Donate
Independent, objective, nonpartisan research

R 601CGR

Authors

R 601CGR

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_601CGR.pdf" ["wpmf_size"]=> string(6) "910554" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(173159) "The Effect of Minority Districts and Minority Representation on Political Participation in California ••• Claudine Gay 2001 PUBLIC POLICY INSTITUTE OF CALIFORNIA Library of Congress Cataloging-in-Publication Data Gay, Claudine, 1970– The effect of minority districts and minority representation on political participation in California / Claudine Gay. p. cm. Includes bibliographical references. ISBN 1-58213-030-2 1. Political participation—California. 2. African Americans— California—Politics and government. 3. Hispanic Americans— California—Politics and government. 4. Election districts— California. 5. United States. Congress. House—Election districts. I. Public Policy Institute of California. II. Title. JK8789 .G39 2001 323'.042'089009794—dc21 2001019718 Copyright © 2001 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 Now that the 2000 Census has shown that California is a majorityminority state, more and more observers have become interested in the political, social, and economic consequences of the state’s shifting demographic profile. For its sheer speed and magnitude, this change is unprecedented in the United States. Since 1960, California has doubled its population, and demographers predict that the population will nearly double again by 2040. A generation ago, 80 percent of Californians were non-Hispanic whites; a generation from now, 50 percent will be Latino. This demographic transformation is leading California and its policymakers into new and uncharted waters. For this reason, PPIC began an intensive study of this transformation and its consequences in 1998. Our work to date includes a portrait of race and ethnicity in California, an assessment of how different ethnic groups react to legal authority, a study of attitudes toward public policy by race and ethnicity, and an investigation of voting behavior in homogeneous and ethnically diverse neighborhoods. Claudine Gay’s study, which examines voter turnout by race and ethnicity in California’s congressional districts, adds a new and significant dimension to that research effort. Her primary question—is voter turnout increased in majority-minority districts—pertains directly to the redistricting process that will be triggered by the 2000 Census. Her findings indicate that turnout among minority voters in the 1994 elections was highest wherever they were able to play a meaningful role in political life. After adjusting for such variables as age, education, and income, she finds that registered voter turnout among Latinos was 33 percentage points higher in majority-Latino districts than in majorityAnglo districts. African-American voting-age turnout was highest in districts where blacks and Latinos were equally matched and together formed the majority of the voting-age population. She also finds that Anglo turnout did not suffer in districts where whites were a minority. iii The next redistricting effort can and will influence voter participation, especially in California’s majority-minority districts. This timely work shows why the redistricting process is so important and how it can be used to increase political engagement throughout the state. David W. Lyon President and CEO Public Policy Institute of California iv Summary Since the passage of the Voting Rights Act of 1965 (VRA), the number of Latino and African-American elected officials at the local, state, and federal levels has increased dramatically. Minority members of the U.S. House of Representatives are among the most influential and prominent class of these new officeholders, and California is home to one of the nation’s most diverse congressional delegations. Benefiting in part from the creation of majority-minority electoral districts under the provisions of the VRA, minority legislators as a group have emerged as visible political actors in an institution traditionally dominated by whites. Considerable debate has emerged over whether this diversity has translated into substantive benefits for California’s minority constituents. Many advocates have argued that, regardless of its influence on legislative outcomes, the creation of majority-minority districts—and the subsequent election of minority representatives—serves an important role in pulling the Latino and African-American communities into the political process. By “creating a climate of inclusion,” these districts are expected to increase political participation and make voters out of previously unengaged minority Californians (New York Times, April 12, 1992). Although these claims have been met with skepticism, neither advocates nor skeptics have offered more than anecdotal evidence and journalistic speculation to clarify the link between race-conscious districting and greater political engagement among ethnic and racial minorities. This report examines the relationship between majority-minority districts, minority representation, and voting participation among Latinos, African Americans, and Anglos in California. In particular, it addresses the following questions: v • Does the level of Latino, African-American, and Anglo voter participation vary with the racial and ethnic balance of a congressional district? If so, how? • Do voter participation rates in the majority-minority congressional districts served by minority members of Congress differ from the rates observed in the majority-minority districts represented by Anglo legislators? If so, how? • What, if any, difference exists between Latinos and African Americans in the relationship between voter turnout and type of congressional district? Drawing on data from the November 1994 congressional elections, the report compares voter turnout rates in the state’s 13 majorityminority congressional districts with turnout rates in majority-Anglo districts. The data assembled for the analysis are aggregated at the precinct level and consist of voting statistics, Census data, and political information. When considering differences in turnout, the analysis takes into account the potentially confounding effects of socioeconomic conditions (e.g., levels of educational attainment, household income, concentrations of new residents) as well as electoral competitiveness (e.g., incumbency, tenure, vote margin) to gauge the independent effects of majority-minority districts and minority representation. The analysis provides empirical support for the claim that minority representation and majority-minority districts are associated with greater involvement in electoral politics among Latinos and African Americans. Four major findings emerge: • Latino and African-American voter participation is highest in congressional districts where Latinos and African Americans, respectively, are able to play prominent roles in deciding political outcomes. For African Americans, who do not constitute a majority in any of the state’s congressional districts, voting-age turnout in districts where they and Latinos are equally matched and, together, form the majority of the voting-age population is on average 6.6 percentage points higher than turnout in majority-Anglo districts. In districts where African Americans vi are one part of a diverse multi-ethnic mix, black voting-age turnout rates exceed those in majority-Anglo districts by 3.9 percentage points. Among Latinos, registered voter turnout rates in any and all majority-minority districts are higher than those observed in majority-Anglo districts. Latino registered voter turnout is at its highest in the state’s six majority-Latino congressional districts. In these districts, registered voter turnout is on average 33 percentage points higher than rates in majority-Anglo districts. Where Latinos and African Americans are more equally matched, Latino registered voter turnout is 30 percentage points above rates in majority-Anglo districts. Finally, in the multiethnic districts where Latinos play a relatively more limited role, their registered voter turnout rates exceed those in majorityAnglo districts by a more modest 6.9 percentage points. • African-American turnout is lowest wherever a single nonblack community (of any race or ethnicity) clearly dominates the electoral space, effectively relegating African Americans to the political margins. Turnout is marginally lower when this “advantaged” community is Latino than when the community is Anglo—the more typical case in the state, and the one to which African Americans have had more time to grow accustomed. • Voter participation among Latinos is particularly high in districts where they enjoy both majority status as well as descriptive representation (i.e., representation by legislators of the same race or ethnicity). In the four majority-Latino districts that were represented in 1994 by Latino members of Congress, Latino registered voter turnout was 36.4 percentage points higher than rates in majority-Anglo districts. By comparison, in the two majority-Latino districts represented by Anglo legislators in 1994, Latino turnout was 26.5 percentage points higher than rates in majority-Anglo districts. For African Americans, there is no discernible independent relationship between black representation, itself, and voting-age participation. vii • Neither the lack of majority status nor the lack of Anglo representation is associated with lower Anglo voting-age participation. In fact, Anglo voter participation is slightly higher in majority-Latino districts than in majority-Anglo districts in the state, possibly reflecting an appreciation for the potential electoral influence that could be wielded by a highly mobilized minority in the context of high Latino noncitizenship and relative youth. These findings have important implications for the decisions and challenges that state lawmakers will confront as they approach redistricting in 2001. First, creating new majority-minority congressional districts is likely to promote political participation in the state’s Latino and African-American communities. In particular, the creation of such districts can compensate for the socioeconomic barriers that so often contribute to low voter participation in minority communities. For example, including a precinct in a majority-black and -Hispanic district rather than in a majority-Anglo district may have a more pronounced and immediate effect on voter participation than increasing the number of college-educated residents by 10 percentage points. As a result of connecting new voters to the political world, the redistricting process may help create a more dynamic electorate that more closely reflects the demographic composition of the state’s citizen population. In creating majority-minority districts, mapmakers should be mindful that turnout among minority voters is high wherever they are able to play a meaningful role in political life. Greater participation among African Americans is observed when they are not relegated to the political margins in a district dominated by a single nonblack racial or ethnic group—white or nonwhite. Whereas Latino registered voter turnout is highest in districts where they enjoy overwhelming majority status, it is also high in districts where they are on roughly equal footing with African Americans. Even in the more heterogeneous areas, Latinos continue to participate at rates that exceed those in majority-Anglo districts. viii In all of these scenarios, Anglo participation does not suffer. Indeed, white voting-age participation is slightly higher in majority-Latino districts than in districts with Anglo majorities. In short, this evidence suggests that there may be some advantage to putting more emphasis on the creation of black-Hispanic and multiethnic districts in the next redistricting round. These districts would allow for multiple racial and ethnic communities to exercise some political leverage—precisely the kind of political environment that encourages participation. Given the democratic value of an engaged citizenry, not to mention the role that politics historically has played in the advancement of racial and ethnic groups in the United States, participation is a worthwhile if not paramount issue. ix Contents Foreword ........................................ Summary ........................................ Figures ......................................... Tables .......................................... iii v xiii xv 1. INTRODUCTION ............................. 2. MAJORITY-MINORITY DISTRICTS AND MINORITY REPRESENTATION ........................... The Voting Rights Act and Its Mandates .............. Majority-Minority Districts in the 1990s .............. Minority Representation in the 1990s................. 3. MINORITY VOTER PARTICIPATION ............. Trends in Minority Voter Participation ............... Constraints on Minority Voter Participation ............ Some Hypotheses About Majority-Minority Districts and Minority Voter Participation ................... The Psychology of Empowerment.................. Targeted Mobilization .......................... An Artifact of the Redistricting Process .............. 4. DATA AND METHODOLOGY ................... Sample Year ................................... Unit of Analysis and Measures ...................... Methodology .................................. 5. RESEARCH FINDINGS ......................... General Findings ............................... Racial Differences in Voter Turnout ................ Aggregate Turnout Differences Across District Type ..... Socioeconomic and Political Influences on Voter Turnout .................................. Majority-Minority Districts and Voter Turnout.......... 1 7 7 8 12 17 17 18 21 22 23 24 27 27 32 37 41 41 42 45 48 50 xi Majority-Minority Districts, Minority Representation, and Voter Turnout ............................. Summary of Findings ............................ 6. POLICY IMPLICATIONS........................ Appendix A. Sample ...................................... B. Methodology .................................. C. District-Level Turnout Estimates with Standard Errors..... D. Regression Output .............................. References ....................................... About the Author .................................. Other Related PPIC Publications ....................... 55 61 63 67 69 75 79 83 95 97 xii Figures 2.1. Majority-Minority Districts in California ............ 11 2.2. Minority Districts and Minority Representatives Nationwide, 1962–1998 ....................... 13 5.1. Voting-Age Turnout by District, November 1994 ..... 42 5.2. Registered Voter Turnout by District, November 1994 ..................................... 43 5.3. Anglo Voting-Age Turnout by District, November 1994 ..................................... 44 5.4. Black Voting-Age Turnout by District, November 1994 ..................................... 45 5.5. Latino Registered Voter Turnout by District, November 1994 ..................................... 46 5.6. District-Level Voter Participation in Majority-Minority Districts, November 1994 ...................... 47 5.7. Predicted Voter Participation in “Average” Precinct, by District Type ............................... 54 5.8. Predicted Voter Turnout in “Average” Precinct, by District Type and Legislator Race................. 60 C.1. District-Level Estimates of Anglo Turnout with Uncertainty ................................ 76 C.2. District-Level Estimates of Black Turnout with Uncertainty ................................ 77 C.3. District-Level Estimates of Latino Turnout with Uncertainty ................................ 78 xiii Tables 2.1. Number of Majority-Minority Districts After 1990 Redistricting, Selected States .................... 3.1. Racial and Ethnic Composition of the Electorate and Adult Population, 1990–1998 ................... 4.1. Majority-Minority Districts and Minority Representatives in California, November 1994 ....... 5.1. Estimates of Voter Participation Differences Across District Types .............................. 5.2. Estimates of Voter Participation Differences Across District Type and Legislator’s Race................ A.1. Number of Precinct Observations in Dataset, by District ................................... B.1. Ecological Inference for Black/Nonblack Voter Turnout .................................. B.2. Ecological Inference for White/Nonwhite Voter Turnout .................................. B.3. Ecological Inference for Latino/Non-Latino Voter Turnout .................................. B.4. Definition of Variables ........................ D.1. Coefficient Estimates for Regression of Majority- Minority Districts on Voting Participation .......... D.2. Coefficient Estimates for Regression of Majority- Minority Districts and Minority Representation on Voting Participation .......................... 10 17 33 50 56 67 70 70 71 73 80 81 xv 1. Introduction Since the passage of the Voting Rights Act of 1965 (VRA), the number of Latino and African-American elected officials at the local, state, and federal levels has increased dramatically.1 According to the most recent figures, more than 14,000 African Americans and Latinos hold political office across the country (Bureau of the Census, 1999). In California alone, there are more than 1,000 African-American and Latino elected officials, up from 700 officeholders in the early 1980s, including 24 Latino and six African-American state legislators and more than 300 Latino city council members—100 from Los Angeles County alone (Guerra, 1998; Bureau of the Census, 1999; Reyes, 2001). Minority members of the U.S. House of Representatives are among the most influential and prominent class of these new officeholders. Benefiting in part from the creation of majority-minority electoral districts under the provisions of the VRA, minority legislators as a group have emerged as visible political actors in an institution traditionally dominated by whites.2 California is home to one of the nation’s most ethnically and racially diverse congressional delegations. In the 106th Congress, seated after the November 1998 congressional elections, California’s U.S. House delegation included four African-American legislators, five Latino legislators, and one Asian legislator. The California delegation alone ____________ 1In this report, I use the terms Latino and Hispanic interchangeably to indicate anyone of Hispanic origin. I use the terms African-American and black interchangeably. I also use the terms white and Anglo to refer to non-Hispanic whites. This report focuses on Latinos and African Americans; Asian legislators and Asian constituents are not discussed in this research, both because there are so few Asian legislators and because the lack of sufficient geographic concentrations of Asian constituents (i.e., comparatively low levels of residential segregation) poses significant methodological challenges. 2Majority-minority electoral districts are defined as any district in which racial or ethnic minorities (Asians, African Americans, and Latinos) constitute more than 50 percent of the total district population. 1 accounted for roughly 18 percent of all minority elected officials serving in the House of Representatives. With one exception, these members of Congress represented districts drawn so that racial and ethnic minorities would constitute the majority of the voting-age population. The rise in the number of minority representatives and the use of race-conscious districting mechanisms to secure their election have prompted considerable debate over whether this greater diversity has translated into substantive benefits for minority constituents. To date, political scientists have focused primarily on the links among minority representation, the legislative process, and public policy outcomes favorable to minority communities (Swain, 1995; Cameron, Epstein, and O’Halloran, 1996; Lublin, 1997; Whitby, 1998; Hutchings, McClerking, and Charles, 1999; Canon, 1999). Of particular concern is the extent to which minority legislators and white legislators of the same political party differ in the quality of issue representation—as measured by bill sponsorship and legislative voting record—they provide for minority constituents.3 On this question the empirical record is mixed. Although several studies have found evidence of greater responsiveness and advocacy among minority legislators—particularly on issues with a racial dimension (Canon, 1999)—others conclude that, on the whole, there are no appreciable racial differences in legislative activism. Complicating the relationship further is mounting evidence that the practice of concentrating minority voters so as to ensure the election of minority legislators may produce deleterious unintended consequences. In particular, adjoining congressional districts, deprived of a significant number of minority voters, may increasingly elect representatives less favorably disposed to minority interests, contributing to an aggregate decline in the responsiveness of the House (Lublin, 1997). In short, not only may direct issue representation (i.e., correspondence between constituents’ policy preferences and a legislator’s policy activism) not depend on descriptive representation (i.e., correspondence on demographic traits), but efforts to secure descriptive representation may ultimately undermine aggregate issue representation. ____________ 3In addition to questions of legislative activism, Swain (1995) and Canon (1999) also consider racial and ethnic differences in the provision of constituency service. 2 After more than a decade of research, scholarly consensus on these questions remains elusive. However, there is broad agreement that minority representation is significant primarily as a guarantor of continued minority influence on the legislative process. This scholarly consensus has produced a body of research that focuses narrowly on legislative behavior and policy outcomes, to the exclusion of the other benefits or costs that may emanate from minority representation. The merits of minority representation and, by extension, the merits of efforts such as race-conscious districting to secure greater minority representation generally have been reduced to a question of policy responsiveness. If white officeholders were found to be equally likely to advocate for minority policy preferences, African-American and Latino officeholding would become less significant. However, such a conclusion may be premature. For the community activists, citizens’ groups, and voting-rights’ attorneys who have worked on redistricting and have pressed for the strict enforcement of the VRA, the creation of majority-minority electoral districts is widely viewed as more than a vehicle for progressive legislation.4 These advocates also believe that majority-minority districts can pull the Latino and AfricanAmerican communities into the political process. In particular, raceconscious districting and the subsequent election of minority representatives may “prime the pump of [minority] voter participation” and “reverse the historic trend” of lower voter registration and turnout (Los Angeles Times, August 26, 1990; see also New York Times, April 12, 1992; Brischetto, 1998). In California, the possibility of greater minority voter participation is especially significant. To many observers, chronically low voter turnout among Latinos is one of the most striking features of the state’s political landscape. As one journalist remarked, “Latino electoral power [exists] more in theory than in fact” (Scott, 2000). Latino leaders have identified this absence at the ballot box as the chief obstacle to real political influence. As long as Latinos remain on the margins of the electoral ____________ 4The redistricting process involves the redrawing of political boundaries (e.g., Assembly, Senate, congressional lines) according to population and other criteria. Redistricting typically accompanies the decennial reapportionment, when congressional seats are redistributed among the various states to reflect population shifts. 3 process, there will be a disjunction between their status as the state’s largest minority group and their actual influence on state politics. With public policy in California increasingly dictated by the “new populism” (Cain, 1992) of the initiative process, rather than by deliberation and compromise in the state legislature, low Latino voter turnout translates directly into limited influence on issues as important and varied as school vouchers, campaign finance, environmental regulation, and affirmative action. If majority-minority electoral districts can lay the groundwork for a more engaged Latino electorate, advocates argue, they may also bring the political dynamics and policy priorities of the state more in line with its demographic trends. These expectations have been met with some skepticism. Because the majority-minority districts served by most minority members of Congress typically have low aggregate turnout rates, many have concluded that neither race-conscious districting nor minority representation has made voters out of previously unengaged minority Americans (Duncan, 1990, 1993a, 1993b; Donovan, 1992). In the 1998 midterm elections, for example, one majority-minority district in Los Angeles County cast fewer than 70,000 ballots in the congressional race—less than half the number of ballots cast in a neighboring district with an Anglo majority (California Journal, 1998).5 This disparity hardly conforms to the small-scale revolution in electoral politics envisioned by advocates. Furthermore, it is unclear what minority districts and minority representation may mean for the willingness of nonminority constituents to remain involved in politics. Neither advocates nor skeptics have offered more than anecdotal evidence or journalistic speculation to bolster their claims regarding the significance of race-conscious districting and minority representation for political participation. Moreover, political scientists have failed to bring their customary empirical rigor to this question of the intersection of representation and political engagement. There has been no systematic investigation of racial and ethnic differences in voter turnout under ____________ 5Of course, advocates rightfully counter that these disparities are confounded by the social and economic dislocation characteristic of majority-minority congressional districts such as the 30th in southeast Los Angeles County. 4 minority representation or in the majority-minority congressional districts in California. Even after a decade that witnessed historic gains in the number of minority members of Congress, and equally important increases in turnout among Latinos, very little is known about the relationship between the two. Yet as the results from Census 2000 are tallied and California again prepares for congressional redistricting, these issues will resurface in the larger debate over efforts to secure greater political representation for Latinos and African Americans in the state. An assessment of the 1990s may afford a fuller picture of the political dynamics in majority-minority districts and thus facilitate a more informed discussion in 2001. This report offers a framework for assessing the relationship between majority-minority districts and minority representation, and voting participation among Latinos, African Americans, and Anglos in California. This report does not take a position in the debate over issue representation. Nor does it suggest that nonpolicy considerations be placed above questions of policy responsiveness as the merits of districting arrangements are evaluated. Instead, the report addresses the following specific questions to illuminate the potential nonpolicy implications of minority districts and minority representation: • How does the propensity to vote vary across racial and ethnic groups? How does it vary across congressional districts in California? • Does the level of Latino, African-American, and Anglo voter participation vary with the racial and ethnic balance of a congressional district? If so, how? • Do voter participation rates in the majority-minority congressional districts served by minority members of Congress differ from the rates observed in the majority-minority districts represented by Anglo legislators? If so, how? • What, if any, difference exists between Latinos and African Americans in the relationship between voter turnout and type of congressional district? The analysis provides empirical support for the claim that minority representation and majority-minority districts are associated with greater 5 involvement in electoral politics among Latinos and African Americans.6 The results suggest that the provisions of the VRA (namely, the creation of majority-minority electoral districts and the subsequent election of minority legislators) go hand in hand with a more dynamic political life and, by extension, an electorate that more closely mirrors the demographic makeup of the state. As such, this research provides a basis for assessing the implications of future efforts to increase minority representation. ____________ 6Throughout this report, I will refer to the relationships among voter turnout, minority districts, and minority representation as just that—a “relationship.” I will avoid the use of terms such as “effect” and “impact,” since the direction of causality is difficult to establish conclusively. However, the evidence favors an interpretation that places minority districts and minority representation causally prior to minority voter participation. 6 2. Majority-Minority Districts and Minority Representation The Voting Rights Act and Its Mandates The Voting Rights Act of 1965 is often cited as one of the most significant pieces of civil rights legislation passed in our nation’s history. Originally designed to give southern blacks access to the voting process by temporarily taking away from certain states the right to determine their own voting procedures, the VRA has been reshaped into a potent tool to expand opportunities for minority representation nationwide and at every level of government. When signed into law by President Lyndon B. Johnson, the VRA (PL 89-110) was narrowly concerned with eliminating such direct obstacles to black political participation as poll taxes and discriminatory literacy tests. The central parts of the measure are Section 2 and Section 5. Section 2 reiterates the guarantees of the 15th amendment, prohibiting any state or political subdivision from adopting voting practices that “deny or abridge the right of any citizen of the United States to vote on account of race or color.” Section 5, imposed only on “covered” jurisdictions with a history of past discrimination, requires Justice Department preclearance of changes in any electoral process or mechanism.1 In 1970, 1975, and 1982, Congress renewed and expanded the provisions of the Voting Rights Act. Three changes have particular relevance for California: ____________ 1“Covered” jurisdictions were originally defined as states in which less than 50 percent of the voting-age population either registered or voted in the 1964 presidential election and that had various discriminatory prerequisites for voting as of November 1, 1964. 7 First, in 1975, the special protections of the act were expanded to include not only African Americans but also Hispanics, Native Americans, Asians, and Alaskan Natives. Second, in 1982, Congress amended Section 2 of the law to stipulate that members of a protected class must have an equal opportunity “to participate in the political process and to elect representatives of their choice” [emphasis added]. Thus, the law was no longer meant to apply strictly to discriminatory practices; it was also to serve as a vehicle for voting mechanisms that could increase minority representation (New York Times, February 14, 1994). The intention was to ensure that “protected” racial and ethnic minorities had a greater voice in governing the country. Subsequently, the Supreme Court issued a ruling (Thornburg v. Gingles) that was widely interpreted to mean that states must create majority-minority districts wherever possible to comply with Section 2 as amended. The Justice Department adjusted its preclearance criteria accordingly. The drawing of new majority-minority electoral districts (from city council districts to congressional districts) became central to the strategy for advancing minority representation. Last, the list of “covered” jurisdictions, at first narrowly targeted to include just the six states of the Deep South (Alabama, Georgia, Louisiana, Mississippi, South Carolina, and Virginia), has continually evolved with various “bailouts” and additions.2 Currently, 22 states have covered jurisdictions, including four counties in California (Merced, Yuba, Kings, and Monterey) with a history of low Latino registration. For these four counties, reapportionment of electoral districts after the 2000 Census will require Justice Department approval. Majority-Minority Districts in the 1990s Congressional efforts to broaden the mandate of the VRA bore fruit in the creation of dozens of new majority-minority congressional districts during the 1990 redistricting cycle. In the aftermath of the 1990 ____________ 2States or jurisdictions can “bail out” from coverage if they can prove the absence of discriminatory practices. 8 reapportionment, the total number of districts nationwide with AfricanAmerican or Hispanic majorities doubled, prompting political scientist David Lublin to characterize the process as “the greatest upheaval in congressional district boundaries since the Supreme Court forced states to minimize population deviations between districts in the mid-1960s” (Lublin, 1997). In California, where reapportionment following the 1990 Census added seven new congressional districts to the state, the new districting plan—developed by a court-appointed panel of retired judges after Governor Wilson failed to reach a compromise with the legislature— placed the state first in the nation in the total number of majorityminority districts (13) and second only to Texas in the number of districts with a Hispanic majority (six, up from three in the 1980s). As was the case in 1981, California’s 1991 congressional districting plan may have been influenced as much by the preferences of the powerful Speaker of the Assembly Willie Brown as it was by pressure to comply with the VRA mandates.3 Table 2.1 presents details on the number of majority-minority districts created nationwide after 1990. In addition to the six congressional districts with Latino majorities, California’s plan created three districts where Latinos and African ____________ 3Since the 1980s, the majority of voting-rights litigation in the state has been concentrated at the local level, where plaintiffs—supported by organizations such as the Mexican-American Legal Defense and Education Fund—have aggressively challenged the use of electoral arrangements that they view as discriminatory in elections to city councils, county boards of supervisors, school boards, and other political offices (Bucy, 1991; Erie, Brackman, and Ingram, 1993; Scott, 1998; Guerra, 1998; Brackman and Erie, 1998). These efforts have enjoyed considerable success. “By the early 1990s,” Brackman and Erie (1998) observe, “the threat and reality of voting rights litigation were largely responsible for the presence of three Latinos on the Los Angeles City Council and one on the County Board of Supervisors.” The same might be said of minority representation in cities such as Watsonville, where a 1987 court challenge to the city’s at-large electoral system eventually led to the district-based election of three Latino council members. Pressure to comply with the expanded mandates of the VRA continued to influence local districting arrangements throughout the 1990s. In fact, in at least one city, Santa Maria, voting-rights litigation initiated at the start of the decade remains unresolved even as we enter the 2001 redistricting cycle. 9 Table 2.1 Number of Majority-Minority Districts After 1990 Redistricting, Selected States Majority- Majority-Black State Hispanic Districts Districts Alabama — 1 Arizona 1— California 6 — Floridaa 2 3 Georgiaa — 3 Illinois 13 Louisianaa — 2 Maryland — 2 Michigan — 2 Mississippi — 1 Missouri — 1 New Jersey — 1 New Mexico — — New York 2 3 North Carolinaa — 2 Ohio —1 Pennsylvania — 2 South Carolina — 1 Tennessee — 1 Texasa 71 Virginia — 1 Multi-Ethnic Districts — — 7 — — — — — — — 1 1 2 — — — — — — 1 — Total MajorityMinority Districts 1 1 13 5 3 4 2 2 2 1 2 2 2 5 2 1 2 1 1 9 1 Total 19 31 12 62 SOURCES: Congressional Quarterly, Inc. (1993); Duncan (1993c). NOTE: Multi-ethnic districts are districts where the majority of the population is not non-Hispanic white and neither blacks nor Hispanics constitute a majority. aSupreme Court challenges eventually led to the elimination of one majority-black district in Florida, Louisiana, North Carolina, and Texas; two majority-black districts in Georgia; and one majority-Hispanic district in Texas. Americans together constituted a majority of the population. And in four congressional districts, Latinos, Asians, and blacks made up a multiethnic majority. In no districts in California do African Americans or Asians, alone, account for the majority of the population. With three exceptions (Oakland’s 9th, San Jose’s 16th, and the Central Valley’s 10 Anglo majority Latino majority Black-Hispanic majority Multi-ethnic majority Figure 2.1—Majority-Minority Districts in California 20th), these majority-minority districts are concentrated primarily in and around the Los Angeles County area.4 (See Figure 2.1.) According to 1990 Census data, 51 percent of all African-American adults in California live in a majority-minority district (60 percent of these adults live in just the three districts with combined black-Hispanic majorities). These same data estimate that 46.5 percent of Latino adults live in majority-minority districts; 60 percent of these adults are concentrated in the six districts with Latino majorities. ____________ 4At the time these district lines were drawn, Los Angeles County was home to 44 percent of the state’s Latino population (Pachon, 1998). Today that percentage is estimated to be 41 percent (California Opinion Index, 2000). 11 The 13 majority-minority districts in California are a diverse group. They vary tremendously not only in their racial and ethnic balance but also in their socioeconomic conditions. At one end of the spectrum are districts such as the 33rd, in “the dreary center of the Los Angeles Basin” (Green, 1995), where noncitizens significantly outnumber registered voters and both unemployment and crime are relatively high. The 35th District faces similar levels of economic dislocation. At the other end of the spectrum are districts such as the 34th, centered in West Covina and the suburbs east of Los Angeles County, which has the distinction of being one of the most racially and ethnically diverse middle-class districts in the state. This group also includes Orange County’s 46th District, which, in the last 15 years, has changed from a largely white area to the home of the nation’s largest concentration of Vietnamese Americans. And, of course, among the oldest multi-ethnic districts in the state is the East Bay’s 9th, a district that includes both struggling Oakland and the more affluent city of Berkeley. The 1990s reapportionment gains extended well beyond federal legislative boundaries to include new majority-minority electoral districts at the state and local levels as well. For example, the plan developed by the court-appointed panel also included 22 Assembly districts and seven state Senate districts where Asians, Latinos, and African Americans, collectively, accounted for the majority of the voting-age population.5 At the local level, where precise summary statistics are difficult to come by, the increase in the number of majority-minority electoral districts— particularly in the most populous cities and counties in the state—was significant enough to lead one observer to conclude that the “1990s reapportionment was the most favorable reapportionment for the Latino community in California’s history” (Guerra, 1998). Minority Representation in the 1990s The creation of majority-minority districts under the aegis of the VRA has contributed to substantial gains in the number of African ____________ 5An additional four Assembly and five state Senate districts have majority-minority total populations but not majority-minority voting-age populations. 12 Americans and Latinos elected to the U.S. House of Representatives, from California and elsewhere. Figure 2.2 demonstrates that the growth in the number of minority legislators has closely tracked the increase in the number of districts with black or Hispanic majorities. The “upheaval” of redistricting in the 1990s, in particular, made the House much more racially and ethnically diverse. (Note the steep rise after 1990 shown in Figure 2.2.) The freshman class of 1992 included 16 African Americans and eight Latinos, all but five of whom were elected from newly created majority-minority congressional districts. So dramatic was the increase in the number of minority legislators that one journalist concluded that the new diversity would represent “a big step in the unfolding of a more democratic and perhaps more tolerant era in elective representation” (New York Times, November 2, 1992). One freshman legislator predicted that “the dialogue will open and issues that members [of Congress] did not have a sensitivity about before will be discussed” (New York Times, November 2, 1992). 