REVIEWING LATINO VOTE CHOICE

Suggesting that some voter characteristics may motivate Latino vote choice more than party identification fits with the recent literature in Latino politics. Although Latinos have consistently voted for the Democratic candidate in presidential elections, they may be motivated to do so beyond simple partisan attachment (DeSipio Reference DeSipio1998; Rodolfo and Cortina, 2007). Zoltan L. Hajnal and Taeku Lee (Reference Hajnal and Lee2011) found that party identification is declining among Latino and Asian Americans as they are not being incorporated into the American party system. Rather, these voters are identifying as independents. Other research finds that U.S.-born and naturalized Latinos showed only slightly higher levels of non-partisanship than non-Hispanic Whites, and that non-citizen Latinos are about twice as likely as their naturalized or U.S.-born counterparts to identify as non-partisans (Sears et al. Reference Sears, Danbold and Zavala2016). It could then be the case that newer Latino immigrants take time to acclimate to the American political environment before affiliating with a political party. To this point, the literature suggests that among Latinos who have been in the United States for less than twenty years, most identified as non-partisans (Alvarez and García Bedolla, Reference Alvarez and Bedolla2003).

Turning to other long held predictors of vote choice, political ideology has been shown to be a strong predictor of Latino vote choice in elections where moral issues have been a focal point (Abrajano et al., Reference Abrajano, Michael Alvarez and Nagler2008; Alvarez and García Bedolla, Reference Alvarez and Bedolla2003), but Hajnal and Lee (Reference Hajnal and Lee2011) argue that the conventional conservative, moderate, and liberal labels may not apply to Latinos. Research on religion and Latino vote choice suggests that denomination and religious attendance are closely tied to ideology and partisanship, as Latino Protestants are more likely to identify as conservatives and hold more conservative views on social issues than Latinos of other religious denominations (Valenzuela Reference Valenzuela2014; Ellison et al., Reference Ellison, Echevarría and Smith2005). Other scholars find that, relative to Catholics and Latinos of other religious denominations, Latino Protestants and evangelicals are much more likely to identify as Republicans and support Republican candidates (Jones-Correa and Leal, Reference Jones-Correa and Leal2001; Kelly and Kelly, Reference Kelly and Kelly2005; Valenzuela Reference Valenzuela2014).

According to Alvarez and García Bedolla (Reference Alvarez and Bedolla2003), Latinos with lower levels of education are more likely to identify as Democrats, although others suggest that education is not associated with party identification and vote choice among Latinos (Uhlaner and Garcia, Reference Uhlaner, Garcia, Segura and Bowler2005). While income does not appear to be associated with party identification for Latinos (Alvarez and García Bedolla, Reference Alvarez and Bedolla2003; Uhlaner and Garcia, Reference Uhlaner, Garcia, Segura and Bowler2005), length of time in the United States is found to shape vote choice, as first- and second-generation Latinos are much less likely to support Republican candidates than third-generation Latinos (DeSipio and Uhlaner, 2007; Nuño Reference Nuño2007). Another consistent finding is that national origin is strongly associated with party identification and vote choice (Alvarez and García Bedolla, Reference Alvarez and Bedolla2003; Uhlaner and Garcia, Reference Uhlaner, Garcia, Segura and Bowler2005). Cubans are much more likely to identify as Republicans and support Republican candidates, while Mexicans, Puerto Ricans, and Central Americans are more likely to be Democrats. To this point, when predicting Republican versus Democratic partisanship among Latinos, research suggests the strongest predictor of identifying as a Republican is being Cuban (Alvarez and García Bedolla, Reference Alvarez and Bedolla2003).

