but the differences were quite small. Among the different types of cities, we achieved the best response rate from central cities (95 percent), although they represent the smallest proportion of cities in California (only 10 percent). Cleaning the Data Several changes were made to fix data errors and minimize missing data. In general, we took respondents’ entries at face value and assumed that they answered questions correctly. If we had substantial concrete evidence contradicting a particular response, we used that information. 82 Table B.2 Geographic Distribution of Cities in Sample Region Greater Los Angeles Bay Area Central Valley Other Percentage Responding to Survey 84.5 84.7 83.0 88.2 City type Central city Suburb Rural 95.9 84.3 83.0 NOTE: significance = 0.763 for region, 0.075 for city type. Such evidence included published newspaper reports, vote tallies listed on official election websites, or phone conversations with city clerks or their staff. The major change we undertook was to minimize the extent of missing data. A substantial minority of respondents did not answer one or more crucial survey questions that would have made their responses effectively unusable. Table B.3 lists the number of missing values for questions regarding the number of voters, registered voters, and Table B.3 Important Missing Values Question Description Mayoral Election q3 Number of voters q4 Registered voters q5 Number of candidates Council Election q12a Number of candidates q13 Number of voters q14 Registered voters Missing 9 5 4 25 41 30 83 candidates for mayoral and council elections. These questions provide the core information for establishing turnout and competitiveness of a given election. To fix missing voter registration tallies, we turned to the Report of Registration published by the California Secretary of State. Although not completed each month, a report is available within two months of any election date, providing an accurate assessment of voter registration for every city in California for a given election. To remedy missing data issues for the number of voters and candidates, we conducted a newspaper search of voting results for each city missing any piece of information. Major newspapers in California publish election returns for both concurrent and nonconcurrent elections and can be easily accessed on-line. This search considerably narrowed the number of missing values for voters and candidates listed. For several cities, we were unable to use this information for council elections because of the method of election used in many California cities. Cities in California often elect more than one council member for a given election. When these elections are conducted at-large, voters are typically asked to select multiple candidates. Because many voters cast ballots for fewer candidates than they are permitted, we cannot obtain an accurate assessment of total turnout for these cities. The newspaper check also allowed us to replace missing values for several other variables: • The number and success of incumbents—questions 6 and 15, • Other offices up for election—question 17, • City propositions on a given ballot—question 18, and • Recall elections—question 21. In general we made very few additions to the data for these questions. Typically, respondents left questions blank when there were no other offices, no recall, no city measures on the ballot, or no incumbents up for reelection. For the handful of cases for which this was not the case, we input information from the newspaper sources. 84 Correcting Specific Data Problems Each survey question was checked for impossible values where necessary. In particular, questions 2 and 11 were checked to see that responses conformed to an actual election date. Questions 7a and 7b were checked to be sure that totals did not add up to more than 100 percent and that 7b was not greater than 50 percent. Questions 6a, 15a, and 21a were made to be missing if respondent indicated a “no” answer on Questions 6, 15, or 21, respectively. These second-part questions should not have been answered if the first part was answered in the negative. We assume that if a respondent answered either question 18a or 18b, but not both, that they intended for the other part to be zero. Also, if a respondent left question 19 blank but wrote in a zero for both 18a and 18b we know that 