Embracing the Status Hierarchy: How Immigration Attitudes, Prejudice, and Sexism Shaped Non-White Support for Trump

Abstract

It is well established that Donald Trump’s rhetoric and actions during his candidacy and presidency endorsed existing group-based social hierarchies, helping to boost his support among white Americans, especially men and those without a college degree. But how did these endorsements shape support for Trump among non-white Americans? Extant theories suggest that these actions should have pushed racial and ethnic minority voter support for the GOP candidate to its lowest observed levels in contemporary political history. Yet Trump outperformed these expectations in 2016 and in 2020 among Black, Latino, and Asian American voters. We propose the same embrace of social hierarchies that motivated white support for Trump also motivated the political preferences and behaviors of a significant number of non-white Americans. Using several national large-N surveys conducted between 2011 and 2021 with large samples of Black, Latino, and Asian Americans, we explore how support for existing status hierarchies—both gender and racial—engendered support for Trump across racial and ethnic groups and discuss implications for the future of electoral politics in a rapidly diversifying United States.

From the announcement of his candidacy in 2015 to his final day in office as president, Donald Trump openly embraced a preservation of America’s traditional status hierarchies of race and gender. From his frequent denigration of immigrants, defense of white supremacists, explicit racial rhetoric, and critiques of feminism, to his “Make America Great Again” slogan, Trump repeatedly signaled to Americans that he aimed to restore traditional status hierarchies in America with white Americans, men in particular, cementing their position on top.

While it’s no surprise that these pronouncements appealed to many white Americans (Tesler 2016), pundits were stunned to see Trump capture significant support from Black, Latino, and Asian voters in 2016 and even greater support in 2020, capturing an estimated 10% of the Black vote, 37% of the Latino vote, and 33% of the Asian American vote in his bid for re-election. ¹

While researchers, analysts, and pundits have acknowledged the political heterogeneity within non-white communities, a majority of extant theories of Black, Latino, and Asian American voting behavior suggest that Trump’s rhetoric and action should have mobilized record non-white support for Democrats. What explains support for Trump among non-white voters? We argue that while scholarship examining the effects of marginalization, discrimination, and xenophobia on non-white voting behavior is invaluable in understanding contemporary politics and the political incorporation, socialization, and behaviors of Latino, Black, and Asian voters, this sole focus on backlash ignores the fact that non-negligible numbers of these voters may not be offended, angered, or turned off by this rhetoric. ² Indeed, we argue that many of these voters support traditional status hierarchies even if they marginalize some members of their group (Jost, Banaji, and Nosek 2004); they oppose expansive immigration policy, have warm feelings toward white Americans but dislike members of other racial and ethnic minority groups, and eschew efforts to narrow gender gaps, preferring maintenance of traditional gender roles instead. These voters may not only be un-alarmed by Donald Trump’s rhetoric and actions; they may actively embrace them. If so, these attitudes should be predictive of support for Trump not just among white Americans but among Black, Latino, and Asian Americans as well.

Using several large-N surveys with sufficiently large and carefully-constructed samples of non-white Americans fielded between 2011 and 2021, we show that substantial numbers of non-white Americans support various facets of traditional status hierarchies and that these attitudes are strongly predictive of support for Donald Trump. Our findings suggest that the Republican Party’s embrace of Trumpism, particularly the overt embrace of traditional American status hierarchies, may not hurt its chances as much with non-white voters as existing theories of non-white voting behavior may suggest. Our findings are replicated across several datasets, each with their own complementary strengths, including large-N repeated cross-sectional and panel survey datasets.

This study makes several contributions to the literature on public opinion and voting behavior. First, we expand upon theories of white political behavior to show how support for traditional status hierarchies is not unique to white Americans; Black, Latino, and Asian Americans not only harbor these views but the views are consequential—strongly predictive of support for reactionary right candidates like Donald Trump. Second, while most studies of voting examine a single racial or ethnic group individually, we use survey samples that are sufficiently large to allow us to comparatively test our theory on all of the largest and most politically salient racial and ethnic groups in American politics. Third, our unique combination of datasets—high frequency cross-sectional surveys, culturally-sensitive multilingual surveys with over-samples of Black, Latino, and Asian respondents, and longitudinal panel data—allow us to robustly and precisely model non-white support for Trump. Finally, rather than focus on a single psychological construct and voting, like modern sexism and its relationship with support for Trump, we broaden the scope of our inquiry to better understand how various related predispositions predict support for candidates who openly embrace a preservation of America’s traditional status hierarchies of gender and race. Our findings help us understand and contextualize Donald Trump’s political success and suggest that the GOP’s recent strategic shift from dog-whistle to bullhorn support for traditional status hierarchies may not prevent the Republican Party from winning nation-wide elections even as the nation’s polity continues to diversify.