45 40 Black members of Congress 35 Majority-black districts Majority-Hispanic districts 30 Latino members of Congress Number 25 20 15 10 5 0 1962 1966 1970 1974 1978 1982 Election year 1986 1990 1994 1998 SOURCES: Data on majority-black and majority-Hispanic districts are from Bureau of the Census (1961, 1963, 1971, 1973, 1983, and 1993). Additional district data and data on minority representatives are from Congressional Quarterly, Inc. (1993, 1996, and 1997); Duncan (1993c, 1995); and Ehrenhalt (1983). Figure 2.2—Minority Districts and Minority Representatives Nationwide, 1962–1998 13 In California, all of the African-American and Latino House members during the 1990s hailed from majority-minority congressional districts. However, the increase in the number of minority legislators in the immediate aftermath of the 1990 redistricting cycle was not as dramatic as that observed elsewhere in the country. Although the state’s new congressional map doubled the number of Hispanic-majority districts, a Latino congressman subsequently represented only one of these three new districts. Observers blame low registration and turnout among Latino residents for the failure to capitalize on the opportunity for minority representation in two of the three new Hispanic-majority districts. Later in the decade, the California delegation gained one additional Latino lawmaker when Loretta Sanchez was elected in 1996— and re-elected in 1998—to represent the 46th District, a majorityHispanic-and-Asian congressional district in Orange County. The racial and ethnic balance of state and local governments also shifted in the aftermath of the 1990s reapportionment, with more favorable implications for Latinos than for African Americans. Within one election cycle (1992-1993), Latino representation in the 300 most significant elective positions in the state had increased by almost 50 percent (Guerra, 1998).6 (By 1997, Latino representation had doubled from its 1989 level.) Some of the greatest gains were in the Assembly where the number of Latino legislators increased from four in 1990 to seven following the 1992 election (eight when Cruz Bustamante’s victory in a 1993 special election is included). The number of Latino held seats in the legislature continued to increase dramatically in the 1990s (to 17 Assembly members and seven senators by 1998) despite, as one observer astutely noted, “only slow growth in the number of Latino voters” (Sacramento Bee, November 30, 2000). At the local level, Latino ____________ 6Guerra (1998) identifies the 300 most significant elective offices (out of 10,000 total) in the state on the basis of “the number of constituents, the budgets under their control, and the strategic position of the office given its perceived prestige by individual officeholders.” These 300 positions include the following: constitutional officers (governor, lieutenant governor, attorney general, etc.); board of equalization; U.S. House and Senate; California Assembly and Senate; board of supervisors for the seven most populous counties; and city council for the seven most populous cities. 14 representation on the city councils of the seven most populous cities in the state increased from five council positions (out of 76 total) in 1989 to 11 after the 1993 municipal elections (Guerra, 1998). (This included new Latino-held seats in San Diego, Long Beach, and Oakland, where Latinos previously had had no representation.) For African Americans, who began the 1990s somewhat overrepresented in elected office (7 percent of the state population but 10 percent of the 300 most significant elective offices), the 1990s reapportionment did not contribute to gains in representation at the state or local level (Guerra, 1998). In fact, by 1996, African Americans held three fewer seats in the state Assembly than they had in 1990. (Each of those three African-American Assembly members was replaced by a white legislator.) 15 3. Minority Voter Participation Trends in Minority Voter Participation Low rates of voter registration and turnout characterize minority political behavior in California, with Latinos in particular remaining largely on the sidelines of electoral politics. Each of the statewide elections in the 1990s witnessed not only low overall turnout (especially compared to the participation “heydays” of the 1950s and 1960s) but also a pronounced difference in the demographic composition of the state’s adult population and its pool of voters (see Table 3.1). For Table 3.1 Racial and Ethnic Composition of the Electorate and Adult Population, 1990–1998 (in percent) Total turnouta 1990 1992 1994 1996 1998 41 54.5 47 52.6 41 Electorate Non-Hispanic white 81 79 78 77 74 Latino 4 10 9 11 13 Black 8 6 7 67 Asian 4 4 6 46 Adult population Non-Hispanic white 63 59 59 57 57 Latino 21 24 24 27 25 Black 7 7 7 67 Asian 9 9 10 10 11 SOURCES: California Opinion Index (1990, 1993, 1995, 1997, 1999); figures for 1990 “Electorate” are from Baldassare (2000). aTotal turnout is measured as a percentage of the state’s citizeneligible population. 17 example, in November 1998, non-Hispanic whites accounted for 57 percent of the state’s adult population and 74 percent of the electorate. At the same time, Latinos constituted 25 percent of the California adult population but accounted for only 13 percent of the ballots cast. African Americans are the only major racial group (minority or nonminority) whose political clout is on a par with its population numbers. Even so, African Americans participate at lower rates than do non-Hispanic whites (Reyes, 2001). For Latinos, the disparity at the ballot box in 1998 actually marked a significant improvement over the first gubernatorial election of the decade, when they accounted for only 4 percent of the electorate. This participation gap is widely recognized as the most fundamental barrier to minority political empowerment in the state. A number of factors contribute to “the [lag] in commensurate influence at the ballot box” (Erie, Brackman, and Ingram, 1993). Constraints on Minority Voter Participation Social scientists have devoted decades of research to identifying the influences on an individual’s decision to participate (Verba and Nie, 1972; Brody, 1978; Wolfinger and Rosenstone, 1980; Teixeira, 1987; Conway, 1991; Verba, Schlozman, and Brady, 1995). Whether concerned exclusively with voting or with participation more generally (including campaign activity, for example), the findings have consistently highlighted the significance of three individual-level factors: age, education, and income. The positive relationship between turnout and these measures of socioeconomic status may well be the most robust finding in the field of political science. The influence of education on participation is thought to derive both from education’s indication of social position and from the tangible resources it provides. “Education imparts citizenship values,” argue Rosenstone and Hansen (1993), “and so the educated . . . [are] composed of people . . . who have been socialized to follow politics and to value and reward political involvement.” Formal schooling also imparts the knowledge and skills necessary to understand and follow political campaigns and to negotiate the bureaucratic requirements of registration and voting. 18 Similarly, income facilitates participation both by what it indicates (social position) and for what it allows (the ability to bear the material and opportunity costs of political involvement). For low-income adults, the money, time, and energy spent making ends meet not only limit the resources (e.g., leisure time) available for political pursuits but also provide payoffs that are more immediate and valuable than the potential benefits of investing in electoral politics (Wolfinger and Rosenstone, 1980). As one journalist put it, in trying to explain low voter turnout among poor adults, “the notion that their lives might be made better by casting a ballot is as fanciful as winning the lottery” (Los Angeles Times, November 28, 1994). Finally, older adults are typically more involved in politics than younger adults. Turnout is highest, in fact, among those who are age 65 and older. As people age, they are thought to become more familiar with the political process, more knowledgeable about politics, and more attached to the parties and their candidates.1 This experience both facilitates and encourages electoral involvement. As pollster Richard Maullin notes, “politics just isn’t as critical” for the young (Lopez and Wahlgren, 1994). On each of these measures, African Americans suffer from a comparative disadvantage compared to non-Hispanic whites in the state. The African-American community as a whole has higher concentrations of young, undereducated, and low-income adults. In 1997, 73 percent of the African-American population were under age 45 compared to 62 percent of non-Hispanic whites (California Department of Finance, 1999). In 1998, the four-year high school dropout rate in California was 17.4 percent for African-American students in the state’s public schools but only 7.5 percent for non-Hispanic whites (Education Data Partnership, 2001). ____________ 1An alternative hypothesis to explain the positive correlation between age and participation is generational, where the differences in participation are artifacts of the socializing experiences of each generation. “People who [came] of age at critical turning points that define political allegiances [Great Depression, New Deal] participate more in electoral politics than people who reach maturity in more ordinary times” (Rosenstone and Hansen, 1993). Generational hypotheses have not fared well in the literature. 19 The Latino population is the state’s youngest, with an average age of 24 (Lopez and Wahlgren, 1994). Not only are many Latinos too young to vote but, among those who are age-eligible, a significant number (36 percent) fall into the age 18–29 cohort that is the least likely to participate (California Opinion Index, 2000). Only one in three Latinos is over age 40. Levels of educational attainment are also low. Forty-four percent of Latino adults are not high school graduates and only 9 percent hold college degrees (California Opinion Index, 2000). In 1998, the four-year dropout rate for Latino public school students was 16.3 percent (Education Data Partnership, 2001). Finally, many Latinos have low incomes. Forty percent of Latino adults report annual household incomes of less than $20,000 (California Opinion Index, 2000). Thus, to a great degree, low rates of minority registration and voting reflect the socioeconomic bias in electoral participation. Even organized registration and voting drives can fail to overcome these barriers: Voter registration campaigns targeted at Latinos in the state have been more successful in stable, working-class suburbs than in the poorer inner-city neighborhoods where many Latinos live (Brackman and Erie, 1998). The Latino community also includes a large noncitizen population. Pachon (1991) speculates that “over 80 percent of the difference in voting between Latinos and black Americans can be explained by noncitizenship among Latino adults.” Estimates place the number of noncitizens between 41 percent and 52 percent of all Latino adults. (The decennial Census does not provide detail on the racial and ethnic makeup of the noncitizen population.) These concentrations create what Brackman and Erie (1998) describe as “an inverted pyramid of electoral participation,” in which “a small minority voting universe rests upon a large minority population base.” Compounding these socioeconomic barriers, some have argued, has been a political environment in which the two major parties for many years spent little time trying to energize Latinos (Maharidge, 1994). As one political observer describes it, Latinos in California have been dismissed as “an occasionally raucous but mostly non-voting constituency” (Frayne, 1998). During the 1996 elections, for example, Republicans in the state legislature backed legislation to distribute official 20 election materials in English only, while Democratic pollsters in one Assembly district hung up the telephone when they reached non-Englishspeaking households—choosing to ignore the opinions of residents they presumed did not vote (Orange County Register, October 22, 1998). Guided by an election calculus that discounted Latinos as a “low propensity ‘expendable’ constituency,” campaign organizers often did not invest the resources necessary to mobilize this population (Scott, 2000). By the 2000 presidential election, there was a noticeable shift in political strategy, with both major party candidates assiduously courting Latino voters. There are some signs that Latino political participation may be on the rise as the Latino proportion of the electorate increases (see Table 3.1). The number of Latino voters increased from 1.35 million in 1990 to 2.35 million in 2000, accounting for most of the 1.1 million net increase in the state’s registered voter rolls over the decade (California Opinion Index, 2000). Much of this growth has taken place in the last six years; nearly half of the Latino adults who were registered to vote in time for the March 2000 primary had registered after 1994. Analysts generally attribute the participation upswing to furor over the passage of Proposition 187, the Illegal Immigration Act. Perhaps not surprisingly, naturalized citizens accounted for a larger share of the voters who registered after 1994 than of those who registered before 1994. Some Hypotheses About Majority-Minority Districts and Minority Voter Participation Advocates of minority representation regard the redistricting process and minority representation as instrumental to the political mobilization of the Latino and African-American communities. This view presumes that political engagement is as much a response to the political environment and the opportunities it is perceived to present as it is a function of individual resources (e.g., education, income). Political scientists, for their part, have articulated similar models of political behavior (although they have not extended them to the study of minority representation), increasingly moving beyond the strict individualist 21 perspective that has dominated the research for decades to consider the influence of political context on the likelihood of voting. The scholarly mission has been to “reconstruct a compelling account of political life” by recognizing the contingent nature of political choices and by identifying the aspects of a political environment “that make people accessible and amenable to the appeals of political leaders” (Huckfeldt and Sprague, 1993). In practice, the environmental constraints and opportunities can be as varied as registration laws that ease or complicate political participation (Nagler, 1991; Timpone, 1998), salient political campaigns that inspire activism (Tate, 1991), or public policy reforms that spur record rates of naturalization (Shaw, de la Garza, and Lee, 2000; Washington Post, April 12, 1995). This logic could be extended to ask whether the political status of African Americans and Latinos correlates with their levels of political engagement. Findings from research on minority officeholding at the local level suggest that this kind of “empowerment” and political participation tend to go together (Bobo and Gilliam, 1990; Gilliam, 1996; Gilliam and Kaufmann, 1996). In the most complete research to date, Bobo and Gilliam (1990) found that African Americans in cities with black mayors are more active than their white counterparts of comparable socioeconomic status and more active than African Americans in cities with white mayors. This pattern suggests that minority representation in local government is significant not only for its implications for public policy but also for its ability to bring minorities in from the political margins. At least two plausible hypotheses lead us to predict a similarly positive relationships among majority-minority districts, minority representation, and levels of political participation. The Psychology of Empowerment Voting-rights advocates link the participatory benefits of the redistricting process to its role in “creating a climate of inclusion” and “showing [minorities] they can elect their own representatives” (New York Times, April 12, 1992 ). Likewise, Bobo and Gilliam (1990) conclude that minority officeholding at the local level influences black political participation by contributing to a more trusting and efficacious 22 orientation toward politics (see also Abney and Hutcheson, 1981; Howell and Fagan, 1988) and by greatly increasing black attentiveness to and interest in political affairs. Thus, majority-minority districts and minority representation may increase political confidence among Latinos and African Americans; this confidence, in turn, encourages greater political engagement. (Conversely, these conditions would be expected to lead non-Hispanic whites to participate less in political life.) This “reservoir of psychic benefits,” as Gilliam and Kaufman (1996) describe this effect, may derive from purely symbolic considerations—for example, the group pride stoked by a “brown face in a high place,” as one activist phrased it—or from an appraisal of policy responsiveness (Lopez and Wahlgren, 1994). Targeted Mobilization An alternative hypothesis is that neither psychological attachments nor the tangible gains associated with minority representation contribute to greater political engagement. Rather, heightened engagement may be the by-product of a new electoral dynamic in which Latino and AfricanAmerican communities become the targets of aggressive mobilization. These communities may benefit when minority political aspirants, guided by an election calculus that defines Latinos and African Americans as their natural constituencies, make special efforts to engage and court this core of voters. Add to this the strategy of independent political organizations, such as the Southwest Voter Registration Education Project, the NAACP, and the National Association of Latino Elected Officials, who, in an attempt to maximize the “bang for the buck,” focus their minority voter registration and turnout efforts in the districts with the highest concentrations of Latinos and African Americans. As research on mobilization has demonstrated, people are more likely to participate in politics if they are asked (Verba, Schlozman, and Brady, 1995), or if their stake in politics is made evident to them (Rosenstone and Hansen, 1993). Majority-minority districts may facilitate and encourage this outreach. If so, higher turnout in majorityminority districts and under minority representation is at least partly the result of the strategic calculations of political entrepreneurs. 23 An Artifact of the Redistricting Process There is a third, but less plausible, reason to predict a positive relationship between majority-minority districts and minority voter participation. A pattern of heightened minority political participation may simply be an artifact of the redistricting process itself, especially if the majority-minority districts in the state have been deliberately constructed around high-propensity minority voters. Thus, voter turnout affects the likelihood of being in a majority-minority district rather than the other way around. If so, there is very little to be said about the participatory benefits of race-conscious districting. Even if voting propensity affected districting, which, in turn, affected voting propensity, that pattern would be difficult to isolate and quantify. Reversing the causal arrow casts doubt on the utility of any empirical investigation into this relationship. To evaluate the merits of this hypothesis, I contacted the Mexican American Legal Defense and Education Fund (MALDEF), which has played a very prominent role in redistricting in California. Given their interest in maximizing the opportunities for Latino officeholding, one would expect MALDEF to be particularly mindful of Latino voter turnout rates and particularly supportive of a strategy that explicitly builds majority-minority districts around high-propensity voters. In an interview with Denise Hulett, a senior litigator who has been with the organization for 17 years, I learned that MALDEF has not pursued such a strategy—if only because of lack of resources.2 Indeed, the chief concern for MALDEF is to avoid creating “phantom” majority-minority districts—districts in which Latinos constitute the majority of the total population but, because of high rates of noncitizenship, do not make up the majority of the voting-eligible population. As a consequence, these “phantom” districts may be majority-minority on paper but majorityAnglo in practice. The challenge facing MALDEF in their efforts to prevent the creation of these districts is that the Census does not provide citizenship data disaggregated by race and ethnicity. In lieu of citizenship data, MALDEF, during the most recent round of ____________ 2Interview with Denise Hulett on January 9, 2001. 24 redistricting, relied on two-year-old (1988) registration data—with Latinos identified on the basis of Spanish surname—to estimate the potential size of the voting-eligible Latino population. According to Hulett, MALDEF never evaluated actual turnout rates. Their districting proposals were not conditioned on the probability that these registered voters would actually vote. Thus, the organization most likely to embrace and pursue a strategy of targeting high-propensity voters did not, in fact, adopt such a strategy. This fact suggests that a relationship between majority-minority districts and political participation is less likely to be an artifact of the redistricting process itself. At least two plausible hypotheses exist, then, that would lead us to predict a positive relationship between race-conscious districting and minority political engagement. Although each attributes the relationship to a different causal mechanism, both suggest that minority districts and minority representation might compensate for the socioeconomic barriers that suppress political participation. However, despite the existence of plausible hypotheses and the importance of political participation, there has been no systematic effort to establish whether there is a relationship between race-conscious districting and political participation. In the following chapters I attempt to do just that. Although the nature of the data I will use precludes any definitive identification of the precise causal mechanism (i.e., attitude change or targeted mobilization), it does allow evaluation of whether a relationship exists at all—and if so, whether it is positive or negative. If the evidence reveals a pattern of distinct political behavior in majority-minority districts and under minority representation, then future research should clarify the factors that contribute to that relationship. 25 4. Data and Methodology The key to clarifying the relationship between race-conscious districting and minority representation, on the one hand, and political participation among Latinos, African Americans, and Anglos, on the other, is to identify (1) the major influences on voting participation, (2) the expected rates of voting participation for various demographic groups given these influences, and (3) the degree to which the actual rates of voting participation differ from expectations wherever racial and ethnic minorities are able to “elect representatives of their choice.” If raceconscious districting and the minority representation that may result compensate for the socioeconomic barriers that suppress minority political participation, higher minority voter turnout rates should be seen in majority-minority districts than would be expected given the demographic profile and resources available. This analysis assembles data from multiple sources and applies a variety of statistical methods to examine voting turnout patterns in every congressional district and precinct in the state. Sample Year This analysis draws on data from the November 1994 congressional elections. Widely regarded as the mark—if not the harbinger—of a political revolution, the 1994 midterm elections shifted the partisan balance of Congress as Republicans assumed majority status. In California, 1994 is best remembered as the year of Proposition 187, the controversial (and ultimately successful) ballot initiative that cut off illegal immigrants from all but emergency health care services and removed their U.S.-born children from public schools. The emotioncharged debate over the proposition spilled over into virtually every campaign for public office in the state. Martis and Block (1994) likened Proposition 187 to Proposition 13, an initiative that “[galvanized] the public to such an extent that it [became] the engine that [drove] an entire 27 election.” The initiative campaign, embraced by then-Governor Pete Wilson, proved to be a watershed moment, igniting a “flame of civic engagement among California Latinos” and “driving tens of thousands of freshly minted citizens into the arms of the Democratic Party” (Scott, 2000). As a result of Wilson’s advocacy on behalf of Proposition 187, and the subsequent surge in naturalization and voter registration among Latinos, some have deemed him the “godfather of Latino California” (Frayne, 1998). Three considerations influenced the decision to use data from the November 1994 elections. First, I chose to focus on voting behavior during midterm elections, when the political dynamics of congressional races—no longer overshadowed by presidential campaigns—figure relatively more prominently. Second, the 1994 midterm election reflected the enormous reapportionment and redistricting changes that went into effect in 1992. Compared to 1990, the 1994 election offers a larger sample of congressional districts (45 vs. 52) and a larger number of majorityminority districts served by minority and Anglo legislators. Finally, the need to leverage available Census data in the analysis of voter participation significantly constrained the final choice of midterm election. In light of the population changes witnessed in California over the course of the 1990s, the data collected in the 1990 Census of Population and Housing—the most current data available until Census 2000 is publicly released—offer a relatively more accurate picture of demographic conditions in 1994 than in 1998. Although the currency of the 1998 midterm elections holds some appeal, the use of 1990 Census data to evaluate voter turnout patterns in that election would require a level of extrapolation that could compromise the validity of the analysis and subsequent conclusions. In lieu of those extrapolations, I rely on data from the November 1994 elections. By doing so, I ensure that the available Census data remain analytically useful. Although November 1994 is a more distant election, the data retain value for understanding the relative differences in voter turnout between districts created to have a majority-minority population and those created to be majority-Anglo. 28 Relying on data from 1994 presents at least one important complication: the potentially confounding effect of Proposition 187 mobilization. Of particular concern are the implications for our interpretation of Latino political behavior in that election. As Pachon (1998) observed, the Latino community “did not see [the “Save Our State” Initiative] as simply an anti-illegal alien measure but as an antiLatino measure.” The initiative has been widely credited with politicizing numerous segments of the Latino population and doing more “in three months to . . . energize [that community] than various outreach efforts had managed in fifteen years” (Block, 1998). The weeks and months leading up to the November elections were dominated by demonstrations, teach-ins, and mass rallies, as Latinos (particularly Latino youth) took to the streets (Los Angeles Times, October 21, 1994, November 8, 1994; Pachon, 1998). Additionally, there are accounts of numerous targeted mobilization efforts: The Sunday before the election, Archbishop Roger Mahoney of Los Angeles encouraged all pastors to instruct their parishioners to vote against Proposition 187; Democrats set up voter registration tables outside citizenship ceremonies in Southern California; the Southwest Voter Registration Education Project spent $100,000 on registration and get-out-the-vote efforts in nine California counties with sizable Latino populations; young Latinos organized precinct walks to register voters; and community-based organizations redoubled their efforts to urge naturalization, registration, and turnout (Lopez and Wahlgren, 1994; Los Angeles Times, October 21, 1994, November 8, 1994, December 4, 1994; Schockman, 1998; Guerra, 1998). How much of this activity carried over to election day is unclear. Overall turnout among all groups increased by six percentage points over the previous midterm election (California Opinion Index, 1995). Additionally, a Los Angeles Times exit poll found that about one-third of white and black voters and nearly half of Latino voters identified Proposition 187 as one of the top two reasons for turning out to vote (Los Angeles Times, December 4, 1994). By comparison, a November 1993 Field Poll found that immigration ranked 19th out of 28 issues of concern to Californians (California Opinion Index, 1994). However, 29 although the Latino community overwhelmingly rejected the initiative, as well as its most prominent spokesman (Pete Wilson), the high turnout that had been anticipated never materialized (Los Angeles Times, November 9, 1994, November 10, 1994, November 14, 1994, November 28, 1994, December 4, 1994; California Political Week, 1994; California Journal Weekly, 1994; Martis and Block, 1994). Former state Senator Art Torres, who had lost his bid to become state insurance commissioner (and the first Latino statewide elected official in a century), invoked the metaphor often used to describe the politically reticent Latino community, “[t]he sleeping giant was in a coma on November 8” (Los Angeles Times, December 4, 1994). In the election postmortems, commentators and activists offered numerous reasons for Latino absence at the polls: anti-187 mobilization efforts that gained momentum only after official voter registration deadlines had elapsed; insufficient targeted outreach to and education of Latino communities, partly because of lack of resources and partly out of fear of provoking white backlash; and a puzzling lack of interest among working-class Latinos. Where there is agreement, however, is that the campaign against Proposition 187 laid the foundation for a mobilized Latino community, guided by a “new political consciousness” (Los Angeles Times, November 11, 1994); but it was only with the passage of Proposition 187—and fears that anti-immigrant backlash might eventually result in social welfare benefits being denied noncitizens—that Latino naturalization, registration, and turnout gained real momentum (Washington Post, April 12, 1995; Scott, 1998; Baldassare, 2000; California Opinion Index, 2000; San Jose Mercury News, October 15, 2000). “With the electoral and policy ramifications of California’s midterm 1994 election,” Schockman (1998) argues, “the bubble of benign non-involvement through noncitizenship burst.” Real behavior change was visible by the 1996 presidential election: Latino voter registration increased by 28.7 percent over its 1992 levels. That year, 50 percent of citizen-eligible Latinos were registered to vote (California Opinion Index, 1997).1 Baldassare (2000) notes that the “sting” of Proposition 187 continued to galvanize ____________ 1By comparison, 74 percent of citizen-eligible non-Hispanic whites and 67 percent of African Americans were registered to vote. 30 Latinos as they approached the 1998 election. In November 1998, the Latino proportion of the electorate rose to 13 percent, from just 9 percent in the 1994 midterms. The Field Institute estimated that by the 2000 presidential elections, Latinos accounted for 16 percent of the state’s registered voters—as mentioned above, 46 percent of those voters had registered since 1994 (California Opinion Index, 2000). Although Proposition 187 appears to have had its most pronounced effect on the political dynamics of subsequent elections, it is still possible that the campaign had some incremental effect on participation in 1994. The California Journal, for example, noted a 2 percent increase in Democratic registration in one majority-Latino congressional district between the spring and fall of 1994 (California Journal, 1994). This pattern has two implications for the analysis. First, the absolute turnout rates observed in 1994 may not be a reliable benchmark for what to expect generally in any midterm election. Of course, this limitation is not unique to 1994 data but applies to any cross-sectional data drawn from a single election year. The proximate influences on overall turnout vary from year to year. Longitudinal data are better suited for generating estimates of average participation. With cross-sectional data, analysts typically limit themselves to accounting for variation across demographic or socioeconomic groups. That is the objective here; as a consequence, this first implication should not be a major concern. Second, the relative differences in turnout (e.g., across different demographic groups) may be affected if these groups differed systematically in their attention to the Proposition 187 campaign.2 For example, if there is evidence that most of the Proposition 187 mobilization efforts were targeted at Latinos or, more important, were concentrated in majority-Latino districts or initiated by Latino elected officials, that might account for higher rates of turnout among Latinos than among other groups and higher turnout among Latinos in majorityLatino districts than among Latinos in majority-Anglo districts. For that reason, the following econometric analysis not only evaluates turnout ____________ 2As indicated by the results from the Los Angeles Times exit poll, groups differed systematically in the influence of Proposition 187 on their voting participation. 31 patterns separately for each racial group but also incorporates a measure to capture differences in the salience of the Proposition 187 campaign across different districts. Unit of Analysis and Measures In 1994, 13 congressional districts had majority-minority populations. On the eve of the November elections, the California congressional delegation included four Latino legislators representing majority-Hispanic districts and three African Americans representing majority-minority districts with roughly equal proportions of blacks and Hispanics. In addition, the delegation included two Anglo lawmakers representing majority-Hispanic districts and four legislators (one African American and three Anglos) representing multi-ethnic districts. All but one of these legislators—Representative Don Edwards of the 16th District—ran for re-election in November 1994. The data assembled for the analysis are aggregated at the precinct level and consist of three major components: • Voting data — Voting-age population turnout: Total ballots cast in the congressional race as a proportion of the total voting-age population in the precinct. — Registered voter turnout: Total ballots cast as a proportion of the total registered voters in the precinct. — Proposition 187 turnout: Total ballots cast on Proposition 187 initiative as proportion of total ballots cast in precinct. • Census data — Demographic characteristics: Racial and ethnic composition, age distribution in each precinct. — Socioeconomic conditions: Education levels, household incomes, proportion of new residents in each precinct. • Political data — District and representative characteristics: Racial and ethnic balance of district, race and ethnicity of representative, party identification of representative. — Electoral competitiveness: Size of winning vote margin, incumbency, tenure of representative. 32 Table 4.1 Majority-Minority Districts and Minority Representatives in California, November 1994 District 30 31 33 34 32 35 37 9 16 20 26 46 50 % Hispanic 61 58 83 62 30 42 44 12 37 55 52 49 41 % Black 4 2 4 2 40 43 34 32 5 6 6 2 14 % Asian 21 23 4 9 8 6 11 16 20 6 7 12 15 Congressional Representative Xavier Becerra Matthew Martinez Lucille Roybal-Allard Esteban Torres Julian Dixon Maxine Waters Walter Tucker Ron Dellums Don Edwards Cal Dooley Howard Berman Robert Dornan Bob Filner SOURCES: Duncan (1993c, 1995); Green (1995). Race/ Year First Ethnicity Elected Latino 1992 Latino 1982 Latino 1992 Latino 1982 Black 1978 Black 1990 Black 1992 Black 1970 Anglo 1962 Anglo 1990 Anglo 1982 Anglo 1976 Anglo 1992 Several variables warrant further clarification. First, to assess the salience of the Proposition 187 ballot initiative, the data include a measure that calculates Proposition 187 turnout as a proportion of the total turnout in the precinct. This measure is intended to serve as a rough proxy for the proportion of voters who would identify Proposition 187 as one of the primary reasons for turning out that November— voters who may have benefited from targeted mobilization efforts or whose own sense of the issues at stake was strong enough to carry them to the polls. One would expect to observe variability in this measure across precincts—reflecting geographic variability in the incidence and intensity of organized campaigns (e.g., higher ratios in majority-Latino districts than in majority-Anglo districts) and demographic variability in the perception of issue importance (e.g., higher ratios in heavily Latino precincts than in heavily Anglo precincts).3 Because of this variability— ____________ 3It is important to emphasize that the absolute value of this measure almost certainly overestimates the number of people in any given precinct who would rank Proposition 187 as one of their top motivations for voting. However, to the extent that this measure varies across precincts, this variation can be assumed to be proportional to the true variation in the number of people for whom Proposition 187 was especially salient. 33 and the role of Proposition 187 as the “engine” of the 1994 elections, influencing the likelihood of turning out at all—including this measure helps account for any differences in levels of participation across congressional districts with and without African-American and Latino majorities or legislators.4 Second, the Census measures of socioeconomic conditions include a variable indicating the proportion of precinct residents who had moved to the state within the preceding five years. Compared to long-term state residents, recent transplants are less likely to be registered to vote and may be less settled in their communities and less invested in the politics of the state (Verba, Schlozman, and Brady, 1995; Rosenstone and Hansen, 1993). Furthermore, this measure may vary systematically across districts, with greater turnover in areas with higher concentrations of minority residents. Finally, the political data include several measures of electoral competitiveness—an important contextual influence on voter turnout (Gilliam, 1985; Barker, Jones, and Tate, 1999). These measures include an indicator of whether the incumbent legislator is running for reelection and, if so, his or her years of service in the House. The third measure is the size of the vote margin in the November 1994 race. Majority-minority districts are generally assumed to be among the least competitive districts in a state, although it is not clear that they are considerably less competitive than districts drawn to maintain a partisan advantage. Nonetheless, this is a factor that must be considered when evaluating the correlation between majority-minority districts and voter participation. The voting data and the Census data used in the analysis were provided by the Statewide Database (SWDB) at the Institute for Governmental Studies at the University of California, Berkeley.5 I ____________ 4Without controlling for the Proposition 187 effect, the subsequent econometric analysis would most likely suffer from omitted variable bias. The result would be to overestimate (i.e., bias away from zero) the effect of majority-minority districts and minority representation on minority voter turnout. 5I would like to thank Karin McDonald and her staff (especially, Michael Wagaman and Gray Chynoweth) for their generous support in making these data available to me. 34 assembled the political data using information available in Green (1995) and Duncan (1993c, 1995). Although the SWDB is the leading source of voting and demographic data in the state—and the official redistricting database for the state of California—the information provided is not without its limitations. First, the data are aggregated at the precinct level. As a result, these data cannot shed light on the individual calculus governing the decision to vote. For that, survey data would be needed. Available survey data have neither the demographic nor the geographic coverage to permit this kind of analysis. Second, to reconcile demographic data (collected for Census blocks whose boundaries are stable for a decade) and voting and registration data (collected for precincts whose boundaries change with each election), the analysts at the SWDB must aggregate each database component up to a common geographical level and then disaggregate them back down to a smaller, common unit of analysis (in this case, the “registration precinct”). This process of conversion necessarily introduces errors. However, the analytical advantage that comes with having data at the precinct level more than compensates for the resulting loss of precision. Furthermore, the California state legislature is sufficiently confident in the procedure to rely on the SWDB as the chief resource in the 2001 reapportionment and redistricting. Third, the SWDB merges 1990 Census data with 1994 voting and registration statistics. As a result, the voting-age population totals used as denominators in the measure of voting participation may underestimate (in the case of population growth) or overestimate (in the case of population loss) the actual number of potential voters in 1994. To address this shortcoming, I used county-level figures from the California Department of Finance (DOF) to adjust the population totals so that they would reflect recorded growth between 1990 and 1994. The county-level data provided by the DOF are disaggregated by racial and ethnic group and by age, allowing for the calculation of differential growth rates for specific voting-age populations. These growth rates are then applied equally to every precinct within a given county. That is, if 35 the DOF estimates a 7.1 percent increase in the size of the black votingage population in Santa Clara County, then 7.1 percent is the growth rate applied to every precinct in that county. This assumption of equal growth across precincts may be flawed but is preferable to a measure that disregards growth in that four-year period. Fourth, the 1990 Census does not include information on the racial and ethnic makeup of the citizen and noncitizen populations.6 This lack of specificity most significantly affects analysis of Latino voter participation. Because 41 to 52 percent of voting-age Latinos in the state are thought to be noncitizens, the total voting-age population figure available in the Census grossly overestimates the pool of potential Latino voters (Stiles et al.,1998; Pachon, 1998; California Opinion Index, 2000; Reyes, 2001). The 33rd congressional district in southeast Los Angeles County, for instance, is home to such a large concentration of noncitizen Latinos that it has the lowest number of registered voters of any district in the country (Los Angeles Times, September 26, 1996). More generally, Brackman and Erie (1998) observe that in the Latino community, unnaturalized but legal aliens outnumber registered voters. Since the 1990 Census data do not allow identification of only voting-age Latino citizens, this analysis measures Latino voter participation as a proportion of the number of Latinos registered to vote in November 1994. 