19 should be coded to indicate that there were no propositions on the ballot. If a respondent incorrectly answered question 19 when both 18a and 18b were zero, we assume that this was an honest misreading of the question and alter their answer to indicate that there were no propositions on the ballot. Another major issue is discrepancies between questions 12b and 28. In both cases respondents were asked to indicate the number of council seats. Twenty respondents answered both questions and indicated different numbers of total seats. All 20 of these were cities that directly elected their mayor. For these cities we assume that the lower number is correct and that the mayor sits as part of the city council. For generallaw cities, this is necessarily true, as they are limited in the number of council members they can have and the mayor, if directly elected, serves on the council. For the five charter cities that had inconsistent responses, we examined city websites to make sure this was the most appropriate strategy. In one case, a city had just changed to a directly elected mayor with the current election. Here we use the number of seats at the time the council was elected to appropriately fit with the other responses. Finally, a small number of cities entered the number of seats in the election for question 12b instead of the number of seats on the council. In these cases we changed the response to correspond to question 28. 85 One smaller problem came because cities entered check marks in response to questions 18a or 18b instead of a number. Although we cannot appropriately include these values for the raw versions of these questions, the analyses presented in this report use only binary versions of these variables (i.e., the presence of any city-level ballot propositions), making it possible to include these cases in our analysis. For tabulated responses to question 30, county refers to services provided either by a county or by another city. We also include a separate code for volunteer fire departments for question 30 when this information was volunteered by the respondent. There were 11 cases of this. 86 Appendix C Voter Turnout Regressions This appendix provides information regarding the voter turnout regressions discussed in Chapter 3. Table C.1 shows the results of estimations for two dependent variables: the percentage of registered voters casting ballots and the estimated percentage of all adults casting ballots. As noted in Chapter 3, city council elections and mayoral elections (for those cities that have a separately elected mayor) are merged in this dataset. The “mayoral election” variable is used to denote the mayoral elections in the sample. Finally, for the 421 observations of council or mayoral elections included in the regressions, Table C.2 provides the mean, high value, and low value for each variable. 87 Table C.1 Turnout in City Elections: Regression Model Timing variables Presidential (compared to off-cycle) Presidential primary (compared to off-cycle) Gubernatorial (compared to off-cycle) Odd-year November (compared to off-cycle) Mayor and council election held same day Other local elections held same day Mayoral election (compared to council election) Institutional variables Charter (compared to general-law) city District (compared to at-large) council election Number of services provided by city staff Office has term limits Electoral context variables One or more propositions on city ballot Election uncontested Candidates per seat Incumbents per seat Percentage of voting-age residents registered (est.) Percentage Casting Ballots Registered Adult Voters Residents 35.72*** (14.60) 20.66*** (6.74) 26.02*** (14.70) 1.97 (0.96) 1.43 (0.92) 0.27 (0.22) –3.33** (2.01) 23.39*** (13.07) 11.19*** (5.00) 15.37*** (11.85) –0.82 (0.54) 1.13 (0.99) 0.25 (0.27) –1.79 (1.47) 0.55 (0.33) –1.03 (0.30) 0.98** (2.02) 0.80 (0.47) 0.64 (0.52) –8.44*** (3.64) 0.37 (1.04) 0.64 (0.51) 4.11*** (3.26) –5.48** (2.09) 0.93* (1.86) 0.41 (0.24) –0.14** (1.98) 3.52*** (3.79) –4.28** (2.22) 0.69* (1.87) 0.13 (0.11) n/i 88 Table C.1 (continued) City population characteristics City population (natural log) Socioeconomic status (factor score) Percentage black Percentage Hispanic Percentage Asian Percentage age 18 to 24 Percentage age 65 or older Percentage lived in same house for 5 years Percentage institutionalized Constant Observations Adjusted R-squared Percentage Casting Ballots Registered Adult Voters Residents –3.28*** (5.31) 3.65*** (3.12) –0.04 (0.38) –0.07 (1.43) –0.25*** (3.34) 0.09 (0.41) 0.31** (2.06) 0.02 (0.29) n/i 65.01*** (6.79) 421 0.61 –2.45*** (5.44) 3.94*** (4.79) –0.00 (0.01) –0.20*** (6.66) –0.34*** (6.48) 0.04 (0.23) 0.26** (2.32) 0.12** (1.98) –0.31*** (2.91) 42.14*** (7.40) 421 0.67 NOTES: *p < 0.10, **p < 0.05, ***p < 0.01. n/i indicates variable is not included in regression. Absolute values of t-values are in parentheses. 