Donald Trump’s Candidacy and Presidency

Donald Trump’s candidacy and presidency were singular in the sheer volume of rhetoric and policy explicitly aimed at targeting and devaluing women and racial, ethnic, and religious out-groups. Trump infamously began his 2016 presidential campaign by stating “when Mexico sends its people; they’re not sending their best. They’re not sending you … They’re sending people that have lots of problems …. They’re bringing drugs. They’re bringing crime. They’re rapists.” ³ Anti-immigrant and anti-Latino rhetoric continued to be featured prominently in Trump’s speeches, remarks, and Tweets. As president, Trump frequently used his podium to espouse rhetoric that denigrated and dehumanized migrants, particularly those from Mexico and Central America. Days prior to the 2018 midterm elections, in Trump’s White House remarks he declared that “large, well-organized caravans of migrants are marching towards our southern border. Some people call it an ‘invasion.’ It’s like an invasion. They have violently overrun the Mexican border”, ⁴ tapping into well-worn migration tropes (Chavez 2013). Trump’s xenophobic rhetoric also targeted Asians during the COVID-19 pandemic, referring often to the virus as the “China Virus” or “Chinese Virus.” ⁵

Policies that targeted immigrants were also a key component of Trump’s domestic agenda. The construction of a wall spanning the entire length of the U.S.–Mexico border, nearly 2,000 miles, became a central policy promise. ⁶ The Trump administration enacted a litany of other policies that limited immigration and oversaw efforts to ramp up Immigration and Customs Enforcement (ICE) deportations and prosecutions. Trump’s “zero-tolerance” immigration policies encouraged the separation of thousands of families resulting in children being placed without their parents in detention centers that became dangerously overcrowded and fraught with abuse and neglect. ⁷ The Trump administration also famously enacted the “Muslim Ban,” an executive order that, among other things, banned travel into the United States from seven predominantly Muslim nations for 90 days and denied entry for all Syrian refugees. ⁸

Trump’s track record on immigrants and immigration echoes a long history of racist business practices and policies. In 1972 the Human Rights Division of New York City discovered that Trump Management was refusing to lease their apartments to Black tenants. ⁹ In 1989 Donald Trump commissioned full-page newspaper advertisements in which he demanded the state adopt the death penalty, a move inspired by the trial of the Central Park Five, five Black and Latino teenagers who were falsely accused of raping a woman in Central Park. ¹⁰ More famously, in 2011, Trump began to publicly question the legitimacy of President Barack Obama’s citizenship and became a loud proponent of the “birther” conspiracy, a clear racial dog whistle. ¹¹ As a candidate, Trump often encouraged violence against Black protesters and those representing Black Lives Matter ¹²—who Trump claimed “(are) looking for trouble” ¹³—and demonstrated a hesitancy or unwillingness to condemn racism, white supremacy, and antisemitism. ¹⁴

Finally, Trump has a long history of degrading women. In addition to serious allegations of sexual assault and rape from multiple women, ¹⁵ Trump has long bragged about entering the dressing room to see young Miss Universe contestants naked, ¹⁶ openly critiqued the appearances of women who he doesn’t like ¹⁷, was caught on an Access Hollywood tape boasting about his ability to sexually assault women and get away with it, and finally was found guilty of sexually assaulting E. Jean Carroll in a 2023 trial. ¹⁸

In short, Trump’s career, candidacy, and presidency were defined by punching down: targeting, maligning, and devaluing marginalized groups in society. And thus, at the heart of Trump’s pitch to voters as a candidate was a question of American identity. Who belongs in the United States? Which groups are deserving of representation and resources? Who should hold power? While identity has always been central to politics and political campaigns in the United States, appeals to racial, national, and gender identities became defining components of the race for the White House in 2016 (Sides, Vavreck, and Tesler 2018) and again in 2020 (Sides, Tausanovitch, and Vavreck 2022). Coupled with Trump’s pronouncements and following eight years of perceived and real progress in narrowing political and economic racial gaps under Barack Obama, “Make America Great Again” became synonymous with a return to stricter racial and gender hierarchies with native-born white men on top and other groups below.