7 The SWDB estimates the number of Latino registered voters by filtering each precinct’s individual registration files through a Spanish surname dictionary. Although surname dictionaries typically include more people than the target population (for example, this method may capture Spanish-surnamed Filipinos), this method is preferable to one that overlooks such a critical feature of the Latino demographic profile. To avoid confusion, I will frequently remind the reader that Latino voter participation—unlike black and Anglo voter ____________ 6Furthermore, the DOF does not provide estimates of these figures, either for the state as a whole or by county. It provides only counts of the total number of new immigrants annually. 7The lack of consensus on the estimate of Latino noncitizenship rates makes me very reluctant to pick a number and apply it across the board to every precinct in the state. 36 participation—is measured as the proportion of registered voters. If rough comparisons are desired, there is a useful (although imperfect) conversion factor: According to the Southwest Voter Research Institute, 1.75 million Latino adults were registered to vote in 1994, about 46 percent of the citizen-eligible Latino population (Los Angeles Times, November 11, 1994; California Opinion Index, 1997). Multiplying this figure by the Latino registered voter turnout rates allows for some comparison to black and Anglo voting-age turnout rates.8 Finally, as of this writing, the SWDB was not able to provide voting data for the precincts in California’s 47th congressional district. As a result, that district is not included in this analysis. Every other precinct and congressional district in the state is included in the dataset.9 Even with its limitations, the SWDB provides the most comprehensive set of voting and socioeconomic variables available for the state of California—and at the greatest level of geographic detail. Methodology To identify the relationship between voting participation, on the one hand, and minority representation and minority districting, on the other, this research applies a two-stage methodology. The first stage uses a model of ecological inference to estimate the rates of Latino, AfricanAmerican, and Anglo voting participation for each precinct and congressional district in California.10 (For African Americans and Anglos, voter participation is measured as a proportion of the group’s voting-age population; for Latinos, voter participation is measured as a proportion of the group’s registered voters.) Ecological inference is a ____________ 8Keep in mind, however, that the black and Anglo measures are calculated as proportions of all voting-age adults, not taking into account the less than 5 percent of these groups who are noncitizens (Reyes, 2001). To “correct” for this, one could simply divide the turnout statistic by 0.95 for a measure of black or Anglo turnout as a proportion of citizen-eligible adults. 9Appendix Table A.1 summarizes the number of precinct observations in the dataset. 10Appendix B presents more detailed information on the model of ecological inference. 37 method of aggregate data analysis, widely applied in voting-rights litigation, in which the goal is to use available data on the racial and ethnic composition of precincts and overall vote tallies to calculate racespecific turnout rates that cannot be observed directly.11 The results from this first stage of analysis are presented in a series of maps that illustrate the varying levels of turnout throughout the state. These maps allow for an initial assessment of turnout patterns, including the relative differences in turnout across ethnic groups and across congressional districts with and without sizable minority populations. In the second stage of the analysis, the estimated measures of Latino, African-American, and Anglo voting participation are used as dependent variables in a series of regressions that model voter turnout as a function of minority representation and minority districting.12 The econometric analysis takes into account the potentially confounding effects of demographic and socioeconomic conditions (e.g., age distribution, levels of educational attainment, household income, concentrations of new residents), as well as the effect of electoral competitiveness (e.g., incumbency, tenure, vote margin), on voter participation in districts with sizable minority populations.13 With these models, one can evaluate the relative importance of each factor on the observed level of voter participation and distinguish the influence of minority representation and minority districting from the influence of other variables closely associated with it. After estimating each of these models, I summarize the relationship of minority districting and minority representation to voter participation for each racial and ethnic group. To more clearly present the results from the multivariate analysis, I generate estimates of voter participation levels for a hypothetical “average” precinct, using the mean value of all ____________ 11These rates are not observed directly because California, unlike some southern states, does not collect racial and ethnic data on citizens who register and vote. This makes it impossible to know with certainty the exact number of Latinos or African Americans who cast ballots. 12Appendix B presents more detailed information on the econometric model. 13The coding of the independent variables included in the econometric model is summarized in Appendix B. 38 independent variables in the models. These simulations allow me to illustrate the difference in voter turnout to be expected between an average precinct located within a majority-Anglo district and that same precinct located in a district with a majority-minority population. 39 5. Research Findings General Findings As is characteristic in nonpresidential election years, voter turnout in November 1994 was relatively low throughout the state. According to the California Secretary of State’s official Statement of Vote, a total of 8.9 million Californians participated in the general elections, representing about 37 percent of the state’s adult population. (By comparison, 11.4 million people voted in the 1992 presidential election.) Not all of these voters cast ballots in the congressional races in their districts; “roll off,” wherein voters cast ballots in higher office elections but opt out of less prominent races, is a common phenomenon and was a factor in this election. In November 1994, a total of only 7.6 million votes were cast in congressional races throughout the state (California Journal, 1994). The more modest levels of congressional turnout are evident in Figure 5.1, which plots the number of congressional ballots cast as a proportion of each district’s total voting-age population. Only one congressional district, the 10th in eastern Contra Costa and Alameda Counties, had more than 20 percent of voting-age adults casting congressional ballots. In seven congressional districts, less than 10 percent of age-eligible constituents voted in the congressional election. Many of these low turnout districts are concentrated in and around Los Angeles County. This pattern is consistent with exit polls conducted by Voter News Service and the Los Angeles Times, which found that, although Los Angeles County was home to 30 percent of all adults in the state, it accounted for only 24 percent of the total 1994 electorate (California Opinion Index, 1995). The picture improves considerably when only those adults (18 and over) who were registered to vote in 1994 are considered (see Figure 5.2). Among registered voters, anywhere from one-third to slightly more than 41 < 10% 10% to 12% 12% to 16% 16% to 20% > 20% SOURCE: Statewide Database. Figure 5.1—Voting-Age Turnout by District, November 1994 50 percent turned out to vote in the various congressional races across California. Again, voting rates were somewhat lower in Southern California than elsewhere. Racial Differences in Voter Turnout Figures 5.3 through 5.5 summarize the district-level estimates of turnout for whites, African Americans, and Latinos, allowing for an initial look at racial and ethnic differences in participation and variation in behavior across congressional districts.1 ____________ 1Appendix C summarizes the district-level turnout estimates in a series of graphs that also report the standard errors for each of these estimates. 42 35% to 40% 40% to 45% 45% to 50% > 50% SOURCE: Statewide Database. Figure 5.2—Registered Voter Turnout by District, November 1994 In the majority of congressional districts, white turnout rates ranged from 20 to 28 percent of the voting-age population (Figure 5.3). Congressional districts in Southern California witnessed some of the lowest Anglo voter participation rates in the state, with turnout in many districts languishing below the 20 percent mark. Turnout was similarly low in the Central Valley’s 18th District. The Bay Area showed greater political participation relative to the rest of the state. This finding is consistent with exit polls showing Bay Area voters accounting for 23 percent of the total 1994 electorate but only 21 percent of all adults in the state (California Opinion Index, 1995). Almost without exception, voter turnout among African Americans significantly trailed turnout among Anglo adults (Figure 5.4). In some 43 < 18% 18% to 21% 21% to 24% > 24% NOTE: The map displays the district-level estimates derived from using ecological inference. Figure 5.3—Anglo Voting-Age Turnout by District, November 1994 districts, the participation gap between the two groups reached 15 to 20 percentage points. Black turnout did not exceed 18 percent of the voting-age population for any congressional district in the state. In more than half of the districts, less than 10 percent of voting-age African Americans cast ballots. Finally, participation among registered Latino voters was generally low (Figure 5.5). In half of the congressional districts, Latino turnout did not reach 15 percent. There were some notable exceptions, however. In 10 of the 20 districts in and around the Los Angeles County area, turnout among registered Latino voters exceeded 20 percent—including a half-dozen districts where 40 percent or more of these constituents cast 44 < 5% 5% to 10% 10% to 15% 15% to 20% NOTE: The map displays the district-level estimates derived from using ecological inference. Figure 5.4—Black Voting-Age Turnout by District, November 1994 ballots in congressional races. As will be seen below, many of these districts have Latino majorities.2 Aggregate Turnout Differences Across District Type Preliminary evidence suggests that minority turnout was above average in some majority-minority districts. The second panel on Figure 5.6 plots black turnout (as a proportion of the black voting-age population) for each of the 13 districts where racial and ethnic minorities ____________ 2 Los Angeles County, where many of these high turnout districts are concentrated, was also the epicenter of much of the Proposition 187 mobilization. 45 < 10% 10% to 25% 25% to 40% 40% to 50% NOTE: The map displays the district-level estimates derived from using ecological inference. Figure 5.5—Latino Registered Voter Turnout by District, November 1994 constituted the majority of the population. The graph also includes a horizontal line indicating the average black turnout for congressional districts in the state. On average, 8 percent of the black adults in any congressional district cast ballots in the congressional race. However, in five of the 13 majority-minority congressional districts, black turnout exceeded this rate. Furthermore, four of those districts (the 9th, 35th, 32nd, and 37th) were represented by black legislators at the time of the November election. For Latinos, registered voter turnout exceeded the 18 percent statewide average in 10 of the 13 districts with majority-minority populations (Figure 5.6, top panel). The four districts with the highest Latino participation rates were majority-Latino and served by Latino legislators. 46 Percentage Percentage Latino registered voter turnout 50 40 30 Average Latino 20 turnout 10 0 9 20 16 46 50 35 32 37 26 30 31 34 33 Majority-minority district Represented by Latino legislator Black voting-age turnout 30 25 20 15 Average black 10 turnout 5 0 46 26 30 31 33 34 20 16 50 9 35 32 37 Majority-minority district Represented by black legislator White voting-age turnout 30 25 Average white turnout 20 15 10 5 0 33 26 50 46 32 20 35 37 30 16 34 9 31 Majority-minority district Percentage Figure 5.6—District-Level Voter Participation in Majority-Minority Districts, November 1994 White turnout in the 13 districts in which they are in the minority trailed, in most cases, the 21 percent statewide average (Figure 5.6, bottom panel). However, in five cases, all of them either majority-Latino 47 or multi-ethnic districts, white voting-age turnout exceeded the statewide average. Socioeconomic and Political Influences on Voter Turnout Figure 5.6 suggests that voting participation may vary systematically with the opportunities for minority congressional officeholding. However, as noted in the previous chapter, the districts and precincts with majority-minority populations typically face both electoral (e.g., less competition) and socioeconomic (e.g., greater poverty) conditions that distinguish them from other areas in the state in ways relevant to participation. To isolate the behavior truly unique to districts where minorities have an opportunity to “elect representatives of their choice,” these areas must be evaluated in light of the patterns observed among constituents in demographically similar precincts facing comparable political conditions. Multivariate regression techniques allow us to disentangle race and ethnicity from other aspects of the social and political environment and, in that way, more accurately assess the interrelationships among minority districting, minority representation, and voter participation. I first consider the effect of socioeconomic factors and other political conditions on these rates (Appendix Table D.1). The results indicate that political participation in 1994 was greatest in precincts with large proportions of older voters (over age 35), few new residents, significant numbers of adults with at least some college education, and many households with annual incomes over $50,000. These results are consistent with exit poll data from November 1994 (California Opinion Index, 1995) and with findings from academic research. For example, each 10 percentage point increment in the proportion of precinct residents who are new to the state translates to a net turnout difference of about 2 percentage points for white voting-age turnout, 1 percentage point for black voting-age turnout, and .3 percentage points for Latino registered voter turnout. Within each racial group, new residency was among the strongest and most consistent predictors of participation. For Latinos, household income was the single strongest socioeconomic correlate. 48 Electoral conditions were also strong predictors of voter turnout in 1994. Both white and Latino voter participation rates were higher in Republican-controlled districts than in districts represented by Democrats. For Latinos, this pattern may reflect the close association of Proposition 187 with the Republican Party. Meanwhile, AfricanAmerican turnout was generally higher in congressional races represented by Democratic legislators. Turnout among all groups was at its lowest in the districts where long-serving incumbents were running for re-election. For every 10 additional years of service in the House, voter participation declined by one percentage point. Latino and Anglo turnout rates also tended to be lower in congressional races that were not particularly competitive, as measured by the size of the winning vote margin.3 By comparison, black voting-age turnout was actually higher in districts with more lopsided electoral outcomes. Finally, Proposition 187 mobilization was not a statistically significant predictor of black or Anglo voter participation. However, it was very strongly associated with Latino registered voter turnout; precincts with high Proposition 187 activity (as measured by the Proposition 187 proportion of the total ballots cast) also had high rates of Latino registered voter participation in the congressional race. ____________ 3The “vote margin” variable has been indexed by quartile: at or below the 25th percentile (vote margin ≤ 57 percent); between the 25th and 50th (57 percent < vote margin ≤ 62 percent); between the 50th and 75th (62 percent < vote margin ≤ 67 percent); above the 75th percentile. The excluded category in the regression is the first quartile. I opted for this specification after a number of diagnostic tests found a nonlinear relationship between turnout and margin of victory, skewed by three “outlier” districts in the highest vote margin quartile. The relationship is strongly negative at vote margins below 70 percent (42 of the 51 districts in the sample) but weakly positive among the nine districts with vote margins above 70 percent. The latter relationship is driven entirely by three Bay Area districts: CD-8 (San Francisco), CD-9 (Oakland/ Berkeley), and CD-7 (Richmond/Vallejo). In each of these districts, the incumbents (Nancy Pelosi, Ron Dellums, and George Miller, respectively) won by overwhelming margins but still witnessed extremely high turnout rates in their districts. Most students of California politics would not be surprised by this finding; it is consistent with the more activist political culture typical of these areas. With these three cases removed, margin of victory is consistently negatively correlated with turnout. 49 Majority-Minority Districts and Voter Turnout I now turn to the effect of the minority “opportunity” variables on observed turnout. The first empirical strategy is to test whether political participation among Latinos and African Americans is greater, on average, in districts where racial and ethnic minorities constitute the majority of the population. Again, I draw on the multivariate analyses presented in Appendix Table D.1. Table 5.1 presents the key findings predicting voter turnout among African Americans, Anglos, and Latinos. The table summarizes the measured effect of majority-minority districts, controlling for the socioeconomic and political variables shown above to influence participation. Each multivariate regression that generated these results distinguished between precincts located in Hispanic-majority, black- and Hispanic-majority, or multi-ethnic majority-minority districts. The entries in the table capture the average difference in voter turnout (for each racial or ethnic group) between precincts located in a particular type of majority-minority district and similar precincts located in a majorityAnglo district. Table 5.1 Estimates of Voter Participation Differences Across District Types District Type Hispanic majority Black and Hispanic majority Multi-ethnic majority African American –1.5%** (.3) 6.6%** (.2) 3.9%** (.2) Anglo 3.4%** (.3) –.3% (.4) 2.4%** (.3) Latino 33.1%** (.3) 30.3%** (.5) 6.9%** (.4) N=9,210 N=17,907 N=14,535 NOTES: Estimates are from regression results presented in Appendix Table C.1. Estimates control for socioceonomic conditions and political factors. Each estimate captures the average difference in voter participation between precincts located in a particular type of majorityminority district and similar precincts in a majority-Anglo district. Standard errors are in parentheses. **p < .01. 50 The econometric results are consistent with the aggregate patterns observed above: minority voter participation tends to be higher in the districts where racial and ethnic minorities enjoy majority status. However, the role of majority-minority districts in predicting minority voter participation is contingent on the group in question and the particular racial and ethnic balance within the districts. Three important findings emerge: 1. African-American adults and Latino registered voters participate at higher levels in congressional districts where their respective communities effectively control electoral outcomes. In no districts in California are African Americans solidly in the majority. As a community, they exercise their greatest electoral influence in the four districts in which they and Latinos, together, constitute the majority of the population. In these districts, turnout among voting-age blacks is on average 6.6 percentage points higher than turnout in similar precincts in majority-Anglo districts. 4 In the more fragmented majorityminority districts, where African Americans, like their Asian and Latino neighbors, play a prominent but not decisive role, turnout rates exceed those in majority-Anglo districts by a more modest 3.9 percentage points. For Latinos, registered voter turnout rates in any and all majorityminority contexts are higher than what one would expect to observe in similar precincts in majority-Anglo districts. Latino voter turnout is at its highest in districts where they alone make up the overwhelming majority of the population. In these majority-Latino districts, registered voter turnout is on average 33 percentage points higher than rates in majority-Anglo districts. In the districts where Latinos and African Americans are more equally matched, Latino turnout is 30 percentage points above majority-Anglo rates. Finally, in the multi-ethnic districts where Latinos play a relatively more limited role, their registered voter participation rates exceed those in majority-Anglo districts by a more modest 6.9 percentage points. ____________ 4“Similar” here refers to similarity in socioeconomic conditions and political characteristics, as defined by the control variables discussed above. 51 2. African-American voting-age turnout is at its lowest in any district dominated by a single nonblack racial or ethnic group. In two types of districts in the state, a single racial or ethnic group enjoys an overwhelming numerical advantage: majority-Latino districts and majority-Anglo districts. Voter participation among African Americans is marginally lower in the six majority-Latino districts in the state than it is in the dozens of majority-Anglo districts. Black voting-age turnout in the districts where Latinos constitute between 52 and 83 percent of the total population was on average 1.5 percentage points below rates in similar precincts in majority-Anglo districts. Although this difference is statistically significant, it is arguably not substantively significant. In fact, this very slight participation gap suggests that there may be more similarities than differences in the political circumstances faced by African Americans who occupy minority status in any district— majority-minority or not—dominated by a single nonblack racial group—Latino or Anglo. It is interesting to compare the patterns in majority-Latino districts to those in multi-ethnic districts—where African Americans remain in the minority but share that status with every other racial and ethnic community in the district. The turnout differential between multi-ethnic and majority-Latino districts is 5.4 percentage points, much greater than the difference between majorityAnglo and majority-Latino districts. 3. Anglo voting-age turnout is highest in congressional districts with either fragmented minority populations or Hispanic majorities. Voter participation among white adults is 2.4 to 3.4 percentage points higher in multi-ethnic and Hispanic-majority districts, respectively, than it is in majority-Anglo congressional districts. In districts where African Americans and Latinos form a solid majority, Anglo voting-age turnout is comparable to rates observed in similar precincts in majority-Anglo districts. One possible explanation for these patterns is that white voters may be responding to the political opportunities presented by a fragmented minority population and a Hispanic population constrained by high rates of noncitizenship and relative youth. These conditions may offer whites the prospect of exercising real influence over electoral outcomes despite the unfavorable racial and ethnic balance in the district. (Compare this to majority- 52 Anglo districts where even low white voting-age turnout would not compromise Anglo control over electoral outcomes.) In districts where blacks and Latinos form a solid majority, accounting for between 70 and 85 percent of the total population, there are few opportunities for even a highly mobilized Anglo minority to exploit successfully. To more clearly illustrate the interrelationship between these different districting arrangements and voter turnout among Latinos, African Americans, and Anglos, I generated estimates of the levels of participation one would predict for an “average” precinct under each of these scenarios.5 These predicted values help translate the independent measured effects summarized in Table 5.1 into more meaningful, concrete terms. The predicted values are presented graphically in Figure 5.7. The vertical distance between each bar captures the difference in turnout to be expected from “shifting” this hypothetical “average” precinct from one districting arrangement into another, holding all other factors constant. The patterns in Figure 5.7 reinforce the findings above. For example, black voting-age turnout would be expected to reach 10.8 percent for the precinct in a district with a black-Hispanic majority but only 2.7 percent in a district with a Hispanic majority. (Again, note the small difference in the height of the majority-Hispanic and majorityAnglo bars for African Americans.) For Latinos, a registered voter turnout rate of 45 percent would be predicted for an “average” precinct ____________ 5The “average” precinct is defined by the following mean values on each of the independent variables: Variable Proportion ≥ age 35 Proportion new residentsa Proportion with some college Household income ≥ $50K African American .55 .11 .57 .34 Proposition 187 turnouta Party of incumbenta Incumbent in racea Tenure (incumbent in race) Vote margin (quartile)a .96 Democrat (1) Yes (1) 9.5 .623 (50th–75th ) aNot race- or ethnicity-specific. Anglo .64 .11 .64 .37 .96 Democrat (1) Yes (1) 9.5 .623 (50th–75th) Latino .40 .11 .42 .31 .96 Democrat (1) Yes (1) 9.5 .623 (50th–75th) 53 Percentage Latino registered turnout in “average” precinct 50 40 30 20 10 0 Black voting-age turnout in “average” precinct 30 Percentage 20 10 0 White voting-age turnout in “average” precinct 30 Percentage 20 10 0 Anglo Latino Black-Latino Majority Multi-ethnic NOTES: Each panel illustrates for either Latinos, African Americans, or whites, the level of voter participation predicted for an “average” precinct under four different districting arrangements. The predicted values are generated by holding all other independent variables (socioeconomic conditions, political factors) constant at their means. Figure 5.7—Predicted Voter Participation in “Average” Precinct, by District Type in a majority-black and -Hispanic context, compared to a rate of 48 percent in a majority-Hispanic district. Finally, for white voting-age adults, an “average” precinct turnout rate of about 24 percent would be 54 predicted in either a majority-Anglo or majority-black and -Hispanic district; in Hispanic and mixed multi-ethnic districts, however, the rate would be closer to 26 or 27 percent of the voting-age population. Majority-Minority Districts, Minority Representation, and Voter Turnout In theory, each of the state’s majority-minority congressional districts affords minority constituents the opportunity to “elect representatives of their choice.” However, minority members of Congress did not represent all of these districts at the time of the November 1994 elections. In fact, Anglo legislators served three of the four multi-ethnic districts and two of the six majority-Hispanic districts. The second empirical strategy considers whether minority representation and minority districts jointly predict voter participation. Table 5.2 summarizes the measured effect of majority-minority districts and minority representation on voter participation, controlling for the same set of socioeconomic and political factors identified above. Each of the multivariate regressions that generated these results distinguished not only among precincts located in Hispanic-majority, black- and Hispanic-majority, or multi-ethnic majority-minority districts, but also among precincts served by African-American, Anglo, or Latino congressional representatives. The entries in this table capture the differences in voter participation, as compared to similar precincts in majority-Anglo districts represented by Anglo members of Congress. Several interesting results emerge with this analysis of the interaction between minority representation and minority districts. 1. Latino representation, independent of majority status within a district, is a significant predictor of Latino registered voter turnout. In the previous empirical analysis, Latino registered voter turnout was seen to be, on average, 33 percentage points higher in the precincts of the state’s six majority-Hispanic districts than in similar precincts in California’s 39 majority-Anglo congressional districts. That summary figure, however, concealed an important distinction: In the four majority-Hispanic congressional districts that were represented in 1994 by Latino members of Congress, Latino registered voter turnout was 36.4 55 Table 5.2 Estimates of Voter Participation Differences Across District Type and Legislator’s Race District Type–Legislator Race Hispanic majority Hispanic legislator White legislator Black and Hispanic majority Black legislator Multi-ethnic majority Black legislator White legislator African American –2.3%** (.3) –.4% (.4) 6.5%** (.2) 3.7%** (.3) 4.0% (.2) Anglo 5.1%** (.4) 1.0%** (.4) –.2% (.4) 6.0%** (.4) .3 (.3) Latino 36.4%** (.4) 26.5%** (.5) 31.1%** (.9) .9% (.8) 9.6%** (.5) N=9,210 N=17,907 N=14,535 NOTES: Estimates are from regression results presented in Appendix Table C.2. Estimates control for socioceonomic conditions and political factors. Each of the estimates captures the average difference in voter participation between precincts located in a particular district-legislator type and similar precincts in a majority-Anglo district with an Anglo representative. Standard errors are in parentheses. **p < .01. percentage points higher than rates in majority-Anglo districts. Meanwhile, in the two majority-Hispanic districts represented by Anglo legislators, Latino turnout was 26.5 percentage points higher than rates in majority-Anglo districts. Thus, Latino representation—quite apart from majority status—was associated with a 10 point difference in the level of Latino voter participation. The fact that high Latino voter participation and Latino congressional representation tend to go together raises the question of causal direction. Namely, is Latino representation the consequence of heightened Latino mobilization in majority-Latino districts, or does the former precede the latter?6 The history of Latino congressional ____________ 6As discussed above, the relationship between majority-Latino districts and Latino voter participation does not present the same chicken-and-egg dilemma. My research 56 representation in the state presents some analytical challenges to clarifying this “chicken-and-egg” question. As of 1994, every sitting Latino member of Congress represented a district that had consistently elected Latino legislators from the moment it had been created. None of these districts has undergone an “ethnic transition” in political leadership. The same goes for the two majority-Latino congressional districts represented by Anglo legislators in 1994. Each of these legislators was already in office when his district was redrawn in 1991 to include a Latino majority. In the absence of any ethnic change in representation, it is impossible to establish conclusively whether heightened mobilization was a consequence of, or a necessary condition for, Latino representation. One possibility for future research might be an analysis of Loretta Sanchez’s 1996 victory over Robert Dornan in Orange County’s 46th District. However, the more fragmented racial and ethnic balance in the 46th might limit the applicability of these findings to the majority-Latino case. 2. For African Americans, there is no discernible independent relationship between black representation itself and voting-age participation. African-American voting-age turnout in the three fragmented multiethnic districts represented by white legislators was comparable to the turnout rates observed in similar precincts in the one multi-ethnic congressional district represented by a black legislator (OaklandBerkeley’s 9th District, represented by Ron Dellums). For each of those cases, black voter turnout was between 3.7 and 4 percentage points higher than turnout in the state’s majority Anglo districts. The only districts in which there was any appreciable difference in the turnout rates observed with a minority legislator as opposed to an Anglo legislator were the districts with majority-Latino populations. In these cases, African-American voting-age turnout was lower in the districts represented by Latino members of Congress than it was in the districts with Anglo legislators. (In fact, there is no statistically significant difference in black voting-age turnout between majority________________________________________________________ suggests that district lines in 1991 were not explicitly drawn to target high-propensity voters (i.e., high Latino registered voter participation did not make majority-Latino districts more likely). 57 Anglo districts and the majority-Latino districts that continue to elect Anglo legislators.) When Latinos enjoy both majority status and descriptive representation, this overlap may reinforce a perception among African Americans that a single group effectively holds a monopoly over politics in the district. The political space may appear relatively more open when the nonblack majority does not also hold the congressional seat. However, it is not clear why lower turnout rates are observed in the districts where Latinos have a lock on politics than in the many more districts where Anglos have a lock on politics (i.e., majority-Anglo districts with Anglo legislators). The explanation may be simple: African Americans have not yet grown accustomed to this particular racial balance of power. Given the demographic momentum of the state, however, this arrangement may become the norm for African Americans over the next decades. 3. There is limited evidence that white voting age participation is positively associated with minority representation. White voting-age turnout was greater in the majority-Latino districts represented by Latino members of Congress than in those majorityLatino districts represented by Anglo legislators. Under Anglo representation, precinct-level white voting-age turnout was only one percentage point higher than rates observed in majority-Anglo districts; with Latino representation, turnout was 5.1 percentage points higher than the majority-Anglo case. As hypothesized above, these turnout differentials relative to majority-Anglo districts may reflect an assessment of the electoral opportunities available to a highly mobilized minority in a context of high Latino noncitizenship. From this perspective, the more pronounced differentials under Latino representation may be correlated with the higher concentrations of noncitizens in districts such as the 30th (downtown Los Angeles/Koreatown) and 33rd (East Los Angeles) in comparison to Anglo-represented districts such as the 20th (North Hollywood/Pacoima/San Fernando). There is also evidence of greater mobilization under black representation in multi-ethnic districts than under white representation. In fact, once multi-ethnic districts are separated by the race of the legislator, voting-age turnout in the fragmented districts with Anglo legislators is seen to be no different from the rates observed in majority- 58 Anglo districts throughout the state. However, the relationship that does remain (between black-represented multi-ethnic districts and white voter participation) is very likely an artifact of the activist political culture that characterizes the state’s one district that falls in this category: Oakland and Berkeley’s 9th. The six percentage point difference between white voting-age turnout in that district and turnout in a majority-Anglo district might reasonably be referred to as the “Berkeley effect.” As such, I am reluctant to interpret this as evidence of a relationship between representation and participation. The bar charts presented in Figure 5.8 reinforce these findings, illustrating the joint contribution of minority districts and minority representation to predicted turnout in an “average” precinct (as defined above in Figure 5.7). Each bar captures the predicted level of turnout for this precinct in each of the possible districting and representation arrangements. From these plots, not only can the absolute rate of voter participation be assessed but also, by measuring the vertical distance between bars, how that rate compares across scenarios. For an “average” precinct located within a majority-Latino district, Latino voter participation would vary between 42 percent of registered voters if that district were represented by an Anglo legislator and 52 percent if the district’s representative were Latino. If the district had a multi-ethnic majority and continued to be represented by an Anglo legislator, a Latino registered voter turnout rate of 25 percent would be predicted. Finally, if this “average” precinct were located within the boundaries of a majority-Anglo district, the predicted level of Latino registered voter turnout would be a relatively modest 15 percent. For Latinos, minority districts and minority representation are each significant predictors of registered voter participation. For African Americans, there is significant district-based variation in turnout but relatively weak associations between turnout and minority representation. For example, note the similarity in bar height for multiethnic districts with and without black legislators; compare this to the difference in bar height between these districts and the majority-Anglo case. In the case of a majority-Latino district, black voting-age turnout in the “average” precinct is predicted to be about 2 percent for a district 59 Percentage Percentage Latino registered turnout in “average” precinct 60 50 40 30 20 10 0 Black voting-age turnout in “average” precinct 30 25 20 15 10 5 0 White voting-age turnout in “average” precinct 30 25 20 15 10 5 0 Latino Anglo Black Anglo Anglo legislator legislator legislator legislator majority Majority-Latino district Multi-ethnic district NOTES: Each panel illustrates for either Latinos, African Americans, or whites, the predicted level of voter participation for an “average” precinct under five different districting-representative arrangements: (1) majority-Latino district with Latino legislator; (2) majority-Latino district with Anglo legislator; (3) multi-ethnic district with black legislator; (4) multi-ethnic district with Anglo-legislator; and (5) majorityAnglo district with Anglo legislator. The predicted values are generated by holding all other independent variables (socioeconomic conditions, political factors) constant at their means. Percentage Figure 5.8—Predicted Voter Turnout in “Average” Precinct, by District Type and Legislator Race 60 served by a Latino member of Congress but closer to 4 percent for a district that continues to be represented by an Anglo legislator. In the latter instance, the predicted turnout does not differ from what one would calculate for a majority-Anglo district. Setting aside the “Berkeley effect,” voting-age participation for white constituents in an “average” precinct is most consistently predicted by the majority or minority status of the Hispanic population and the presence or absence of a Hispanic legislator. If included in the boundaries of a majority-Latino district, white voting-age turnout is predicted to be about 26 percent with a Hispanic legislator and about 22 percent with an Anglo legislator. With the exception of the Berkeley case, no other representation or districting arrangement generates turnout estimates significantly different from what one would expect from a majority-Anglo district. Summary of Findings In sum, there is clearly an interrelationship among minority districting, minority representation, and political engagement among African Americans, Latinos, and Anglos in California. Latino registered voter participation and African-American voting-age turnout are highest in districts where Latinos and African Americans, respectively, are able to play prominent roles in deciding electoral outcomes. For African Americans, this includes districts where they and Latinos are equally matched and, together, constitute the majority of the district’s population. It also includes the more racially fragmented districts where African Americans may account for only a minority of the population but find themselves on equal footing with the district’s other racial and ethnic communities—white and nonwhite. For Latinos, registered voter turnout is high not only in the districts where they collectively share majority status with one or more nonwhite communities but also (and especially) in the six majority-Latino districts in the state. Furthermore, voter participation among Latinos is greatest in districts where they both enjoy an outright numerical advantage and are represented by Latino legislators. A Latino legislator can make the difference between turnout rates that are 27 percentage points above those observed in majority-Anglo districts in the state and turnout rates 61 that are 36 percentage points above those observed at these majorityAnglo polling places. African-American turnout is lowest wherever a single nonblack community clearly dominates the electoral space, effectively relegating African Americans to the political margins. Turnout is somewhat lower when this “advantaged” community is Latino (majority-Latino district and Latino legislator) than when the community is Anglo—the more typical case in the state and the one to which African Americans have had more time to grow accustomed. Finally, neither the lack of majority status nor the lack of descriptive representation is associated with lower Anglo voter participation. The only instance in which Anglo turnout differs significantly from what one would expect to observe in a majority-Anglo district (taking into account the local political culture, e.g., the “Berkeley effect”) is in majority-Latino districts, particularly those represented by Latino legislators. In these districts, Anglo voter participation is higher than expected, possibly reflecting an appreciation for the political opportunities available to a highly mobilized minority in that context. As discussed above, one real limitation of aggregate data analysis is that it does not provide a window into individual-level motivations. Consequently, one can only speculate about the reasons for the observed variation in political behavior across districts and legislators. In Chapter 3, I outlined two possible hypotheses: one focused on attitude change, emphasizing the effect of majority status and minority officeholding on levels of trust and political efficacy; the other focused on targeted mobilization, emphasizing the strategic calculations of minority political aspirants and independent political organizations. The results here cannot distinguish effectively between these competing explanations— although the model specification did allow us to control for Proposition 187 mobilization. The observable implications of either of these hypotheses is the same: greater political participation in majorityminority districts and under minority representation. What can be concluded, in light of these findings, is that the political dynamics present in majority-minority districts and under minority representation largely favor minority political engagement and do not adversely influence Anglo political participation. 