89 Table C.2 Descriptive Statistics of Variables for Observations in the Regression Dependent variables Turnout of registered voters Turnout of adult residents Mean Value Low Value High Value 47.16 30.04 7.32 88.64 4.46 78.79 Timing variables Presidential Presidential primary Gubernatorial Odd-year November Mayor and council election held same day Other local elections held same day 0.10 0.05 0.49 0.18 0.49 0.54 0 0 0 0 0 0 1 1 1 1 1 1 Institutional variables Charter city District council election Number of services provided by city staff Office has term limits 0.24 0.04 2.57 0.22 0 0 0 0 1 1 5 1 Electoral context variables Any ballot measures Uncontested election Candidates per seat Incumbents per seat Percentage registered, of voting-age residents (est.) 0.34 0.07 2.60 0.69 63.00 0 0 1 0 12.47 1 1 14 1 97.64 Type of election Mayoral 0.30 0 1 City population characteristics City population (natural log) Socioeconomic status (factor score) Percentage black Percentage Hispanic Percentage Asian Percentage age 18 to 24 Percentage age 65 or older Percentage lived in same house for 5 years Percentage institutionalized 10.27 0.01 4.23 30.57 9.09 9.28 11.61 43.70 1.49 5.25 –2.02 0 2.15 0 2.36 3.65 11.35 0 15.13 4.62 46.42 98.27 61.51 33.56 45.06 70.81 39.98 90 Appendix D Candidate Competition Regression Table D.1 displays results for the regression described in Chapter 4. This model attempts to account for the number of candidates running for office (whose names are listed on the ballot). The dependent variable is the natural logarithm of the number of candidates (for mayoral elections) or of the number of total candidates divided by the number of seats (for council elections). Mayoral and council elections are merged in the regression. The “mayoral election” variable is an indicator used to differentiate between mayoral and council elections in the sample. 91 Table D.1 Determinants of Competition for Office Mayoral election (compared to council election) Concurrent with statewide election (compared to nonconcurrent) Mayoral and council election held same day Other local offices elected same day District council elections (compared to at-large) Full-time mayor (compared to part-time) Four-year mayoral term (compared to two-year term) Charter (compared to general-law) city Number of services provided by city staff Office has term limits Percentage voting-age citizens registered (est.) Number of voter initiatives, 1997–1999 Typical degree of controversy of city elections (3-point scale) Incumbents per seat Percentage independent/third party (of registered voters, 1999) City population (natural log) Residential stability Percentage black Percentage Hispanic –0.21*** (3.48) 0.03 (0.69) 0.02 (0.49) –0.08** (2.20) –0.12 (1.07) 0.35** (2.57) 0.27*** (3.68) 0.05 (0.86) –0.00 (0.27) –0.10* (1.77) 0.00 (1.52) 0.03** (2.40) 0.08** (2.45) –0.28*** (5.47) 0.03*** (4.02) 0.12*** (6.43) –0.01** (2.25) 0.01** (2.22) 0.00*** (3.32) 92 Table D.1 (continued) Percentage Asian Percentage age 18 to 24 Percentage age 65 or older Constant Observations Adjusted R-squared –0.00 (0.13) –0.01 (1.45) 0.01 (1.21) –0.90** (2.54) 400 0.40 NOTE: *p < 0.1, **p < 0.05, ***p < 0.01. Absolute values of t-values are in parentheses. Dependent variable is the log of the number of candidates per available council seat (for council elections) and the log of the number of candidates for mayor (for mayoral elections). A potential statistical issue with the regression in Table D.1 is that it treats mayoral and council elections from the same city as independent observations. However, running the model with the observations grouped, or clustered, by city yields the same results and significance levels, with one exception: The percentage of young adults is negative and significant at the 0.05 significance level. We