Urban Affairs Review

Urban Affairs Review http://uar.sagepub.com

Voting for Minority Candidates in Multiracial/Multiethnic Communities Robert M. Stein, Stacy G. Ulbig and Stephanie Shirley Post Urban Affairs Review 2005; 41; 157 DOI: 10.1177/1078087405280311 The online version of this article can be found at: http://uar.sagepub.com/cgi/content/abstract/41/2/157

Published by: http://www.sagepublications.com

Additional services and information for Urban Affairs Review can be found at: Email Alerts: http://uar.sagepub.com/cgi/alerts Subscriptions: http://uar.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations http://uar.sagepub.com/cgi/content/refs/41/2/157

Downloaded from http://uar.sagepub.com by Sandra Hopps on October 15, 2008

10.1177/1078087405280311 Stein et al. / VOTING FOR MINORITY URBAN AFFAIRS CANDID REVIEW ATES / November 2005

VOTING FOR MINORITY CANDIDATES IN MULTIRACIAL/ MULTIETHNIC COMMUNITIES ROBERT M. STEIN Rice University

STACY G. ULBIG Southwest Missouri State University

STEPHANIE SHIRLEY POST Rice University

Recent research suggests that over time the performance of minority officeholders rivals racebased attitudes and group membership as the primary determinant of citizen evaluations of minority officeholders. Here, we examine the determinants of electoral support for an AfricanAmerican mayor in a multiracial/multiethnic venue. We test alternative explanations (race, social distance, and performance-based models) of voter support for an African-American mayor in a setting where no ethnic or racial group represents more than half the electorate. Our findings indicate that approval ratings coexist with racial-group identification as a determinant of voter support for minority mayors, with one important caveat. Racial voting appears to be more influential in minority candidates’ first electoral bids. While race strongly influences voter support for minority mayors during their initial run for office, job approval becomes more important when the minority candidate runs for reelection.

Keywords: race; ethnicity; voting behavior; urban politics

The rate at which non-Anglo candidates have succeeded at the ballot box continues to increase. Some communities have more than 20 years of experience with African-American mayors, council members, and state and congressional representatives. More importantly, minority candidates are increasingly winning elections in districts and communities where minority voters are not a majority of the electorate, a fact especially important in a nation trending toward a demographic distribution in which Anglos constiURBAN AFFAIRS REVIEW, Vol. 41, No. 2, November 2005 157-181 DOI: 10.1177/1078087405280311 © 2005 Sage Publications

157


158

URBAN AFFAIRS REVIEW / November 2005

tute a smaller proportion of the population and electorate (U.S. Census 2000). Increasingly, elections across the country will be played out in settings with candidates from various racial and ethnic groups competing for votes among a multiracial/multiethnic electorate. Consequently, researchers have begun to ask whether the continued election of minority candidates, especially with the support of Anglo voters, signals a change in race-based voting. Several researchers (Howell and Perry 2001; Howell and McLean 2001; Hajnal 2001, 2001a) have found that over time the performance of minority officeholders rivals race-based attitudes and group membership as a determinant of citizen evaluations of minority officeholders. To date, these studies have focused on citizen evaluations of minority officeholders rather than their vote choice for these candidates. Furthermore, most of these studies have been conducted in biracial settings with either a majority Anglo or majority African-American electorate. In this paper we examine the determinants of electoral support for an African-American mayor in a multiracial/ multiethnic, rather than biracial, venue. We test alternative explanations of voter support for an African-American mayor in a setting where no ethnic or racial group represents more than half the electorate.

RACIAL AND PERFORMANCE EXPLANATIONS OF ELECTORAL SUPPORT Previous research on voting for minority officeholders has focused on three explanations: racial voting, racism, and performance/information. Bullock and Campbell (1984) and Henig (1993) draw an important distinction between racial group voting and racist voting. The frequently observed tendency of racial and ethnic groups to vote for “their own kind” when given the opportunity raises the question whether these cleavages reflect deep-seated biases or simply the pragmatic assessment that candidates of one’s own race or ethnicity are more likely to promote values similar to one’s own. Bullock and Campbell (1984) label the former possibility racist voting and reserve the term racial voting for the latter (Henig 1993, 545). Racial voting is normally measured in terms of self-identified group membership or by residence in a racially or ethnically homogenous precinct, census group, or community. A strong correlation between these measures and voting for a minority candidate is taken as evidence of voting for “someone like me,” or racial group voting.1 To date, the evidence of racial voting is quite strong. An army of researchers (Browning, Marshall, and Tabb 1984;


