The Hispanic voter--and I want to say this very carefully--has not shown a lot of willingness or affinity to support black candidates.
--Clinton Pollster Sergio Bendixen to Ryan Lizza, The New Yorker, January 21, 2008
The Democratic Party's nomination of an African American as its standard-bearer in the 2008 presidential election created an unprecedented opportunity to examine the role of race in electoral choice. Barack Obama managed to win an Electoral College landslide and, for the first time in a generation, a majority of the popular vote for the Democratic nominee. But Obama's historic nomination and election served as shocks to the political system and to political science research investigating the role of race in electoral politics. On the one hand, some observers were quick to announce the end of race as a determinative force in American electoral politics, suggesting that Obama's race was unrelated to two-party choice and the dawn of a postracial society. By contrast, others raised suggestions during the primary that racial sentiments would hobble an Obama effort in the general election--specifically, that interminority conflict might suppress Obama's vote among Hispanics. (1) With the election behind us, we now can ask what, if any, evidence is available for these claims?
In this effort, we explore the role of race in the 2008 election, with particular attention to the possible importance of interminority dynamics in shaping the two-party vote. In so doing, we employ multiple indicators of racial sentiment--explicit, indirect, and implicit--to ascertain the predictive validity of each, using data from the 2008 American National Election Study (ANES). We argue, first, that racial sentiments vary considerably across voters when grouped by partisan attachment and two-party vote. Second, we observe that there is far less variation between Latinos and non-Hispanic whites, suggesting that the attitudinal basis for a strong racially motivated vote exists among Latinos. Third, beyond its effects on partisan attachments, racial sentiment had a unique additional effect on the voting behavior of Americans in the last election, but this effect was significantly attenuated or altogether absent among Latinos. Notwithstanding their similar distributions on measures of race sentiment, Latino voters do not appear to have applied those sentiments to an evaluation of then-candidate Obama. Finally, we show that a similar pattern of results obtains when examining the full range of evaluation and intensity of views on Obama, rather than solely the dichotomous vote choice.
Race and American Elections
The notion that Obama's election was in any way postracial is belied by the facts on the ground. National exit polls estimate that Obama received only 43% of the white vote at a time when the United States was engaged in two hot wars overseas, faced an economy in complete freefall, and had an incumbent president with some of the lowest popularity ratings in the history of polling on the matter. By contrast, Obama received clear majorities among Asian Americans (62%), "others" (66%), Latinos or Hispanics (67%), and African Americans (95%). By any measure, the electorate remains racially polarized and, in light of the economic and political circumstances of the country during the fall campaign, Obama may well have underperformed what we might expect from another Democrat, although such a counterfactual is not available.
Bowler and Segura (forthcoming) argue that race can be realized in a vote choice in two manners. First, they suggest, because race is deeply embedded in the structure of American party coalitions (Carmines and Stimson 1989), racial sentiment is ever present in the two-party vote. If, for example, we examined the racial and ethnic breakdown of the vote between George W. Bush and John Kerry in 2004, we would also find considerable (albeit less) racial polarization. Democratic presidential candidates have not won a majority of white votes since the 1964 election. Considering Asian Americans, Latinos, and African Americans in total, there are 64 nonwhite members of the U.S. House of Representatives (in 2009) from the Democratic Party, compared with 5 from the GOP. At the state legislative level (as of 2008), nonwhites represent only a scant portion of the GOP in office: only 2.1% of Republicans in lower chambers are persons of color, compared with 22.4% of Democrats, and only 1.3% of Republicans in state senates, compared with 22.6% of Democrats. As a practical matter, the GOP is an all-white political party.
Clearly, there is an association of party and race in American politics irrespective of the race of particular candidates. But the presence of a nonwhite candidate of a major party allows us to explore whether race of candidate effects created a uniquely difficult hurdle for the Obama candidacy. By accounting for partisanship, multivariate estimations of vote choice allow us to isolate the distinct effect of Obama's race.