“Make America Great Again” and White Support for Trump

Researchers have established that this rhetoric assisted in attracting record levels of white support for Donald Trump, particularly among those without a college degree. In 2016, white voter turnout increased by 2.4 percentage points relative to 2012, while Black, Latino, and Asian American turnout fell by 4.7, 3.8, and 3 percentage points, respectively. ¹⁹ According to an analysis from the Pew Research Center, 88% of Trump’s 2016 voters were white; 44% of the entire electorate were white Americans without a college degree. ²⁰

What explained variation in white voting for Trump? With the 2016 election framed around competing visions of America with respect to immigration, race, and gender, it is little surprise that immigration attitudes, racial prejudice, and sexism were activated (Tesler 2015) and became uniquely predictive of support for Trump in both 2016 and 2020 (Sides, Vavreck, and Tesler 2018). With his hard-line immigration rhetoric, Trump tapped into strong opposition in certain segments of the electorate to the nation’s expansive immigration regime. Immigration attitudes, in turn, became a potent predictor of support for Trump in both the primary and general election (Newman, Shah, and Collingwood 2018; Hooghe and Dassonneville 2018) as well as defection from the Democratic Party (Sides, Vavreck, and Tesler 2018; Reny, Collingwood, and Valenzuela 2019). Along the same lines, Trump’s rhetoric and policy proposals fueled racial polarization initiated by Obama’s presidency in 2008 (Tesler 2016), increasing the predictive power of racial prejudice in voting for the Republican candidate (Schaffner, MacWilliams, and Nteta 2018; Abramowitz and McCoy 2019; Reny, Collingwood, and Valenzuela 2019; Sides, Vavreck, and Tesler 2018; Green and Mcelwee 2018; Mutz 2018). Finally, Trump’s rhetoric and behaviors, coupled with Hillary Clinton’s historic candidacy opposite Trump in 2016, sexism similarly became highly predictive of support for Trump (Schaffner, MacWilliams, and Nteta 2018; Sides, Vavreck, and Tesler 2018; Cassese and Holman 2019; Frasure-Yokley 2018; Ratliff et al. 2019; Glick 2019; Bracic, Israel-Trummel, and Shortle 2019; Valentino,Wayne, and Oceno 2018).

Non-White Support for Trump

In contrast to literature on white voting, extant theories of Black, Latino, and Asian American attitudes and voting suggest that Trump’s rhetoric would mobilize non-white voters against Trump and the Republican Party.

Black Americans’ loyalty to the Democratic Party and overwhelming support for Democratic Party presidential candidates is a well-documented and long-running fact of American politics (Tate 1993). The reasons for this steadfast support are numerous but scholars generally point to the two parties’ positions on race—either rooted in early twentieth-century labor politics (Schickler 2016) or political posturing during the 1960s civil rights movement (Carmines and Stimson 1989; Rigueur 2015)—as predominantly responsible. Continued group support for the Democratic Party is facilitated, at least in part, by perceived hostility of the Republican Party and its policies to the interests of the Black community (e.g. Dawson 1994; though see White and Laird 2020). It is easy to see how Trump’s long history of racially controversial positions and actions, his campaign-trail rhetoric, and his policy proposals would be perceived as particularly threatening and harmful to the Black community, and driving record levels of opposition to his candidacy relative to past Republican candidates.

Latino and Asian political behavior, similarly, is motivated by perceived discrimination and threats to group interests. In the mid-1990s, for example, a series of ballot propositions in California that targeted undocumented immigrants, dismantled affirmative action in the state, and outlawed bilingual instruction in public schools, have been linked to increased Latino political knowledge (Pantoja and Segura 2003), naturalization rates, and voting (Pantoja, Ramirez, and Segura 2001), helping to solidify long-term Democratic dominance in California (Bowler, Nicholson, and Segura 2006). Policy threats in the 2000s and 2010s similarly mobilized Latino voters via an increase in activism and protest (White 2016; Barreto and Nuno 2009; Zepeda-Millan 2017). For Asian Americans, social exclusion remains an important precursor of both partisan identity (Kuo, Malhotra, and Mo 2017) and political participation, particularly for immigrants (Chan, Nguy, and Masuoka 2022). ²¹ Given that Trump’s rhetoric was perceived as a direct threat to Latinos directly, and immigrants, more broadly, it follows that both Latinos and Asian Americans might be especially mobilized to vote against Trump (Haney-Lopez 2016). ²² Indeed there is some evidence that this is true for many Latino voters, particularly those who strongly identify with their ethnic group (Sanchez and Gomez-Aguinaga 2017; Gutierrez et al. 2019), and Black Americans as well (Towler and Parker 2018). ²³

