Do poor whites vote for the wrong party more than other voters? Scholars and popular media have highlighted the ways poor whites may cast ballots against their beliefs and policy interests (Bartels 2006; Brooks 2001; Frank 2005; Hutchinson 2012). The supposed outcome of this irrational behavior is a group of voters who systemically vote against their own interests. This question is of particular concern because there is not yet an accepted answer in both academic and media accounts. The dominant claim in the media each presidential election season, however, is that poor whites are being duped into voting for Republican presidential candidates (see, e.g., Frank 2005; Hutchinson 2012).
Empirical tests of these claims show that poor whites still vote for Democrats more than Republicans (Bartels 2006) and that there are state-level differences that need to be taken into account when considering these voters' behavior (Gelman 2009; Gelman et al. 2007). Essentially, the academic literature suggests that poor white voters are not voting for Republicans to the extent that popular media figures suggest. Discussion about poor whites and their role in presidential elections has been intense in both political science and popular media since Nixon's Southern Strategy, despite the fact that the migration from the Democratic Party to the Republican Party began much before that (Shafer and Johnston 2009). (1) For example, focus on this group of voters has given rise to talk of NASCAR dads who would benefit from the economic policies of Democrats, but they are erroneously convinced to vote for the Republican Party (Elder and Greene 2007). There are many reasons suggested for why poor whites vote against their self-interest to support Republican candidates, including race baiting, fear mongering, or other values-based manipulations in which Republican candidates supposedly engage (Frank 2005). The fundamental interest in this phenomenon stems from the general perception that the Republican Party now needs--and indeed uses--these poor white voters to win elections (Peterson and Chinni 2014).
Popular accounts suggest that poor white voters are voting en masse for candidates who do not reflect their policy preferences, while academic accounts focus on the empirical data, which suggest that poor whites vote Democratic just as they always have. Missing from both journalists' accounts and political scientists' accounts is a measure matching the political preferences of voters with the candidates for whom they vote. We account for this gap in the debate by using Lau and Redlawsk's (1997, 2006) measure of voting correctly from 1972 to 2008 American National Election Survey (ANES) data as the dependent variable. Correct voting in presidential elections is a better theoretical tool because the claims by both camps in the debate rest on notions of rational behavior. This is the crucial missing piece of the existing literature, as that research does not measure whether these poor whites are voting against their stated policy views but simply shows that their vote totals are higher for Democratic candidates. With this in mind, we test the definitive question in the debate: are poor whites less likely to vote correctly when they vote for Republican presidential candidates?
We add to this debate by showing how the interaction of economics and race--namely, being a poor white voter--is an insignificant predictor of voting correctly for Republican presidential candidates between 1972 and 2008 in presidential elections. This finding is important because it is the first time this debate has been adjudicated using this direct measure of voter preference and choice, but also because we highlight the availability of bias producing this red herring and how it does not hold up when other factors are taken into account. We conclude by suggesting ways to overcome this debate by turning to more substantively interesting aspects of voting behavior by this demographic group.
Voting Behavior Among Poor Whites
There are two discussions about the impact of race and economics on voting behavior in American presidential elections. The first side is exemplified by Thomas Frank's (2005) book, What's the Matter with Kansas. Frank claims conservatives are winning presidential elections in the United States by essentially duping lower-class whites into voting for Republicans. According to Frank, poor whites are convinced to vote for Republicans because of socially conservative policies focusing on abortion and same-sex marriage. Moreover, Frank maintains that these voters should be voting for Democrats because the Democratic Party represents lower-class economic interests. This trend, according to Frank, is detrimental because the vote choices of poor whites produce conditions contrary to that of their stated opinions. David Brooks (2001) also claims that voters' behavior and their economic preferences are divorced from one another, but--rather than just poor whites--Brooks also cites upper-middle-class whites in the Northeast. Brooks claims that these voters should be voting for Republicans as their platform supports policies that would likely benefit these voters more than the Democratic policies. The key claim for both authors is that voters--particularly poor whites--are voting incorrectly and thus skewing the electoral outcome from its rational result. Political scientists, however, have not left these claims unattended.
Analyses by Bartels (2006) and Gelman et al. (2007) suggest that claims by Brooks (2001) and Frank (2005) do not stand up to empirical scrutiny. Bartels (2006) subjects Frank's arguments--which are anecdotal accounts--to empirical examinations using ANES data concluding that Frank overestimates the effects of social issues in voters' minds. Contrary to a wholesale shift to social issues as the basis of voting as suggested by Frank, Bartels shows that only the wealthiest Americans attach more weight to social issues and use them as primary voting decision mechanisms. Gelman et al. (2007) find similar results using multilevel modeling to account for the differences between states. For low-income voters, economic issues remain a top priority (Bartels 2006; Gelman et al. 2007). Furthermore, Bartels confirms that Democratic presidential votes declined as a whole in the past fifty years; however, the decline is confined to the South. Even accounting for the nested structure of presidential voting--state, county, and individual--research suggests that individual voters are no different today than they have been in the last fifty years as poorer voters still support Democrats at higher rates that wealthier voters (Gelman et al. 2007). Ansolabehere, Rodden, and Snyder (2006) echo Bartels's findings. They use General Social Survey data to estimate the effect of economic and social issues on vote choice. They claim, like Bartels, that Frank's (2005) argument is far too simplistic to account for the shift of poor whites into the Republican Party. Also, Ansolabehere, Rodden, and Snyder (2006) confirm Bartels's claims that the voting shift is primarily taking place in the southeastern United States.
Another reason the South may have seen the shift in voting patterns as described by Frank and Brooks is the proximity in which poor and wealthier voters live in this region. Galbraith and Hale (2008) show that states with lower levels of class segregation have higher levels of polarization in their voting. In states with higher levels of class segregation (i.e., Connecticut), there is more uniform support for the Democratic Party. Gelman et al. (2007) and Gelman (2009) suggest similar findings as they claim that the rich are the ones who are different by state, not the poor. Yet, the focus of the debate continues to revolve around the poor and, specifically, in the South. There are numerous issue-related mechanisms put forward by scholars for this shift toward Republicans for poor whites, which might also explain decreases in correct voting. There is a large literature debating this issue with some claiming the cause is racialized politics (Carmines and Stimson 1989), wedge issues generally (Hillygus and Shields 2014), or religious affiliations and social policies (Wald and Calhoun-Brown 2011).
Given the forefront nature of the South in this debate (Bartels 2006; Gelman et al. 2007; Shafer and Johnston 2009), it is possible that racial politics are responsible for this shift. Carmines and Stimson's (1989) seminal work on race issue politics demonstrates that enterprising Republican politicians are able to use racial issues to make significant inroads among Southern whites, which finally culminated in a Republican majority for the House after 1994 midterm elections. Relying on race, Republicans are able to turn voters away from salient economic issues and use symbolic politics to encourage a different, less instrumentally rational, voting calculus. Furthermore, Edsall and Edsall (1992) and Gilens (2009) argue that race has been the focal point in almost all policy debates since the 1960s. This frame, according to Gilens, shifts the focus of the debate away from policy and toward the issue of race. The shift to race priming in political discourse is one reason why poorer white voters might vote against Democrats, but the thrust of the debate rests squarely with economics.
What is missing in these accounts of erroneous voting behavior among economically disadvantaged white voters is an objective measure of their preferences. Gelman et al...