also tried running the competition model with uncontested elections omitted, resulting in 370 observations rather than 400, and an adjusted R-squared of 0.39. Results were again basically the same, with the following exceptions: District council elections had a slightly significant, negative relationship to council competition, as did the percentage of young adults in the city. More interesting, the difference in competition levels between mayoral and council elections vanished. This indicates that the lower level of competition for mayoralties is primarily attributable to the substantial number of uncontested mayoral elections. As noted in Chapter 4, 17 percent of mayoral elections in our data were uncontested, compared to 4 percent of council elections. 93 References Alford, Robert R., and Eugene C. Lee, “Voting Turnout in American Cities,” American Political Science Review, Vol. 62, 1968, pp. 796– 813. Baldassare, Mark, California in the New Millennium: The Changing Social and Political Landscape, University of California Press, Berkeley, California, 2000. Bennett, Stephen Earl, and David Resnick, “The Implications of Nonvoting for Democracy in the United States,” American Journal of Political Science, Vol. 34, No. 3, 1990, pp. 771–802. Bobo, Lawrence, and Franklin D. Gilliam, Jr., “Race, Sociopolitical Participation, and Black Empowerment,” American Political Science Review, Vol. 84, No. 2, 1990, pp. 377–393. Bridges, Amy, Morning Glories: Municipal Reform in the Southwest, Princeton University Press, Princeton, New Jersey, 1997. Bullock, Charles S., “Turnout in Municipal Elections,” Policy Studies Review, Vol. 9, No. 3, 1990, pp. 539–549. Bullock, Charles S., and Susan A. MacManus, “Structural Features of Municipalities and the Incidence of Hispanic Councilmembers,” Social Science Quarterly, Vol. 71, No. 4, 1990, pp. 665–681. Cain, Bruce E., Megan Mullin, and Gillian Peele, “City Caesars? An Examination of Mayoral Power in California,” paper presented at the Annual Meeting of the American Political Science Association, San Francisco, California, 2001. California Secretary of State, “Registration and Participation, November 7, 2000,” 2001, available at http://www.ss.ca.gov/elections/ elections.htm. 95 Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes, The American Voter: The Unabridged Edition, University of Chicago Press, Chicago, Illinois, 1980 (orig. 1960). Casel, Carol A., “The Nonpartisan Ballot in the United States,” in Bernard Grofman and Arend Lijphart (eds.), Electoral Laws and Their Political Consequences, Agathon Press, New York, 1986, pp. 226–241. Copeland, Gary W., “Term Limits for Urban Settings: One Size Fits All?” Urban News, Vol. 11, No. 1, 1997, pp. 1–10. Cox, Gary, and Michael C. Munger, “Closeness, Expenditures and Turnout in the 1982 House Elections,” American Political Science Review, Vol. 83, 1989, pp. 217–231. Curtin, Daniel J., Jr., Curtin’s California Land Use and Planning Law, 20th ed., Solano Press Books, Point Arena, California, 2000. Davidson, Chandler, and Bernard Grofman (eds.), Quiet Revolution in the South: The Impact of the Voting Rights Act, 1965–1990, Princeton University Press, Princeton, New Jersey, 1994. Dennis, Jack, “Theories of Turnout: An Empirical Comparison of Alienationist and Rationalist Perspectives,” in William Crotty (ed.), Political Participation and American Democracy, Greenwood Press, New York, 1991, pp. 23–65. Engstrom, Richard L., and Michael D. McDonald, “The Underrepresentation of Blacks on City Councils: Comparing the Structural and Socioeconomic Explanations for South/Non-South Differences,” Journal of Politics, Vol. 44, 1982, pp. 1088–1099. Espino, Rodolfo, “The Political Legacy of Municipal Reform,” paper presented at the Annual Meeting of the Midwestern Political Science Association, Chicago, Illinois, 2001. Finkel, Steven E., “Reciprocal Effects of Participation and Political Efficacy: A Panel Analysis,” American Journal of Political Science, Vol. 29, 1985, pp. 891–913. Gerber, Elisabeth R., Prospects for Reforming the Initiative Process, report prepared for the University of Virginia Project on Initiatives, 1999. 