Yet overall, Black, Latino, and Asian American turnout was underwhelming in 2016 and 2020, and there was not a large vote margin swing toward the Democratic candidate in either election. According to a Pew study, 2016 Black voter turnout dropped by nearly 7 percentage points, held steady for Latinos, and increased only slightly for Asian Americans relative to 2012. ²⁴ Further, of those who voted, support for Trump was higher in the Black, Latino, and Asian American communities than it was for Romney in 2012. ²⁵ The pattern was repeated in 2020. Relative to 2016, Latino voters swung an additional estimated 8 percentage points, Black voters by 3 percentage points, and Asian American voters by 1 percentage point toward Trump in the 2020 election. ²⁶

These voting outcomes, which defied expectations, have been to this point under-explored in the literature. What motivated non-white support for Trump? While the previously discussed extant theories of the effects of marginalization and xenophobia on racial and ethnic minority voting behavior is invaluable in contextualizing and understanding contemporary politics in the United States, they do not explain the sizable non-white voting bloc that backed Donald Trump as a candidate in 2016 and 2020. We argue that some non-white voters supported Trump not despite his xenophobic, racist, and sexist comments, but because of them. More specifically, we argue that a sizable number of non-white Americans support traditional status hierarchies even if these power structures marginalize members of their own groups (Jost, Banaji, and Nosek 2004). For those who oppose expansive immigration policies, feel favorable toward white Americans but not other racial and ethnic minority groups, are high in racial resentment, and prefer traditional gender roles, Trump’s rhetoric could have been attractive and motivated support for his candidacy among Black, Latino, and Asian Americans much in the way it did among white Americans. Indeed, extant literature suggests that there are segments of the Black, Latino, and Asian American populations who hold these beliefs.

Xenophobia, Racism, and Sexism in Non-White Communities

The same social identity theory (Tajfel and Turner 1979) that provides the theoretical foundation for expectations of backlash, anger, and opposition to Trump among members of marginalized social groups also provides a theoretical roadmap for their support. The impact of group membership on reactions to political stimuli like xenophobic rhetoric or policies that may harm a group will depend heavily on pre-existing levels of attachment to this group.

Those who do not have strong group identities tend to disassociate from a “low-status” group in the face of xenophobic rhetoric (Perez 2014; Garcia-Bedolla and Michelson 2012; Bedolla 2003) and pursue other identities that carry higher levels of social prestige (Garcia-Rios, Pedraza, and Wilcox-Archuleta 2019), what scholars call social mobility (Jackson et al. 1996; Wright, Taylor, and Moghaddam 1990). In the case of Latinos this might mean moving from being “Mexican and Brown” to being “American and White” (Basler 2008) which serves a social psychological need for a broader community and protection from threats (Lipsitz 1996).

Which groups are perceived as being higher status is itself a product of social norms created and reinforced by dominant group members (white, native born, male, etc.). Entrenched societal norms can lead to hierarchy-reinforcing beliefs and stereotypes about subordinate groups among dominant group members and subordinate group members alike (Ashburn-Nardo, Knowles, and Monteith 2003; Bobo and Massagli 2001; Jost and Banaji 1994; Sidanius and Pratto 1999).

While members of some groups may be able to actually adopt higher-status identities (i.e., become “white”) others may simply adopt the attitudes and beliefs of dominant groups in order to satisfy a psychological need to belong and be accepted by the dominant group (Frankenberg 1993; Basler 2008; Ignatiev 1995; Roediger 1991), or to cope with their own marginalization (Pérez, Robertson, and Vicuña 2023; Carter 2019). These attitudes and beliefs might include conservative immigration policy views, racial prejudice toward one’s own group or other marginalized groups, and sexist attitudes.

There is broad evidence that these attitudes exist in Black, Latino, and Asian American communities. Many Black Americans, for example, harbor anti-immigrant attitudes (Carter and King-Meadows 2019). A Pew Research Center poll shows that a non-trivial percentage of Latino respondents in the United States hold conservative immigration views. In 2018, 25% of Latinos indicated that they believed that there were too many immigrants in the United States, 10% opposed the DREAM Act, and 19% indicated support for building more border wall on the U.S.–Mexican border (Lopez, Gonzalez-Barrera, and Krogstad 2018). These immigration-based policy views are strongly correlated with Latino votes for Trump (Galbraith and Callister 2020).