For those Latinos with low party attachment, it stands to reason that they may look to candidates for cues (Wattenberg Reference Wattenberg1991). The research finds that co-ethnic cues are a strong predictor of Latino vote choice, with low information voters, or those voters who have little to no understanding of politics, being especially receptive to these cues (Barreto 2007, Reference Barreto2010; Nicholson et al. Reference Nicholson, Pantoja and Segura2006). For Latino voters, the presence of a co-ethnic on the ballot functions as a shortcut that helps them judge how well the candidate will represent the interests of their ethnic group (Nicholson et al., Reference Nicholson, Pantoja and Segura2006). There are limits to shared ethnicity, however; more educated Latinos are less likely to favor co-ethnic candidates solely because of their ethnicity (Manzano and Sanchez, Reference Manzano and Sanchez2010). Despite their shared ethnicity, Latino Republican candidates in California have not been able to win significant shares of the Latino vote, likely because they are not able to overcome the Republican Party’s past support of the anti-Latino Proposition 187(Michelson Reference Michelson2005). Latino candidates still make up a relatively small share of all candidates in the United States, so enterprising candidates from both parties have had to rely on cross-racial appeals to attract Latino voters.

Cross-racial appeals to Latinos often take the form of Spanish-language radio or television advertisements, which date as far back as Dwight Eisenhower’s campaign (Abrajano Reference Abrajano2010; Collingwood et al., Reference Collingwood, Barreto and Garcia-Rios2014). Jessica Lavariega Monforti and colleagues (Reference Lavariega Monforti, Michelson and Franco2013) found that Latino voters, including Republicans, in Texas generally favor Spanish speaking candidates at the polls. The literature suggests these cross-racial appeals may be effective among Latino voters as they demonstrate the candidate’s respect for the Latino community (Barreto and Collingwood, 2014). Candidates can also demonstrate respect and concern for the Latino community via policy-based outreach, as Obama was able do on immigration and the DREAM (Development, Relief, and Education for Alien Minors) Act in 2012 (Barreto and Collingwood, 2014; Collingwood et al., Reference Collingwood, Barreto and Garcia-Rios2014). The literature on Latino vote choice suggests that racial and ethnic variables may at times trump traditional predictors of vote choice, but little has been written in political science on how denial of racism, and racism in general, among Latinos affects vote choice. In other words, do appeals traditionally designed for racially resentful Whites (e.g., Trump’s rhetoric) also work with a subset of Latino voters who might hold similar beliefs?

EXPLORING LATINO RACISM AND DENIAL OF RACISM

Much of the work on racism and Latinos has explored the effects of White racism on Latinos. This is understandable given that for many Latinos, their experience in the United States has been shaped by racism in one form or another (Brewster et al., Reference Brewster, Foster and Stephenson2016; Florido Reference Florido2017). However, leaving out measures of racism among Latinos when predicting vote choice, particularly Republican vote choice, may limit our understanding of why some Latinos vote Republican, especially when recent work suggests racism was a strong predictor of support for Trump among Whites (Miller Reference Miller2017; Schaffner et al., Reference Schaffner, MacWilliams and Nteta2018).

The literature on Latino racism has largely focused on racial resentment, which has been defined as a conjunction of anti-Black feelings and American moral traditionalism (Carmines Reference Carmines, Sniderman and Easter2011; Kinder and Sears, Reference Kinder and Sears1981). One major theme of this literature is that it questions the emerging Democratic majority posited by John B. Judis and Ruy Teixeira (2004). Judis and Teixeira (Reference Judis and Teixeira2004) argued that given current demographic and political trends, the U.S. political arena would eventually be ruled by a multicultural, multiethnic Democratic coalition. However, this emerging Democratic majority depends in part on Latinos not becoming White as Italian and Irish Americans did (Murguia and Forman, Reference Murguia, Forman, Doane and Bonilla-Silva2003). The work of Edward Murguia and Tyrone Forman (2003) recognizes the fact that Italian and Irish Americans may have had an easier path to Whiteness than Latinos given their lighter complexions and European backgrounds, but they did not rule out the possibility that some Latinos may want to become White to escape persecution and racism. Just as Italian and Irish Americans did, the fastest path to Whiteness for Latinos may be to minimize the social space between themselves and Whites and adopt White views of race and racism (Gans Reference Gans2012; Jones-Correa Reference Jones-Correa1998; Murguia and Forman, Reference Murguia, Forman, Doane and Bonilla-Silva2003; Warren and Twine, Reference Warren and Twine1997).

