Do Americans prefer coethnic representation? The impact of race on House incumbent evaluations.

• Author: Ansolabehere, Stephen

Table of Contents Introduction I. Race and Representation A. Conceptual Foundations B. Prior Empirical Research II. Methodology and Data III. Results A. Coethnicity and Evaluations of Representatives B. Effects of Race and Party C. Accounting for Racial Preferences of White Democrats: A Multivariate Analysis IV. Discussion Conclusion Introduction

Social groups, such as races and religions, are undeniably important in the election of representatives in a democracy. Group identities can help solve collective action problems, such as voting, (1) and group politics shape the distribution of and responses to public goods provision and other public policies. (2) Furthermore, group identities are wrapped in symbols that can have particularly powerful appeal to individuals of that group or can alienate individuals from a competing or hostile group. (3) In U.S. politics, racial and ethnic identity creates one of the most enduring political and social groupings. (4) An extensive and multifaceted literature examines how race influences our understanding of political choices and voting, with scholars repeatedly finding substantial racial and ethnic differences in voters' political beliefs and preferences. (5) More controversial, however, is whether those differences translate into a preference for coethnic representation--that is, whether people prefer representatives who are the same ethnic background as they are, and whether that preference is inherently racial or reflects some other factor, such as party, that is correlated with race or ethnicity.

The conjecture that people prefer coethnic representation has driven legislation and litigation concerning voting rights for over half a century in the United States. The theory behind the Voting Rights Act (VRA) (6) and litigation on behalf of minority voters posits that black voters and Hispanic voters want representation by people of the same race or ethnicity as themselves. (7) Under such a conjecture, minorities' preferences for their "own" candidates mean that they have distinctive preferences or interests, and when combined with whites' preference for whites under a plurality system, minority voters would be unable to elect their preferred candidates without intentionally constructing districts around minority interests. (8) As a result, one of the most striking effects of the VRA has been the creation of a substantial number of congressional districts in which African Americans and Hispanics win seats in the U.S. House of Representatives and in state legislatures. (9) Such districts have been required in areas where (1) there is highly polarized voting along racial lines and (2) there are sufficient numbers of blacks and Hispanics to create districts where these groups can elect candidates they prefer. (10) Advocates and scholars argue that it is necessary to create majority or near majority-minority districts in order to ensure minorities have an equal opportunity to elect their preferred candidates or candidates of their own race, regardless of minority support for white candidates or the intentions of districtors. (11)

These majority-minority districts offer an important research opportunity. The significant number of minority representatives in Congress makes it possible to measure the degree of satisfaction that whites, blacks, and Hispanics express with representatives of their own race and of other races. After the 2010 elections, there were 68 black or Hispanic members of Congress and 359 white members of Congress. (12) How do blacks or Hispanics living in districts represented by whites feel about their members of Congress? How do whites living in districts represented by blacks or Hispanics feel about their members of Congress?

This Article examines two distinct, but often conflated, questions. First, to what extent does a representative's race or ethnicity affect how citizens evaluate that representative and whether citizens support that representative? Specifically, do coethnics support their representative more than individuals from other ethnic or racial groups do? Second, is a citizen's support for a representative more reflective of the racial identity of the representative per se (apart from any policy or ideology), or is it more reflective of ideological or policy preferences that the candidates present?

A substantial literature examines the effects of race on voter choice, but there has been less attention to preferences about representation. The main reason for the neglect of this question is methodological. (13) The primary surveys for studying political representation have been the American National Election Studies (ANES) (14) and the National Black Election Study Series. (15) These surveys typically have up to 2000 respondents, with very small numbers in districts represented by minority politicians. (16) In a national sample survey of 2000 people, one would expect only 300 respondents in districts represented by blacks or Hispanics. Of those 300 people, approximately 100 are black, 100 are Hispanic, and 100 are white. So in a national sample survey, any inferences about minority voting preferences are based on only 100 people within a given group, and any inferences about white attitudes toward minority representatives are based on only 100 respondents as well. To address this issue, we leverage the large-sample Cooperative Congressional Election Study (CCES) surveys conducted in 2008 and 2010, in which respondents were asked to evaluate their incumbent House members via approval ratings along with their willingness to reelect their representative. (17) These surveys combined have more than 80,000 respondents. Beyond the larger sample size offered by the CCES, the surveys also asked a battery of questions asking constituents to describe the racial and ethnic background, party, and ideology of their elected officials, allowing for a clear measure of the respondent's perceptions of these key variables and how they align with the respondent's own identity. Therefore, we can be sure that the results we report are not due to misperceptions about a legislator's background or political positions on the part of the constituent. As a result, we are able to accurately gauge the extent to which racial groups favor coethnic representation and whether race or some other factor drives those preferences.

To summarize our main findings, we show that race alone serves as a predictor of whether a citizen approves of the job a representative is doing and whether the citizen voted for the representative in the most recent election or intends to vote for that representative in the next election. White, black, and Latino constituents express much higher levels of approval and electoral support for politicians who are the same race as them. However, turning to the second hypothesis, we also find that the effect of race on support is much smaller after controlling for party and ideology, and that the effect of party controlling for race (i.e., within racial groups) is much larger than the effect of race controlling for party (i.e., within partisan groups). Party and ideology do not completely explain racial preferences in representation, but they do account for much of the observed support for coethnic representation. As shown in other work, a modest racial difference persists after controlling for party. (18) Interestingly, the effect seems to be concentrated among whites. Blacks and Hispanics express approximately the same level of approval for white representatives of a given party as they do for black and Hispanic representatives of that same party. However, white citizens express higher levels of approval of white representatives than they do of black or Hispanic representatives of a given party. Owing to the small number of minority Republican representatives, we focus on Democratic voters with Democratic representatives. Investigating further the opinions of white Democrats, we see that the only factor that appears to account for whites' tendency to rate coethnic copartisans more highly is the respondents' racial attitudes. Since we account for other types of political beliefs throughout the study, this points to race as a small but significant factor in evaluating elected officials.

1. Race and Representation

   1. Conceptual Foundations

      At its core, our research analyzes race and representation within Hanna Pitkin's theoretical framework of descriptive, symbolic, and substantive representation. (19) According to Pitkin's classification, descriptive representation occurs when a voter is represented by someone of the same social identity as the voter, such as gender, occupation, or race: men represented by men, farmers represented by farmers, blacks represented by blacks. (20) Symbolic representation occurs when the representative stands for or carries a particularly important meaning for those being represented. (21) For example, a war hero might symbolically represent the struggles the society endured during the war. Substantive representation occurs when a representative takes action on behalf of the people represented, such as sponsoring legislation that benefits the district or voting for a law that the people in the district support. (22)

      Debates about the meaning and application of the Voting Rights Act concern what mode or style of representation voters seek. Do blacks, Hispanics, and whites want representatives who are of the same race as they are as a matter of descriptive representation? Are whites opposed to the preferences of blacks and Hispanics because of the race of the representative? Or do people seek representatives who reflect their policy or partisan preferences? These questions can be understood by examining the opinions, attitudes, and vote preferences of people when facing a choice between race and policy or race and party. Might white Republicans represent Hispanic voters in South Florida as well as Hispanic Republicans...