Similarly, there is broad evidence of inter-minority racial tension and prejudice (Carter and King-Meadows 2019; Pérez, Robertson, and Vicuña 2023; Zou and Cheryan 2017; Kim 1999; Tokeshi 2021; Krupnikov and Piston 2016), particularly under conditions of inter-group resource competition (Mcclain 1993; Gay 2006; Meier et al. 2004). This prejudice is linked to attitudes and voting behavior among Black (Carter and King-Meadows 2019), Asian (Tokeshi 2021), and Latino adults (Krupnikov and Piston 2016; Alamillo 2019).

Finally, it is well documented that sexist attitudes exist (Barnett 1993; hooks 1981; Tate 1993) and shape political attitudes and behaviors in non-white communities (though Black women are more likely to reject white dominant views of gender and vote for women; see Crenshaw 1989, and Sigelman and Welch 1984). Sexism and support for strict gender roles, for example, is correlated with greater Latino (Hickel and Deckman 2022) and Black (Cassino 2020) support for Trump (though see Frasure-Yokley 2018).

In sum, we propose a broader theory of support for Trump that bridges a fractured literature on the antecedents of voting behavior in both white and non-white communities. More specifically, we posit that support of existing status hierarchies was activated by Donald Trump’s rhetoric and policies in both white and non-white Americans and that this activation uniquely motivated support for Donald Trump.

Data and Methods

To assess the relationship between support for status hierarchy and Donald Trump, we begin with data gathered through the Nationscape Survey (NS) conducted by the Democracy Fund + UCLA (Tausanovitch and Vavreck 2021). The NS is a large-scale (N=465,297) weekly repeated cross-sectional survey that began in July 2019 and ended in January 2021 and was sampled and weighted to approximate a representative sample of the U.S. adult population (for more see Holliday et al. 2021). Relative to most public opinion surveys used in political science (e.g., the American National Election Study ²⁷), this extremely large sample allows us to run analyses on racial and ethnic subgroups with high levels of precision.

We replicate our findings using two additional large national opinion surveys. First, we replicate our analysis using the 2020 Cooperative Election Study ²⁸ (CES; N=61,000), a highly respected public opinion survey fielded by YouGov that uses a novel two-stage sample matching process to obtain an approximately representative national sample. Similar to NS, the large sample size in the CES provides sufficiently large samples of Black, Latino, and Asian American respondents to conduct subgroup analyses. Second, to address potential racial and ethnic subgroup cultural competency concerns with these two survey instruments and samples (Barreto, Reny, and Wilcox-Archuleta 2017), we also replicate our findings using the 2020 Collaborative Multiracial Post-Election Survey ²⁹ (CMPS; N=14,988). The CMPS fields culturally-sensitive multilingual surveys with over-samples of Black, Latino, and Asian respondents. The sample is benchmarked to national demographics for each group. For more information on replication materials and surveys used, please refer to Geiger and Reny (2024) and online appendix A.

Our main independent variables are different operationalizations of support for the status hierarchy in the United States across three-broad dimensions—1) immigration; 2) out-group prejudice; and 3) sexism.

IV1: Immigration Attitudes

Throughout U.S. history, restricting the flow and naturalization rights of immigrants from areas that might threaten white institutional and social supremacy has been an important tool to uphold the status hierarchy (King and Smith 2005). At various points in American history, laws were enacted that targeted various immigrant-based groups, including Mexicans in the Southwest and Asians in California. Further, laws like the 1882 Chinese Exclusion Act, the 1924 Johnson-Reed Immigration Act, and the Naturalization Act of 1952 all enacted race-based immigration quotas aimed at shaping the racial characteristics of the nation’s newcomers (Ngai 2005; Tichenor 2002). These laws were often drafted and enacted by the same architects of southern Black segregation (Jacobson 1999) and were seen by white supremacists as a key tool to achieve greater national “whitening” (King 2002, 153-155).

We thus view contemporary support for conservative immigration policies as signaling a strong preference for native- over foreign-born groups—an upholding of a traditional status hierarchy. To measure preferences for conservative immigration policy in the NS data, we created an additive scale of support for a variety of immigration policies including 1) building a wall along the Mexico-U.S. border, 2) opposition to the DREAM Act, and 3) opposition to creating a pathway to citizenship for undocumented immigrants (weighted mean = 0.31; sd = 0.36). ³⁰

IV2: Prejudice

Prejudice toward marginalized out-groups is among the clearest measures of support for the status hierarchy. Much of the public support for the institutions of white supremacy throughout American history was buttressed by both the recognition of the superior status of white Americans and psychological aversion to people of color (King and Smith 2005). Indeed, many theories of prejudice argue that prejudice is motivated, in part, by group-based competition and a desire to uphold group hierarchies (Blumer 1958; Bobo 1983).