Stein et al. / VOTING FOR MINORITY CANDIDATES

159

Huckfeldt and Kohfeld 1989; Kleppner 1985; Vanderleeuw 1990; Henig 1993; Giles and Buckner 1993; Kaufman 1998; Howell and McLean 2001; Hajnal 2001) have firmly established the strong positive relationship between these measures and balloting for minority candidates. All of these studies, however, have been conducted in settings where the population is either majority White or majority African-American. Racist attitudes are normally measured in terms of survey-generated measures of racial attitudes toward majority and minority group members. These measures are often an index based on a series of questions that tap antiminority feelings and negative stereotypes of racial and ethnic minorities. Evidence that racist attitudes influence voting for minority candidates is mixed and problematic. First, the use of survey-generated measures poses several serious obstacles for researchers. Responses to questions that attempt to measure racial attitudes and motivations may “lead respondents to rationalize and deny even to themselves the role that race plays in their decision processes” (Henig 1993, 546). Researchers have employed several devices to correct for this measurement problem, including less obtrusive questions about members of other races and ethnic groups and group conflict (cf. Sears and Henry 1999). When measures of racist attitudes are included with indicators of racial group membership in an analysis of evaluations of minority candidates and officeholders, racial group explains a significant portion of variation in mayoral approval (Howell and Perry 2001b). Moreover, these researchers find that racial group membership is more influential than racist attitudes in explaining minority mayoral approval. Howell and Perry explain this finding in two ways. “First, is the problem of measurement error in the attitude scales. Second, is the power of racial identity, stripped of any overlay of racial attitudes. Social identity theory tells us that we develop a sense of who we are socially very early in life, long before race-related attitudes become meaningful” (2001, pp. 22–23). Collectively this body of research demonstrates that racial group membership rather than racist attitudes is a more important determinant of predicting approval of minority mayors.

INFORMATION AND PERFORMANCE EXPLANATIONS OF ELECTORAL SUPPORT An alternative explanation of voting for minority candidates is the experience and knowledge voters—especially nonminority voters—obtain from their experience with minority elected officials. Hajnal suggests “that black political representation should significantly improve white attitudes toward African-Americans and increase the likelihood that whites will vote for the


160


black incumbent in the next election (even if she/he runs against a white challenger)” (2001a, 603). The logic behind this hypothesis is simple. Anglo voters’ experience with minority officeholders imparts critical information that greatly reduces uncertainty and dispels White fears about AfricanAmericans and African-American leadership. Over time, Anglo voters begin to evaluate minority officeholders based on their performance in office rather than their race. Hajnal’s (2001; 2001a) analysis of Anglo racial attitudes and support for African-American incumbents supports his informational story. He finds that in African-American challenger elections Anglo voters are more likely to vote on the basis of racial group membership. In AfricanAmerican incumbent elections, he finds that Anglo voters “tend to vote in a more ‘reasoned’ fashion as issues, candidates, and endorsements become more important” (2001, 603). 2 Howell and McLean (2001) and Howell and Perry (2001) provide additional support for Hajnal’s information model of voting for minority candidates. Drawing from research that examines voter behavior in incumbent elections at the national and state level, they explain voting behavior in mayoral elections as a function of incumbent job performance. These authors hypothesize that incumbents are reelected when they have performed up to or in excess of voter expectations. The job performance model easily outperforms the racial model in explaining voter approval for incumbent AfricanAmerican mayors. It appears that over time voters are more concerned about mayoral performance than race when at the ballot box. Howell and Perry (2001) also find that racial attitudes have little explanatory power in a model of approval ratings for African-American incumbent mayors. Rather, racial group membership is an important determinant of approval ratings for African-American incumbents. The information and performance models of minority incumbent voting suggest that over time we should observe a change in the relative explanatory power of racial group membership, racist attitudes, and incumbent job performance. Specifically, the explanatory power of race-based voting, attitudinal or group membership, should decline and incumbent job performance should increase in explanatory power. To date, confirmation of this hypothesis has been found in settings where minority mayors service a biracial electorate, either a majority African-American or a majority Anglo electorate. This of course is the modal condition in most central cities and larger metropolitan areas. This is not, however, the direction the nation is headed. The Census Bureau estimates that by 2050 less than half the U.S. population will be Anglo (U.S. Census 2000). Studying voting for incumbent minority candidates in a multiracial and multiethnic setting provides the basis for general-



161

izing beyond urban and metropolitan politics to include state and national political settings.