96 Guinier, Lani, The Tyranny of the Majority: Fundamental Fairness in Representative Democracy, Free Press, New York, 1994. Hahn, Harlan, and Sheldon Kamieniecki, Referenda Voting: Social Status and Policy Preferences, Greenwood Press, New York, 1987. Hajnal, Zoltan L., and Mark Baldassare, Finding Common Ground: Racial and Ethnic Attitudes in California, Public Policy Institute of California, San Francisco, California, 2001. Hampton, Gloria J., and Katherine Tate, “Political Participation in Urban Elections: The SES Baseline Model Reexamined,” paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, Illinois, 1996. Hill, John, “Voter Turnout Falling in Both U.S. and State,” Sacramento Bee, December 22, 2000. International City/County Management Association (ICMA), The Municipal Year Book, International City/County Management Association, Chicago, Illinois, 1996. Jacobson, Gary C., The Politics of Congressional Elections, Little, Brown, Boston, Massachusetts, 1983. Johnson, Ted, “Cities Delay Elections to Avoid 1996 Primary Conflict,” Los Angeles Times, December 22, 1994. Karnig, Albert K., and B. Oliver Walter, “Decline in Municipal Voter Turnout: A Function of Changing Structure,” American Politics Quarterly, Vol. 11, No. 4, 1983, pp. 491–505. Karnig, Albert K., and B. Oliver Walter, “Municipal Voter Turnout During the 1980s: The Case of Continued Decline,” in Harry A. Bailey and Jay M. Shafritz (eds.), State and Local Government and Politics: Essential Reading, F. E. Peacock Publishers, Itasca, Illinois, 1993. Krebs, Timothy B., “The Determinants of Candidates’ Vote Share and the Advantages of Incumbency in City Council Elections,” American Journal of Political Science, Vol. 42, No. 3, 1998, pp. 921–935. 97 Lelchuk, Irene, “Record-Low Vote Came at High Price,” San Francisco Chronicle, December 13, 2001. Lien, Pei-te, “Ethnicity and Political Participation: A Comparison Between Asian and Mexican Americans,” Political Behavior, Vol. 16, 1994, pp. 237–264. Lupia, Arthur, “Shortcuts Versus Encyclopedias: Information and Voting Behavior in California Insurance Reform Elections,” American Political Science Review, Vol. 88, No. 1, 1994, pp. 63– 76. Magleby, David B., “Direct Legislation in the American States,” in David Butler and Austin Ranney (eds.), Referendums Around the World, AEI Press, Washington, D.C., 1994. Mitchell, Greg, The Campaign of the Century: Upton Sinclair’s Race for Governor of California and the Birth of Media Politics, Random House, New York, 1992. Morlan, Robert L., “Municipal Versus National Election Voter Turnout: Europe and the United States,” Political Science Quarterly, Vol. 99, No. 3, 1984, pp. 457–470. Oliver, J. Eric, “City Size and Civic Involvement in Metropolitan America,” American Political Science Review, Vol. 94, No. 2, 2000, pp. 361–372. Pantoja, Adrian D., and Gary M. Segura, “Citizens by Choice, Voters by Necessity: Patterns in Political Mobilization by Naturalized Latinos,” paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, Illinois, 2000. Parker, Frank R., Black Votes Count: Political Empowerment in Mississippi After 1965, University of North Carolina Press, Chapel Hill, North Carolina, 1990. Powell, G. Bingham, “American Voter Turnout in Comparative Perspective,” American Political Science Review, Vol. 81, 1986, pp. 45–65. Putnam, Robert D., Bowling Alone: The Collapse and Revival of American Community, Simon & Schuster, New York, 2000. 98 Rohrs, Sarah, “Growing Numbers Voting Through Mail Instead of Booth,” San Jose Mercury News, October 20, 2000. Rosales, Rodolfo, The Illusion of Inclusion: The Untold Political Story of San Antonio, University of Texas Press, Austin, Texas, 2000. Rose, Roger (ed.), The California Initiative Process: Current Controversies and Prescriptions, California Policy Seminar, University of California, Berkeley, California, 1990. Rosenstone, Steven J., and John Mark Hansen, Mobilization, Participation, and Democracy in America, Macmillan Publishing Company, New York, 1993. Schaffner, Brian F., Gerald Wright, and Matthew Streb, “Teams Without Uniforms: The Nonpartisan Ballot in State and Local Elections,” Political Research Quarterly, Vol. 54, 2001, pp. 7–30. Schneider, Mark, and Paul Teske, Public Entrepreneurs: Agents for Change in American Government, Princeton University Press, Princeton, New Jersey, 1995. Schrag, Peter, Paradise Lost: California’s Experience, America’s Future. New Press, New York, 1998. Simerman, John, “Debate Surrounds Plan to Move Local Elections,” Contra Costa Times, June 8, 1998. Sokolow, Alvin D., and Peter M. Detwiler, “California,” in Dale Krane, Platon Rigos, and Melvin Hill (eds.), Home Rule in America: A Fifty-State Handbook, CQ Press, Washington, D.C., 2001, pp. 58–68. Squire, Peverill, et al., Dynamics of Democracy, Brown & Benchmark, Madison, Wisconsin, 1995. Squire, Peverill, and Raymond E. Wolfinger, “Residential Mobility and Voter Turnout,” American Political Science Review, Vol. 81, 1987, pp. 45–65. Teixeira, Ruy, The Disappearing American Voter, Brookings Institution Press, Washington, D.C., 1992. 