Of course, not every Latino will want to pursue Whiteness given the racism and treatment as second-class citizens many Latinos have experienced (López Reference López1997; Garcia and Sanchez, 2008). At the same time though, given the racial diversity that exists among Latinos, every Latino will not be subjected to the same set of racialized experiences (Kaufmann Reference Kaufmann2003; Sanchez and Masuoka, Reference Sanchez and Masuoka2010). Even when Latinos are subjected to racism, research suggests many will deny or explain the racism away (Florido Reference Florido2017; O’Brien Reference O’Brien2008; Rojas-Sosa Reference Rojas-Sosa2016). In her study of denial of racism among Latino college students, Deyanira Rojas-Sosa (Reference Rojas-Sosa2016) found that Latino students are often hesitant to recognize that they were the subject of discrimination or to qualify experiences as racist. Rojas-Sosa (Reference Rojas-Sosa2016) argued that Latinos, and other minority groups, may deny the existence of racism to not challenge the ideologies of majority groups (e.g., Whites) on issues of race and immigration. Beyond denying racism in their own lives, some Latinos will even deny racism on the part of their countrymen, especially when they are trying to defend their image of the United States as a place that is free of racism (Rojas-Sosa Reference Rojas-Sosa2016). To this point, Rojas-Sosa (Reference Rojas-Sosa2016) demonstrated that a common strategy used to explain away racism is that society has overcome racism as a social problem. This logic is similar to that of the colorblind ideology, which argues that race no longer has a key impact on the lives of Americans as the United States has transcended racism and become a post-racial society (Bonilla-Silva Reference Bonilla-Silva2017).

Other research also suggests many Latinos and Asians try to deny racism in their lives and treat racism as something of the past (O’Brien Reference O’Brien2008). However, when minorities espouse colorblind ideologies and deny racism, this does not necessarily mean that they are embracing these positions; rather, they may be acting instrumentally as complaining about racism could call attention to their otherness (Jones-Correa Reference Jones-Correa1998; O’Brien Reference O’Brien2008). For those Latinos who want to minimize the social space that exists between themselves and non-Hispanic Whites, Herbert J. Gans (Reference Gans2012) argues they can modify their behavior and lifestyle to resemble Whites socially and culturally. This could be adopting White views on racism and society, such as the colorblind ideology, or voting for White candidates who espouse these views, such as Donald Trump (Gans Reference Gans2012).

Research also suggests that for some Latinos, voting for Republican candidates and restrictionist policies is a conscious effort to emulate White behavior in order to achieve inclusion, validation, and protection (Basler Reference Basler2008). Based on interviews with Mexican Americans, Carleen Basler (Reference Basler2008) identifies California’s anti-immigrant and anti-Mexican Proposition 187 as a turning point, which gave naturalized Mexican Americans a chance to align themselves with non-Hispanic Whites against their fellow Mexicans and Latinos in order to obtain Whiteness and the social benefits that come with it. Many Mexican Americans in Basler’s (Reference Basler2008) interviews were afraid or tired of being profiled as undocumented immigrants by Whites, so they voted in favor of restrictionist policies to improve their own standing and signal their allegiance to Whites. Echoing Gans’ (Reference Gans2012) findings, some Mexican Americans told Basler (Reference Basler2008) that by voting Republican, they could distance themselves from Blacks, the reference category against which Whiteness is determined, and further ingratiate themselves to Whites.

THEORY

Trump is just the most recent in a long line of candidates to use dog-whistle politics to provoke racial animus (López Reference López2016). Dog whistles are appeals that may seem innocuous to most individuals while communicating a coded appeal to a specific subgroup. In the case of racial appeals, dog whistles can shield a candidate and their supporters from criticism, as we see when Trump reassures his supporters that he and they are not racist. When Trump was attacked by Romney for his comments about a Mexican American judge, Trump tweeted “Mitt Romney had his chance to beat a failed president but he choked like a dog. Now he calls me racist—but I am the least racist person there is” (Trump Reference Trump2016b). After reported comments about Haiti and African countries, Trump said, “No, no, I’m not a racist. I am the least racist person you have ever interviewed. That I can tell you” (Covucci Reference Covucci2018).