VOTING FOR MINORITY INCUMBENTS IN MULTIRACIAL SETTINGS We suspect that voting for minority candidates and incumbents may vary with the presence or absence of an ethnic or racial majority. There is strong evidence in the literature to show that electoral support for incumbent minority candidates in majority-Anglo communities is shaped by voter job evaluation of the incumbent. For several reasons we are less confident that this explanation will hold unambiguously in multiracial/multiethnic settings. In majority-Anglo and majority-minority electorates, voter choices and dynamics center on whether majority-group members ballot for minority candidates (Browing, Marchall, and Tabb 1984). In multiracial/multiethnic settings electoral coalition partners increase to include three potentially winning electoral dyads: Anglo/African-American, Anglo/Hispanic, and African-American/Hispanic. Any of these dyads can provide electoral victory over the remaining group. Meier and Stewart (1991) demonstrate that voter choices and electoral coalitions are shaped by social distance: “cooperation is more likely if two groups are similar in social, economic and racial terms” (1991, 1125). Meier and Stewart and others (Dyer, Vedlitz, and Worchel 1989) believe Anglos are likely to perceive Hispanics as more similar than African-Americans to themselves, leading them to hypothesize that ceribus paribus Anglos are more likely to vote for Latino rather than AfricanAmerican candidates. These authors and others (Hero 1987; Giles and Evans 1985) find that social distance as measured by class (i.e., income) explains the greater likelihood of Anglo/Hispanic rather than African-American/Hispanic electoral coalitions. Multiracial/multiethnic electorates provide voters and candidates with more potential winning electoral coalitions than are present in majorityAnglo or majority-minority electorates. Our test of the information hypothesis in multiracial elections will examine these alternative electoral coalitions as they affect the relationship between voters’ job evaluation of incumbent minority candidates and support for the incumbent. Given the opportunities for alternative racial/ethnic coalitions to form in nonmajority communities we might expect that voting on the basis of incumbent job performance will be tempered by the social distance between the racial group affiliation of the incumbent mayor and the voter’s racial/ethnic group membership. Most importantly, we expect that racial and job performance explanations of


162


voting for incumbents will empirically coexist and vary in importance with the racial makeup of the contesting candidates. We expect that a social distance model influences voting behavior when a voter’s racial group is not represented among the contending mayoral candidates. In this case incumbent job performance and race coexist as determinants of mayoral vote choice. Furthermore, we expect that with each successive election vote choice for the incumbent minority mayor will be influenced more by the voters’ job evaluation of the incumbent mayor than either social distance or racial-group voting.

HYPOTHESES Given our review of the literature and expectations about voter behavior in racially/ethnically plural communities we identify several hypotheses for testing: RACIAL VOTING HYPOTHESIS

Hypothesis 1: Support for a mayoral candidate is significantly related to congruence between the race of the voter and candidate. SOCIAL DISTANCE VOTING HYPOTHESIS

Hypothesis 2: In electoral contests between Hispanic and African-American mayoral candidates Anglo voters are more likely to vote for the Hispanic candidate. Hypothesis 3: In electoral contests between Anglo and African-American mayoral candidates Hispanic voters are more likely to vote for the Anglo candidate. JOB-APPROVAL VOTING HYPOTHESIS

Hypothesis 4: Support for an incumbent mayoral candidate is positively related to a voter’s job rating of the incumbent mayor. RACIAL AND JOB-APPROVAL VOTING HYPOTHESES

Hypothesis 5: The strength of job approval on mayoral vote choice increases with successive incumbent reelection campaigns. The strength of racial group affiliation on vote choice decreases with successive incumbent reelection campaigns.

THE RESEARCH SETTING, DATA, AND MEASURES The setting for this study is three Houston, Texas, mayoral elections held between 1997 and 2001. Houston is a multiracial/multiethnic community with 39% of its population Anglo, 36% Hispanic, and 25% AfricanAmerican. The current racial composition of the city reflects the predicted national racial composition by the year 2020. The electorate’s profile in