62 6. Policy Implications Since the 1992 redistricting “upheaval,” there has been a steady erosion in popular, political, and judicial support for the mechanisms of the Voting Rights Act. The 2001 reapportionment and redistricting process will take shape against a backdrop of recent Supreme Court decisions (e.g., Shaw v. Reno, Miller v. Johnson, Bush v. Vera, Abrams v. Johnson, Reno v. Bossier Parish) that not only threaten the electoral gains of the past decade but may severely restrict the prospects for greater minority representation in the next. The findings presented here have important implications for the decisions and challenges that policymakers will confront in the next redistricting cycle. First, expanding political opportunities for minorities through the creation of new majority-minority districts may be an effective tool for encouraging participation among groups that otherwise might remain on the margins. What this analysis makes clear is that, for racial and ethnic minorities, the opportunity to “elect representatives of their choice” can compensate for the socioeconomic barriers that so often contribute to low voter participation. Including a precinct in a majority-black and -Hispanic district rather than in a majority-Anglo district may have a more pronounced and immediate effect on voter participation than increasing the number of college-educated residents by 10 percentage points. As a result of connecting new voters to the political world, the redistricting process may help create a more dynamic electorate that more closely reflects the demographic composition of the state’s citizen population. Second, in creating majority-minority districts, mapmakers should be mindful that turnout among minority voters is high wherever they are able to play a meaningful role in political life. Greater participation among African Americans is observed when they are not relegated to the political margins in a district dominated by a single nonblack racial or ethnic group—white or nonwhite. Latino registered voter turnout is 63 highest in districts where Latinos overwhelmingly enjoy majority status, but it is also quite high in districts where they are on roughly equal footing with African Americans. Even in the more heterogeneous areas, Latinos continue to participate at rates that exceed those in majorityAnglo districts. Furthermore, even where Anglo voters are consigned to a modest role compared to their position in majority-Anglo districts, there is no evidence that participation suffers. All of this suggests that there may be some advantage to putting more emphasis on the creation of black-Hispanic and multi-ethnic districts in the next redistricting round. These districts would allow for multiple racial and ethnic communities to exercise some political leverage; such a political environment encourages participation. The state’s demographic momentum may challenge efforts to create more black-Hispanic majority congressional districts. With Latinos constituting the fastest growing population in the state and with the shift in Latino population out of the “barrio” and into historically black neighborhoods, it is widely anticipated that the Southern California districts in which blacks and Hispanics were once equally matched may have clear Hispanic majorities when the lines are redrawn (Scott, 2000; McClain and Stewart, 1999). In places such as the 37th congressional district, anchored in Compton, this demographic trend is already contributing to political tensions between Latinos and African Americans (Sample, 1998; Scott, 1998; San Francisco Examiner, February 13, 2000; Los Angeles Times, October 9, 2000). African Americans will prefer districting arrangements that safeguard the delicate racial and ethnic balance in these areas and allow for them to retain their elected posts. Latinos will prefer districting arrangements that enhance their political leverage and offer new opportunities for Latino officeholding. State lawmakers will have to strike a balance between the benefits of expanding political opportunities for Latinos and the potential costs of redistricting African Americans out of power. Obviously, voter participation is only one of many issues raised by the practice of minority districting. As outlined in the introduction, the practice has provoked considerable debate over substantive representation and the responsiveness of legislative institutions as a whole. The intention in this report is not to elevate participation above all other 64 considerations but to expand the set of considerations to include participation. Furthermore, the observation that Latino registered voter turnout, for example, is higher in the majority-Latino districts served by Latino legislators than it is in those served by Anglo legislators should not be interpreted as a categorical statement about the merits of Anglo representation of Latino constituencies. This finding speaks only to the issue of whether minority districts and minority representation can deliver a more active body politic. Given the democratic value of an engaged citizenry—and the role that politics historically has played as a vehicle in the mobility and “mainstreaming” of racial and ethnic groups in the United States—participation itself is a worthwhile issue to consider. As mapmakers take up the challenge presented by decennial reapportionment and capitalize on the opportunities created by the one additional district California will gain (Giroux, 2000), they should appreciate fully the implications of efforts to expand, reduce, or maintain the existing political opportunities for African Americans and Latinos in the state. 65 Appendix A Sample Table A.1 summarizes the number of precinct observations for each congressional district included in the dataset. Table A.1 Number of Precinct Observations in Dataset, by District District 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Number of Precincts 521 650 485 627 461 461 500 498 353 622 494 460 342 509 509 417 345 410 121 313 456 540 233 376 395 233 District 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 48 49 50 51 52 Number of Precincts 360 382 431 184 256 344 154 311 268 435 280 361 233 412 230 277 465 489 333 105 251 389 276 459 428 NOTES: This list includes all observations for which it was possible to generate estimates of either Latino, black, or Anglo voter participation. The actual number of cases in the regression analyses will vary with patterns of residential segregation (i.e., African Americans are concentrated in relatively few precincts). 67 Appendix B Methodology Stage One: Ecological Inference In the first stage of the analysis, I use an ecological inference model to estimate the rates of Latino, African-American, and Anglo turnout in each precinct in the state. Estimation on the basis of aggregate statistics for the voting-age population (or statistics on the total population of registered voters) is necessary because the state of California does not collect racial and ethnic data on citizens who register or turn out to vote.1 That is, it is not possible to know with certainty the exact number of Latinos, African Americans, and Anglos who cast ballots in any given election. In the absence of such precise and validated individual-level statistics, we must calculate these voter participation figures using available data on the racial and ethnic composition of precincts. Table B.1 illustrates what is involved in using the available aggregate statistics to infer back to specific quantities of interest. In this particular example, Xi represents the proportion of voting-age adults who are black, and Ti represents the proportion of voting-age adults turning out to vote, in precinct i. The goal of ecological inference is to use these marginals, the only observed data, to estimate the quantities within the body of the table, which are not observed directly. Specifically, the quantity of interest turning in this out to vexoatem(ptolepi-sleBftibc,etlhl)e.2prDouproirntgiosntaogfevoontien, gB-aibg(eanbdlacks ____________ 1As mentioned above, the SWDB makes use of a Spanish surname dictionary to identify registered voters of Hispanic origin. 2In this case, Binb is not directly of interest. Rather than the catch-all “nonblack” category, the voting behavior of specific nonblack demographic groups (Latinos, Anglos) is of interest. In the event a precinct consists of precisely two demographic groups (e.g., only blacks and whites), then both Bib and Binb would be of interest. 69 Table B.1 Ecological Inference for Black/Nonblack Voter Turnout Race Black Nonblack Voting Decision Vote No Vote Bib 1 – Bib Binb 1 – Binb Ti 1 – Ti All VotingAge Adults Xi 1 – Xi accompanying standard error) is calculated for each precinct in the state, using an estimation procedure developed in King (1997). This estimation procedure is repeated in a similar analysis of Anglo voter participation, structured around the following contingency table (Table B.2). In Table B.2, Xi is the proportion of the voting-age population in precinct i that is in precinct i that Anglo; Ti is turned out the proportion of to vote. The goal the voting-age population is to estimate Biw , the proportion of Anglos who turned out to vote. Last, the estimation of Latino voter participation is structured around Table B.3. Note that this table differs slightly from Tables B.1 and B.2, reflecting the use of SWDB information on Latino registered voters. As discussed in Chapter 4, these alternative measures compensate for the high noncitizen ratio among Latino voting-age adults: In Table B.3, Xi is the proportion of registered voters (as opposed to voting-age adults) in precinct i that is Latino; Ti is the proportion of registered Table B.2 Ecological Inference for White/Nonwhite Voter Turnout Race White Nonwhite Voting Decision Vote No Vote Biw 1 – Biw Binw 1 – Binw Ti 1 – Ti All VotingAge Adults Xi 1 – Xi 70 Table B.3 Ecological Inference for Latino/Non-Latino Voter Turnout Race Latino Non-Latino Voting Decision Vote No Vote Bil 1 – Bil Binl 1 – Binl Ti 1 – Ti All Registered Voters Xi 1 – Xi voters in precinct i that turned out to vote. The goal is to estimate Bil , the proportion of Latinos who turned out to vote. To estimate Bib, Biw , and Bil for each precinct in California, I apply the ecological inference model developed in King (1997). The model (known as “EI” after the software that implements the procedure) couples the deterministic method of bounds with the maximum likelihood approach, drawing on statistical, deterministic, and diagnostic sources to improve the quality of the resulting estimates. In the method of bounds, the available precinct-level information (i.e., Xi and Ti) is used to restrict the range of possible values for each turnout estimate in each precinct. With these initial bounds defined, the procedure (using maximum likelihood) leverages the information available across all precincts to narrow the range of possible values for each turnout estimate still further. Finally, the procedure selects from this probability distribution the turnout estimates with the highest probability. Stage Two: Feasible Generalized Least Squares (FGLS) Regression Analysis The second-stage analysis uses the precinct estimates of Latino, African-American, and Anglo turnout as dependent variables in regressions predicting turnout as a function of minority representation and minority districting. These second-stage regressions use Lewis’s (2000) FGLS estimator.3 Like the more common variance weighted least ____________ 3The FGLS procedure is implemented in STATA using a program written and generously provided by Jeff Lewis. 71 squares (WLS) approach, the FGLS estimator explicitly takes into account the uncertainty in the estimates of Bib, Biw , and Bil ; however, Lewis (2000) demonstrates that FGLS significantly improves on the inefficiency and overconfidence that can result from a WLS approach to models with estimated dependent variables. The independent variables included in the FGLS regression analysis are summarized in Table B.4. 72 Table B.4 Definition of Variables 73 Variable Name Description Congressional District Type Hispanic majority Hispanics constitute 51 percent or more of population Black and Hispanic majority Hispanics and blacks constitute 51 percent or more of population Multi-ethnic majority Hispanics, Asians, and blacks constitute 51 percent or more of population Hispanic-Hispanic majority Hispanic-majority district with a Hispanic legislator White-Hispanic majority Hispanic-majority district with a white legislator Black-black and Hispanic majority Black and Hispanic majority district with a black legislator Black-multi-ethnic majority Multi-ethnic majority district with a black legislator White-multi-ethnic majority Multi-ethnic majority district with a white legislator SES Conditions Proportion over age 35 Proportion of voting-age population over age 35 (raceand ethnicity-specific) Proportion new residents Proportion of residents who moved to precinct within last five years (race- and ethnicity-specific) Proportion with some college Proportion of adults (age 25 and over) with at least some college education (race- and ethnicity-specific) Proportion of households earning Proportion of households with 1989 income of at least over $50K $50K (race- and ethnicity-specific) Coding 1 if district = 30, 31, 33, 34, 20, 26; otherwise 0 1 if district = 32, 35, 37; otherwise 0 1 if district = 9, 16, 46, 50; otherwise 0 1 if district = 30, 31, 33, 34; otherwise 0 1 if district = 20, 26; otherwise 0 1 if district = 32, 35, 37; otherwise 0 1 if district = 9; otherwise 0 1 if district = 16, 46, 50; otherwise 0 0 to 1 0 to 1 0 to 1 0 to 1 Table B.4 (continued) 74 Variable Name Party of incumbent Description Electoral Conditions Political party of incumbent U.S. House representative in November 1994 Coding 1 if party = Democrat; 0 if party = Republican Incumbent in race Is House incumbent running for re-election in November 1 if running; otherwise 0 1994 Tenure * incumbent Interaction of “incumbent” and years of service in the U.S. House, as of November 1994 Vote margin quartiles Winning proportion of the vote in November 1994 Raw variable ranges from .49 to .82; transformed variable indexed by quartile: 1st quartile (.49 – .57); 2nd quartile (.58 – .62); 3rd quartile (.63 – .67), 4th quartile (.68 – .82) NOTE: For the socioeconomic variables, there are separate measures for each racial and ethnic group. Appendix C District-Level Turnout Estimates with Standard Errors The ecological inference procedure applied in the estimation of Latino, African-American, and white turnout rates generates not only point estimates for these parameters but also standard errors. Figures C.1 through C.3 present the (weighted) district-level estimates for turnout; the figures also include standard error bars, capturing the uncertainty in these district estimates. (Each box is centered on the point estimate and extends one standard error above and below that estimate.) The overall pattern across these three figures reflects the far greater certainty contained in the estimates of Anglo turnout as compared to black turnout and (to a more limited extent) Latino turnout. With most congressional districts dominated by precincts that are overwhelmingly Anglo, the data typically contain more information on which to make inferences about Anglo voting behavior than they do information about black or Latino voting behavior.1 The various majority-minority congressional districts (identified by “*”) are the exceptions, consisting as they do of precincts that are predominantly Latino or predominantly black. In these districts, the estimates of black and Latino turnout have the narrow error bars more characteristic of the white turnout estimates in Figure C.1. ____________ 1For example, imagine a precinct (or congressional district) that is 80 percent Anglo and 20 percent African American. If 40 percent of the voting-age population turned out to vote on election day, that is consistent with a black turnout rate of anywhere between 0 and 100 percent. The composition of the geographic unit leaves us with very little information about African Americans—and thus greater uncertainty in the subsequent estimate. By comparison, there is enough information in the data to allow one to conclude that at least 25 percent, and not more than 50 percent, of Anglos turned out to vote. This would lead to comparatively narrow error bars around the estimate. 75 Congressional district 31* 9* 8 34* 39 28 23 15 14 22 36 16* 17 24 38 2 6 19 12 45 21 3 30* 51 37* 4 5 25 1 40 35* 27 13 10 20* 41 11 7 29 49 32* 48 43 52 46* 44 50* 26* 33* 42 18 0 15 20 25 Estimated Anglo voter turnout (% voting-age population) 30 NOTE: Each bar represents the estimated rate of Anglo voting participation in the congressional district, plus or minus one standard error. The width of the bars reflects the degree of statistical uncertainty in the estimate. * = majority-minority district. Figure C.1—District-Level Estimates of Anglo Turnout with Uncertainty 76 Congressional district 37* 38 36 29 5 32* 35* 13 12 11 10 28 27 25 9* 4 3 2 1 50* 8 7 6 42 17 16* 15 14 21 20* 19 18 34* 33* 31* 30* 26* 52 51 49 41 40 39 48 46* 45 44 43 24 23 22 0 5 10 15 Estimated black voter turnout (% voting-age population) 20 NOTE: Each bar represents the estimated rate of black voting participation in the congressional district, plus or minus one standard error. The width of the bars reflects the degree of statistical uncertainty in the estimate. * = majority-minority district. Figure C.2—District-Level Estimates of Black Turnout with Uncertainty 77 Congressional district 33* 34* 52 31* 30* 26* 37* 17 32* 35* 23 50* 8 48 44 46* 12 13 41 16* 42 38 18 3 20* 39 4 2 22 45 43 40 25 28 19 21 7 51 10 6 9* 36 27 1 11 29 49 5 24 15 14 0 10 20 30 40 Estimated Latino voter turnout (% registered) 50 NOTE: Each bar represents the estimated rate of Latino voting participation in the congressional district, plus or minus one standard error. The width of the bars reflects the degree of statistical uncertainty in the estimate. * = majority-minority district. Figure C.3—District-Level Estimates of Latino Turnout with Uncertainty 78 Appendix D Regression Output To evaluate the relationship of minority districts and minority representation to voter participation among Latinos, African Americans, and Anglos, I estimated a series of multivariate regression models. The first set of models analyzes the relationship between majority-minority districts and voting participation; the second set of models identifies how minority representation and majority-minority districts interact to predict turnout. Both sets of multivariate models include controls for socioeconomic conditions and political factors. The only differences in the specifications of the control variables are across racial and ethnic groups; the measures of age, education, and income are all race- and ethnicity-specific. Tables D.1 and D.2 report the results from this regression analysis. 79 Table D.1 Coefficient Estimates for Regression of Majority-Minority Districts on Voting Participation Variables Constant African American Anglo Latino .016 (.019) .171 (.026)** –.060 (.037) District type Hispanic majority Black and Hispanic majority Multi-ethnic majority –.015 (.003)** .034 (.003)** .331 (.003)** .066 (.002)**–.003 (.004) .303 (.005)** .039 (.002)** .024 (.003)** .069 (.004)** SES conditions Proportion over age 35 .011 (.002)** .088 (.005)** .031 (.006)** Proportion new residents –.101 (.006)**–.196 (.009)** –.032 (.013)* Proportion with some college .011 (.002)** .082 (.005)** .040 (.005)** Proportion of households earning over $50K .000 (.002) .090 (.004)** .051 (.005)** Electoral conditions Proposition 187 turnout –.009 (.019) –.036 (.026) .172 (.038)** Party of incumbent .026 (.001)**–.006 (.001)** –.054 (.002)** Incumbent in race .012 (.001)** .002 (.002) .012 (.004)** Tenure * incumbent –.001 (.000)**–.001 (.000)** –.001 (.000)** Vote margin 2nd quartile (58%–62%) .031 (.002)**–.006 (.002)** –.063 (.002)** Vote margin 3rd quartile (62%–67%) .005 (.001)**–.008 (.001)** .057 (.002)** Vote margin 4th quartile (67%–82%) .050 (.001)** .009 (.001)** –.026 (.002)** N 9,210 17,907 14,535 SER .036 .075 .104 R2 .73 .20 .55 NOTES: For African Americans and Anglos, voter turnout rates are calculated as a proportion of voting-age adults. For Latinos, voter turnout rates are calculated as a proportion of registered voters. The dependent variables range from 0 to 1. District type variables are dichotomous, coded 0 (not in district type), 1 (in district type). The excluded comparison district type is Anglo majority. “Party of incumbent” is coded 0 (Republican) and 1 (Democrat). Measures of age, education, and income are race- and ethnicity-specific. “Vote margin” is indexed by quartile; the model includes dichotomous variables for each quartile, coded 0 (not in quartile) or 1 (in quartile). The excluded comparison “Vote margin” category is the fist quartile: cases with winning vote margins at or below 57 percent. Standard errors are in parentheses. **p < .01, *p < .05, two-tailed. 80 Table D.2 Coefficient Estimates for Regression of Majority-Minority Districts and Minority Representation on Voting Participation Variables Constant African American .022 (.019) Anglo Latino .173 (.026)** –.131 (.036)** District type-legislator race Hispanic majority-Hispanic Hispanic majority-white Black-Hispanic majority-black Multi-ethnic majority-black Multi-ethnic majority-white –.023 (.003)** .051 (.004)** –.004 (.004) .010 (.004)** .065 (.002)** –.002 (.004) .037 (.003)** .060 (.004)** .040 (.002)** .003 (.003) .364 (.004)** .265 (.005)** .311 (.005)** .009 (.008) .096 (.005)** SES conditions Proportion over age 35 .012 (.002)** .085 (.005)** .023 (.006)** Proportion new residents –.100 (.006)** –.193 (.009)** –.044 (.013)** Proportion with some college .012 (.002)** .073 (.005)** .039 (.005)** Proportion of households earning over $50K .000 (.016) .097 (.004)** .043 (.005)** Electoral conditions Proposition 187 turnout –.016 (.019) –.057 (.026)* .239 (.037)** Party of incumbent .026 (.001)** –.005 (.001)** –.059 (.002)** Incumbent in race .013 (.001)** –.001 (.002) .019 (.004)** Tenure * incumbent –.001 (.000)** –.001 (.000)** .000 (.000) Vote margin 2nd quartile (58%–62%) .030 (.002)** –.008 (.002)** –.059 (.002)** Vote margin 3rd quartile (62%–67%) .004 (.001)** –.007 (.001)** .059 (.002)** Vote margin 4th quartile (67%–82%) .051 (.001)** .006 (.002)** –.024 (.002)** N 9,210 17,907 14,535 SER .04 .074 .10 R2 .74 .21 .58 NOTES: For African Americans and Anglos, voter turnout rates are calculated as a proportion of voting-age adults. For Latinos, voter turnout rates are calculated as a proportion of registered voters. The dependent variables range from 0 to 1. District type-legislator race variables are dichotomous, coded 0 (matches district type-legislator race), 1 (matches district type-legislator race). The excluded comparison district type is Anglo majority-Anglo legislator. “Party of incumbent” is coded 0 (Republican) and 1 (Democrat). Measures of age, education, and income are race- and ethnicity-specific. “Vote margin” is indexed by quartile; the model includes dichotomous variables for each quartile, coded 0 (not in quartile) or 1 (in quartile). The excluded comparison “Vote margin” category is the fist quartile: cases with winning vote margins at or below 57 percent. Standard errors are in parentheses. **p < .01, *p < .05, two-tailed. 81 References Abney, Glen, and John D. Hutcheson, “Race, Representation and Trust,” Public Opinion Quarterly, Vol. 45, 1981, pp. 91–101. Baldassare, Mark, California in the New Millennium: The Changing Social and Political Landscape, University of California Press, Berkeley, California, 2000. Barker, Lucius J., Mack Jones, and Katherine Tate, African Americans and the American Political System, Prentice Hall, New Jersey, 1999. Barone, Michael, and Grant Ujifusa, The Almanac of American Politics, 2000, National Journal, Washington, D.C., 1999. Block, A. G., “Wilson Legacy,” California Journal, November 1998, pp. 6–10. Bobo, Lawrence, and Franklin Gilliam, “Race, Sociopolitical Participation, and Black Empowerment,” American Political Science Review, Vol. 84, 1990, pp. 377–393. Brackman, Harold, and Steven P. Erie, “At Rainbow’s End: Empowerment Prospects for Latinos and Asian Pacific Americans in Los Angeles,” in Michael Preston, Bruce Cain, and Sandra Bass (eds.), Racial and Ethnic Politics in California, Vol. 2, Institute of Governmental Studies, Berkeley, California, 1998. Brischetto, Robert R., “Latino Voters and Redistricting in the New Millennium,” in David A. Bositis (ed.), Redistricting and Minority Representation: Learning from the Past, Preparing for the Future, Joint Center for Political and Economic Studies, Washington, D.C., 1998. 83 Brody, Richard, “The Puzzle of Political Participation in America,” in Anthony King (ed.), The New American Political System, American Enterprise Institute, Washington, D.C, 1978. Bucy, E. Page, “Redistricting by Lawsuit,” California Journal, January 1991. Bureau of the Census, Congressional District Data Book (Districts of the 87th Congress), U.S. Department of Commerce, Washington, D.C., 1961. Bureau of the Census, Congressional District Data Book (Districts of the 88th Congress), U.S. Department of Commerce, Washington, D.C., 1963. Bureau of the Census, Congressional District Data Book (Districts of the 92nd Congress), U.S. Department of Commerce, Washington, D.C., 1971. Bureau of the Census, Congressional District Data Book (Districts of the 93rd Congress), U.S. Department of Commerce, Washington, D.C., 1973. Bureau of the Census, 1980 Census of Population and Housing: Congressional Districts of the 98th Congress, U.S. Department of Commerce, Washington, D.C., 1983. Bureau of the Census, 1990 Census of Population and Housing: Population and Housing Characteristics for Congressional Districts of the 103rd Congress, U.S. Department of Commerce, Washington, D.C., 1993. Bureau of the Census, Statistical Abstract of the United States, U.S. Department of Commerce, Washington, D.C., 1999. Cain, Bruce, “Voting Rights and Democratic Theory: Toward a ColorBlind Society?” in Bernard Grofman and Chandler Davidson 84 (eds.), Controversies in Minority Voting, Brookings Institution, Washington, D.C., 1992. California Department of Finance, California Statistical Abstract, Sacramento, California, 1999. California Journal, Congress, December 1994. California Journal, State Assembly, December 1998, pp. 25–28. California Journal Weekly, “Proposition 187: Services of Illegal Immigrants,” November 14, 1994. California Opinion Index, Field Institute, San Francisco, California, August 1990. California Opinion Index, Field Institute, San Francisco, California, January 1993. California Opinion Index, Field Institute, San Francisco, California, June 1994. California Opinion Index, Field Institute, San Francisco, California, January 1995. California Opinion Index, Field Institute, San Francisco, California, February 1997. California Opinion Index, Field Institute, San Francisco, California, January 1999. California Opinion Index, Field Institute, San Francisco, California, May 2000. California Political Week, Inc., California Political Week, Beverly Hills, California, November 14, 1994. California Student Trends, 2000: www.ed-data.k12.ca.us. 85 Cameron, Charles, David Epstein, and Sharon O’Halloran, “Do Majority-Minority Districts Maximize Substantive Black Representation in Congress?” American Political Science Review, Vol. 90, 1996, pp. 794–812. Canon, David T., Race, Redistricting, and Representation, University of Chicago, Chicago, Illinois, 1999. Chavez, Linda, “Party Lines: The Republicans’ Racial Quotas,” The New Republic, Vol. 204, No. 25, June 24, 1991, p.14. Chicago Tribune, “The South’s New Reality,” August 2, 1992. Cho, W. T., “If the Assumption Fits . . . : A Comment on King’s Ecological Inference Solution,” Political Analysis, Vol. 7, 1998. Congressional Quarterly, Inc., Congressional Districts in the 1990s: A Portrait of America, Washington, D.C., 1993. Congressional Quarterly, Inc., CQ Almanac, 1996, Washington, D.C., 1996. Congressional Quarterly, Inc., CQ Almanac, 1997, Washington, D.C., 1997. Conway, Margaret, Political Participation in the United States, 2d. ed., Congressional Quarterly, Inc., Washington, D.C., 1991. Donovan, Beth, “New ‘Majority Minority’ Districts May Mean Lower Black Turnout,” Congressional Quarterly Weekly Report, Vol. 50, No. 10, March 7, 1992, pp. 563–564. Duncan, Phil, “Creating Black Districts May Segregate Voters,” Congressional Quarterly Weekly Report, Vol. 48, No. 30, July 28, 1990, p. 2462. 86 Duncan, Phil, “Minority Districts Fail to Enhance Turnout,” Congressional Quarterly Weekly Report, Vol. 51, No. 13, March 27, 1993a, p. 798. Duncan, Phil, “New Minority Districts: A Conflict of Goals,” Congressional Quarterly Weekly Report, Vol. 51, No. 30, July 24, 1993b, p. 2010. Duncan, Phil, Politics in America, 1994: The 103rd Congress, Congressional Quarterly, Inc., Washington, D.C., 1993c. Duncan, Phil, Politics in America, 1996: The 104th Congress, Congressional Quarterly, Inc., Washington, D.C., 1995. Education Data Partnership, California Student Trends Report, 2001: http://www.ed-data.k12.ca.us/state. Ehrenhalt, Alan (ed.), Politics in America: Members of Congress in Washington and at Home, 1984, Congressional Quarterly, Inc., Washington, D.C., 1983. Erie, Steven P., Harold Brackman, and James Ingram, Paths to Political Incorporation for Latinos and Asian Pacifics in California, California Policy Seminar, University of California, Berkeley, California, 1993. Frayne, Gabriel, “Race for Latino California,” The Nation, Vol. 267, No. 13, October 26, 1998. Freedman, D. A., S. P. Klein, M. Ostland, and M. R. Roberts, “Review of A Solution to the Ecological Inference Problem,” Journal of the American Statistical Association, Vol. 93, 1998. Gilliam, Frank, “Influences on Voter Turnout for U.S. House Elections in Non-Presidential Years,” Legislative Studies Quarterly, Vol. 10, 1985, pp. 339–351. 87 Gilliam, Frank, “Exploring Minority Empowerment: Symbolic Politics, Governing Coalitions, and Traces of Political Style in Los Angeles,” American Journal of Political Science, Vol. 40, 1996, pp. 56–81. Gilliam, Frank, and Karen Kaufmann, “Is There an Empowerment Lifecycle?” paper presented at the Annual Meeting of the Western Political Science Association, San Francisco, California, 1996. Giroux, Gregory, “The Hidden Election: Day of the Mapmaker,” Congressional Quarterly Weekly Report, Vol. 58, No. 8, February 19, 2000, pp. 344–350. Green, Stephen, California Political Almanac, 1995–1996, California Journal, Sacramento, California, 1995. Guerra, Fernando, “Latino Politics in California: The Necessary Conditions for Success,” in Michael Preston, Bruce Cain, and Sandra Bass (eds.), Racial and Ethnic Politics in California, Vol. 2, Institute of Governmental Studies, Berkeley, California, 1998. Handley, Lisa, Bernard Grofman, and Wayne Arden, “Electing Minority-Preferred Candidates to Legislative Office: The Relationship Between Minority Percentages in Districts and the Election of Minority-Preferred Candidates,” in Bernard Grofman (ed.), Race and Redistricting in the 1990s, Agathon Press, New York, 1998. Howell, Susan, and Deborah Fagan, “Race and Trust in Government,” Public Opinion Quarterly, Vol. 52, 1988, pp. 343–350. Huckfeldt, Robert, and John Sprague,“Citizens, Context, and Politics,” The State of the Discipline II, American Political Science Association, Washington, D.C., 1993. Hutchings, Vincent, Harwood McClerking, and Guy-Uriel Charles, “When Does Race Matter for Congressional Responsiveness? An Examination of Support for Black Interests Among Southern 88 Members of Congress,” paper prepared for the Annual Meeting of the Southern Political Science Association, Savannah, Georgia, 1999. King, Gary, A Solution to the Ecological Inference Problem, Princeton University, Princeton, New Jersey, 1997. King, Gary, “The Future of Ecological Inference Research: A Reply to Freedman et al.,” Journal of the American Statistical Association, Vol. 94, 1999, pp. 352–355. Lewis, Jeffrey,“Two-Stage Approaches to Regression Models in Which the Dependent Variable Is Based on Estimates,” unpublished working paper, 2000. Lopez, Elisabeth, and Eric Wahlgren, “The Latino Vote,” California Journal, November 1994, pp. 29–31. Los Angeles Times, “Latino Activists Zero in on County Representation,” August 26, 1990. Los Angeles Times, “Lawyer Views Remap Case As Only 1st Step,” October 11, 1990. Los Angeles Times, “74 Cities Targeted for Voting Rights Fights,” January 21, 1991. Los Angeles Times, “1 Million Voters: Latino Groups Launch Effort to Achieve Goal,” April 30, 1992. Los Angeles Times, “Political Activists Launch Extensive Campaign to Get Latinos to the Polls,” September 6, 1992. Los Angeles Times, “Latinos Unite in Self-Defense on Prop. 187,” October 21, 1994. Los Angeles Times, “Emotions High at Last-Minute Protests,” November 8, 1994. 89 Los Angeles Times, “Crime, Immigration Issues Helped Wilson, Poll Finds,” November 9, 1994. Los Angeles Times, “State’s Diversity Doesn’t Reach Voting Booth,” November 10, 1994. Los Angeles Times, “Perspectives on Election Fallout,” November 11, 1994. Los Angeles Times, “Let Election Setbacks Be Call for Action,” November 14, 1994. Los Angeles Times, “O.C. Precinct Typifies Latino Voter Apathy,” November 28, 1994. Los Angeles Times, “Despite Gains, Latino Voters Still Lack Clout,” December 4, 1994. Los Angeles Times, “Massive Tide of Naturalized Citizens Swells Voter Rolls,” September 29, 1996. Los Angeles Times, “In Contests Big and Small, Latinos Take Historic Leap for the Record,” November 5, 1998. Los Angeles Times, “Latino Revolution Leaves Some City Councils Untouched,” October 9, 2000. Lublin, David, The Paradox of Representation, Princeton University, Princeton, New Jersey, 1997. Maharidge, Dale, “Did 1992 Herald the Dawn of Latino Political Power?” in Thomas Hoeber and Charles Price (eds.), California Government and Politics Annual, 1994–1995, California Journal, Sacramento, California, 1994. Martis, Nancy, and A. G. Block, “Proposition 187,” California Journal, December 1994, pp. 20–25. 90 McClain, Paula D., and Joseph Stewart, Jr., Can We All Get Along? Racial and Ethnic Minorities in American Politics, Westview Press, Boulder, Colorado, 1999. Nagler, Jonathan, “The Effect of Registration Laws and Education on U.S. Voter Turnout,” American Political Science Review, Vol. 85, 1991, pp.1393–1405. New York Times, “Hispanic Politicians Seek a Recipe for Raw Numbers,” April 12, 1992. New York Times, “Redistricting Expected to Bring Surge in Minority Lawmakers,” August 3, 1992. New York Times, “The Old Order Changes in Congress—A Little,” November 2, 1992. New York Times, “Suits Challenging Redrawn Districts That Help Blacks,” February 14, 1994. Orange County Register, “Hispanics See a Political Sea Change,” October 22, 1998. Pachon, Harry P., “Latino Participation in California Politics,” in Byan O. Jackson and Michael Preston (eds.), Racial and Ethnic Politics in California, Vol. 1, Institute of Governmental Studies, Berkeley, California, 1991. Pachon, Harry P., “Latino Politics in the Golden State: Ready for the 21st Century?” in Michael Preston, Bruce Cain, and Sandra Bass (eds.), Racial and Ethnic Politics in California, Vol. 2, Institute of Governmental Studies, Berkeley, California, 1998. Palmquist, B. L., and D. Stephen Voss, “Racial Polarization and Turnout in Louisiana: New Insights from Aggregate Data Analysis,” paper presented at the Annual Meeting of the American Political Science Association, San Francisco, California, 1996. 91 Reyes, Belinda (ed.), A Portrait of Race and Ethnicity in California: The Social and Economic Well-Being of Racial and Ethnic Groups, Public Policy Institute of California, San Francisco, California, 2001. Rodriguez, Gregory, “The Latino Century,” California Journal, January 2000, pp. 8–14. Rosenstone, Steven J., and John Mark Hansen, Mobilization, Participation, and Democracy in America, MacMillan, New York, 1993. Sacramento Bee, “The Paradox of Remapping,” November 30, 2000. Sample, Herbert, “Washington Notes,” California Journal, January 1998, p. 44. San Francisco Chronicle, “New Voters in State Are Mostly Latino,” May 1, 2000. San Francisco Examiner, “A New California: A Growing Latino Presence in Compton Shows Signs of Reshaping Culture, Politics,” February 13, 2000. San Jose Mercury News, “Established Latinos More Likely to Support Curbing Immigration,” October 15, 2000. Schockman, H. Eric, “California’s Ethnic Experiment and the Unsolvable Immigration Issue: Proposition 187 and Beyond,” in Michael Preston, Bruce Cain, and Sandra Bass (eds.), Racial and Ethnic Politics in California, Vol. 2, Institute of Governmental Studies, Berkeley, California, 1998. Scott, Steve, “Reality Votes,” California Journal, January 1998, pp. 25– 29. Scott, Steve, “Competing for the New Majority Vote,” California Journal, January 2000, pp. 16–23. 92 Shaw, Daron, Rodolfo O. de la Garza, and Jongho Lee, “Examining Latino Turnout in 1996: A Three-State, Validated Survey Approach,” American Journal of Political Science, Vol. 44, 2000, pp. 332–340. Stiles, Jon, Jonathan Cohen, Zachary Elkins, and Frederic Gey, California Latino Demographic Databook, California Policy Seminar, University of California, Berkeley, California, 1998. Swain, Carol, Black Faces, Black Interests, Harvard University, Cambridge, Massachusetts, 1995. Tate, Katherine, “Black Political Participation in the 1984 and 1988 Presidential Elections,” American Political Science Review, Vol. 85, 1991, pp. 1159–1176. Teixeira, Ruy, Why Americans Don’t Vote, Greenwood Press, Westport, Connecticut, 1987. Timpone, Richard J., “Structure, Behavior and Voter Turnout in the United States,” American Political Science Review, Vol. 92, 1998, pp. 145–158. Verba, Sidney, and Norman Nie, Participation in America, University of Chicago Press, Chicago, Illinois, 1972. Verba, Sidney, Kay Schlozman, and Henry Brady, Voice and Equality: Civic Voluntarism in American Politics, Harvard University, Cambridge, Massachusetts, 1995. Washington Post, “INS Sees Surge in Citizenship Seekers,” April 12, 1995. Whitby, Kenny J., The Color of Representation, University of Michigan Press, Ann Arbor, Michigan, 1998. 93 Wolfinger, Raymond E., and Steven Rosenstone, Who Votes? Yale, New Haven, Connecticut, 1980. 94 About the Author CLAUDINE GAY Claudine Gay is an assistant professor of political science at Stanford University and an adjunct fellow at the Public Policy Institute of California. Her recent publications include “Doubly Bound: The Impact of Gender and Race on the Politics of Black Women ” (with Katherine Tate), which appeared in Political Psychology in 1997. A graduate of Stanford University in 1992, she received a Ph.D. from Harvard University, where she won the Toppan Prize in 1998 for best dissertation in political science. 95 Other Related PPIC Publications A Portrait of Race and Ethnicity in California: An Assessment of Social and Economic Well-Being Belinda I. Reyes (editor) et al. Ethnic Context, Race Relations, and California Politics Bruce Cain, Jack Citrin, Cara Wong Finding Common Ground: Racial and Ethnic Attitudes in California Zoltan Hajnal, Mark Baldassare How Different Ethnic Groups React to Legal Authority Yuen J. Huo, Tom R. Tyler 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 97" } ["___content":protected]=> string(102) "

R 601CGR