Trump’s rhetoric allows his White supporters to act on their racial biases while denying their own racism, and racism in America at large, in line with the colorblind ideology. For most Latinos, Trump’s rhetoric will signal a threat, which likely increases in-group solidarity and decreases support for Trump (LeVine and Campbell, Reference LeVine and Campbell1972). For Latinos who seek to become White though, Trump’s anti-Latino and anti-immigrant rhetoric signals a litmus test of one’s allegiance to Whites. I predict Latinos who deny racism, as a means to reduce the social space between themselves and Whites (Murguia and Forman, Reference Murguia, Forman, Doane and Bonilla-Silva2003; Warren and Twine, Reference Warren and Twine1997), will support Trump, just as Latinos who voted in support of California’s Proposition 187 did in 1994 (Basler Reference Basler2008). Thus, for Latinos, denial of racism serves as a measure of an individual’s desire to achieve Whiteness (Gans Reference Gans2012; O’Brien Reference O’Brien2008; Rojas-Sosa Reference Rojas-Sosa2016).

Could cognitive dissonance be at work when Latinos who purport to deny racism support race-baiting candidates like Trump? Based on the work of Gans (Reference Gans2012) and Basler (Reference Basler2008), I argue not, as denying racism while supporting race-baiting candidates is a strategy some Latinos will undertake to climb the racial hierarchy. Like Gans (Reference Gans2012), I argue that to fully embrace Whiteness, minorities will have to adopt commonly-held White viewpoints, such as the colorblind ideology, even in the face of racism, to signal their allegiance to Whites (Bonilla-Silva Reference Bonilla-Silva2017). I contend that in the presence of a candidate like Trump, who made explicitly anti-immigrant and anti-Mexican appeals, Latinos are presented with a choice: they can acknowledge that racism is still a pressing concern in the United States and reject the candidate who engages in race-baiting, or they can deny that racism exists and support the candidate who seeks to defend the racial hierarchy from which they hope to benefit. Beyond voting for party or ideology, some Latinos in 2016 saw Trump as their chance to signal their allegiance to Whites, as Basler (Reference Basler2008) demonstrated was the case with California’s Proposition 187. As one Mexican American told Basler, “voting for 187 was a ‘way to make sure the whites [at his job] knew he was with them’ so they wouldn’t think he was ‘some Mexican traitor who felt sorry for the wetbacks’” (2008, p. 148).

What then, is driving Latinos who deny racism to support a candidate like Trump or policies like Proposition 187? According to research, Latinos may be consciously manipulating their identities to fulfill their social psychological need for protection in the face of a potential threat (Smith Reference Smith1992; Basler Reference Basler2008). Should Whites accept Latinos who deny racism as White, a Latino’s need for validation and inclusion could also be fulfilled (Smith Reference Smith1992; Basler Reference Basler2008). We should not expect most Latinos to deny racism in an attempt to embrace Whiteness, as realistic conflict theory argues that out-group threat and hostility often lead to higher levels of in-group identification (Levine and Campbell, Reference LeVine and Campbell1972). However, among members of stigmatized groups, it is possible that some may attempt to distance themselves from their in-group when facing a threat (Howard Reference Howard2000). This is in line with self-affirmation theory, which argues that individuals are motivated to maintain self-integrity (Sherman and Cohen, Reference Sherman, Cohen and Zanna2006). Indeed, Sherman and Cohen (Reference Sherman, Cohen and Zanna2006) argue some individuals will protect their self-integrity through the affirmation of alternative sources of identity, as the literature shows minorities who adopt Whiteness have done (Gans Reference Gans2012; Basler Reference Basler2008).

Thus, I argue that in the presence of a candidate like Trump, who focused his attacks on Latinos, denial of racism should be a stronger predictor of support for Trump among Latinos than Whites. We can expect this because due to his anti-Latino rhetoric, Trump should be starting at a lower point of support among Latinos relative to Whites, all else held equal. However, Latinos high in denial of racism, which I treat as a proxy for one’s desire for Whiteness, should see Trump as a chance to signal their allegiance to Whites and should be highly motivated to support Trump, as Basler (Reference Basler2008) argues some Latinos did when California’s Proposition 187 was on the ballot.