99 Thompson, Joel A., and Gary F. Moncrief, “The Implications of Term Limits for Women and Minorities: Some Evidence from the States,” Social Science Quarterly, Vol. 74, 1993, pp. 300–309. Uhlaner, Carole Jean, Bruce E. Cain, and D. Roderick Kiewiet, “Political Participation of Ethnic Minorities in the 1980s,” Political Behavior, Vol. 11, 1989, pp. 195–221. Verba, Sidney, Kay Lehman Schlozman, and Henry E. Brady, Voice and Equality: Civic Voluntarism in American Politics, Harvard University Press, Cambridge, Massachusetts, 1995. Wattenberg, Martin P., Turnout Decline in the U.S. and Other Advanced Industrial Democracies, Center for the Study of Democracy, University of California, Irvine, California, 1998. Welch, Susan, “The Impact of At-Large Elections on the Representation of Blacks and Hispanics,” Journal of Politics, Vol. 52, 1990, pp. 1050–1076. Welch, Susan, and Timothy Bledsoe, Urban Reform and Its Consequences: A Study in Representation, University of Chicago Press, Chicago, Illinois, 1988. Wolfinger, Raymond E., and Steven J. Rosenstone, Who Votes? Yale University Press, New Haven, Connecticut, 1980. Wolman, Harold, Edward Page, and Martha Reauley, “Mayors and Mayoral Careers,” Urban Affairs Quarterly, Vol. 25, No. 3, 1990, pp. 500–513. 100 About the Authors ZOLTAN L. HAJNAL Zoltan L. Hajnal is an assistant professor of political science at the University of California, San Diego. Before joining UCSD, he was a research fellow at the Public Policy Institute of California and a Mellon Postdoctoral Fellow at Brandeis University. His research interests include racial and ethnic relations, urban politics, and direct democracy. He holds a B.S. in geophysics and a B.A. in political science from Yale University and an M.A. and Ph.D. in political science from the University of Chicago. PAUL G. LEWIS Paul G. Lewis is a research fellow at the Public Policy Institute of California, where he studies local government and urban policy issues. He has authored or co-authored publications on local government structure, local sales taxes, urban development, transportation policy, and regional governance, and has made presentations on such topics to a variety of governmental bodies and civic organizations. He holds an A.B. in political science from Indiana University and an M.A. and Ph.D. in politics from Princeton. HUGH LOUCH Hugh Louch is an analyst at Cambridge Systematics and a former research associate at the Public Policy Institute of California. His research interests include urban transportation, housing, and planning. He received a B.A. in sociology from the University of California, Santa Barbara, and an M.A. in sociology from Princeton University. 101 Other Related PPIC Publications The Effect of Minority Districts and Minority Representation on Political Participation in California Claudine Gay Interest Group Influence in the California Initiative Process Elisabeth R. Gerber Are There Winners and Losers? Race, Ethnicity, and California’s Initiative Process Zoltan Hajnal, Hugh Louch PPIC publications may be ordered by phone or from our website (800) 232-5343 [mainland U.S.] (415) 291-4400 [Canada, Hawaii, overseas] www.ppic.org 103" ["post_date_gmt"]=> string(19) "2017-05-20 09:35:24" ["comment_status"]=> string(4) "open" ["ping_status"]=> string(6) "closed" ["post_password"]=> string(0) "" ["post_name"]=> string(8) "r_302zhr" ["to_ping"]=> string(0) "" ["pinged"]=> string(0) "" ["post_modified"]=> string(19) "2017-05-20 02:35:24" ["post_modified_gmt"]=> string(19) "2017-05-20 09:35:24" ["post_content_filtered"]=> string(0) "" ["guid"]=> string(50) "http://148.62.4.17/wp-content/uploads/R_302ZHR.pdf" ["menu_order"]=> int(0) ["post_mime_type"]=> string(15) "application/pdf" ["comment_count"]=> string(1) "0" ["filter"]=> string(3) "raw" ["status"]=> string(7) "inherit" ["attachment_authors"]=> bool(false) }