" ["_permalink":protected]=> string(144) "https://www.ppic.org/publication/the-effect-of-minority-districts-and-minority-representation-on-political-participation-in-california/r_601cgr/" ["_next":protected]=> array(0) { } ["_prev":protected]=> array(0) { } ["_css_class":protected]=> NULL ["id"]=> int(8140) ["ID"]=> int(8140) ["post_author"]=> string(1) "1" ["post_content"]=> string(0) "" ["post_date"]=> string(19) "2017-05-20 02:35:11" ["post_excerpt"]=> string(0) "" ["post_parent"]=> int(3260) ["post_status"]=> string(7) "inherit" ["post_title"]=> string(8) "R 601CGR" ["post_type"]=> string(10) "attachment" ["slug"]=> string(8) "r_601cgr" ["__type":protected]=> NULL ["_wp_attached_file"]=> string(12) "R_601CGR.pdf" ["wpmf_size"]=> string(6) "910554" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(173159) "The Effect of Minority Districts and Minority Representation on Political Participation in California ••• Claudine Gay 2001 PUBLIC POLICY INSTITUTE OF CALIFORNIA Library of Congress Cataloging-in-Publication Data Gay, Claudine, 1970– The effect of minority districts and minority representation on political participation in California / Claudine Gay. p. cm. Includes bibliographical references. ISBN 1-58213-030-2 1. Political participation—California. 2. African Americans— California—Politics and government. 3. Hispanic Americans— California—Politics and government. 4. Election districts— California. 5. United States. Congress. House—Election districts. I. Public Policy Institute of California. II. Title. JK8789 .G39 2001 323'.042'089009794—dc21 2001019718 Copyright © 2001 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 Now that the 2000 Census has shown that California is a majorityminority state, more and more observers have become interested in the political, social, and economic consequences of the state’s shifting demographic profile. For its sheer speed and magnitude, this change is unprecedented in the United States. Since 1960, California has doubled its population, and demographers predict that the population will nearly double again by 2040. A generation ago, 80 percent of Californians were non-Hispanic whites; a generation from now, 50 percent will be Latino. This demographic transformation is leading California and its policymakers into new and uncharted waters. For this reason, PPIC began an intensive study of this transformation and its consequences in 1998. Our work to date includes a portrait of race and ethnicity in California, an assessment of how different ethnic groups react to legal authority, a study of attitudes toward public policy by race and ethnicity, and an investigation of voting behavior in homogeneous and ethnically diverse neighborhoods. Claudine Gay’s study, which examines voter turnout by race and ethnicity in California’s congressional districts, adds a new and significant dimension to that research effort. Her primary question—is voter turnout increased in majority-minority districts—pertains directly to the redistricting process that will be triggered by the 2000 Census. Her findings indicate that turnout among minority voters in the 1994 elections was highest wherever they were able to play a meaningful role in political life. After adjusting for such variables as age, education, and income, she finds that registered voter turnout among Latinos was 33 percentage points higher in majority-Latino districts than in majorityAnglo districts. African-American voting-age turnout was highest in districts where blacks and Latinos were equally matched and together formed the majority of the voting-age population. She also finds that Anglo turnout did not suffer in districts where whites were a minority. iii The next redistricting effort can and will influence voter participation, especially in California’s majority-minority districts. This timely work shows why the redistricting process is so important and how it can be used to increase political engagement throughout the state. David W. Lyon President and CEO Public Policy Institute of California iv Summary Since the passage of the Voting Rights Act of 1965 (VRA), the number of Latino and African-American elected officials at the local, state, and federal levels has increased dramatically. Minority members of the U.S. House of Representatives are among the most influential and prominent class of these new officeholders, and California is home to one of the nation’s most diverse congressional delegations. Benefiting in part from the creation of majority-minority electoral districts under the provisions of the VRA, minority legislators as a group have emerged as visible political actors in an institution traditionally dominated by whites. Considerable debate has emerged over whether this diversity has translated into substantive benefits for California’s minority constituents. Many advocates have argued that, regardless of its influence on legislative outcomes, the creation of majority-minority districts—and the subsequent election of minority representatives—serves an important role in pulling the Latino and African-American communities into the political process. By “creating a climate of inclusion,” these districts are expected to increase political participation and make voters out of previously unengaged minority Californians (New York Times, April 12, 1992). Although these claims have been met with skepticism, neither advocates nor skeptics have offered more than anecdotal evidence and journalistic speculation to clarify the link between race-conscious districting and greater political engagement among ethnic and racial minorities. This report examines the relationship between majority-minority districts, minority representation, and voting participation among Latinos, African Americans, and Anglos in California. In particular, it addresses the following questions: v • Does the level of Latino, African-American, and Anglo voter participation vary with the racial and ethnic balance of a congressional district? If so, how? • Do voter participation rates in the majority-minority congressional districts served by minority members of Congress differ from the rates observed in the majority-minority districts represented by Anglo legislators? If so, how? • What, if any, difference exists between Latinos and African Americans in the relationship between voter turnout and type of congressional district? Drawing on data from the November 1994 congressional elections, the report compares voter turnout rates in the state’s 13 majorityminority congressional districts with turnout rates in majority-Anglo districts. The data assembled for the analysis are aggregated at the precinct level and consist of voting statistics, Census data, and political information. When considering differences in turnout, the analysis takes into account the potentially confounding effects of socioeconomic conditions (e.g., levels of educational attainment, household income, concentrations of new residents) as well as electoral competitiveness (e.g., incumbency, tenure, vote margin) to gauge the independent effects of majority-minority districts and minority representation. The analysis provides empirical support for the claim that minority representation and majority-minority districts are associated with greater involvement in electoral politics among Latinos and African Americans. Four major findings emerge: • Latino and African-American voter participation is highest in congressional districts where Latinos and African Americans, respectively, are able to play prominent roles in deciding political outcomes. For African Americans, who do not constitute a majority in any of the state’s congressional districts, voting-age turnout in districts where they and Latinos are equally matched and, together, form the majority of the voting-age population is on average 6.6 percentage points higher than turnout in majority-Anglo districts. In districts where African Americans vi are one part of a diverse multi-ethnic mix, black voting-age turnout rates exceed those in majority-Anglo districts by 3.9 percentage points. Among Latinos, registered voter turnout rates in any and all majority-minority districts are higher than those observed in majority-Anglo districts. Latino registered voter turnout is at its highest in the state’s six majority-Latino congressional districts. In these districts, registered voter turnout is on average 33 percentage points higher than rates in majority-Anglo districts. Where Latinos and African Americans are more equally matched, Latino registered voter turnout is 30 percentage points above rates in majority-Anglo districts. Finally, in the multiethnic districts where Latinos play a relatively more limited role, their registered voter turnout rates exceed those in majorityAnglo districts by a more modest 6.9 percentage points. • African-American turnout is lowest wherever a single nonblack community (of any race or ethnicity) clearly dominates the electoral space, effectively relegating African Americans to the political margins. Turnout is marginally lower when this “advantaged” community is Latino than when the community is Anglo—the more typical case in the state, and the one to which African Americans have had more time to grow accustomed. • Voter participation among Latinos is particularly high in districts where they enjoy both majority status as well as descriptive representation (i.e., representation by legislators of the same race or ethnicity). In the four majority-Latino districts that were represented in 1994 by Latino members of Congress, Latino registered voter turnout was 36.4 percentage points higher than rates in majority-Anglo districts. By comparison, in the two majority-Latino districts represented by Anglo legislators in 1994, Latino turnout was 26.5 percentage points higher than rates in majority-Anglo districts. For African Americans, there is no discernible independent relationship between black representation, itself, and voting-age participation. vii • Neither the lack of majority status nor the lack of Anglo representation is associated with lower Anglo voting-age participation. In fact, Anglo voter participation is slightly higher in majority-Latino districts than in majority-Anglo districts in the state, possibly reflecting an appreciation for the potential electoral influence that could be wielded by a highly mobilized minority in the context of high Latino noncitizenship and relative youth. These findings have important implications for the decisions and challenges that state lawmakers will confront as they approach redistricting in 2001. First, creating new majority-minority congressional districts is likely to promote political participation in the state’s Latino and African-American communities. In particular, the creation of such districts can compensate for the socioeconomic barriers that so often contribute to low voter participation in minority communities. For example, including a precinct in a majority-black and -Hispanic district rather than in a majority-Anglo district may have a more pronounced and immediate effect on voter participation than increasing the number of college-educated residents by 10 percentage points. As a result of connecting new voters to the political world, the redistricting process may help create a more dynamic electorate that more closely reflects the demographic composition of the state’s citizen population. In creating majority-minority districts, mapmakers should be mindful that turnout among minority voters is high wherever they are able to play a meaningful role in political life. Greater participation among African Americans is observed when they are not relegated to the political margins in a district dominated by a single nonblack racial or ethnic group—white or nonwhite. Whereas Latino registered voter turnout is highest in districts where they enjoy overwhelming majority status, it is also high in districts where they are on roughly equal footing with African Americans. Even in the more heterogeneous areas, Latinos continue to participate at rates that exceed those in majority-Anglo districts. viii In all of these scenarios, Anglo participation does not suffer. Indeed, white voting-age participation is slightly higher in majority-Latino districts than in districts with Anglo majorities. In short, this evidence suggests that there may be some advantage to putting more emphasis on the creation of black-Hispanic and multiethnic districts in the next redistricting round. These districts would allow for multiple racial and ethnic communities to exercise some political leverage—precisely the kind of political environment that encourages participation. Given the democratic value of an engaged citizenry, not to mention the role that politics historically has played in the advancement of racial and ethnic groups in the United States, participation is a worthwhile if not paramount issue. ix Contents Foreword ........................................ Summary ........................................ Figures ......................................... Tables .......................................... iii v xiii xv 1. INTRODUCTION ............................. 2. MAJORITY-MINORITY DISTRICTS AND MINORITY REPRESENTATION ........................... The Voting Rights Act and Its Mandates .............. Majority-Minority Districts in the 1990s .............. Minority Representation in the 1990s................. 3. MINORITY VOTER PARTICIPATION ............. Trends in Minority Voter Participation ............... Constraints on Minority Voter Participation ............ Some Hypotheses About Majority-Minority Districts and Minority Voter Participation ................... The Psychology of Empowerment.................. Targeted Mobilization .......................... An Artifact of the Redistricting Process .............. 4. DATA AND METHODOLOGY ................... Sample Year ................................... Unit of Analysis and Measures ...................... Methodology .................................. 5. RESEARCH FINDINGS ......................... General Findings ............................... Racial Differences in Voter Turnout ................ Aggregate Turnout Differences Across District Type ..... Socioeconomic and Political Influences on Voter Turnout .................................. Majority-Minority Districts and Voter Turnout.......... 1 7 7 8 12 17 17 18 21 22 23 24 27 27 32 37 41 41 42 45 48 50 xi Majority-Minority Districts, Minority Representation, and Voter Turnout ............................. Summary of Findings ............................ 6. POLICY IMPLICATIONS........................ Appendix A. Sample ...................................... B. Methodology .................................. C. District-Level Turnout Estimates with Standard Errors..... D. Regression Output .............................. References ....................................... About the Author .................................. Other Related PPIC Publications ....................... 55 61 63 67 69 75 79 83 95 97 xii Figures 2.1. Majority-Minority Districts in California ............ 11 2.2. Minority Districts and Minority Representatives Nationwide, 1962–1998 ....................... 13 5.1. Voting-Age Turnout by District, November 1994 ..... 42 5.2. Registered Voter Turnout by District, November 1994 ..................................... 43 5.3. Anglo Voting-Age Turnout by District, November 1994 ..................................... 44 5.4. Black Voting-Age Turnout by District, November 1994 ..................................... 45 5.5. Latino Registered Voter Turnout by District, November 1994 ..................................... 46 5.6. District-Level Voter Participation in Majority-Minority Districts, November 1994 ...................... 47 5.7. Predicted Voter Participation in “Average” Precinct, by District Type ............................... 54 5.8. Predicted Voter Turnout in “Average” Precinct, by District Type and Legislator Race................. 60 C.1. District-Level Estimates of Anglo Turnout with Uncertainty ................................ 76 C.2. District-Level Estimates of Black Turnout with Uncertainty ................................ 77 C.3. District-Level Estimates of Latino Turnout with Uncertainty ................................ 78 xiii Tables 2.1. Number of Majority-Minority Districts After 1990 Redistricting, Selected States .................... 3.1. Racial and Ethnic Composition of the Electorate and Adult Population, 1990–1998 ................... 4.1. Majority-Minority Districts and Minority Representatives in California, November 1994 ....... 5.1. Estimates of Voter Participation Differences Across District Types .............................. 5.2. Estimates of Voter Participation Differences Across District Type and Legislator’s Race................ A.1. Number of Precinct Observations in Dataset, by District ................................... B.1. Ecological Inference for Black/Nonblack Voter Turnout .................................. B.2. Ecological Inference for White/Nonwhite Voter Turnout .................................. B.3. Ecological Inference for Latino/Non-Latino Voter Turnout .................................. B.4. Definition of Variables ........................ D.1. Coefficient Estimates for Regression of Majority- Minority Districts on Voting Participation .......... D.2. Coefficient Estimates for Regression of Majority- Minority Districts and Minority Representation on Voting Participation .......................... 10 17 33 50 56 67 70 70 71 73 80 81 xv 1. Introduction Since the passage of the Voting Rights Act of 1965 (VRA), the number of Latino and African-American elected officials at the local, state, and federal levels has increased dramatically.1 According to the most recent figures, more than 14,000 African Americans and Latinos hold political office across the country (Bureau of the Census, 1999). In California alone, there are more than 1,000 African-American and Latino elected officials, up from 700 officeholders in the early 1980s, including 24 Latino and six African-American state legislators and more than 300 Latino city council members—100 from Los Angeles County alone (Guerra, 1998; Bureau of the Census, 1999; Reyes, 2001). Minority members of the U.S. House of Representatives are among the most influential and prominent class of these new officeholders. Benefiting in part from the creation of majority-minority electoral districts under the provisions of the VRA, minority legislators as a group have emerged as visible political actors in an institution traditionally dominated by whites.2 California is home to one of the nation’s most ethnically and racially diverse congressional delegations. In the 106th Congress, seated after the November 1998 congressional elections, California’s U.S. House delegation included four African-American legislators, five Latino legislators, and one Asian legislator. The California delegation alone ____________ 1In this report, I use the terms Latino and Hispanic interchangeably to indicate anyone of Hispanic origin. I use the terms African-American and black interchangeably. I also use the terms white and Anglo to refer to non-Hispanic whites. This report focuses on Latinos and African Americans; Asian legislators and Asian constituents are not discussed in this research, both because there are so few Asian legislators and because the lack of sufficient geographic concentrations of Asian constituents (i.e., comparatively low levels of residential segregation) poses significant methodological challenges. 2Majority-minority electoral districts are defined as any district in which racial or ethnic minorities (Asians, African Americans, and Latinos) constitute more than 50 percent of the total district population. 1 accounted for roughly 18 percent of all minority elected officials serving in the House of Representatives. With one exception, these members of Congress represented districts drawn so that racial and ethnic minorities would constitute the majority of the voting-age population. The rise in the number of minority representatives and the use of race-conscious districting mechanisms to secure their election have prompted considerable debate over whether this greater diversity has translated into substantive benefits for minority constituents. To date, political scientists have focused primarily on the links among minority representation, the legislative process, and public policy outcomes favorable to minority communities (Swain, 1995; Cameron, Epstein, and O’Halloran, 1996; Lublin, 1997; Whitby, 1998; Hutchings, McClerking, and Charles, 1999; Canon, 1999). Of particular concern is the extent to which minority legislators and white legislators of the same political party differ in the quality of issue representation—as measured by bill sponsorship and legislative voting record—they provide for minority constituents.3 On this question the empirical record is mixed. Although several studies have found evidence of greater responsiveness and advocacy among minority legislators—particularly on issues with a racial dimension (Canon, 1999)—others conclude that, on the whole, there are no appreciable racial differences in legislative activism. Complicating the relationship further is mounting evidence that the practice of concentrating minority voters so as to ensure the election of minority legislators may produce deleterious unintended consequences. In particular, adjoining congressional districts, deprived of a significant number of minority voters, may increasingly elect representatives less favorably disposed to minority interests, contributing to an aggregate decline in the responsiveness of the House (Lublin, 1997). In short, not only may direct issue representation (i.e., correspondence between constituents’ policy preferences and a legislator’s policy activism) not depend on descriptive representation (i.e., correspondence on demographic traits), but efforts to secure descriptive representation may ultimately undermine aggregate issue representation. ____________ 3In addition to questions of legislative activism, Swain (1995) and Canon (1999) also consider racial and ethnic differences in the provision of constituency service. 2 After more than a decade of research, scholarly consensus on these questions remains elusive. However, there is broad agreement that minority representation is significant primarily as a guarantor of continued minority influence on the legislative process. This scholarly consensus has produced a body of research that focuses narrowly on legislative behavior and policy outcomes, to the exclusion of the other benefits or costs that may emanate from minority representation. The merits of minority representation and, by extension, the merits of efforts such as race-conscious districting to secure greater minority representation generally have been reduced to a question of policy responsiveness. If white officeholders were found to be equally likely to advocate for minority policy preferences, African-American and Latino officeholding would become less significant. However, such a conclusion may be premature. For the community activists, citizens’ groups, and voting-rights’ attorneys who have worked on redistricting and have pressed for the strict enforcement of the VRA, the creation of majority-minority electoral districts is widely viewed as more than a vehicle for progressive legislation.4 These advocates also believe that majority-minority districts can pull the Latino and AfricanAmerican communities into the political process. In particular, raceconscious districting and the subsequent election of minority representatives may “prime the pump of [minority] voter participation” and “reverse the historic trend” of lower voter registration and turnout (Los Angeles Times, August 26, 1990; see also New York Times, April 12, 1992; Brischetto, 1998). In California, the possibility of greater minority voter participation is especially significant. To many observers, chronically low voter turnout among Latinos is one of the most striking features of the state’s political landscape. As one journalist remarked, “Latino electoral power [exists] more in theory than in fact” (Scott, 2000). Latino leaders have identified this absence at the ballot box as the chief obstacle to real political influence. As long as Latinos remain on the margins of the electoral ____________ 4The redistricting process involves the redrawing of political boundaries (e.g., Assembly, Senate, congressional lines) according to population and other criteria. Redistricting typically accompanies the decennial reapportionment, when congressional seats are redistributed among the various states to reflect population shifts. 3 process, there will be a disjunction between their status as the state’s largest minority group and their actual influence on state politics. With public policy in California increasingly dictated by the “new populism” (Cain, 1992) of the initiative process, rather than by deliberation and compromise in the state legislature, low Latino voter turnout translates directly into limited influence on issues as important and varied as school vouchers, campaign finance, environmental regulation, and affirmative action. If majority-minority electoral districts can lay the groundwork for a more engaged Latino electorate, advocates argue, they may also bring the political dynamics and policy priorities of the state more in line with its demographic trends. These expectations have been met with some skepticism. Because the majority-minority districts served by most minority members of Congress typically have low aggregate turnout rates, many have concluded that neither race-conscious districting nor minority representation has made voters out of previously unengaged minority Americans (Duncan, 1990, 1993a, 1993b; Donovan, 1992). In the 1998 midterm elections, for example, one majority-minority district in Los Angeles County cast fewer than 70,000 ballots in the congressional race—less than half the number of ballots cast in a neighboring district with an Anglo majority (California Journal, 1998).5 This disparity hardly conforms to the small-scale revolution in electoral politics envisioned by advocates. Furthermore, it is unclear what minority districts and minority representation may mean for the willingness of nonminority constituents to remain involved in politics. Neither advocates nor skeptics have offered more than anecdotal evidence or journalistic speculation to bolster their claims regarding the significance of race-conscious districting and minority representation for political participation. Moreover, political scientists have failed to bring their customary empirical rigor to this question of the intersection of representation and political engagement. There has been no systematic investigation of racial and ethnic differences in voter turnout under ____________ 5Of course, advocates rightfully counter that these disparities are confounded by the social and economic dislocation characteristic of majority-minority congressional districts such as the 30th in southeast Los Angeles County. 4 minority representation or in the majority-minority congressional districts in California. Even after a decade that witnessed historic gains in the number of minority members of Congress, and equally important increases in turnout among Latinos, very little is known about the relationship between the two. Yet as the results from Census 2000 are tallied and California again prepares for congressional redistricting, these issues will resurface in the larger debate over efforts to secure greater political representation for Latinos and African Americans in the state. An assessment of the 1990s may afford a fuller picture of the political dynamics in majority-minority districts and thus facilitate a more informed discussion in 2001. This report offers a framework for assessing the relationship between majority-minority districts and minority representation, and voting participation among Latinos, African Americans, and Anglos in California. This report does not take a position in the debate over issue representation. Nor does it suggest that nonpolicy considerations be placed above questions of policy responsiveness as the merits of districting arrangements are evaluated. Instead, the report addresses the following specific questions to illuminate the potential nonpolicy implications of minority districts and minority representation: • How does the propensity to vote vary across racial and ethnic groups? How does it vary across congressional districts in California? • Does the level of Latino, African-American, and Anglo voter participation vary with the racial and ethnic balance of a congressional district? If so, how? • Do voter participation rates in the majority-minority congressional districts served by minority members of Congress differ from the rates observed in the majority-minority districts represented by Anglo legislators? If so, how? • What, if any, difference exists between Latinos and African Americans in the relationship between voter turnout and type of congressional district? The analysis provides empirical support for the claim that minority representation and majority-minority districts are associated with greater 5 involvement in electoral politics among Latinos and African Americans.6 The results suggest that the provisions of the VRA (namely, the creation of majority-minority electoral districts and the subsequent election of minority legislators) go hand in hand with a more dynamic political life and, by extension, an electorate that more closely mirrors the demographic makeup of the state. As such, this research provides a basis for assessing the implications of future efforts to increase minority representation. ____________ 6Throughout this report, I will refer to the relationships among voter turnout, minority districts, and minority representation as just that—a “relationship.” I will avoid the use of terms such as “effect” and “impact,” since the direction of causality is difficult to establish conclusively. However, the evidence favors an interpretation that places minority districts and minority representation causally prior to minority voter participation. 6 2. Majority-Minority Districts and Minority Representation The Voting Rights Act and Its Mandates The Voting Rights Act of 1965 is often cited as one of the most significant pieces of civil rights legislation passed in our nation’s history. Originally designed to give southern blacks access to the voting process by temporarily taking away from certain states the right to determine their own voting procedures, the VRA has been reshaped into a potent tool to expand opportunities for minority representation nationwide and at every level of government. When signed into law by President Lyndon B. Johnson, the VRA (PL 89-110) was narrowly concerned with eliminating such direct obstacles to black political participation as poll taxes and discriminatory literacy tests. The central parts of the measure are Section 2 and Section 5. Section 2 reiterates the guarantees of the 15th amendment, prohibiting any state or political subdivision from adopting voting practices that “deny or abridge the right of any citizen of the United States to vote on account of race or color.” Section 5, imposed only on “covered” jurisdictions with a history of past discrimination, requires Justice Department preclearance of changes in any electoral process or mechanism.1 In 1970, 1975, and 1982, Congress renewed and expanded the provisions of the Voting Rights Act. Three changes have particular relevance for California: ____________ 1“Covered” jurisdictions were originally defined as states in which less than 50 percent of the voting-age population either registered or voted in the 1964 presidential election and that had various discriminatory prerequisites for voting as of November 1, 1964. 7 First, in 1975, the special protections of the act were expanded to include not only African Americans but also Hispanics, Native Americans, Asians, and Alaskan Natives. Second, in 1982, Congress amended Section 2 of the law to stipulate that members of a protected class must have an equal opportunity “to participate in the political process and to elect representatives of their choice” [emphasis added]. Thus, the law was no longer meant to apply strictly to discriminatory practices; it was also to serve as a vehicle for voting mechanisms that could increase minority representation (New York Times, February 14, 1994). The intention was to ensure that “protected” racial and ethnic minorities had a greater voice in governing the country. Subsequently, the Supreme Court issued a ruling (Thornburg v. Gingles) that was widely interpreted to mean that states must create majority-minority districts wherever possible to comply with Section 2 as amended. The Justice Department adjusted its preclearance criteria accordingly. The drawing of new majority-minority electoral districts (from city council districts to congressional districts) became central to the strategy for advancing minority representation. Last, the list of “covered” jurisdictions, at first narrowly targeted to include just the six states of the Deep South (Alabama, Georgia, Louisiana, Mississippi, South Carolina, and Virginia), has continually evolved with various “bailouts” and additions.2 Currently, 22 states have covered jurisdictions, including four counties in California (Merced, Yuba, Kings, and Monterey) with a history of low Latino registration. For these four counties, reapportionment of electoral districts after the 2000 Census will require Justice Department approval. Majority-Minority Districts in the 1990s Congressional efforts to broaden the mandate of the VRA bore fruit in the creation of dozens of new majority-minority congressional districts during the 1990 redistricting cycle. In the aftermath of the 1990 ____________ 2States or jurisdictions can “bail out” from coverage if they can prove the absence of discriminatory practices. 8 reapportionment, the total number of districts nationwide with AfricanAmerican or Hispanic majorities doubled, prompting political scientist David Lublin to characterize the process as “the greatest upheaval in congressional district boundaries since the Supreme Court forced states to minimize population deviations between districts in the mid-1960s” (Lublin, 1997). In California, where reapportionment following the 1990 Census added seven new congressional districts to the state, the new districting plan—developed by a court-appointed panel of retired judges after Governor Wilson failed to reach a compromise with the legislature— placed the state first in the nation in the total number of majorityminority districts (13) and second only to Texas in the number of districts with a Hispanic majority (six, up from three in the 1980s). As was the case in 1981, California’s 1991 congressional districting plan may have been influenced as much by the preferences of the powerful Speaker of the Assembly Willie Brown as it was by pressure to comply with the VRA mandates.3 Table 2.1 presents details on the number of majority-minority districts created nationwide after 1990. In addition to the six congressional districts with Latino majorities, California’s plan created three districts where Latinos and African ____________ 3Since the 1980s, the majority of voting-rights litigation in the state has been concentrated at the local level, where plaintiffs—supported by organizations such as the Mexican-American Legal Defense and Education Fund—have aggressively challenged the use of electoral arrangements that they view as discriminatory in elections to city councils, county boards of supervisors, school boards, and other political offices (Bucy, 1991; Erie, Brackman, and Ingram, 1993; Scott, 1998; Guerra, 1998; Brackman and Erie, 1998). These efforts have enjoyed considerable success. “By the early 1990s,” Brackman and Erie (1998) observe, “the threat and reality of voting rights litigation were largely responsible for the presence of three Latinos on the Los Angeles City Council and one on the County Board of Supervisors.” The same might be said of minority representation in cities such as Watsonville, where a 1987 court challenge to the city’s at-large electoral system eventually led to the district-based election of three Latino council members. Pressure to comply with the expanded mandates of the VRA continued to influence local districting arrangements throughout the 1990s. In fact, in at least one city, Santa Maria, voting-rights litigation initiated at the start of the decade remains unresolved even as we enter the 2001 redistricting cycle. 9 Table 2.1 Number of Majority-Minority Districts After 1990 Redistricting, Selected States Majority- Majority-Black State Hispanic Districts Districts Alabama — 1 Arizona 1— California 6 — Floridaa 2 3 Georgiaa — 3 Illinois 13 Louisianaa — 2 Maryland — 2 Michigan — 2 Mississippi — 1 Missouri — 1 New Jersey — 1 New Mexico — — New York 2 3 North Carolinaa — 2 Ohio —1 Pennsylvania — 2 South Carolina — 1 Tennessee — 1 Texasa 71 Virginia — 1 Multi-Ethnic Districts — — 7 — — — — — — — 1 1 2 — — — — — — 1 — Total MajorityMinority Districts 1 1 13 5 3 4 2 2 2 1 2 2 2 5 2 1 2 1 1 9 1 Total 19 31 12 62 SOURCES: Congressional Quarterly, Inc. (1993); Duncan (1993c). NOTE: Multi-ethnic districts are districts where the majority of the population is not non-Hispanic white and neither blacks nor Hispanics constitute a majority. aSupreme Court challenges eventually led to the elimination of one majority-black district in Florida, Louisiana, North Carolina, and Texas; two majority-black districts in Georgia; and one majority-Hispanic district in Texas. Americans together constituted a majority of the population. And in four congressional districts, Latinos, Asians, and blacks made up a multiethnic majority. In no districts in California do African Americans or Asians, alone, account for the majority of the population. With three exceptions (Oakland’s 9th, San Jose’s 16th, and the Central Valley’s 10 Anglo majority Latino majority Black-Hispanic majority Multi-ethnic majority Figure 2.1—Majority-Minority Districts in California 20th), these majority-minority districts are concentrated primarily in and around the Los Angeles County area.4 (See Figure 2.1.) According to 1990 Census data, 51 percent of all African-American adults in California live in a majority-minority district (60 percent of these adults live in just the three districts with combined black-Hispanic majorities). These same data estimate that 46.5 percent of Latino adults live in majority-minority districts; 60 percent of these adults are concentrated in the six districts with Latino majorities. ____________ 4At the time these district lines were drawn, Los Angeles County was home to 44 percent of the state’s Latino population (Pachon, 1998). Today that percentage is estimated to be 41 percent (California Opinion Index, 2000). 11 The 13 majority-minority districts in California are a diverse group. They vary tremendously not only in their racial and ethnic balance but also in their socioeconomic conditions. At one end of the spectrum are districts such as the 33rd, in “the dreary center of the Los Angeles Basin” (Green, 1995), where noncitizens significantly outnumber registered voters and both unemployment and crime are relatively high. The 35th District faces similar levels of economic dislocation. At the other end of the spectrum are districts such as the 34th, centered in West Covina and the suburbs east of Los Angeles County, which has the distinction of being one of the most racially and ethnically diverse middle-class districts in the state. This group also includes Orange County’s 46th District, which, in the last 15 years, has changed from a largely white area to the home of the nation’s largest concentration of Vietnamese Americans. And, of course, among the oldest multi-ethnic districts in the state is the East Bay’s 9th, a district that includes both struggling Oakland and the more affluent city of Berkeley. The 1990s reapportionment gains extended well beyond federal legislative boundaries to include new majority-minority electoral districts at the state and local levels as well. For example, the plan developed by the court-appointed panel also included 22 Assembly districts and seven state Senate districts where Asians, Latinos, and African Americans, collectively, accounted for the majority of the voting-age population.5 At the local level, where precise summary statistics are difficult to come by, the increase in the number of majority-minority electoral districts— particularly in the most populous cities and counties in the state—was significant enough to lead one observer to conclude that the “1990s reapportionment was the most favorable reapportionment for the Latino community in California’s history” (Guerra, 1998). Minority Representation in the 1990s The creation of majority-minority districts under the aegis of the VRA has contributed to substantial gains in the number of African ____________ 5An additional four Assembly and five state Senate districts have majority-minority total populations but not majority-minority voting-age populations. 12 Americans and Latinos elected to the U.S. House of Representatives, from California and elsewhere. Figure 2.2 demonstrates that the growth in the number of minority legislators has closely tracked the increase in the number of districts with black or Hispanic majorities. The “upheaval” of redistricting in the 1990s, in particular, made the House much more racially and ethnically diverse. (Note the steep rise after 1990 shown in Figure 2.2.) The freshman class of 1992 included 16 African Americans and eight Latinos, all but five of whom were elected from newly created majority-minority congressional districts. So dramatic was the increase in the number of minority legislators that one journalist concluded that the new diversity would represent “a big step in the unfolding of a more democratic and perhaps more tolerant era in elective representation” (New York Times, November 2, 1992). One freshman legislator predicted that “the dialogue will open and issues that members [of Congress] did not have a sensitivity about before will be discussed” (New York Times, November 2, 1992). 