Following this logic, we should expect Latinos who deny racism to be most supportive of Donald Trump. Latinos who acknowledge racism should then be the least supportive of Donald Trump. While I use Latino support for Trump as my test case, I also argue that denial of racism should be a strong predictor of Latino support for Mitt Romney, although to a lesser degree than Trump given the tenor of Trump’s rhetoric relative to Romney’s. To this point, denial of racism should not be a statistically significant predictor of support for non-race baiting candidates, whether they be Democrats, Republicans, or independents. Finally, although I argue denial of racism should be a strong predictor of non-Hispanic White support for Trump, I argue that denial of racism will be a stronger predictor of Latino support for Trump. Based on my theory, I will test the following hypotheses in this paper:

1. H₁: Denial of racism will be positively associated with support for Donald Trump among Latinos.

2. H₂: Denial of racism will be positively associated with support for Mitt Romney among Latinos.

3. H₃: Denial of racism will be a stronger predictor of support for Donald Trump among Latinos than non-Hispanic Whites.

DATA AND METHODS

To test whether denial of racism is a predictor of Latino support for Trump, I use the 2016 CCES, a nationally representative survey of American adults. The CCES includes 46,289 non-Hispanic White respondents and 7495 Latino respondents. The data were weighted using post-election sample weights, and weights were included in each analysis. The primary dependent variable is the two-party vote for president. This variable includes only those who reported voting in the presidential election and voted for either the Democratic or Republican candidate. Individuals who voted for a candidate other than Clinton or Trump in 2016, or Obama or Romney in 2012, are not included in the main analyses. For the sake of comparison on the key independent variable, the 2012 vote models use a measure of presidential vote choice taken during the 2016 CCES as no measure of denial of racism is available on the 2012 CCES.

The key independent variable measures denial of racism. Denial of racism is measured on a scale made up of three items used by Brian F. Schaffner and colleagues (2018) to measure acknowledgement and empathy with racism. The scale draws largely from the colorblind racial attitudes scale, which measures the belief that race does not affect one’s life chances (Neville et al., Reference Neville, Lilly, Duran, Lee and Browne2000). Colorblind racial attitudes are problematic as race still plays a large role in determining life outcomes (Neville et al., Reference Neville, Lilly, Duran, Lee and Browne2000; Schaffner et al., Reference Schaffner, MacWilliams and Nteta2018). The two items used to measure colorblind racial attitudes are:

1. 1. White people in the U.S. have certain advantages because of the color of their skin.

2. 2. Racial problems in the U.S. are rare, isolated situations.

But a scale of racism that measures only one’s awareness of racial inequality is lacking (DeSante and Smith, Reference DeSante and Smith2016; Schaffner et al., Reference Schaffner, MacWilliams and Nteta2018). Thus, a third item, drawn from the Psychosocial Costs of Racism to Whites battery (Poteat and Spanierman, Reference Poteat and Spanierman2008; Spanierman and Heppner, Reference Spanierman and Heppner2004; Spanierman et al., Reference Spanierman, Paul Poteat, Beer and Armstrong2006) is used to measure the degree of empathy individuals feel about racial inequalities:

1. 3. I am angry that racism exists.

On each of these three items, respondents used a five-point scale to measure how much they agreed or disagreed with each statement (strongly agree, somewhat agree, neither agree nor disagree, somewhat disagree, strongly disagree). Responses to these items were scaled and standardized to create a measure of denial of racism with a mean of 0 and standard deviation of 1. Negative scores on the denial of racism scale signify that the respondent is more acknowledging of racism, while positive scores signify the respondent is more denying of racism. The denial of racism scale has an alpha of 0.67 for non-Hispanic Whites, and .63 for Latinos. Figure 1 shows the distribution of responses of denial of racism, with all Latinos and Latino Republicans on the first row, and all Whites and White Republicans on the second row. Table 1 shows the average values of denial of racism for Latinos, Whites, Latino Republicans, and White Republicans. As one might expect, Latinos have a lower mean denial of racism than Whites, and Latino Republicans have a lower mean denial of racism than White Republicans. Latinos at large are the only observed group to have a negative mean denial of racism, indicating that most Latinos fall below the mean denial of racism.