163

Houston is similar to that of the general population but skewed toward Anglo voters.3 All Houston elective offices (i.e., Mayor, Controller, and City Council members) are nonpartisan4 and term limited to three two-year terms of office. In 1997 Houston elected its first African-American mayor, Lee P. Brown. Brown, a former Houston Chief of Police (1981–1991), defeated Anglo businessman Robert Mosbacher. Brown was reelected in 1999 where he faced weak Anglo opposition. He won a third term in 2001 defeating Orlando Sanchez in a runoff election by less than 10,000 votes out of 284,000 votes cast. Sanchez, a Cuban-American, term-limited, three-term city council member, was the first Hispanic to run for the office of mayor. This election was the first mayoral election in a major city where a Hispanic and an African-American candidate faced each other in a runoff. Lee P. Brown’s three consecutive terms of office provide an opportunity to test competing explanations of voting for minority candidates in a multiracial/multiethnic electorate. The racial and ethnic makeup of candidates in these mayoral elections provides a unique opportunity to study voter behavior in elections with several different racial and ethnic pairings of mayoral candidates. Data come from three cross-sectional surveys of registered voters conducted before the Houston mayoral elections in 1997, 1999, and 2001.5 Voters were queried about their vote intention, mayoral-job approval, racial group membership, partisan affiliation, ideology, selected demographic traits, and racial attitudes (see Table 4). To test our hypotheses about voter behavior we have constructed a pooled cross-sectional database of voter responses to preelection surveys conducted in 1997, 1999, and 2001. This design allows us to test alternative explanations of voter behavior within and across elections. Dummy measures for each election (i.e., year) and their interaction with the race/ethnicity of the voter and mayoral-job approval are included to measure the effects of these independent variables in each election. Our initial model includes measures of voters’ racial group membership, approval rating of the African-American mayoral candidate, partisan identification, ideology, and education. Several alternative models are reported that include alternative explanations of vote choice including racial attitudes and a test for the endogeneity of job approval.6 The dependent variable in our analysis is vote intention: 1 = intend to vote for Brown, 0 = do not intend to vote for Brown.7 The independent variables include: mayoral-job approval (1 = unfavorable, 2 = neutral, 3 = favorable), party identification (1 = Democrat, 2 = Independent, 3 = Republican), ideology (1 = liberal, 2 = moderate, 3 = conservative), educational level (1 = less than high school, 2 = high school graduate, 3 = business/technical school, 4 = some college, 5 = college graduate, 6 = post-graduate education), and the


164


race/ethnicity of the respondent (dichotomous variables indicating whether respondent was Asian, African-American, or Hispanic).8 In 1997 and 2001, survey respondents were asked whether they have a “favorable or unfavorable opinion of Lee P. Brown.” In 1999 and 2001 survey respondents were asked: “Overall, do you approve or disapprove of the job Lee P. Brown is doing as mayor?”9

Findings Table 1 reports the logit regression coefficients for a pooled crosssectional analysis of mayoral voting in the 1997, 1999, and 2001 mayoral elections (N = 1,774).10 The independent regressors include racial group affiliation, incumbent mayoral-job approval, party identification, ideology, education, year, and several interaction terms. Logit coefficients are not easily interpreted. Consequently, the magnitude and significance of the effect of any independent variable on voting is determined by calculating the probability of voting for incumbent African-American mayor Lee P. Brown, for specific values of independent variables of interest, controlling for all other independent variables at their modal values.11 We use CLARIFY software (Tomz, Wittenberg, and King 1999) to judge the statistical significance of differences in the probabilities of voting for Brown. The model of vote intention reported in Table 1 performs well. The model accounts for 40% of the variation in vote intention over the three mayoral elections. Moreover, the model correctly predicts 60% of the cases for the three elections for a 50% proportional reduction in error.

RACIAL VOTING AND SOCIAL-DISTANCE HYPOTHESES The year by race/ethnicity interaction terms allows us to test the racial voting and social-distance hypotheses (Hypothesis 1–Hyopthesis 3) and our expectations about these relationships over time (Hypothesis 5). As expected, the coefficient for African-American voters in 1997 is positive and statistically significant (p < .05). African-American voters were significantly more likely than other voters to ballot for the African-American mayoral candidate Lee Brown (see Table 2). Compared to Anglos and Hispanics, African-American voters were 66% and 47% more likely to vote for Brown respectively. As expected, the likelihood that African-American voters balloted for Brown, independent of other determinants, diminished in his two



165

TABLE 1: Logit Estimates for Vote Intention in Houston Mayoral Elections: 1997, 1999, 2001

Constant Approval rating African-American Asian Hispanic Party Ideology Education Year = 1997 Year = 2001 African-American *1997 African-American *2001 Hispanic *1997 Hispanic *2001 Asian *1997 Asian *2001 Approval *1997 Approval *2001

Coefficient

Standard Error

Z

–1.31 1.13* .954* 1.77 –4.06 –.787* –.290* .057 –5.76 –1.60* 2.35* .392 1.28* –.553 –1.49 –.685 .174 .688*

.462 .136 .313 1.15 .421 .088 .099 .047 .361 .590 .673 .486 .581 .537 1.50 1.47 .230 .223

–2.83 8.27 3.05 1.54 –.96 –8.95 –2.94 1.22 –1.60 –2.71 3.50 .810 2.20 –1.03 –.990 –.460 .750 3.09

NOTE: N = 1,774. Lr Ch 12 (17): 943.61. Pseudo R-square: .395. % case correctly predicted: 60. Proportion reduction in error: 50%. *p