45 40 Black members of Congress 35 Majority-black districts Majority-Hispanic districts 30 Latino members of Congress Number 25 20 15 10 5 0 1962 1966 1970 1974 1978 1982 Election year 1986 1990 1994 1998 SOURCES: Data on majority-black and majority-Hispanic districts are from Bureau of the Census (1961, 1963, 1971, 1973, 1983, and 1993). Additional district data and data on minority representatives are from Congressional Quarterly, Inc. (1993, 1996, and 1997); Duncan (1993c, 1995); and Ehrenhalt (1983). Figure 2.2—Minority Districts and Minority Representatives Nationwide, 1962–1998 13 In California, all of the African-American and Latino House members during the 1990s hailed from majority-minority congressional districts. However, the increase in the number of minority legislators in the immediate aftermath of the 1990 redistricting cycle was not as dramatic as that observed elsewhere in the country. Although the state’s new congressional map doubled the number of Hispanic-majority districts, a Latino congressman subsequently represented only one of these three new districts. Observers blame low registration and turnout among Latino residents for the failure to capitalize on the opportunity for minority representation in two of the three new Hispanic-majority districts. Later in the decade, the California delegation gained one additional Latino lawmaker when Loretta Sanchez was elected in 1996— and re-elected in 1998—to represent the 46th District, a majorityHispanic-and-Asian congressional district in Orange County. The racial and ethnic balance of state and local governments also shifted in the aftermath of the 1990s reapportionment, with more favorable implications for Latinos than for African Americans. Within one election cycle (1992-1993), Latino representation in the 300 most significant elective positions in the state had increased by almost 50 percent (Guerra, 1998).6 (By 1997, Latino representation had doubled from its 1989 level.) Some of the greatest gains were in the Assembly where the number of Latino legislators increased from four in 1990 to seven following the 1992 election (eight when Cruz Bustamante’s victory in a 1993 special election is included). The number of Latino held seats in the legislature continued to increase dramatically in the 1990s (to 17 Assembly members and seven senators by 1998) despite, as one observer astutely noted, “only slow growth in the number of Latino voters” (Sacramento Bee, November 30, 2000). At the local level, Latino ____________ 6Guerra (1998) identifies the 300 most significant elective offices (out of 10,000 total) in the state on the basis of “the number of constituents, the budgets under their control, and the strategic position of the office given its perceived prestige by individual officeholders.” These 300 positions include the following: constitutional officers (governor, lieutenant governor, attorney general, etc.); board of equalization; U.S. House and Senate; California Assembly and Senate; board of supervisors for the seven most populous counties; and city council for the seven most populous cities. 14 representation on the city councils of the seven most populous cities in the state increased from five council positions (out of 76 total) in 1989 to 11 after the 1993 municipal elections (Guerra, 1998). (This included new Latino-held seats in San Diego, Long Beach, and Oakland, where Latinos previously had had no representation.) For African Americans, who began the 1990s somewhat overrepresented in elected office (7 percent of the state population but 10 percent of the 300 most significant elective offices), the 1990s reapportionment did not contribute to gains in representation at the state or local level (Guerra, 1998). In fact, by 1996, African Americans held three fewer seats in the state Assembly than they had in 1990. (Each of those three African-American Assembly members was replaced by a white legislator.) 15 3. Minority Voter Participation Trends in Minority Voter Participation Low rates of voter registration and turnout characterize minority political behavior in California, with Latinos in particular remaining largely on the sidelines of electoral politics. Each of the statewide elections in the 1990s witnessed not only low overall turnout (especially compared to the participation “heydays” of the 1950s and 1960s) but also a pronounced difference in the demographic composition of the state’s adult population and its pool of voters (see Table 3.1). For Table 3.1 Racial and Ethnic Composition of the Electorate and Adult Population, 1990–1998 (in percent) Total turnouta 1990 1992 1994 1996 1998 41 54.5 47 52.6 41 Electorate Non-Hispanic white 81 79 78 77 74 Latino 4 10 9 11 13 Black 8 6 7 67 Asian 4 4 6 46 Adult population Non-Hispanic white 63 59 59 57 57 Latino 21 24 24 27 25 Black 7 7 7 67 Asian 9 9 10 10 11 SOURCES: California Opinion Index (1990, 1993, 1995, 1997, 1999); figures for 1990 “Electorate” are from Baldassare (2000). aTotal turnout is measured as a percentage of the state’s citizeneligible population. 17 example, in November 1998, non-Hispanic whites accounted for 57 percent of the state’s adult population and 74 percent of the electorate. At the same time, Latinos constituted 25 percent of the California adult population but accounted for only 13 percent of the ballots cast. African Americans are the only major racial group (minority or nonminority) whose political clout is on a par with its population numbers. Even so, African Americans participate at lower rates than do non-Hispanic whites (Reyes, 2001). For Latinos, the disparity at the ballot box in 1998 actually marked a significant improvement over the first gubernatorial election of the decade, when they accounted for only 4 percent of the electorate. This participation gap is widely recognized as the most fundamental barrier to minority political empowerment in the state. A number of factors contribute to “the [lag] in commensurate influence at the ballot box” (Erie, Brackman, and Ingram, 1993). Constraints on Minority Voter Participation Social scientists have devoted decades of research to identifying the influences on an individual’s decision to participate (Verba and Nie, 1972; Brody, 1978; Wolfinger and Rosenstone, 1980; Teixeira, 1987; Conway, 1991; Verba, Schlozman, and Brady, 1995). Whether concerned exclusively with voting or with participation more generally (including campaign activity, for example), the findings have consistently highlighted the significance of three individual-level factors: age, education, and income. The positive relationship between turnout and these measures of socioeconomic status may well be the most robust finding in the field of political science. The influence of education on participation is thought to derive both from education’s indication of social position and from the tangible resources it provides. “Education imparts citizenship values,” argue Rosenstone and Hansen (1993), “and so the educated . . . [are] composed of people . . . who have been socialized to follow politics and to value and reward political involvement.” Formal schooling also imparts the knowledge and skills necessary to understand and follow political campaigns and to negotiate the bureaucratic requirements of registration and voting. 18 Similarly, income facilitates participation both by what it indicates (social position) and for what it allows (the ability to bear the material and opportunity costs of political involvement). For low-income adults, the money, time, and energy spent making ends meet not only limit the resources (e.g., leisure time) available for political pursuits but also provide payoffs that are more immediate and valuable than the potential benefits of investing in electoral politics (Wolfinger and Rosenstone, 1980). As one journalist put it, in trying to explain low voter turnout among poor adults, “the notion that their lives might be made better by casting a ballot is as fanciful as winning the lottery” (Los Angeles Times, November 28, 1994). Finally, older adults are typically more involved in politics than younger adults. Turnout is highest, in fact, among those who are age 65 and older. As people age, they are thought to become more familiar with the political process, more knowledgeable about politics, and more attached to the parties and their candidates.1 This experience both facilitates and encourages electoral involvement. As pollster Richard Maullin notes, “politics just isn’t as critical” for the young (Lopez and Wahlgren, 1994). On each of these measures, African Americans suffer from a comparative disadvantage compared to non-Hispanic whites in the state. The African-American community as a whole has higher concentrations of young, undereducated, and low-income adults. In 1997, 73 percent of the African-American population were under age 45 compared to 62 percent of non-Hispanic whites (California Department of Finance, 1999). In 1998, the four-year high school dropout rate in California was 17.4 percent for African-American students in the state’s public schools but only 7.5 percent for non-Hispanic whites (Education Data Partnership, 2001). ____________ 1An alternative hypothesis to explain the positive correlation between age and participation is generational, where the differences in participation are artifacts of the socializing experiences of each generation. “People who [came] of age at critical turning points that define political allegiances [Great Depression, New Deal] participate more in electoral politics than people who reach maturity in more ordinary times” (Rosenstone and Hansen, 1993). Generational hypotheses have not fared well in the literature. 19 The Latino population is the state’s youngest, with an average age of 24 (Lopez and Wahlgren, 1994). Not only are many Latinos too young to vote but, among those who are age-eligible, a significant number (36 percent) fall into the age 18–29 cohort that is the least likely to participate (California Opinion Index, 2000). Only one in three Latinos is over age 40. Levels of educational attainment are also low. Forty-four percent of Latino adults are not high school graduates and only 9 percent hold college degrees (California Opinion Index, 2000). In 1998, the four-year dropout rate for Latino public school students was 16.3 percent (Education Data Partnership, 2001). Finally, many Latinos have low incomes. Forty percent of Latino adults report annual household incomes of less than $20,000 (California Opinion Index, 2000). Thus, to a great degree, low rates of minority registration and voting reflect the socioeconomic bias in electoral participation. Even organized registration and voting drives can fail to overcome these barriers: Voter registration campaigns targeted at Latinos in the state have been more successful in stable, working-class suburbs than in the poorer inner-city neighborhoods where many Latinos live (Brackman and Erie, 1998). The Latino community also includes a large noncitizen population. Pachon (1991) speculates that “over 80 percent of the difference in voting between Latinos and black Americans can be explained by noncitizenship among Latino adults.” Estimates place the number of noncitizens between 41 percent and 52 percent of all Latino adults. (The decennial Census does not provide detail on the racial and ethnic makeup of the noncitizen population.) These concentrations create what Brackman and Erie (1998) describe as “an inverted pyramid of electoral participation,” in which “a small minority voting universe rests upon a large minority population base.” Compounding these socioeconomic barriers, some have argued, has been a political environment in which the two major parties for many years spent little time trying to energize Latinos (Maharidge, 1994). As one political observer describes it, Latinos in California have been dismissed as “an occasionally raucous but mostly non-voting constituency” (Frayne, 1998). During the 1996 elections, for example, Republicans in the state legislature backed legislation to distribute official 20 election materials in English only, while Democratic pollsters in one Assembly district hung up the telephone when they reached non-Englishspeaking households—choosing to ignore the opinions of residents they presumed did not vote (Orange County Register, October 22, 1998). Guided by an election calculus that discounted Latinos as a “low propensity ‘expendable’ constituency,” campaign organizers often did not invest the resources necessary to mobilize this population (Scott, 2000). By the 2000 presidential election, there was a noticeable shift in political strategy, with both major party candidates assiduously courting Latino voters. There are some signs that Latino political participation may be on the rise as the Latino proportion of the electorate increases (see Table 3.1). The number of Latino voters increased from 1.35 million in 1990 to 2.35 million in 2000, accounting for most of the 1.1 million net increase in the state’s registered voter rolls over the decade (California Opinion Index, 2000). Much of this growth has taken place in the last six years; nearly half of the Latino adults who were registered to vote in time for the March 2000 primary had registered after 1994. Analysts generally attribute the participation upswing to furor over the passage of Proposition 187, the Illegal Immigration Act. Perhaps not surprisingly, naturalized citizens accounted for a larger share of the voters who registered after 1994 than of those who registered before 1994. Some Hypotheses About Majority-Minority Districts and Minority Voter Participation Advocates of minority representation regard the redistricting process and minority representation as instrumental to the political mobilization of the Latino and African-American communities. This view presumes that political engagement is as much a response to the political environment and the opportunities it is perceived to present as it is a function of individual resources (e.g., education, income). Political scientists, for their part, have articulated similar models of political behavior (although they have not extended them to the study of minority representation), increasingly moving beyond the strict individualist 21 perspective that has dominated the research for decades to consider the influence of political context on the likelihood of voting. The scholarly mission has been to “reconstruct a compelling account of political life” by recognizing the contingent nature of political choices and by identifying the aspects of a political environment “that make people accessible and amenable to the appeals of political leaders” (Huckfeldt and Sprague, 1993). In practice, the environmental constraints and opportunities can be as varied as registration laws that ease or complicate political participation (Nagler, 1991; Timpone, 1998), salient political campaigns that inspire activism (Tate, 1991), or public policy reforms that spur record rates of naturalization (Shaw, de la Garza, and Lee, 2000; Washington Post, April 12, 1995). This logic could be extended to ask whether the political status of African Americans and Latinos correlates with their levels of political engagement. Findings from research on minority officeholding at the local level suggest that this kind of “empowerment” and political participation tend to go together (Bobo and Gilliam, 1990; Gilliam, 1996; Gilliam and Kaufmann, 1996). In the most complete research to date, Bobo and Gilliam (1990) found that African Americans in cities with black mayors are more active than their white counterparts of comparable socioeconomic status and more active than African Americans in cities with white mayors. This pattern suggests that minority representation in local government is significant not only for its implications for public policy but also for its ability to bring minorities in from the political margins. At least two plausible hypotheses lead us to predict a similarly positive relationships among majority-minority districts, minority representation, and levels of political participation. The Psychology of Empowerment Voting-rights advocates link the participatory benefits of the redistricting process to its role in “creating a climate of inclusion” and “showing [minorities] they can elect their own representatives” (New York Times, April 12, 1992 ). Likewise, Bobo and Gilliam (1990) conclude that minority officeholding at the local level influences black political participation by contributing to a more trusting and efficacious 22 orientation toward politics (see also Abney and Hutcheson, 1981; Howell and Fagan, 1988) and by greatly increasing black attentiveness to and interest in political affairs. Thus, majority-minority districts and minority representation may increase political confidence among Latinos and African Americans; this confidence, in turn, encourages greater political engagement. (Conversely, these conditions would be expected to lead non-Hispanic whites to participate less in political life.) This “reservoir of psychic benefits,” as Gilliam and Kaufman (1996) describe this effect, may derive from purely symbolic considerations—for example, the group pride stoked by a “brown face in a high place,” as one activist phrased it—or from an appraisal of policy responsiveness (Lopez and Wahlgren, 1994). Targeted Mobilization An alternative hypothesis is that neither psychological attachments nor the tangible gains associated with minority representation contribute to greater political engagement. Rather, heightened engagement may be the by-product of a new electoral dynamic in which Latino and AfricanAmerican communities become the targets of aggressive mobilization. These communities may benefit when minority political aspirants, guided by an election calculus that defines Latinos and African Americans as their natural constituencies, make special efforts to engage and court this core of voters. Add to this the strategy of independent political organizations, such as the Southwest Voter Registration Education Project, the NAACP, and the National Association of Latino Elected Officials, who, in an attempt to maximize the “bang for the buck,” focus their minority voter registration and turnout efforts in the districts with the highest concentrations of Latinos and African Americans. As research on mobilization has demonstrated, people are more likely to participate in politics if they are asked (Verba, Schlozman, and Brady, 1995), or if their stake in politics is made evident to them (Rosenstone and Hansen, 1993). Majority-minority districts may facilitate and encourage this outreach. If so, higher turnout in majorityminority districts and under minority representation is at least partly the result of the strategic calculations of political entrepreneurs. 23 An Artifact of the Redistricting Process There is a third, but less plausible, reason to predict a positive relationship between majority-minority districts and minority voter participation. A pattern of heightened minority political participation may simply be an artifact of the redistricting process itself, especially if the majority-minority districts in the state have been deliberately constructed around high-propensity minority voters. Thus, voter turnout affects the likelihood of being in a majority-minority district rather than the other way around. If so, there is very little to be said about the participatory benefits of race-conscious districting. Even if voting propensity affected districting, which, in turn, affected voting propensity, that pattern would be difficult to isolate and quantify. Reversing the causal arrow casts doubt on the utility of any empirical investigation into this relationship. To evaluate the merits of this hypothesis, I contacted the Mexican American Legal Defense and Education Fund (MALDEF), which has played a very prominent role in redistricting in California. Given their interest in maximizing the opportunities for Latino officeholding, one would expect MALDEF to be particularly mindful of Latino voter turnout rates and particularly supportive of a strategy that explicitly builds majority-minority districts around high-propensity voters. In an interview with Denise Hulett, a senior litigator who has been with the organization for 17 years, I learned that MALDEF has not pursued such a strategy—if only because of lack of resources.2 Indeed, the chief concern for MALDEF is to avoid creating “phantom” majority-minority districts—districts in which Latinos constitute the majority of the total population but, because of high rates of noncitizenship, do not make up the majority of the voting-eligible population. As a consequence, these “phantom” districts may be majority-minority on paper but majorityAnglo in practice. The challenge facing MALDEF in their efforts to prevent the creation of these districts is that the Census does not provide citizenship data disaggregated by race and ethnicity. In lieu of citizenship data, MALDEF, during the most recent round of ____________ 2Interview with Denise Hulett on January 9, 2001. 24 redistricting, relied on two-year-old (1988) registration data—with Latinos identified on the basis of Spanish surname—to estimate the potential size of the voting-eligible Latino population. According to Hulett, MALDEF never evaluated actual turnout rates. Their districting proposals were not conditioned on the probability that these registered voters would actually vote. Thus, the organization most likely to embrace and pursue a strategy of targeting high-propensity voters did not, in fact, adopt such a strategy. This fact suggests that a relationship between majority-minority districts and political participation is less likely to be an artifact of the redistricting process itself. At least two plausible hypotheses exist, then, that would lead us to predict a positive relationship between race-conscious districting and minority political engagement. Although each attributes the relationship to a different causal mechanism, both suggest that minority districts and minority representation might compensate for the socioeconomic barriers that suppress political participation. However, despite the existence of plausible hypotheses and the importance of political participation, there has been no systematic effort to establish whether there is a relationship between race-conscious districting and political participation. In the following chapters I attempt to do just that. Although the nature of the data I will use precludes any definitive identification of the precise causal mechanism (i.e., attitude change or targeted mobilization), it does allow evaluation of whether a relationship exists at all—and if so, whether it is positive or negative. If the evidence reveals a pattern of distinct political behavior in majority-minority districts and under minority representation, then future research should clarify the factors that contribute to that relationship. 25 4. Data and Methodology The key to clarifying the relationship between race-conscious districting and minority representation, on the one hand, and political participation among Latinos, African Americans, and Anglos, on the other, is to identify (1) the major influences on voting participation, (2) the expected rates of voting participation for various demographic groups given these influences, and (3) the degree to which the actual rates of voting participation differ from expectations wherever racial and ethnic minorities are able to “elect representatives of their choice.” If raceconscious districting and the minority representation that may result compensate for the socioeconomic barriers that suppress minority political participation, higher minority voter turnout rates should be seen in majority-minority districts than would be expected given the demographic profile and resources available. This analysis assembles data from multiple sources and applies a variety of statistical methods to examine voting turnout patterns in every congressional district and precinct in the state. Sample Year This analysis draws on data from the November 1994 congressional elections. Widely regarded as the mark—if not the harbinger—of a political revolution, the 1994 midterm elections shifted the partisan balance of Congress as Republicans assumed majority status. In California, 1994 is best remembered as the year of Proposition 187, the controversial (and ultimately successful) ballot initiative that cut off illegal immigrants from all but emergency health care services and removed their U.S.-born children from public schools. The emotioncharged debate over the proposition spilled over into virtually every campaign for public office in the state. Martis and Block (1994) likened Proposition 187 to Proposition 13, an initiative that “[galvanized] the public to such an extent that it [became] the engine that [drove] an entire 27 election.” The initiative campaign, embraced by then-Governor Pete Wilson, proved to be a watershed moment, igniting a “flame of civic engagement among California Latinos” and “driving tens of thousands of freshly minted citizens into the arms of the Democratic Party” (Scott, 2000). As a result of Wilson’s advocacy on behalf of Proposition 187, and the subsequent surge in naturalization and voter registration among Latinos, some have deemed him the “godfather of Latino California” (Frayne, 1998). Three considerations influenced the decision to use data from the November 1994 elections. First, I chose to focus on voting behavior during midterm elections, when the political dynamics of congressional races—no longer overshadowed by presidential campaigns—figure relatively more prominently. Second, the 1994 midterm election reflected the enormous reapportionment and redistricting changes that went into effect in 1992. Compared to 1990, the 1994 election offers a larger sample of congressional districts (45 vs. 52) and a larger number of majorityminority districts served by minority and Anglo legislators. Finally, the need to leverage available Census data in the analysis of voter participation significantly constrained the final choice of midterm election. In light of the population changes witnessed in California over the course of the 1990s, the data collected in the 1990 Census of Population and Housing—the most current data available until Census 2000 is publicly released—offer a relatively more accurate picture of demographic conditions in 1994 than in 1998. Although the currency of the 1998 midterm elections holds some appeal, the use of 1990 Census data to evaluate voter turnout patterns in that election would require a level of extrapolation that could compromise the validity of the analysis and subsequent conclusions. In lieu of those extrapolations, I rely on data from the November 1994 elections. By doing so, I ensure that the available Census data remain analytically useful. Although November 1994 is a more distant election, the data retain value for understanding the relative differences in voter turnout between districts created to have a majority-minority population and those created to be majority-Anglo. 28 Relying on data from 1994 presents at least one important complication: the potentially confounding effect of Proposition 187 mobilization. Of particular concern are the implications for our interpretation of Latino political behavior in that election. As Pachon (1998) observed, the Latino community “did not see [the “Save Our State” Initiative] as simply an anti-illegal alien measure but as an antiLatino measure.” The initiative has been widely credited with politicizing numerous segments of the Latino population and doing more “in three months to . . . energize [that community] than various outreach efforts had managed in fifteen years” (Block, 1998). The weeks and months leading up to the November elections were dominated by demonstrations, teach-ins, and mass rallies, as Latinos (particularly Latino youth) took to the streets (Los Angeles Times, October 21, 1994, November 8, 1994; Pachon, 1998). Additionally, there are accounts of numerous targeted mobilization efforts: The Sunday before the election, Archbishop Roger Mahoney of Los Angeles encouraged all pastors to instruct their parishioners to vote against Proposition 187; Democrats set up voter registration tables outside citizenship ceremonies in Southern California; the Southwest Voter Registration Education Project spent $100,000 on registration and get-out-the-vote efforts in nine California counties with sizable Latino populations; young Latinos organized precinct walks to register voters; and community-based organizations redoubled their efforts to urge naturalization, registration, and turnout (Lopez and Wahlgren, 1994; Los Angeles Times, October 21, 1994, November 8, 1994, December 4, 1994; Schockman, 1998; Guerra, 1998). How much of this activity carried over to election day is unclear. Overall turnout among all groups increased by six percentage points over the previous midterm election (California Opinion Index, 1995). Additionally, a Los Angeles Times exit poll found that about one-third of white and black voters and nearly half of Latino voters identified Proposition 187 as one of the top two reasons for turning out to vote (Los Angeles Times, December 4, 1994). By comparison, a November 1993 Field Poll found that immigration ranked 19th out of 28 issues of concern to Californians (California Opinion Index, 1994). However, 29 although the Latino community overwhelmingly rejected the initiative, as well as its most prominent spokesman (Pete Wilson), the high turnout that had been anticipated never materialized (Los Angeles Times, November 9, 1994, November 10, 1994, November 14, 1994, November 28, 1994, December 4, 1994; California Political Week, 1994; California Journal Weekly, 1994; Martis and Block, 1994). Former state Senator Art Torres, who had lost his bid to become state insurance commissioner (and the first Latino statewide elected official in a century), invoked the metaphor often used to describe the politically reticent Latino community, “[t]he sleeping giant was in a coma on November 8” (Los Angeles Times, December 4, 1994). In the election postmortems, commentators and activists offered numerous reasons for Latino absence at the polls: anti-187 mobilization efforts that gained momentum only after official voter registration deadlines had elapsed; insufficient targeted outreach to and education of Latino communities, partly because of lack of resources and partly out of fear of provoking white backlash; and a puzzling lack of interest among working-class Latinos. Where there is agreement, however, is that the campaign against Proposition 187 laid the foundation for a mobilized Latino community, guided by a “new political consciousness” (Los Angeles Times, November 11, 1994); but it was only with the passage of Proposition 187—and fears that anti-immigrant backlash might eventually result in social welfare benefits being denied noncitizens—that Latino naturalization, registration, and turnout gained real momentum (Washington Post, April 12, 1995; Scott, 1998; Baldassare, 2000; California Opinion Index, 2000; San Jose Mercury News, October 15, 2000). “With the electoral and policy ramifications of California’s midterm 1994 election,” Schockman (1998) argues, “the bubble of benign non-involvement through noncitizenship burst.” Real behavior change was visible by the 1996 presidential election: Latino voter registration increased by 28.7 percent over its 1992 levels. That year, 50 percent of citizen-eligible Latinos were registered to vote (California Opinion Index, 1997).1 Baldassare (2000) notes that the “sting” of Proposition 187 continued to galvanize ____________ 1By comparison, 74 percent of citizen-eligible non-Hispanic whites and 67 percent of African Americans were registered to vote. 30 Latinos as they approached the 1998 election. In November 1998, the Latino proportion of the electorate rose to 13 percent, from just 9 percent in the 1994 midterms. The Field Institute estimated that by the 2000 presidential elections, Latinos accounted for 16 percent of the state’s registered voters—as mentioned above, 46 percent of those voters had registered since 1994 (California Opinion Index, 2000). Although Proposition 187 appears to have had its most pronounced effect on the political dynamics of subsequent elections, it is still possible that the campaign had some incremental effect on participation in 1994. The California Journal, for example, noted a 2 percent increase in Democratic registration in one majority-Latino congressional district between the spring and fall of 1994 (California Journal, 1994). This pattern has two implications for the analysis. First, the absolute turnout rates observed in 1994 may not be a reliable benchmark for what to expect generally in any midterm election. Of course, this limitation is not unique to 1994 data but applies to any cross-sectional data drawn from a single election year. The proximate influences on overall turnout vary from year to year. Longitudinal data are better suited for generating estimates of average participation. With cross-sectional data, analysts typically limit themselves to accounting for variation across demographic or socioeconomic groups. That is the objective here; as a consequence, this first implication should not be a major concern. Second, the relative differences in turnout (e.g., across different demographic groups) may be affected if these groups differed systematically in their attention to the Proposition 187 campaign.2 For example, if there is evidence that most of the Proposition 187 mobilization efforts were targeted at Latinos or, more important, were concentrated in majority-Latino districts or initiated by Latino elected officials, that might account for higher rates of turnout among Latinos than among other groups and higher turnout among Latinos in majorityLatino districts than among Latinos in majority-Anglo districts. For that reason, the following econometric analysis not only evaluates turnout ____________ 2As indicated by the results from the Los Angeles Times exit poll, groups differed systematically in the influence of Proposition 187 on their voting participation. 31 patterns separately for each racial group but also incorporates a measure to capture differences in the salience of the Proposition 187 campaign across different districts. Unit of Analysis and Measures In 1994, 13 congressional districts had majority-minority populations. On the eve of the November elections, the California congressional delegation included four Latino legislators representing majority-Hispanic districts and three African Americans representing majority-minority districts with roughly equal proportions of blacks and Hispanics. In addition, the delegation included two Anglo lawmakers representing majority-Hispanic districts and four legislators (one African American and three Anglos) representing multi-ethnic districts. All but one of these legislators—Representative Don Edwards of the 16th District—ran for re-election in November 1994. The data assembled for the analysis are aggregated at the precinct level and consist of three major components: • Voting data — Voting-age population turnout: Total ballots cast in the congressional race as a proportion of the total voting-age population in the precinct. — Registered voter turnout: Total ballots cast as a proportion of the total registered voters in the precinct. — Proposition 187 turnout: Total ballots cast on Proposition 187 initiative as proportion of total ballots cast in precinct. • Census data — Demographic characteristics: Racial and ethnic composition, age distribution in each precinct. — Socioeconomic conditions: Education levels, household incomes, proportion of new residents in each precinct. • Political data — District and representative characteristics: Racial and ethnic balance of district, race and ethnicity of representative, party identification of representative. — Electoral competitiveness: Size of winning vote margin, incumbency, tenure of representative. 32 Table 4.1 Majority-Minority Districts and Minority Representatives in California, November 1994 District 30 31 33 34 32 35 37 9 16 20 26 46 50 % Hispanic 61 58 83 62 30 42 44 12 37 55 52 49 41 % Black 4 2 4 2 40 43 34 32 5 6 6 2 14 % Asian 21 23 4 9 8 6 11 16 20 6 7 12 15 Congressional Representative Xavier Becerra Matthew Martinez Lucille Roybal-Allard Esteban Torres Julian Dixon Maxine Waters Walter Tucker Ron Dellums Don Edwards Cal Dooley Howard Berman Robert Dornan Bob Filner SOURCES: Duncan (1993c, 1995); Green (1995). Race/ Year First Ethnicity Elected Latino 1992 Latino 1982 Latino 1992 Latino 1982 Black 1978 Black 1990 Black 1992 Black 1970 Anglo 1962 Anglo 1990 Anglo 1982 Anglo 1976 Anglo 1992 Several variables warrant further clarification. First, to assess the salience of the Proposition 187 ballot initiative, the data include a measure that calculates Proposition 187 turnout as a proportion of the total turnout in the precinct. This measure is intended to serve as a rough proxy for the proportion of voters who would identify Proposition 187 as one of the primary reasons for turning out that November— voters who may have benefited from targeted mobilization efforts or whose own sense of the issues at stake was strong enough to carry them to the polls. One would expect to observe variability in this measure across precincts—reflecting geographic variability in the incidence and intensity of organized campaigns (e.g., higher ratios in majority-Latino districts than in majority-Anglo districts) and demographic variability in the perception of issue importance (e.g., higher ratios in heavily Latino precincts than in heavily Anglo precincts).3 Because of this variability— ____________ 3It is important to emphasize that the absolute value of this measure almost certainly overestimates the number of people in any given precinct who would rank Proposition 187 as one of their top motivations for voting. However, to the extent that this measure varies across precincts, this variation can be assumed to be proportional to the true variation in the number of people for whom Proposition 187 was especially salient. 33 and the role of Proposition 187 as the “engine” of the 1994 elections, influencing the likelihood of turning out at all—including this measure helps account for any differences in levels of participation across congressional districts with and without African-American and Latino majorities or legislators.4 Second, the Census measures of socioeconomic conditions include a variable indicating the proportion of precinct residents who had moved to the state within the preceding five years. Compared to long-term state residents, recent transplants are less likely to be registered to vote and may be less settled in their communities and less invested in the politics of the state (Verba, Schlozman, and Brady, 1995; Rosenstone and Hansen, 1993). Furthermore, this measure may vary systematically across districts, with greater turnover in areas with higher concentrations of minority residents. Finally, the political data include several measures of electoral competitiveness—an important contextual influence on voter turnout (Gilliam, 1985; Barker, Jones, and Tate, 1999). These measures include an indicator of whether the incumbent legislator is running for reelection and, if so, his or her years of service in the House. The third measure is the size of the vote margin in the November 1994 race. Majority-minority districts are generally assumed to be among the least competitive districts in a state, although it is not clear that they are considerably less competitive than districts drawn to maintain a partisan advantage. Nonetheless, this is a factor that must be considered when evaluating the correlation between majority-minority districts and voter participation. The voting data and the Census data used in the analysis were provided by the Statewide Database (SWDB) at the Institute for Governmental Studies at the University of California, Berkeley.5 I ____________ 4Without controlling for the Proposition 187 effect, the subsequent econometric analysis would most likely suffer from omitted variable bias. The result would be to overestimate (i.e., bias away from zero) the effect of majority-minority districts and minority representation on minority voter turnout. 5I would like to thank Karin McDonald and her staff (especially, Michael Wagaman and Gray Chynoweth) for their generous support in making these data available to me. 34 assembled the political data using information available in Green (1995) and Duncan (1993c, 1995). Although the SWDB is the leading source of voting and demographic data in the state—and the official redistricting database for the state of California—the information provided is not without its limitations. First, the data are aggregated at the precinct level. As a result, these data cannot shed light on the individual calculus governing the decision to vote. For that, survey data would be needed. Available survey data have neither the demographic nor the geographic coverage to permit this kind of analysis. Second, to reconcile demographic data (collected for Census blocks whose boundaries are stable for a decade) and voting and registration data (collected for precincts whose boundaries change with each election), the analysts at the SWDB must aggregate each database component up to a common geographical level and then disaggregate them back down to a smaller, common unit of analysis (in this case, the “registration precinct”). This process of conversion necessarily introduces errors. However, the analytical advantage that comes with having data at the precinct level more than compensates for the resulting loss of precision. Furthermore, the California state legislature is sufficiently confident in the procedure to rely on the SWDB as the chief resource in the 2001 reapportionment and redistricting. Third, the SWDB merges 1990 Census data with 1994 voting and registration statistics. As a result, the voting-age population totals used as denominators in the measure of voting participation may underestimate (in the case of population growth) or overestimate (in the case of population loss) the actual number of potential voters in 1994. To address this shortcoming, I used county-level figures from the California Department of Finance (DOF) to adjust the population totals so that they would reflect recorded growth between 1990 and 1994. The county-level data provided by the DOF are disaggregated by racial and ethnic group and by age, allowing for the calculation of differential growth rates for specific voting-age populations. These growth rates are then applied equally to every precinct within a given county. That is, if 35 the DOF estimates a 7.1 percent increase in the size of the black votingage population in Santa Clara County, then 7.1 percent is the growth rate applied to every precinct in that county. This assumption of equal growth across precincts may be flawed but is preferable to a measure that disregards growth in that four-year period. Fourth, the 1990 Census does not include information on the racial and ethnic makeup of the citizen and noncitizen populations.6 This lack of specificity most significantly affects analysis of Latino voter participation. Because 41 to 52 percent of voting-age Latinos in the state are thought to be noncitizens, the total voting-age population figure available in the Census grossly overestimates the pool of potential Latino voters (Stiles et al.,1998; Pachon, 1998; California Opinion Index, 2000; Reyes, 2001). The 33rd congressional district in southeast Los Angeles County, for instance, is home to such a large concentration of noncitizen Latinos that it has the lowest number of registered voters of any district in the country (Los Angeles Times, September 26, 1996). More generally, Brackman and Erie (1998) observe that in the Latino community, unnaturalized but legal aliens outnumber registered voters. Since the 1990 Census data do not allow identification of only voting-age Latino citizens, this analysis measures Latino voter participation as a proportion of the number of Latinos registered to vote in November 1994. 