Fig. 1. Distribution of Latinos and Whites on denial of racism scale.

Table 1. Average values of denial of racism.

Values are standardized. Standard errors in parentheses.

Figure 1 shows that 40.23% of Latinos are above the mean score on the denial of racism scale, and only 11.58% of Latinos score above one. While only a relatively small number of Latinos will have their vote choice for Trump increased by a significant amount, it is important to examine the power of denial of racism relative to party identification, which the literature tells us should be a strong predictor of support for Trump (Campbell et al., Reference Campbell, Converse, Miller and Stokes1960; Lewis-Beck et al., Reference Lewis-Beck, Jacoby, Norpoth and Weisberg2008). In the 2016 CCES, only 17.09% of Latinos identify as Republicans. Per the post-estimation simulations in Table 4, a Latino who identifies as an independent has a 31.9% chance of voting for Trump. Identifying as a Republican and holding all other variables at their observed values increases the likelihood a Latino will vote for Trump from 31.9% to 46.6%. If we assume that only 46.6% of Latino Republicans are voting for Trump, that doesn’t account for even one-third of the 30% of Latinos who voted for Trump. Something else beyond Republican party identification must be accounting for Trump’s share of the Latino vote, and I argue that denial of racism is an important part of that explanation.

For those Latinos who score above one on the denial of racism scale, their probability of voting for Trump increases by a minimum of 14.1%; in other words, for these Latinos, their denial of racism has nearly as much predictive power when predicting vote choice for Trump as Republican party identification, as it increases their likelihood of voting for Trump to at least 46%. For those Latinos who score above 1.5 on the denial of racism scale, their probability of voting for Trump increases by a minimum of 22.6%, at which point they become more likely to vote for Trump than Clinton. As a point of comparison, even Republican Latinos are still more likely to vote for Clinton than Trump. To be clear, relatively few Latinos will have their denial of racism increase their likelihood of voting for Trump by a large amount, but per Table 4, only slightly more Latinos will have their vote choice influenced by Republican party identification. So, while denial of racism likely only influenced the vote choice of a relatively small number of Latinos, it is important to remember that only 30% of all Latinos who voted did so for Trump.

The limits of the data prevent a full exploration of what predicts Latino denial of racism, but the available data allow for exploratory analyses. In these analyses, I control for multiple variables given that what predicts denial of racism is an unexplored topic. I control for age, as older adults have been shown to express more prejudice than younger adults (Stewart et al., Reference Stewart, von Hippel and Radvansky2009), as well as income, as individuals with higher incomes have been found to have lower levels of prejudice (Carvacho et al., Reference Carvacho, Zick, Haye, González, Manzi, Kocik and Bertl2013). I control for education as well because, while individuals with higher levels of education have been shown to report lower levels of explicit racism, they are more likely to score higher on implicit measures of racism (Kuppens and Spears, Reference Kuppens and Spears2014). While few differences in racial attitudes have been found among White men and women (Hughes and Tuch, Reference Hughes and Tuch2003), I control for gender as there may be differences in Latino denial of racism. I also control for national origin group, using Cuban as the reference category, as the various racial hierarchies that exist across Latin America may influence Latino denial of racism differently for each group in the United States (Telles et al., Reference Telles, Flores and Urrea-Giraldo2015). There may also be differences among national origin groups when predicting support for Trump, as research suggests non-Mexican heritage Latinos supported Trump at slightly higher levels than Mexican heritage Latinos (Garcia-Rios et al., Reference Garcia-Rios, Francisco and Bryan2018).