Oddly, Obama's performance among whites, at 43%, exceeded that of Kerry (41%), and Obama won the majority of the white vote in 16 states, a considerable improvement over Kerry. At first blush, this suggests that his candidacy had little exacerbating effect on the racial dynamics between the major parties. On the other hand, Obama's share of the white vote was lower than Kerry's, and it was shockingly small in several states, including Louisiana (14%), Mississippi (11%), and Alabama (10%). These declines were of significant magnitude, with a 10% decline in Louisiana, a 9% decline in Alabama, and an 8% decline in Texas. Moreover, even in states where Obama appears to have outperformed Kerry, we cannot tell at a glance what the counterfactual might be. That is, might Obama have done even better in the states where he won the white vote had racial sentiment not played a role?
The literature on the race of candidate effect is substantial and has a broad consensus. Candidates of color fare poorly with white voters, all else being equal (Grofman and Handley 1989; Lublin 1997; Parker 1990). Indeed, the entire literature on the implementation of the Voting Rights Act and its effect on representation is premised on the long-standing empirical claim that voting is racially polarized; indeed, racial polarization is a pillar of the prevailing judicial standard set forth in Thornburgh v. Gingles (1986). (2) For Latinos and African Americans, the overwhelming empirical record is that persons of color are elected almost solely from jurisdictions or districts with a majority of the electorate of that racial or ethnic group, and the rare exceptions frequently involve interminority cooperation or the presence of an incumbent. The election of nonwhites to high public office from majority-white constituencies is so rare that they can be recounted anecdotally (Bowler and Segura, forthcoming).
Black-Brown Relations and the 2008 Campaign
Despite their similar policy preferences and ideological beliefs (Hero and Preuhs 2009; Kaufmann 2003b), African Americans and Latinos are frequently in comparable socioeconomic situations, and thus frequently in competition for scare economic and political resources (Gay 2006; McClain et al. 2007; Morris 2000; Vaca 2004). As Kaufmann writes, such "competition over jobs, educational resources, housing, and political power often places blacks and Latinos in conflict against one another, and this conflict can act as a powerful barrier to political alliance" (2003b, 199). Extant research has shown the limited extent or transitory nature of "rainbow coalitions"--the electoral alignment of Latinos and blacks in support of the same candidate or policy--and emphasizes the frequent divergence between Latino and black voters in racially salient elections (Kaufmann 2003a, 2004; Morris 2000).
Other research approaches the question of Latino--black coalition potential by examining individual Latino or black attitudes about the other groups' members, targeted policies, or degree of political and economic competition (Barreto, Gonzalez, and Sanchez, forthcoming; Barreto, Sanchez, and Morin, forthcoming; Gay 2006; Kaufmann 2007; Lopez and Pantoja 2004; McClain 2008; McClain et al. 2008). This line of research has generated ample evidence of Latino hostility toward blacks and black-targeted policies (cf. McClain et al. 2008; McClain et al. 2007; Vaca 2004), although several scholars contend that such racial animus and perceptions of competition among Latinos is limited to the U.S. South (Barreto and Sanchez 2009; Barreto, Sanchez, and Morin, forthcoming). Nonetheless, extant research shows that some Latinos do hold negative racial stereotypes about blacks, perceive greater commonality with whites than with blacks (McClain et al. 2008), continue to see black political gains in zero-sum terms (Kaufmann 2007), and frequently support opportunity-enhancing policies at lower rates than blacks, all else being equal (Lopez and Pantoja 2004). As a consequence, a common trope has emerged about Republicans' ability to peel Latino voters away from the Democratic Party in spite of broad Latino agreement with Democratic policy priorities (Nicholson and Segura 2005; Uhlaner and Garcia 2005).
However, research that explores mass attitudes among Latinos without accounting for political context and the important role of political leadership can lead to erroneous conclusions about the willingness of Latinos to support black candidates, or vice versa (Barreto 2007; Segura and Rodrigues 2006). In fact, minority leaders and activist groups can bridge the frequent attitudinal and electoral divide among Latinos and blacks under some conditions (Hero and Preuhs 2009). Other aspects of political competition, such as party structure and the primary process, can also generate outcomes that, by appearances, reflect Latino-black differences that are not, in fact, substantial. For example, in light of Hillary...