7 The SWDB estimates the number of Latino registered voters by filtering each precinct’s individual registration files through a Spanish surname dictionary. Although surname dictionaries typically include more people than the target population (for example, this method may capture Spanish-surnamed Filipinos), this method is preferable to one that overlooks such a critical feature of the Latino demographic profile. To avoid confusion, I will frequently remind the reader that Latino voter participation—unlike black and Anglo voter ____________ 6Furthermore, the DOF does not provide estimates of these figures, either for the state as a whole or by county. It provides only counts of the total number of new immigrants annually. 7The lack of consensus on the estimate of Latino noncitizenship rates makes me very reluctant to pick a number and apply it across the board to every precinct in the state. 36 participation—is measured as the proportion of registered voters. If rough comparisons are desired, there is a useful (although imperfect) conversion factor: According to the Southwest Voter Research Institute, 1.75 million Latino adults were registered to vote in 1994, about 46 percent of the citizen-eligible Latino population (Los Angeles Times, November 11, 1994; California Opinion Index, 1997). Multiplying this figure by the Latino registered voter turnout rates allows for some comparison to black and Anglo voting-age turnout rates.8 Finally, as of this writing, the SWDB was not able to provide voting data for the precincts in California’s 47th congressional district. As a result, that district is not included in this analysis. Every other precinct and congressional district in the state is included in the dataset.9 Even with its limitations, the SWDB provides the most comprehensive set of voting and socioeconomic variables available for the state of California—and at the greatest level of geographic detail. Methodology To identify the relationship between voting participation, on the one hand, and minority representation and minority districting, on the other, this research applies a two-stage methodology. The first stage uses a model of ecological inference to estimate the rates of Latino, AfricanAmerican, and Anglo voting participation for each precinct and congressional district in California.10 (For African Americans and Anglos, voter participation is measured as a proportion of the group’s voting-age population; for Latinos, voter participation is measured as a proportion of the group’s registered voters.) Ecological inference is a ____________ 8Keep in mind, however, that the black and Anglo measures are calculated as proportions of all voting-age adults, not taking into account the less than 5 percent of these groups who are noncitizens (Reyes, 2001). To “correct” for this, one could simply divide the turnout statistic by 0.95 for a measure of black or Anglo turnout as a proportion of citizen-eligible adults. 9Appendix Table A.1 summarizes the number of precinct observations in the dataset. 10Appendix B presents more detailed information on the model of ecological inference. 37 method of aggregate data analysis, widely applied in voting-rights litigation, in which the goal is to use available data on the racial and ethnic composition of precincts and overall vote tallies to calculate racespecific turnout rates that cannot be observed directly.11 The results from this first stage of analysis are presented in a series of maps that illustrate the varying levels of turnout throughout the state. These maps allow for an initial assessment of turnout patterns, including the relative differences in turnout across ethnic groups and across congressional districts with and without sizable minority populations. In the second stage of the analysis, the estimated measures of Latino, African-American, and Anglo voting participation are used as dependent variables in a series of regressions that model voter turnout as a function of minority representation and minority districting.12 The econometric analysis takes into account the potentially confounding effects of demographic and socioeconomic conditions (e.g., age distribution, levels of educational attainment, household income, concentrations of new residents), as well as the effect of electoral competitiveness (e.g., incumbency, tenure, vote margin), on voter participation in districts with sizable minority populations.13 With these models, one can evaluate the relative importance of each factor on the observed level of voter participation and distinguish the influence of minority representation and minority districting from the influence of other variables closely associated with it. After estimating each of these models, I summarize the relationship of minority districting and minority representation to voter participation for each racial and ethnic group. To more clearly present the results from the multivariate analysis, I generate estimates of voter participation levels for a hypothetical “average” precinct, using the mean value of all ____________ 11These rates are not observed directly because California, unlike some southern states, does not collect racial and ethnic data on citizens who register and vote. This makes it impossible to know with certainty the exact number of Latinos or African Americans who cast ballots. 12Appendix B presents more detailed information on the econometric model. 13The coding of the independent variables included in the econometric model is summarized in Appendix B. 38 independent variables in the models. These simulations allow me to illustrate the difference in voter turnout to be expected between an average precinct located within a majority-Anglo district and that same precinct located in a district with a majority-minority population. 39 5. Research Findings General Findings As is characteristic in nonpresidential election years, voter turnout in November 1994 was relatively low throughout the state. According to the California Secretary of State’s official Statement of Vote, a total of 8.9 million Californians participated in the general elections, representing about 37 percent of the state’s adult population. (By comparison, 11.4 million people voted in the 1992 presidential election.) Not all of these voters cast ballots in the congressional races in their districts; “roll off,” wherein voters cast ballots in higher office elections but opt out of less prominent races, is a common phenomenon and was a factor in this election. In November 1994, a total of only 7.6 million votes were cast in congressional races throughout the state (California Journal, 1994). The more modest levels of congressional turnout are evident in Figure 5.1, which plots the number of congressional ballots cast as a proportion of each district’s total voting-age population. Only one congressional district, the 10th in eastern Contra Costa and Alameda Counties, had more than 20 percent of voting-age adults casting congressional ballots. In seven congressional districts, less than 10 percent of age-eligible constituents voted in the congressional election. Many of these low turnout districts are concentrated in and around Los Angeles County. This pattern is consistent with exit polls conducted by Voter News Service and the Los Angeles Times, which found that, although Los Angeles County was home to 30 percent of all adults in the state, it accounted for only 24 percent of the total 1994 electorate (California Opinion Index, 1995). The picture improves considerably when only those adults (18 and over) who were registered to vote in 1994 are considered (see Figure 5.2). Among registered voters, anywhere from one-third to slightly more than 41 < 10% 10% to 12% 12% to 16% 16% to 20% > 20% SOURCE: Statewide Database. Figure 5.1—Voting-Age Turnout by District, November 1994 50 percent turned out to vote in the various congressional races across California. Again, voting rates were somewhat lower in Southern California than elsewhere. Racial Differences in Voter Turnout Figures 5.3 through 5.5 summarize the district-level estimates of turnout for whites, African Americans, and Latinos, allowing for an initial look at racial and ethnic differences in participation and variation in behavior across congressional districts.1 ____________ 1Appendix C summarizes the district-level turnout estimates in a series of graphs that also report the standard errors for each of these estimates. 42 35% to 40% 40% to 45% 45% to 50% > 50% SOURCE: Statewide Database. Figure 5.2—Registered Voter Turnout by District, November 1994 In the majority of congressional districts, white turnout rates ranged from 20 to 28 percent of the voting-age population (Figure 5.3). Congressional districts in Southern California witnessed some of the lowest Anglo voter participation rates in the state, with turnout in many districts languishing below the 20 percent mark. Turnout was similarly low in the Central Valley’s 18th District. The Bay Area showed greater political participation relative to the rest of the state. This finding is consistent with exit polls showing Bay Area voters accounting for 23 percent of the total 1994 electorate but only 21 percent of all adults in the state (California Opinion Index, 1995). Almost without exception, voter turnout among African Americans significantly trailed turnout among Anglo adults (Figure 5.4). In some 43 < 18% 18% to 21% 21% to 24% > 24% NOTE: The map displays the district-level estimates derived from using ecological inference. Figure 5.3—Anglo Voting-Age Turnout by District, November 1994 districts, the participation gap between the two groups reached 15 to 20 percentage points. Black turnout did not exceed 18 percent of the voting-age population for any congressional district in the state. In more than half of the districts, less than 10 percent of voting-age African Americans cast ballots. Finally, participation among registered Latino voters was generally low (Figure 5.5). In half of the congressional districts, Latino turnout did not reach 15 percent. There were some notable exceptions, however. In 10 of the 20 districts in and around the Los Angeles County area, turnout among registered Latino voters exceeded 20 percent—including a half-dozen districts where 40 percent or more of these constituents cast 44 < 5% 5% to 10% 10% to 15% 15% to 20% NOTE: The map displays the district-level estimates derived from using ecological inference. Figure 5.4—Black Voting-Age Turnout by District, November 1994 ballots in congressional races. As will be seen below, many of these districts have Latino majorities.2 Aggregate Turnout Differences Across District Type Preliminary evidence suggests that minority turnout was above average in some majority-minority districts. The second panel on Figure 5.6 plots black turnout (as a proportion of the black voting-age population) for each of the 13 districts where racial and ethnic minorities ____________ 2 Los Angeles County, where many of these high turnout districts are concentrated, was also the epicenter of much of the Proposition 187 mobilization. 45 < 10% 10% to 25% 25% to 40% 40% to 50% NOTE: The map displays the district-level estimates derived from using ecological inference. Figure 5.5—Latino Registered Voter Turnout by District, November 1994 constituted the majority of the population. The graph also includes a horizontal line indicating the average black turnout for congressional districts in the state. On average, 8 percent of the black adults in any congressional district cast ballots in the congressional race. However, in five of the 13 majority-minority congressional districts, black turnout exceeded this rate. Furthermore, four of those districts (the 9th, 35th, 32nd, and 37th) were represented by black legislators at the time of the November election. For Latinos, registered voter turnout exceeded the 18 percent statewide average in 10 of the 13 districts with majority-minority populations (Figure 5.6, top panel). The four districts with the highest Latino participation rates were majority-Latino and served by Latino legislators. 46 Percentage Percentage Latino registered voter turnout 50 40 30 Average Latino 20 turnout 10 0 9 20 16 46 50 35 32 37 26 30 31 34 33 Majority-minority district Represented by Latino legislator Black voting-age turnout 30 25 20 15 Average black 10 turnout 5 0 46 26 30 31 33 34 20 16 50 9 35 32 37 Majority-minority district Represented by black legislator White voting-age turnout 30 25 Average white turnout 20 15 10 5 0 33 26 50 46 32 20 35 37 30 16 34 9 31 Majority-minority district Percentage Figure 5.6—District-Level Voter Participation in Majority-Minority Districts, November 1994 White turnout in the 13 districts in which they are in the minority trailed, in most cases, the 21 percent statewide average (Figure 5.6, bottom panel). However, in five cases, all of them either majority-Latino 47 or multi-ethnic districts, white voting-age turnout exceeded the statewide average. Socioeconomic and Political Influences on Voter Turnout Figure 5.6 suggests that voting participation may vary systematically with the opportunities for minority congressional officeholding. However, as noted in the previous chapter, the districts and precincts with majority-minority populations typically face both electoral (e.g., less competition) and socioeconomic (e.g., greater poverty) conditions that distinguish them from other areas in the state in ways relevant to participation. To isolate the behavior truly unique to districts where minorities have an opportunity to “elect representatives of their choice,” these areas must be evaluated in light of the patterns observed among constituents in demographically similar precincts facing comparable political conditions. Multivariate regression techniques allow us to disentangle race and ethnicity from other aspects of the social and political environment and, in that way, more accurately assess the interrelationships among minority districting, minority representation, and voter participation. I first consider the effect of socioeconomic factors and other political conditions on these rates (Appendix Table D.1). The results indicate that political participation in 1994 was greatest in precincts with large proportions of older voters (over age 35), few new residents, significant numbers of adults with at least some college education, and many households with annual incomes over $50,000. These results are consistent with exit poll data from November 1994 (California Opinion Index, 1995) and with findings from academic research. For example, each 10 percentage point increment in the proportion of precinct residents who are new to the state translates to a net turnout difference of about 2 percentage points for white voting-age turnout, 1 percentage point for black voting-age turnout, and .3 percentage points for Latino registered voter turnout. Within each racial group, new residency was among the strongest and most consistent predictors of participation. For Latinos, household income was the single strongest socioeconomic correlate. 48 Electoral conditions were also strong predictors of voter turnout in 1994. Both white and Latino voter participation rates were higher in Republican-controlled districts than in districts represented by Democrats. For Latinos, this pattern may reflect the close association of Proposition 187 with the Republican Party. Meanwhile, AfricanAmerican turnout was generally higher in congressional races represented by Democratic legislators. Turnout among all groups was at its lowest in the districts where long-serving incumbents were running for re-election. For every 10 additional years of service in the House, voter participation declined by one percentage point. Latino and Anglo turnout rates also tended to be lower in congressional races that were not particularly competitive, as measured by the size of the winning vote margin.3 By comparison, black voting-age turnout was actually higher in districts with more lopsided electoral outcomes. Finally, Proposition 187 mobilization was not a statistically significant predictor of black or Anglo voter participation. However, it was very strongly associated with Latino registered voter turnout; precincts with high Proposition 187 activity (as measured by the Proposition 187 proportion of the total ballots cast) also had high rates of Latino registered voter participation in the congressional race. ____________ 3The “vote margin” variable has been indexed by quartile: at or below the 25th percentile (vote margin ≤ 57 percent); between the 25th and 50th (57 percent < vote margin ≤ 62 percent); between the 50th and 75th (62 percent < vote margin ≤ 67 percent); above the 75th percentile. The excluded category in the regression is the first quartile. I opted for this specification after a number of diagnostic tests found a nonlinear relationship between turnout and margin of victory, skewed by three “outlier” districts in the highest vote margin quartile. The relationship is strongly negative at vote margins below 70 percent (42 of the 51 districts in the sample) but weakly positive among the nine districts with vote margins above 70 percent. The latter relationship is driven entirely by three Bay Area districts: CD-8 (San Francisco), CD-9 (Oakland/ Berkeley), and CD-7 (Richmond/Vallejo). In each of these districts, the incumbents (Nancy Pelosi, Ron Dellums, and George Miller, respectively) won by overwhelming margins but still witnessed extremely high turnout rates in their districts. Most students of California politics would not be surprised by this finding; it is consistent with the more activist political culture typical of these areas. With these three cases removed, margin of victory is consistently negatively correlated with turnout. 49 Majority-Minority Districts and Voter Turnout I now turn to the effect of the minority “opportunity” variables on observed turnout. The first empirical strategy is to test whether political participation among Latinos and African Americans is greater, on average, in districts where racial and ethnic minorities constitute the majority of the population. Again, I draw on the multivariate analyses presented in Appendix Table D.1. Table 5.1 presents the key findings predicting voter turnout among African Americans, Anglos, and Latinos. The table summarizes the measured effect of majority-minority districts, controlling for the socioeconomic and political variables shown above to influence participation. Each multivariate regression that generated these results distinguished between precincts located in Hispanic-majority, black- and Hispanic-majority, or multi-ethnic majority-minority districts. The entries in the table capture the average difference in voter turnout (for each racial or ethnic group) between precincts located in a particular type of majority-minority district and similar precincts located in a majorityAnglo district. Table 5.1 Estimates of Voter Participation Differences Across District Types District Type Hispanic majority Black and Hispanic majority Multi-ethnic majority African American –1.5%** (.3) 6.6%** (.2) 3.9%** (.2) Anglo 3.4%** (.3) –.3% (.4) 2.4%** (.3) Latino 33.1%** (.3) 30.3%** (.5) 6.9%** (.4) N=9,210 N=17,907 N=14,535 NOTES: Estimates are from regression results presented in Appendix Table C.1. Estimates control for socioceonomic conditions and political factors. Each estimate captures the average difference in voter participation between precincts located in a particular type of majorityminority district and similar precincts in a majority-Anglo district. Standard errors are in parentheses. **p < .01. 50 The econometric results are consistent with the aggregate patterns observed above: minority voter participation tends to be higher in the districts where racial and ethnic minorities enjoy majority status. However, the role of majority-minority districts in predicting minority voter participation is contingent on the group in question and the particular racial and ethnic balance within the districts. Three important findings emerge: 1. African-American adults and Latino registered voters participate at higher levels in congressional districts where their respective communities effectively control electoral outcomes. In no districts in California are African Americans solidly in the majority. As a community, they exercise their greatest electoral influence in the four districts in which they and Latinos, together, constitute the majority of the population. In these districts, turnout among voting-age blacks is on average 6.6 percentage points higher than turnout in similar precincts in majority-Anglo districts. 4 In the more fragmented majorityminority districts, where African Americans, like their Asian and Latino neighbors, play a prominent but not decisive role, turnout rates exceed those in majority-Anglo districts by a more modest 3.9 percentage points. For Latinos, registered voter turnout rates in any and all majorityminority contexts are higher than what one would expect to observe in similar precincts in majority-Anglo districts. Latino voter turnout is at its highest in districts where they alone make up the overwhelming majority of the population. In these majority-Latino districts, registered voter turnout is on average 33 percentage points higher than rates in majority-Anglo districts. In the districts where Latinos and African Americans are more equally matched, Latino turnout is 30 percentage points above majority-Anglo rates. Finally, in the multi-ethnic districts where Latinos play a relatively more limited role, their registered voter participation rates exceed those in majority-Anglo districts by a more modest 6.9 percentage points. ____________ 4“Similar” here refers to similarity in socioeconomic conditions and political characteristics, as defined by the control variables discussed above. 51 2. African-American voting-age turnout is at its lowest in any district dominated by a single nonblack racial or ethnic group. In two types of districts in the state, a single racial or ethnic group enjoys an overwhelming numerical advantage: majority-Latino districts and majority-Anglo districts. Voter participation among African Americans is marginally lower in the six majority-Latino districts in the state than it is in the dozens of majority-Anglo districts. Black voting-age turnout in the districts where Latinos constitute between 52 and 83 percent of the total population was on average 1.5 percentage points below rates in similar precincts in majority-Anglo districts. Although this difference is statistically significant, it is arguably not substantively significant. In fact, this very slight participation gap suggests that there may be more similarities than differences in the political circumstances faced by African Americans who occupy minority status in any district— majority-minority or not—dominated by a single nonblack racial group—Latino or Anglo. It is interesting to compare the patterns in majority-Latino districts to those in multi-ethnic districts—where African Americans remain in the minority but share that status with every other racial and ethnic community in the district. The turnout differential between multi-ethnic and majority-Latino districts is 5.4 percentage points, much greater than the difference between majorityAnglo and majority-Latino districts. 3. Anglo voting-age turnout is highest in congressional districts with either fragmented minority populations or Hispanic majorities. Voter participation among white adults is 2.4 to 3.4 percentage points higher in multi-ethnic and Hispanic-majority districts, respectively, than it is in majority-Anglo congressional districts. In districts where African Americans and Latinos form a solid majority, Anglo voting-age turnout is comparable to rates observed in similar precincts in majority-Anglo districts. One possible explanation for these patterns is that white voters may be responding to the political opportunities presented by a fragmented minority population and a Hispanic population constrained by high rates of noncitizenship and relative youth. These conditions may offer whites the prospect of exercising real influence over electoral outcomes despite the unfavorable racial and ethnic balance in the district. (Compare this to majority- 52 Anglo districts where even low white voting-age turnout would not compromise Anglo control over electoral outcomes.) In districts where blacks and Latinos form a solid majority, accounting for between 70 and 85 percent of the total population, there are few opportunities for even a highly mobilized Anglo minority to exploit successfully. To more clearly illustrate the interrelationship between these different districting arrangements and voter turnout among Latinos, African Americans, and Anglos, I generated estimates of the levels of participation one would predict for an “average” precinct under each of these scenarios.5 These predicted values help translate the independent measured effects summarized in Table 5.1 into more meaningful, concrete terms. The predicted values are presented graphically in Figure 5.7. The vertical distance between each bar captures the difference in turnout to be expected from “shifting” this hypothetical “average” precinct from one districting arrangement into another, holding all other factors constant. The patterns in Figure 5.7 reinforce the findings above. For example, black voting-age turnout would be expected to reach 10.8 percent for the precinct in a district with a black-Hispanic majority but only 2.7 percent in a district with a Hispanic majority. (Again, note the small difference in the height of the majority-Hispanic and majorityAnglo bars for African Americans.) For Latinos, a registered voter turnout rate of 45 percent would be predicted for an “average” precinct ____________ 5The “average” precinct is defined by the following mean values on each of the independent variables: Variable Proportion ≥ age 35 Proportion new residentsa Proportion with some college Household income ≥ $50K African American .55 .11 .57 .34 Proposition 187 turnouta Party of incumbenta Incumbent in racea Tenure (incumbent in race) Vote margin (quartile)a .96 Democrat (1) Yes (1) 9.5 .623 (50th–75th ) aNot race- or ethnicity-specific. Anglo .64 .11 .64 .37 .96 Democrat (1) Yes (1) 9.5 .623 (50th–75th) Latino .40 .11 .42 .31 .96 Democrat (1) Yes (1) 9.5 .623 (50th–75th) 53 Percentage Latino registered turnout in “average” precinct 50 40 30 20 10 0 Black voting-age turnout in “average” precinct 30 Percentage 20 10 0 White voting-age turnout in “average” precinct 30 Percentage 20 10 0 Anglo Latino Black-Latino Majority Multi-ethnic NOTES: Each panel illustrates for either Latinos, African Americans, or whites, the level of voter participation predicted for an “average” precinct under four different districting arrangements. The predicted values are generated by holding all other independent variables (socioeconomic conditions, political factors) constant at their means. Figure 5.7—Predicted Voter Participation in “Average” Precinct, by District Type in a majority-black and -Hispanic context, compared to a rate of 48 percent in a majority-Hispanic district. Finally, for white voting-age adults, an “average” precinct turnout rate of about 24 percent would be 54 predicted in either a majority-Anglo or majority-black and -Hispanic district; in Hispanic and mixed multi-ethnic districts, however, the rate would be closer to 26 or 27 percent of the voting-age population. Majority-Minority Districts, Minority Representation, and Voter Turnout In theory, each of the state’s majority-minority congressional districts affords minority constituents the opportunity to “elect representatives of their choice.” However, minority members of Congress did not represent all of these districts at the time of the November 1994 elections. In fact, Anglo legislators served three of the four multi-ethnic districts and two of the six majority-Hispanic districts. The second empirical strategy considers whether minority representation and minority districts jointly predict voter participation. Table 5.2 summarizes the measured effect of majority-minority districts and minority representation on voter participation, controlling for the same set of socioeconomic and political factors identified above. Each of the multivariate regressions that generated these results distinguished not only among precincts located in Hispanic-majority, black- and Hispanic-majority, or multi-ethnic majority-minority districts, but also among precincts served by African-American, Anglo, or Latino congressional representatives. The entries in this table capture the differences in voter participation, as compared to similar precincts in majority-Anglo districts represented by Anglo members of Congress. Several interesting results emerge with this analysis of the interaction between minority representation and minority districts. 1. Latino representation, independent of majority status within a district, is a significant predictor of Latino registered voter turnout. In the previous empirical analysis, Latino registered voter turnout was seen to be, on average, 33 percentage points higher in the precincts of the state’s six majority-Hispanic districts than in similar precincts in California’s 39 majority-Anglo congressional districts. That summary figure, however, concealed an important distinction: In the four majority-Hispanic congressional districts that were represented in 1994 by Latino members of Congress, Latino registered voter turnout was 36.4 55 Table 5.2 Estimates of Voter Participation Differences Across District Type and Legislator’s Race District Type–Legislator Race Hispanic majority Hispanic legislator White legislator Black and Hispanic majority Black legislator Multi-ethnic majority Black legislator White legislator African American –2.3%** (.3) –.4% (.4) 6.5%** (.2) 3.7%** (.3) 4.0% (.2) Anglo 5.1%** (.4) 1.0%** (.4) –.2% (.4) 6.0%** (.4) .3 (.3) Latino 36.4%** (.4) 26.5%** (.5) 31.1%** (.9) .9% (.8) 9.6%** (.5) N=9,210 N=17,907 N=14,535 NOTES: Estimates are from regression results presented in Appendix Table C.2. Estimates control for socioceonomic conditions and political factors. Each of the estimates captures the average difference in voter participation between precincts located in a particular district-legislator type and similar precincts in a majority-Anglo district with an Anglo representative. Standard errors are in parentheses. **p < .01. percentage points higher than rates in majority-Anglo districts. Meanwhile, in the two majority-Hispanic districts represented by Anglo legislators, Latino turnout was 26.5 percentage points higher than rates in majority-Anglo districts. Thus, Latino representation—quite apart from majority status—was associated with a 10 point difference in the level of Latino voter participation. The fact that high Latino voter participation and Latino congressional representation tend to go together raises the question of causal direction. Namely, is Latino representation the consequence of heightened Latino mobilization in majority-Latino districts, or does the former precede the latter?6 The history of Latino congressional ____________ 6As discussed above, the relationship between majority-Latino districts and Latino voter participation does not present the same chicken-and-egg dilemma. My research 56 representation in the state presents some analytical challenges to clarifying this “chicken-and-egg” question. As of 1994, every sitting Latino member of Congress represented a district that had consistently elected Latino legislators from the moment it had been created. None of these districts has undergone an “ethnic transition” in political leadership. The same goes for the two majority-Latino congressional districts represented by Anglo legislators in 1994. Each of these legislators was already in office when his district was redrawn in 1991 to include a Latino majority. In the absence of any ethnic change in representation, it is impossible to establish conclusively whether heightened mobilization was a consequence of, or a necessary condition for, Latino representation. One possibility for future research might be an analysis of Loretta Sanchez’s 1996 victory over Robert Dornan in Orange County’s 46th District. However, the more fragmented racial and ethnic balance in the 46th might limit the applicability of these findings to the majority-Latino case. 2. For African Americans, there is no discernible independent relationship between black representation itself and voting-age participation. African-American voting-age turnout in the three fragmented multiethnic districts represented by white legislators was comparable to the turnout rates observed in similar precincts in the one multi-ethnic congressional district represented by a black legislator (OaklandBerkeley’s 9th District, represented by Ron Dellums). For each of those cases, black voter turnout was between 3.7 and 4 percentage points higher than turnout in the state’s majority Anglo districts. The only districts in which there was any appreciable difference in the turnout rates observed with a minority legislator as opposed to an Anglo legislator were the districts with majority-Latino populations. In these cases, African-American voting-age turnout was lower in the districts represented by Latino members of Congress than it was in the districts with Anglo legislators. (In fact, there is no statistically significant difference in black voting-age turnout between majority________________________________________________________ suggests that district lines in 1991 were not explicitly drawn to target high-propensity voters (i.e., high Latino registered voter participation did not make majority-Latino districts more likely). 57 Anglo districts and the majority-Latino districts that continue to elect Anglo legislators.) When Latinos enjoy both majority status and descriptive representation, this overlap may reinforce a perception among African Americans that a single group effectively holds a monopoly over politics in the district. The political space may appear relatively more open when the nonblack majority does not also hold the congressional seat. However, it is not clear why lower turnout rates are observed in the districts where Latinos have a lock on politics than in the many more districts where Anglos have a lock on politics (i.e., majority-Anglo districts with Anglo legislators). The explanation may be simple: African Americans have not yet grown accustomed to this particular racial balance of power. Given the demographic momentum of the state, however, this arrangement may become the norm for African Americans over the next decades. 3. There is limited evidence that white voting age participation is positively associated with minority representation. White voting-age turnout was greater in the majority-Latino districts represented by Latino members of Congress than in those majorityLatino districts represented by Anglo legislators. Under Anglo representation, precinct-level white voting-age turnout was only one percentage point higher than rates observed in majority-Anglo districts; with Latino representation, turnout was 5.1 percentage points higher than the majority-Anglo case. As hypothesized above, these turnout differentials relative to majority-Anglo districts may reflect an assessment of the electoral opportunities available to a highly mobilized minority in a context of high Latino noncitizenship. From this perspective, the more pronounced differentials under Latino representation may be correlated with the higher concentrations of noncitizens in districts such as the 30th (downtown Los Angeles/Koreatown) and 33rd (East Los Angeles) in comparison to Anglo-represented districts such as the 20th (North Hollywood/Pacoima/San Fernando). There is also evidence of greater mobilization under black representation in multi-ethnic districts than under white representation. In fact, once multi-ethnic districts are separated by the race of the legislator, voting-age turnout in the fragmented districts with Anglo legislators is seen to be no different from the rates observed in majority- 58 Anglo districts throughout the state. However, the relationship that does remain (between black-represented multi-ethnic districts and white voter participation) is very likely an artifact of the activist political culture that characterizes the state’s one district that falls in this category: Oakland and Berkeley’s 9th. The six percentage point difference between white voting-age turnout in that district and turnout in a majority-Anglo district might reasonably be referred to as the “Berkeley effect.” As such, I am reluctant to interpret this as evidence of a relationship between representation and participation. The bar charts presented in Figure 5.8 reinforce these findings, illustrating the joint contribution of minority districts and minority representation to predicted turnout in an “average” precinct (as defined above in Figure 5.7). Each bar captures the predicted level of turnout for this precinct in each of the possible districting and representation arrangements. From these plots, not only can the absolute rate of voter participation be assessed but also, by measuring the vertical distance between bars, how that rate compares across scenarios. For an “average” precinct located within a majority-Latino district, Latino voter participation would vary between 42 percent of registered voters if that district were represented by an Anglo legislator and 52 percent if the district’s representative were Latino. If the district had a multi-ethnic majority and continued to be represented by an Anglo legislator, a Latino registered voter turnout rate of 25 percent would be predicted. Finally, if this “average” precinct were located within the boundaries of a majority-Anglo district, the predicted level of Latino registered voter turnout would be a relatively modest 15 percent. For Latinos, minority districts and minority representation are each significant predictors of registered voter participation. For African Americans, there is significant district-based variation in turnout but relatively weak associations between turnout and minority representation. For example, note the similarity in bar height for multiethnic districts with and without black legislators; compare this to the difference in bar height between these districts and the majority-Anglo case. In the case of a majority-Latino district, black voting-age turnout in the “average” precinct is predicted to be about 2 percent for a district 59 Percentage Percentage Latino registered turnout in “average” precinct 60 50 40 30 20 10 0 Black voting-age turnout in “average” precinct 30 25 20 15 10 5 0 White voting-age turnout in “average” precinct 30 25 20 15 10 5 0 Latino Anglo Black Anglo Anglo legislator legislator legislator legislator majority Majority-Latino district Multi-ethnic district NOTES: Each panel illustrates for either Latinos, African Americans, or whites, the predicted level of voter participation for an “average” precinct under five different districting-representative arrangements: (1) majority-Latino district with Latino legislator; (2) majority-Latino district with Anglo legislator; (3) multi-ethnic district with black legislator; (4) multi-ethnic district with Anglo-legislator; and (5) majorityAnglo district with Anglo legislator. The predicted values are generated by holding all other independent variables (socioeconomic conditions, political factors) constant at their means. Percentage Figure 5.8—Predicted Voter Turnout in “Average” Precinct, by District Type and Legislator Race 60 served by a Latino member of Congress but closer to 4 percent for a district that continues to be represented by an Anglo legislator. In the latter instance, the predicted turnout does not differ from what one would calculate for a majority-Anglo district. Setting aside the “Berkeley effect,” voting-age participation for white constituents in an “average” precinct is most consistently predicted by the majority or minority status of the Hispanic population and the presence or absence of a Hispanic legislator. If included in the boundaries of a majority-Latino district, white voting-age turnout is predicted to be about 26 percent with a Hispanic legislator and about 22 percent with an Anglo legislator. With the exception of the Berkeley case, no other representation or districting arrangement generates turnout estimates significantly different from what one would expect from a majority-Anglo district. Summary of Findings In sum, there is clearly an interrelationship among minority districting, minority representation, and political engagement among African Americans, Latinos, and Anglos in California. Latino registered voter participation and African-American voting-age turnout are highest in districts where Latinos and African Americans, respectively, are able to play prominent roles in deciding electoral outcomes. For African Americans, this includes districts where they and Latinos are equally matched and, together, constitute the majority of the district’s population. It also includes the more racially fragmented districts where African Americans may account for only a minority of the population but find themselves on equal footing with the district’s other racial and ethnic communities—white and nonwhite. For Latinos, registered voter turnout is high not only in the districts where they collectively share majority status with one or more nonwhite communities but also (and especially) in the six majority-Latino districts in the state. Furthermore, voter participation among Latinos is greatest in districts where they both enjoy an outright numerical advantage and are represented by Latino legislators. A Latino legislator can make the difference between turnout rates that are 27 percentage points above those observed in majority-Anglo districts in the state and turnout rates 61 that are 36 percentage points above those observed at these majorityAnglo polling places. African-American turnout is lowest wherever a single nonblack community clearly dominates the electoral space, effectively relegating African Americans to the political margins. Turnout is somewhat lower when this “advantaged” community is Latino (majority-Latino district and Latino legislator) than when the community is Anglo—the more typical case in the state and the one to which African Americans have had more time to grow accustomed. Finally, neither the lack of majority status nor the lack of descriptive representation is associated with lower Anglo voter participation. The only instance in which Anglo turnout differs significantly from what one would expect to observe in a majority-Anglo district (taking into account the local political culture, e.g., the “Berkeley effect”) is in majority-Latino districts, particularly those represented by Latino legislators. In these districts, Anglo voter participation is higher than expected, possibly reflecting an appreciation for the political opportunities available to a highly mobilized minority in that context. As discussed above, one real limitation of aggregate data analysis is that it does not provide a window into individual-level motivations. Consequently, one can only speculate about the reasons for the observed variation in political behavior across districts and legislators. In Chapter 3, I outlined two possible hypotheses: one focused on attitude change, emphasizing the effect of majority status and minority officeholding on levels of trust and political efficacy; the other focused on targeted mobilization, emphasizing the strategic calculations of minority political aspirants and independent political organizations. The results here cannot distinguish effectively between these competing explanations— although the model specification did allow us to control for Proposition 187 mobilization. The observable implications of either of these hypotheses is the same: greater political participation in majorityminority districts and under minority representation. What can be concluded, in light of these findings, is that the political dynamics present in majority-minority districts and under minority representation largely favor minority political engagement and do not adversely influence Anglo political participation. 