There may be an immigration gap in how Latinos perceive discrimination, as immigrant Latinos are much less likely to report experiencing discrimination (Florido Reference Florido2017). Neighborhood demographics could also influence one’s denial of racism, as Gans (Reference Gans2012) suggests that minorities who live in mostly White neighborhoods may be pushed to deny racism to not draw attention to their otherness. Research has shown ideological conservatives are more likely than liberals to display racial prejudice (Sniderman et al., Reference Sniderman, Phillip E. Tetlock and Kendrick1991), while more recent work suggests a correlation between anti-immigrant sentiment and racial resentment (Ingraham 2018). I control for identifying as Born Again, identifying as Protestant, and religious attendance, as a correlation between religious fundamentalism, and prejudice has been demonstrated (Allport and Ross, Reference Allport and Ross1967; Altemeyer and Hunsberger, Reference Altemeyer and Hunsberger1992; Hall et al., 2010; Hunsberger and Jackson, Reference Hunsberger and Jackson2005). Finally, I control for Republican and Democratic party identification with independents as the reference category given the Republican Party’s history of racial appeals, coded and explicit, dating back to Barry Goldwater (Carr Reference Carr1997; Kousser Reference Kousser2000; López Reference López2015).

Table 2 contains four OLS models predicting denial of racism among Latinos. Model 1, a demographic model, excludes measures of party identification as, although Donald P. Green et al. (2003) argue party identification is a stable construct, I cannot definitively say that partisan identification predates denial of racism in Latinos. In Model 1, being male and older are positively associated with denial of racism. Although Michael Hughes and Steven A. Tuch (2003) found no significant differences between White males and females on the racial resentment scale, it seems that Latino males are more likely to have higher levels of denial of racism than Latino females. While being older is associated with higher levels of denial of racism among Latinos, Model 1 suggests that no relationship exists between a Latino’s immigration generation and their level of denial of racism. This finding runs counter to Adrian Florido’s (Reference Florido2017) work suggesting that immigrant Latinos were more likely than U.S.-born Latinos to deny racism, but there is no clear consensus suggesting that foreign-born Latinos deny racism more than U.S.-born Latinos, as Rojas-Sosa (Reference Rojas-Sosa2016) demonstrates U.S.-born Latinos deny racism as well. Latinos with middle and high incomes are not significantly different than those with low incomes when predicting denial of racism, but not reporting one’s income is positively associated with higher levels of denial of racism. Higher levels of education are negatively associated with denial of racism, which runs counter to Tom Kuppens and Russell Sears’ (2014) work, and being Mexican, Puerto Rican, and Other Hispanic American are associated with lower levels of denial of racism relative to being Cuban American.

Table 2. OLS regression model predicting denial of racism among Latinos.

Robust standard errors in parentheses

***p<0.01, **p<0.05, *p<0.1

In Model 2, which controls for ideology and anti-immigrant attitudes, being older is no longer a statistically significant predictor of denial of racism among Latinos, but all the other significant variables from Model 1 retain their significance and previous positive or negative associations. Identifying as a conservative is positively associated with denial of racism among Latinos, while identifying as a liberal is negatively associated with denial of racism. Although the relationship between denial of racism and ideology has not been previously studied, these results suggest denial of racism operates similarly to other forms of prejudice studied in the literature (Sniderman et al., Reference Sniderman, Phillip E. Tetlock and Kendrick1991). Anti-immigrant sentiment is also positively associated with denial of racism, which aligns with findings demonstrating a link between racial resentment and anti-immigrant attitudes (Ingraham Reference Ingraham2018). In Model 3, which controls for identifying as Born Again, identifying as Protestant, and religious attendance, all the significant variables from Model 2 retain and their significance and associations. None of the religious controls are statistically significant predictors of denial of racism among Latinos, which runs counter to the literature’s findings that religious fundamentalism is associated with increased prejudice (Hall et al., 2010; Hunsberger and Jackson, Reference Hunsberger and Jackson2005).

In Model 4, which controls for party identification, results are substantively similar to Model 3, although identifying as a Hispanic other than Puerto Rican or Mexican, relative to identifying as a Cuban, is no longer a statistically significant predictor of denial of racism. These results suggest that relative to other Hispanic Americans, those of Mexican and Puerto Rican descent face different, perhaps more racialized, experiences that lead to lower levels of denial of racism. Relative to independent identifiers, being a Republican is positively associated with denial of racism, while being a Democrat is negatively associated with denial of racism. This finding is not surprising given the Republican Party’s history of racial appeals (Carr Reference Carr1997; Kousser Reference Kousser2000; López Reference López2015). I reiterate these models are exploratory analyses, as they lack measures of American identity, Latino linked fate, and skin color, which could potentially play a large role in predicting Latino denial of racism. Future research should investigate the role these variables play in predicting Latino denial of racism.

For the article’s main analyses, I will use logistic regressions predicting support for the Republican candidate for president in a given election. To test alternative hypotheses for why Latinos might support Trump or Romney, I will control for numerous rival explanations based on the literature. Beyond partisanship and ideology, which the literature suggests should be strong predictors of vote choice (Abrajano et al., Reference Abrajano, Michael Alvarez and Nagler2008; Alvarez and García Bedolla, Reference Alvarez and Bedolla2003; Campbell et al., Reference Campbell, Converse, Miller and Stokes1960; Lewis-Beck et al., Reference Lewis-Beck, Jacoby, Norpoth and Weisberg2008), I control for income, which the literature argues is not influential for Latinos (Alvarez and García Bedolla, Reference Alvarez and Bedolla2003; Uhlaner and Garcia, Reference Uhlaner, Garcia, Segura and Bowler2005). I control for education as there is no clear consensus on the importance of education to Latino vote choice (Alvarez and García Bedolla, Reference Alvarez and Bedolla2003; Uhlaner and Garcia, Reference Uhlaner, Garcia, Segura and Bowler2005). As Gans (Reference Gans2012) suggests that a Latino’s local context may influence their desire to support a Republican candidate, I control for region using the four Census regions with the west as the reference category, as well as ZIP Code level demographics based on the 2014 American Community Survey.

As there is a gender gap in partisan affiliation among Latinos, with Latinas being slightly more likely to identify as Democrats than Latinos (Pew Research Center 2018), I control for gender using those who identify as female as the reference category. I control for age and generation as research suggests older, immigrant Latinos are less likely to vote Republican than younger and third generation Latinos (Nuño Reference Nuño2007; DeSipio and Uhlaner, Reference DeSipio and Uhlaner2007). Controlling for national origin, with Cuban as the reference category, is also important as Cubans are more likely than other Latinos to vote Republican (Alvarez and García Bedolla, Reference Alvarez and Bedolla2003; Uhlaner and Garcia, Reference Uhlaner, Garcia, Segura and Bowler2005). As the literature suggests Protestants and Born-Again Christians are more likely to vote Republican than other denominations, I control for these identifiers. I also control for religious attendance, as recent surveys suggest those who attended religious services weekly or more, regardless of denomination, were more likely to vote for Trump than not (Smith and Martínez, Reference Smith and Martínez2016).

I control for political knowledge, as research suggests that for Latinos with lower political knowledge, partisan preferences are particularly important in determining vote choice (Nicholson et al., Reference Nicholson, Pantoja and Segura2006). Finally, I control for anti-immigrant sentiment, as Basler (Reference Basler2008) shows that for some Latinos, being anti-immigrant is a litmus test for inclusion into the Republican Party. Beyond these individual controls and alternative explanations, I also test for several interactions with denial of racism. I interact denial of racism and party identification, with identifying an independent serving as the reference category, as being Republican might amplify the effect of denial of racism on one’s support for Trump, while being a Democrat may have the opposite effect. Similarly, I interact denial of racism and anti-immigrant sentiment, as among those who deny racism, those with anti-immigrant sentiments might be even more likely to support Trump given his rhetoric, while those who deny racism but are not anti-immigrant may be less likely to support Trump. Lastly, I interact denial of racism and my three religious measures, as Robert P. Jones (Reference Jones2017) finds that relative to minority Protestants and Evangelicals, White Protestants and Evangelicals perceived much less discrimination against immigrants and Blacks. When taking this finding into account with the literature linking religious fundamentalism and prejudice (Allport and Ross, Reference Allport and Ross1967; Altemeyer and Hunsberger, Reference Altemeyer and Hunsberger1992), it may be the case that religious fundamentalism and denial of racism interact to increase support for Trump beyond only being a religious fundamentalist or racism denier.

Coding for these variables is available in the Appendix. Supplementary models predicting Latino support for various candidates in the 2016 Republican primary use the same control variables.

SUPPLEMENTARY MATERIAL

To view supplementary material for this article, please visit https://doi.org/10.1017/S1742058X19000328