62 6. Policy Implications Since the 1992 redistricting “upheaval,” there has been a steady erosion in popular, political, and judicial support for the mechanisms of the Voting Rights Act. The 2001 reapportionment and redistricting process will take shape against a backdrop of recent Supreme Court decisions (e.g., Shaw v. Reno, Miller v. Johnson, Bush v. Vera, Abrams v. Johnson, Reno v. Bossier Parish) that not only threaten the electoral gains of the past decade but may severely restrict the prospects for greater minority representation in the next. The findings presented here have important implications for the decisions and challenges that policymakers will confront in the next redistricting cycle. First, expanding political opportunities for minorities through the creation of new majority-minority districts may be an effective tool for encouraging participation among groups that otherwise might remain on the margins. What this analysis makes clear is that, for racial and ethnic minorities, the opportunity to “elect representatives of their choice” can compensate for the socioeconomic barriers that so often contribute to low voter participation. Including a precinct in a majority-black and -Hispanic district rather than in a majority-Anglo district may have a more pronounced and immediate effect on voter participation than increasing the number of college-educated residents by 10 percentage points. As a result of connecting new voters to the political world, the redistricting process may help create a more dynamic electorate that more closely reflects the demographic composition of the state’s citizen population. Second, in creating majority-minority districts, mapmakers should be mindful that turnout among minority voters is high wherever they are able to play a meaningful role in political life. Greater participation among African Americans is observed when they are not relegated to the political margins in a district dominated by a single nonblack racial or ethnic group—white or nonwhite. Latino registered voter turnout is 63 highest in districts where Latinos overwhelmingly enjoy majority status, but it is also quite high in districts where they are on roughly equal footing with African Americans. Even in the more heterogeneous areas, Latinos continue to participate at rates that exceed those in majorityAnglo districts. Furthermore, even where Anglo voters are consigned to a modest role compared to their position in majority-Anglo districts, there is no evidence that participation suffers. All of this suggests that there may be some advantage to putting more emphasis on the creation of black-Hispanic and multi-ethnic districts in the next redistricting round. These districts would allow for multiple racial and ethnic communities to exercise some political leverage; such a political environment encourages participation. The state’s demographic momentum may challenge efforts to create more black-Hispanic majority congressional districts. With Latinos constituting the fastest growing population in the state and with the shift in Latino population out of the “barrio” and into historically black neighborhoods, it is widely anticipated that the Southern California districts in which blacks and Hispanics were once equally matched may have clear Hispanic majorities when the lines are redrawn (Scott, 2000; McClain and Stewart, 1999). In places such as the 37th congressional district, anchored in Compton, this demographic trend is already contributing to political tensions between Latinos and African Americans (Sample, 1998; Scott, 1998; San Francisco Examiner, February 13, 2000; Los Angeles Times, October 9, 2000). African Americans will prefer districting arrangements that safeguard the delicate racial and ethnic balance in these areas and allow for them to retain their elected posts. Latinos will prefer districting arrangements that enhance their political leverage and offer new opportunities for Latino officeholding. State lawmakers will have to strike a balance between the benefits of expanding political opportunities for Latinos and the potential costs of redistricting African Americans out of power. Obviously, voter participation is only one of many issues raised by the practice of minority districting. As outlined in the introduction, the practice has provoked considerable debate over substantive representation and the responsiveness of legislative institutions as a whole. The intention in this report is not to elevate participation above all other 64 considerations but to expand the set of considerations to include participation. Furthermore, the observation that Latino registered voter turnout, for example, is higher in the majority-Latino districts served by Latino legislators than it is in those served by Anglo legislators should not be interpreted as a categorical statement about the merits of Anglo representation of Latino constituencies. This finding speaks only to the issue of whether minority districts and minority representation can deliver a more active body politic. Given the democratic value of an engaged citizenry—and the role that politics historically has played as a vehicle in the mobility and “mainstreaming” of racial and ethnic groups in the United States—participation itself is a worthwhile issue to consider. As mapmakers take up the challenge presented by decennial reapportionment and capitalize on the opportunities created by the one additional district California will gain (Giroux, 2000), they should appreciate fully the implications of efforts to expand, reduce, or maintain the existing political opportunities for African Americans and Latinos in the state. 65 Appendix A Sample Table A.1 summarizes the number of precinct observations for each congressional district included in the dataset. Table A.1 Number of Precinct Observations in Dataset, by District District 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Number of Precincts 521 650 485 627 461 461 500 498 353 622 494 460 342 509 509 417 345 410 121 313 456 540 233 376 395 233 District 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 48 49 50 51 52 Number of Precincts 360 382 431 184 256 344 154 311 268 435 280 361 233 412 230 277 465 489 333 105 251 389 276 459 428 NOTES: This list includes all observations for which it was possible to generate estimates of either Latino, black, or Anglo voter participation. The actual number of cases in the regression analyses will vary with patterns of residential segregation (i.e., African Americans are concentrated in relatively few precincts). 67 Appendix B Methodology Stage One: Ecological Inference In the first stage of the analysis, I use an ecological inference model to estimate the rates of Latino, African-American, and Anglo turnout in each precinct in the state. Estimation on the basis of aggregate statistics for the voting-age population (or statistics on the total population of registered voters) is necessary because the state of California does not collect racial and ethnic data on citizens who register or turn out to vote.1 That is, it is not possible to know with certainty the exact number of Latinos, African Americans, and Anglos who cast ballots in any given election. In the absence of such precise and validated individual-level statistics, we must calculate these voter participation figures using available data on the racial and ethnic composition of precincts. Table B.1 illustrates what is involved in using the available aggregate statistics to infer back to specific quantities of interest. In this particular example, Xi represents the proportion of voting-age adults who are black, and Ti represents the proportion of voting-age adults turning out to vote, in precinct i. The goal of ecological inference is to use these marginals, the only observed data, to estimate the quantities within the body of the table, which are not observed directly. Specifically, the quantity of interest turning in this out to vexoatem(ptolepi-sleBftibc,etlhl)e.2prDouproirntgiosntaogfevoontien, gB-aibg(eanbdlacks ____________ 1As mentioned above, the SWDB makes use of a Spanish surname dictionary to identify registered voters of Hispanic origin. 2In this case, Binb is not directly of interest. Rather than the catch-all “nonblack” category, the voting behavior of specific nonblack demographic groups (Latinos, Anglos) is of interest. In the event a precinct consists of precisely two demographic groups (e.g., only blacks and whites), then both Bib and Binb would be of interest. 69 Table B.1 Ecological Inference for Black/Nonblack Voter Turnout Race Black Nonblack Voting Decision Vote No Vote Bib 1 – Bib Binb 1 – Binb Ti 1 – Ti All VotingAge Adults Xi 1 – Xi accompanying standard error) is calculated for each precinct in the state, using an estimation procedure developed in King (1997). This estimation procedure is repeated in a similar analysis of Anglo voter participation, structured around the following contingency table (Table B.2). In Table B.2, Xi is the proportion of the voting-age population in precinct i that is in precinct i that Anglo; Ti is turned out the proportion of to vote. The goal the voting-age population is to estimate Biw , the proportion of Anglos who turned out to vote. Last, the estimation of Latino voter participation is structured around Table B.3. Note that this table differs slightly from Tables B.1 and B.2, reflecting the use of SWDB information on Latino registered voters. As discussed in Chapter 4, these alternative measures compensate for the high noncitizen ratio among Latino voting-age adults: In Table B.3, Xi is the proportion of registered voters (as opposed to voting-age adults) in precinct i that is Latino; Ti is the proportion of registered Table B.2 Ecological Inference for White/Nonwhite Voter Turnout Race White Nonwhite Voting Decision Vote No Vote Biw 1 – Biw Binw 1 – Binw Ti 1 – Ti All VotingAge Adults Xi 1 – Xi 70 Table B.3 Ecological Inference for Latino/Non-Latino Voter Turnout Race Latino Non-Latino Voting Decision Vote No Vote Bil 1 – Bil Binl 1 – Binl Ti 1 – Ti All Registered Voters Xi 1 – Xi voters in precinct i that turned out to vote. The goal is to estimate Bil , the proportion of Latinos who turned out to vote. To estimate Bib, Biw , and Bil for each precinct in California, I apply the ecological inference model developed in King (1997). The model (known as “EI” after the software that implements the procedure) couples the deterministic method of bounds with the maximum likelihood approach, drawing on statistical, deterministic, and diagnostic sources to improve the quality of the resulting estimates. In the method of bounds, the available precinct-level information (i.e., Xi and Ti) is used to restrict the range of possible values for each turnout estimate in each precinct. With these initial bounds defined, the procedure (using maximum likelihood) leverages the information available across all precincts to narrow the range of possible values for each turnout estimate still further. Finally, the procedure selects from this probability distribution the turnout estimates with the highest probability. Stage Two: Feasible Generalized Least Squares (FGLS) Regression Analysis The second-stage analysis uses the precinct estimates of Latino, African-American, and Anglo turnout as dependent variables in regressions predicting turnout as a function of minority representation and minority districting. These second-stage regressions use Lewis’s (2000) FGLS estimator.3 Like the more common variance weighted least ____________ 3The FGLS procedure is implemented in STATA using a program written and generously provided by Jeff Lewis. 71 squares (WLS) approach, the FGLS estimator explicitly takes into account the uncertainty in the estimates of Bib, Biw , and Bil ; however, Lewis (2000) demonstrates that FGLS significantly improves on the inefficiency and overconfidence that can result from a WLS approach to models with estimated dependent variables. The independent variables included in the FGLS regression analysis are summarized in Table B.4. 72 Table B.4 Definition of Variables 73 Variable Name Description Congressional District Type Hispanic majority Hispanics constitute 51 percent or more of population Black and Hispanic majority Hispanics and blacks constitute 51 percent or more of population Multi-ethnic majority Hispanics, Asians, and blacks constitute 51 percent or more of population Hispanic-Hispanic majority Hispanic-majority district with a Hispanic legislator White-Hispanic majority Hispanic-majority district with a white legislator Black-black and Hispanic majority Black and Hispanic majority district with a black legislator Black-multi-ethnic majority Multi-ethnic majority district with a black legislator White-multi-ethnic majority Multi-ethnic majority district with a white legislator SES Conditions Proportion over age 35 Proportion of voting-age population over age 35 (raceand ethnicity-specific) Proportion new residents Proportion of residents who moved to precinct within last five years (race- and ethnicity-specific) Proportion with some college Proportion of adults (age 25 and over) with at least some college education (race- and ethnicity-specific) Proportion of households earning Proportion of households with 1989 income of at least over $50K $50K (race- and ethnicity-specific) Coding 1 if district = 30, 31, 33, 34, 20, 26; otherwise 0 1 if district = 32, 35, 37; otherwise 0 1 if district = 9, 16, 46, 50; otherwise 0 1 if district = 30, 31, 33, 34; otherwise 0 1 if district = 20, 26; otherwise 0 1 if district = 32, 35, 37; otherwise 0 1 if district = 9; otherwise 0 1 if district = 16, 46, 50; otherwise 0 0 to 1 0 to 1 0 to 1 0 to 1 Table B.4 (continued) 74 Variable Name Party of incumbent Description Electoral Conditions Political party of incumbent U.S. House representative in November 1994 Coding 1 if party = Democrat; 0 if party = Republican Incumbent in race Is House incumbent running for re-election in November 1 if running; otherwise 0 1994 Tenure * incumbent Interaction of “incumbent” and years of service in the U.S. House, as of November 1994 Vote margin quartiles Winning proportion of the vote in November 1994 Raw variable ranges from .49 to .82; transformed variable indexed by quartile: 1st quartile (.49 – .57); 2nd quartile (.58 – .62); 3rd quartile (.63 – .67), 4th quartile (.68 – .82) NOTE: For the socioeconomic variables, there are separate measures for each racial and ethnic group. Appendix C District-Level Turnout Estimates with Standard Errors The ecological inference procedure applied in the estimation of Latino, African-American, and white turnout rates generates not only point estimates for these parameters but also standard errors. Figures C.1 through C.3 present the (weighted) district-level estimates for turnout; the figures also include standard error bars, capturing the uncertainty in these district estimates. (Each box is centered on the point estimate and extends one standard error above and below that estimate.) The overall pattern across these three figures reflects the far greater certainty contained in the estimates of Anglo turnout as compared to black turnout and (to a more limited extent) Latino turnout. With most congressional districts dominated by precincts that are overwhelmingly Anglo, the data typically contain more information on which to make inferences about Anglo voting behavior than they do information about black or Latino voting behavior.1 The various majority-minority congressional districts (identified by “*”) are the exceptions, consisting as they do of precincts that are predominantly Latino or predominantly black. In these districts, the estimates of black and Latino turnout have the narrow error bars more characteristic of the white turnout estimates in Figure C.1. ____________ 1For example, imagine a precinct (or congressional district) that is 80 percent Anglo and 20 percent African American. If 40 percent of the voting-age population turned out to vote on election day, that is consistent with a black turnout rate of anywhere between 0 and 100 percent. The composition of the geographic unit leaves us with very little information about African Americans—and thus greater uncertainty in the subsequent estimate. By comparison, there is enough information in the data to allow one to conclude that at least 25 percent, and not more than 50 percent, of Anglos turned out to vote. This would lead to comparatively narrow error bars around the estimate. 75 Congressional district 31* 9* 8 34* 39 28 23 15 14 22 36 16* 17 24 38 2 6 19 12 45 21 3 30* 51 37* 4 5 25 1 40 35* 27 13 10 20* 41 11 7 29 49 32* 48 43 52 46* 44 50* 26* 33* 42 18 0 15 20 25 Estimated Anglo voter turnout (% voting-age population) 30 NOTE: Each bar represents the estimated rate of Anglo voting participation in the congressional district, plus or minus one standard error. The width of the bars reflects the degree of statistical uncertainty in the estimate. * = majority-minority district. Figure C.1—District-Level Estimates of Anglo Turnout with Uncertainty 76 Congressional district 37* 38 36 29 5 32* 35* 13 12 11 10 28 27 25 9* 4 3 2 1 50* 8 7 6 42 17 16* 15 14 21 20* 19 18 34* 33* 31* 30* 26* 52 51 49 41 40 39 48 46* 45 44 43 24 23 22 0 5 10 15 Estimated black voter turnout (% voting-age population) 20 NOTE: Each bar represents the estimated rate of black voting participation in the congressional district, plus or minus one standard error. The width of the bars reflects the degree of statistical uncertainty in the estimate. * = majority-minority district. Figure C.2—District-Level Estimates of Black Turnout with Uncertainty 77 Congressional district 33* 34* 52 31* 30* 26* 37* 17 32* 35* 23 50* 8 48 44 46* 12 13 41 16* 42 38 18 3 20* 39 4 2 22 45 43 40 25 28 19 21 7 51 10 6 9* 36 27 1 11 29 49 5 24 15 14 0 10 20 30 40 Estimated Latino voter turnout (% registered) 50 NOTE: Each bar represents the estimated rate of Latino voting participation in the congressional district, plus or minus one standard error. The width of the bars reflects the degree of statistical uncertainty in the estimate. * = majority-minority district. Figure C.3—District-Level Estimates of Latino Turnout with Uncertainty 78 Appendix D Regression Output To evaluate the relationship of minority districts and minority representation to voter participation among Latinos, African Americans, and Anglos, I estimated a series of multivariate regression models. The first set of models analyzes the relationship between majority-minority districts and voting participation; the second set of models identifies how minority representation and majority-minority districts interact to predict turnout. Both sets of multivariate models include controls for socioeconomic conditions and political factors. The only differences in the specifications of the control variables are across racial and ethnic groups; the measures of age, education, and income are all race- and ethnicity-specific. Tables D.1 and D.2 report the results from this regression analysis. 79 Table D.1 Coefficient Estimates for Regression of Majority-Minority Districts on Voting Participation Variables Constant African American Anglo Latino .016 (.019) .171 (.026)** –.060 (.037) District type Hispanic majority Black and Hispanic majority Multi-ethnic majority –.015 (.003)** .034 (.003)** .331 (.003)** .066 (.002)**–.003 (.004) .303 (.005)** .039 (.002)** .024 (.003)** .069 (.004)** SES conditions Proportion over age 35 .011 (.002)** .088 (.005)** .031 (.006)** Proportion new residents –.101 (.006)**–.196 (.009)** –.032 (.013)* Proportion with some college .011 (.002)** .082 (.005)** .040 (.005)** Proportion of households earning over $50K .000 (.002) .090 (.004)** .051 (.005)** Electoral conditions Proposition 187 turnout –.009 (.019) –.036 (.026) .172 (.038)** Party of incumbent .026 (.001)**–.006 (.001)** –.054 (.002)** Incumbent in race .012 (.001)** .002 (.002) .012 (.004)** Tenure * incumbent –.001 (.000)**–.001 (.000)** –.001 (.000)** Vote margin 2nd quartile (58%–62%) .031 (.002)**–.006 (.002)** –.063 (.002)** Vote margin 3rd quartile (62%–67%) .005 (.001)**–.008 (.001)** .057 (.002)** Vote margin 4th quartile (67%–82%) .050 (.001)** .009 (.001)** –.026 (.002)** N 9,210 17,907 14,535 SER .036 .075 .104 R2 .73 .20 .55 NOTES: For African Americans and Anglos, voter turnout rates are calculated as a proportion of voting-age adults. For Latinos, voter turnout rates are calculated as a proportion of registered voters. The dependent variables range from 0 to 1. District type variables are dichotomous, coded 0 (not in district type), 1 (in district type). The excluded comparison district type is Anglo majority. “Party of incumbent” is coded 0 (Republican) and 1 (Democrat). Measures of age, education, and income are race- and ethnicity-specific. “Vote margin” is indexed by quartile; the model includes dichotomous variables for each quartile, coded 0 (not in quartile) or 1 (in quartile). The excluded comparison “Vote margin” category is the fist quartile: cases with winning vote margins at or below 57 percent. Standard errors are in parentheses. **p < .01, *p < .05, two-tailed. 80 Table D.2 Coefficient Estimates for Regression of Majority-Minority Districts and Minority Representation on Voting Participation Variables Constant African American .022 (.019) Anglo Latino .173 (.026)** –.131 (.036)** District type-legislator race Hispanic majority-Hispanic Hispanic majority-white Black-Hispanic majority-black Multi-ethnic majority-black Multi-ethnic majority-white –.023 (.003)** .051 (.004)** –.004 (.004) .010 (.004)** .065 (.002)** –.002 (.004) .037 (.003)** .060 (.004)** .040 (.002)** .003 (.003) .364 (.004)** .265 (.005)** .311 (.005)** .009 (.008) .096 (.005)** SES conditions Proportion over age 35 .012 (.002)** .085 (.005)** .023 (.006)** Proportion new residents –.100 (.006)** –.193 (.009)** –.044 (.013)** Proportion with some college .012 (.002)** .073 (.005)** .039 (.005)** Proportion of households earning over $50K .000 (.016) .097 (.004)** .043 (.005)** Electoral conditions Proposition 187 turnout –.016 (.019) –.057 (.026)* .239 (.037)** Party of incumbent .026 (.001)** –.005 (.001)** –.059 (.002)** Incumbent in race .013 (.001)** –.001 (.002) .019 (.004)** Tenure * incumbent –.001 (.000)** –.001 (.000)** .000 (.000) Vote margin 2nd quartile (58%–62%) .030 (.002)** –.008 (.002)** –.059 (.002)** Vote margin 3rd quartile (62%–67%) .004 (.001)** –.007 (.001)** .059 (.002)** Vote margin 4th quartile (67%–82%) .051 (.001)** .006 (.002)** –.024 (.002)** N 9,210 17,907 14,535 SER .04 .074 .10 R2 .74 .21 .58 NOTES: For African Americans and Anglos, voter turnout rates are calculated as a proportion of voting-age adults. For Latinos, voter turnout rates are calculated as a proportion of registered voters. The dependent variables range from 0 to 1. District type-legislator race variables are dichotomous, coded 0 (matches district type-legislator race), 1 (matches district type-legislator race). The excluded comparison district type is Anglo majority-Anglo legislator. “Party of incumbent” is coded 0 (Republican) and 1 (Democrat). Measures of age, education, and income are race- and ethnicity-specific. “Vote margin” is indexed by quartile; the model includes dichotomous variables for each quartile, coded 0 (not in quartile) or 1 (in quartile). The excluded comparison “Vote margin” category is the fist quartile: cases with winning vote margins at or below 57 percent. Standard errors are in parentheses. **p < .01, *p < .05, two-tailed. 81 References Abney, Glen, and John D. Hutcheson, “Race, Representation and Trust,” Public Opinion Quarterly, Vol. 45, 1981, pp. 91–101. Baldassare, Mark, California in the New Millennium: The Changing Social and Political Landscape, University of California Press, Berkeley, California, 2000. Barker, Lucius J., Mack Jones, and Katherine Tate, African Americans and the American Political System, Prentice Hall, New Jersey, 1999. Barone, Michael, and Grant Ujifusa, The Almanac of American Politics, 2000, National Journal, Washington, D.C., 1999. Block, A. G., “Wilson Legacy,” California Journal, November 1998, pp. 6–10. Bobo, Lawrence, and Franklin Gilliam, “Race, Sociopolitical Participation, and Black Empowerment,” American Political Science Review, Vol. 84, 1990, pp. 377–393. Brackman, Harold, and Steven P. Erie, “At Rainbow’s End: Empowerment Prospects for Latinos and Asian Pacific Americans in Los Angeles,” in Michael Preston, Bruce Cain, and Sandra Bass (eds.), Racial and Ethnic Politics in California, Vol. 2, Institute of Governmental Studies, Berkeley, California, 1998. Brischetto, Robert R., “Latino Voters and Redistricting in the New Millennium,” in David A. Bositis (ed.), Redistricting and Minority Representation: Learning from the Past, Preparing for the Future, Joint Center for Political and Economic Studies, Washington, D.C., 1998. 83 Brody, Richard, “The Puzzle of Political Participation in America,” in Anthony King (ed.), The New American Political System, American Enterprise Institute, Washington, D.C, 1978. Bucy, E. Page, “Redistricting by Lawsuit,” California Journal, January 1991. Bureau of the Census, Congressional District Data Book (Districts of the 87th Congress), U.S. Department of Commerce, Washington, D.C., 1961. Bureau of the Census, Congressional District Data Book (Districts of the 88th Congress), U.S. Department of Commerce, Washington, D.C., 1963. Bureau of the Census, Congressional District Data Book (Districts of the 92nd Congress), U.S. Department of Commerce, Washington, D.C., 1971. Bureau of the Census, Congressional District Data Book (Districts of the 93rd Congress), U.S. Department of Commerce, Washington, D.C., 1973. Bureau of the Census, 1980 Census of Population and Housing: Congressional Districts of the 98th Congress, U.S. Department of Commerce, Washington, D.C., 1983. Bureau of the Census, 1990 Census of Population and Housing: Population and Housing Characteristics for Congressional Districts of the 103rd Congress, U.S. Department of Commerce, Washington, D.C., 1993. Bureau of the Census, Statistical Abstract of the United States, U.S. Department of Commerce, Washington, D.C., 1999. Cain, Bruce, “Voting Rights and Democratic Theory: Toward a ColorBlind Society?” in Bernard Grofman and Chandler Davidson 84 (eds.), Controversies in Minority Voting, Brookings Institution, Washington, D.C., 1992. California Department of Finance, California Statistical Abstract, Sacramento, California, 1999. California Journal, Congress, December 1994. California Journal, State Assembly, December 1998, pp. 25–28. California Journal Weekly, “Proposition 187: Services of Illegal Immigrants,” November 14, 1994. California Opinion Index, Field Institute, San Francisco, California, August 1990. California Opinion Index, Field Institute, San Francisco, California, January 1993. California Opinion Index, Field Institute, San Francisco, California, June 1994. California Opinion Index, Field Institute, San Francisco, California, January 1995. California Opinion Index, Field Institute, San Francisco, California, February 1997. California Opinion Index, Field Institute, San Francisco, California, January 1999. California Opinion Index, Field Institute, San Francisco, California, May 2000. California Political Week, Inc., California Political Week, Beverly Hills, California, November 14, 1994. California Student Trends, 2000: www.ed-data.k12.ca.us. 85 Cameron, Charles, David Epstein, and Sharon O’Halloran, “Do Majority-Minority Districts Maximize Substantive Black Representation in Congress?” American Political Science Review, Vol. 90, 1996, pp. 794–812. Canon, David T., Race, Redistricting, and Representation, University of Chicago, Chicago, Illinois, 1999. Chavez, Linda, “Party Lines: The Republicans’ Racial Quotas,” The New Republic, Vol. 204, No. 25, June 24, 1991, p.14. Chicago Tribune, “The South’s New Reality,” August 2, 1992. Cho, W. T., “If the Assumption Fits . . . : A Comment on King’s Ecological Inference Solution,” Political Analysis, Vol. 7, 1998. Congressional Quarterly, Inc., Congressional Districts in the 1990s: A Portrait of America, Washington, D.C., 1993. Congressional Quarterly, Inc., CQ Almanac, 1996, Washington, D.C., 1996. Congressional Quarterly, Inc., CQ Almanac, 1997, Washington, D.C., 1997. Conway, Margaret, Political Participation in the United States, 2d. ed., Congressional Quarterly, Inc., Washington, D.C., 1991. Donovan, Beth, “New ‘Majority Minority’ Districts May Mean Lower Black Turnout,” Congressional Quarterly Weekly Report, Vol. 50, No. 10, March 7, 1992, pp. 563–564. Duncan, Phil, “Creating Black Districts May Segregate Voters,” Congressional Quarterly Weekly Report, Vol. 48, No. 30, July 28, 1990, p. 2462. 86 Duncan, Phil, “Minority Districts Fail to Enhance Turnout,” Congressional Quarterly Weekly Report, Vol. 51, No. 13, March 27, 1993a, p. 798. Duncan, Phil, “New Minority Districts: A Conflict of Goals,” Congressional Quarterly Weekly Report, Vol. 51, No. 30, July 24, 1993b, p. 2010. Duncan, Phil, Politics in America, 1994: The 103rd Congress, Congressional Quarterly, Inc., Washington, D.C., 1993c. Duncan, Phil, Politics in America, 1996: The 104th Congress, Congressional Quarterly, Inc., Washington, D.C., 1995. Education Data Partnership, California Student Trends Report, 2001: http://www.ed-data.k12.ca.us/state. Ehrenhalt, Alan (ed.), Politics in America: Members of Congress in Washington and at Home, 1984, Congressional Quarterly, Inc., Washington, D.C., 1983. Erie, Steven P., Harold Brackman, and James Ingram, Paths to Political Incorporation for Latinos and Asian Pacifics in California, California Policy Seminar, University of California, Berkeley, California, 1993. Frayne, Gabriel, “Race for Latino California,” The Nation, Vol. 267, No. 13, October 26, 1998. Freedman, D. A., S. P. Klein, M. Ostland, and M. R. Roberts, “Review of A Solution to the Ecological Inference Problem,” Journal of the American Statistical Association, Vol. 93, 1998. Gilliam, Frank, “Influences on Voter Turnout for U.S. House Elections in Non-Presidential Years,” Legislative Studies Quarterly, Vol. 10, 1985, pp. 339–351. 87 Gilliam, Frank, “Exploring Minority Empowerment: Symbolic Politics, Governing Coalitions, and Traces of Political Style in Los Angeles,” American Journal of Political Science, Vol. 40, 1996, pp. 56–81. Gilliam, Frank, and Karen Kaufmann, “Is There an Empowerment Lifecycle?” paper presented at the Annual Meeting of the Western Political Science Association, San Francisco, California, 1996. Giroux, Gregory, “The Hidden Election: Day of the Mapmaker,” Congressional Quarterly Weekly Report, Vol. 58, No. 8, February 19, 2000, pp. 344–350. Green, Stephen, California Political Almanac, 1995–1996, California Journal, Sacramento, California, 1995. Guerra, Fernando, “Latino Politics in California: The Necessary Conditions for Success,” in Michael Preston, Bruce Cain, and Sandra Bass (eds.), Racial and Ethnic Politics in California, Vol. 2, Institute of Governmental Studies, Berkeley, California, 1998. Handley, Lisa, Bernard Grofman, and Wayne Arden, “Electing Minority-Preferred Candidates to Legislative Office: The Relationship Between Minority Percentages in Districts and the Election of Minority-Preferred Candidates,” in Bernard Grofman (ed.), Race and Redistricting in the 1990s, Agathon Press, New York, 1998. Howell, Susan, and Deborah Fagan, “Race and Trust in Government,” Public Opinion Quarterly, Vol. 52, 1988, pp. 343–350. Huckfeldt, Robert, and John Sprague,“Citizens, Context, and Politics,” The State of the Discipline II, American Political Science Association, Washington, D.C., 1993. Hutchings, Vincent, Harwood McClerking, and Guy-Uriel Charles, “When Does Race Matter for Congressional Responsiveness? An Examination of Support for Black Interests Among Southern 88 Members of Congress,” paper prepared for the Annual Meeting of the Southern Political Science Association, Savannah, Georgia, 1999. King, Gary, A Solution to the Ecological Inference Problem, Princeton University, Princeton, New Jersey, 1997. King, Gary, “The Future of Ecological Inference Research: A Reply to Freedman et al.,” Journal of the American Statistical Association, Vol. 94, 1999, pp. 352–355. Lewis, Jeffrey,“Two-Stage Approaches to Regression Models in Which the Dependent Variable Is Based on Estimates,” unpublished working paper, 2000. Lopez, Elisabeth, and Eric Wahlgren, “The Latino Vote,” California Journal, November 1994, pp. 29–31. Los Angeles Times, “Latino Activists Zero in on County Representation,” August 26, 1990. Los Angeles Times, “Lawyer Views Remap Case As Only 1st Step,” October 11, 1990. Los Angeles Times, “74 Cities Targeted for Voting Rights Fights,” January 21, 1991. Los Angeles Times, “1 Million Voters: Latino Groups Launch Effort to Achieve Goal,” April 30, 1992. Los Angeles Times, “Political Activists Launch Extensive Campaign to Get Latinos to the Polls,” September 6, 1992. Los Angeles Times, “Latinos Unite in Self-Defense on Prop. 187,” October 21, 1994. Los Angeles Times, “Emotions High at Last-Minute Protests,” November 8, 1994. 89 Los Angeles Times, “Crime, Immigration Issues Helped Wilson, Poll Finds,” November 9, 1994. Los Angeles Times, “State’s Diversity Doesn’t Reach Voting Booth,” November 10, 1994. Los Angeles Times, “Perspectives on Election Fallout,” November 11, 1994. Los Angeles Times, “Let Election Setbacks Be Call for Action,” November 14, 1994. Los Angeles Times, “O.C. Precinct Typifies Latino Voter Apathy,” November 28, 1994. Los Angeles Times, “Despite Gains, Latino Voters Still Lack Clout,” December 4, 1994. Los Angeles Times, “Massive Tide of Naturalized Citizens Swells Voter Rolls,” September 29, 1996. Los Angeles Times, “In Contests Big and Small, Latinos Take Historic Leap for the Record,” November 5, 1998. Los Angeles Times, “Latino Revolution Leaves Some City Councils Untouched,” October 9, 2000. Lublin, David, The Paradox of Representation, Princeton University, Princeton, New Jersey, 1997. Maharidge, Dale, “Did 1992 Herald the Dawn of Latino Political Power?” in Thomas Hoeber and Charles Price (eds.), California Government and Politics Annual, 1994–1995, California Journal, Sacramento, California, 1994. Martis, Nancy, and A. G. Block, “Proposition 187,” California Journal, December 1994, pp. 20–25. 90 McClain, Paula D., and Joseph Stewart, Jr., Can We All Get Along? Racial and Ethnic Minorities in American Politics, Westview Press, Boulder, Colorado, 1999. Nagler, Jonathan, “The Effect of Registration Laws and Education on U.S. Voter Turnout,” American Political Science Review, Vol. 85, 1991, pp.1393–1405. New York Times, “Hispanic Politicians Seek a Recipe for Raw Numbers,” April 12, 1992. New York Times, “Redistricting Expected to Bring Surge in Minority Lawmakers,” August 3, 1992. New York Times, “The Old Order Changes in Congress—A Little,” November 2, 1992. New York Times, “Suits Challenging Redrawn Districts That Help Blacks,” February 14, 1994. Orange County Register, “Hispanics See a Political Sea Change,” October 22, 1998. Pachon, Harry P., “Latino Participation in California Politics,” in Byan O. Jackson and Michael Preston (eds.), Racial and Ethnic Politics in California, Vol. 1, Institute of Governmental Studies, Berkeley, California, 1991. Pachon, Harry P., “Latino Politics in the Golden State: Ready for the 21st Century?” in Michael Preston, Bruce Cain, and Sandra Bass (eds.), Racial and Ethnic Politics in California, Vol. 2, Institute of Governmental Studies, Berkeley, California, 1998. Palmquist, B. L., and D. Stephen Voss, “Racial Polarization and Turnout in Louisiana: New Insights from Aggregate Data Analysis,” paper presented at the Annual Meeting of the American Political Science Association, San Francisco, California, 1996. 91 Reyes, Belinda (ed.), A Portrait of Race and Ethnicity in California: The Social and Economic Well-Being of Racial and Ethnic Groups, Public Policy Institute of California, San Francisco, California, 2001. Rodriguez, Gregory, “The Latino Century,” California Journal, January 2000, pp. 8–14. Rosenstone, Steven J., and John Mark Hansen, Mobilization, Participation, and Democracy in America, MacMillan, New York, 1993. Sacramento Bee, “The Paradox of Remapping,” November 30, 2000. Sample, Herbert, “Washington Notes,” California Journal, January 1998, p. 44. San Francisco Chronicle, “New Voters in State Are Mostly Latino,” May 1, 2000. San Francisco Examiner, “A New California: A Growing Latino Presence in Compton Shows Signs of Reshaping Culture, Politics,” February 13, 2000. San Jose Mercury News, “Established Latinos More Likely to Support Curbing Immigration,” October 15, 2000. Schockman, H. Eric, “California’s Ethnic Experiment and the Unsolvable Immigration Issue: Proposition 187 and Beyond,” in Michael Preston, Bruce Cain, and Sandra Bass (eds.), Racial and Ethnic Politics in California, Vol. 2, Institute of Governmental Studies, Berkeley, California, 1998. Scott, Steve, “Reality Votes,” California Journal, January 1998, pp. 25– 29. Scott, Steve, “Competing for the New Majority Vote,” California Journal, January 2000, pp. 16–23. 92 Shaw, Daron, Rodolfo O. de la Garza, and Jongho Lee, “Examining Latino Turnout in 1996: A Three-State, Validated Survey Approach,” American Journal of Political Science, Vol. 44, 2000, pp. 332–340. Stiles, Jon, Jonathan Cohen, Zachary Elkins, and Frederic Gey, California Latino Demographic Databook, California Policy Seminar, University of California, Berkeley, California, 1998. Swain, Carol, Black Faces, Black Interests, Harvard University, Cambridge, Massachusetts, 1995. Tate, Katherine, “Black Political Participation in the 1984 and 1988 Presidential Elections,” American Political Science Review, Vol. 85, 1991, pp. 1159–1176. Teixeira, Ruy, Why Americans Don’t Vote, Greenwood Press, Westport, Connecticut, 1987. Timpone, Richard J., “Structure, Behavior and Voter Turnout in the United States,” American Political Science Review, Vol. 92, 1998, pp. 145–158. Verba, Sidney, and Norman Nie, Participation in America, University of Chicago Press, Chicago, Illinois, 1972. Verba, Sidney, Kay Schlozman, and Henry Brady, Voice and Equality: Civic Voluntarism in American Politics, Harvard University, Cambridge, Massachusetts, 1995. Washington Post, “INS Sees Surge in Citizenship Seekers,” April 12, 1995. Whitby, Kenny J., The Color of Representation, University of Michigan Press, Ann Arbor, Michigan, 1998. 93 Wolfinger, Raymond E., and Steven Rosenstone, Who Votes? Yale, New Haven, Connecticut, 1980. 94 About the Author CLAUDINE GAY Claudine Gay is an assistant professor of political science at Stanford University and an adjunct fellow at the Public Policy Institute of California. Her recent publications include “Doubly Bound: The Impact of Gender and Race on the Politics of Black Women ” (with Katherine Tate), which appeared in Political Psychology in 1997. A graduate of Stanford University in 1992, she received a Ph.D. from Harvard University, where she won the Toppan Prize in 1998 for best dissertation in political science. 95 Other Related PPIC Publications A Portrait of Race and Ethnicity in California: An Assessment of Social and Economic Well-Being Belinda I. Reyes (editor) et al. Ethnic Context, Race Relations, and California Politics Bruce Cain, Jack Citrin, Cara Wong Finding Common Ground: Racial and Ethnic Attitudes in California Zoltan Hajnal, Mark Baldassare How Different Ethnic Groups React to Legal Authority Yuen J. Huo, Tom R. Tyler 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 97" ["post_date_gmt"]=> string(19) "2017-05-20 09:35:11" ["comment_status"]=> string(4) "open" ["ping_status"]=> string(6) "closed" ["post_password"]=> string(0) "" ["post_name"]=> string(8) "r_601cgr" ["to_ping"]=> string(0) "" ["pinged"]=> string(0) "" ["post_modified"]=> string(19) "2017-05-20 02:35:11" ["post_modified_gmt"]=> string(19) "2017-05-20 09:35:11" ["post_content_filtered"]=> string(0) "" ["guid"]=> string(50) "http://148.62.4.17/wp-content/uploads/R_601CGR.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) }