We measure prejudice in two ways. First, we create an additive racial resentment scale using two items from the traditional racial resentment scale (“Generations of slavery and discrimination have created conditions that make it difficult for blacks to work their way out of the lower class.” and “Irish, Italians, Jews and many other minorities overcame prejudice and worked their way up. Blacks should do the same without any special favors.” ³¹) that was re-coded to range between 0 and 1 (weighted mean=0.53, sd=0.29). ³² Second, we created a scale of white ethnocentrism that is not only applicable and valid across groups but does a better job of approximating our conceptualization of a racial hierarchy. We construct the scale by subtracting average favorability of all racial out-groups, excluding the respondents’ own, from the favorability of white groups. ³³ Higher values on the white ethnocentrism scale demonstrate a preference for white racial groups compared with other groups, consistent with Kim’s (1999) argument that immigrant groups face the difficult challenge of “racial triangulation” by which they enter a racial field defined by white and Black identities at the top and bottom and are pressured to choose sides between the nation’s racial orders. ³⁴ This scale is recoded to range between 0 and 1 (weighted mean = 0.50, sd = 0.16).

IV3: Sexism

Finally, we examine attitudes about gender as a third measure of support for the status hierarchy. According to social dominance theory (Sidanius and Pratto 1999), the gender system, in which men have disproportionate social, political, and military power, relative to women, forms a central pillar of the trimorphic structure of group-based social hierarchy.

We measure our final predictor variable using a modified Modern Sexism Scale (Swim et al. 1995) ³⁵ crafted from two items in the NS: “Increased opportunities for women have significantly improved the quality of life in the United States,” and “Women who complain about harassment often cause more problems than they solve.” Like all of our other predictor variables, this additive scale was recoded to range between 0 and 1, where 1 is the more conservative attitude (weighted mean =0.33, sd = 0.22). Additional details on all scales can be found in online appendix B.

DV: Support for Trump

Support for Trump, our outcome variable, is operationalized as self-reported support for Trump in a head-to-head match up with Biden (weighted mean = 0.37) and, in a robustness check, a 4-point favorability Likert scale (weighted mean = 0.59, sd = 0.42). Following King and Smith 2005, we view support for Trump as a clear example of support for a political entrepreneur who is embedded within and advocating for an institutional order that promotes, maintains, and reifies racial and gender hierarchies and social institutions.

For our primary models, we run separate logistic regressions for each racial group and each independent variable (a total of 16 models). Our statistical models control for standard individual-level demographic factors such as college education, household income, sex, age, ideology, and partisanship. Latino and AAPI models include dummy variables for the largest country-of-origin groups. For AAPI models we include Indian, Korean, and Other, leaving Chinese as our reference category. For the Latino models, we include Cuban, Puerto Rican, and Other leaving Mexico as our reference category.

Results

We begin by comparing our measures of status hierarchy support, our key independent variables, across groups. In Figure 1 we show mean support for conservative immigration policy (Panel A), mean racial resentment (B), mean white ethnocentrism (C) and mean sexism (D) for white, Asian, Latino, and Black respondents from the NS survey. Consistent with expectations, white Americans have the most conservative immigration attitudes and the highest levels of racial resentment and white ethnocentrism. Asians, Latinos, and Black Americans follow an ordering that roughly corresponds with each group’s placement in the racial hierarchy (Kim 1999; Masuoka and Junn 2013). While support for conservative immigration policy is relatively low among racial and ethnic minority groups, and racial resentment particularly low among Black respondents, Asians and Latinos look quite similar to whites on many measures. Racial resentment levels are substantively identical between white, Asian, and Latino respondents. The same is true of white ethnocentrism and sexism. We break these scales down to their component items and display group means for each across racial groups in online appendix B. It is hardly the case, then, that support for the status hierarchy is solely restricted to white respondents. Across various measures, Asian, Latino, and in fewer cases, Black Americans harbor similar attitudes to white Americans. Whether these attitudes translate into support for Trump, though, remains an open question that we explore next.

Figure 1 Support for status hierarchies across racial groups

Notes: The figure displays weighted means for each independent variable scale across racial/ethnic groups in the Nationscape Survey (2019–2021). In online appendix figure B1, we display individual items composing these scales broken out by racial groups.

We move next to our regression models. Rather than present logistic coefficients, which are difficult to interpret, we simulate the predicted probability of support for Trump moving from lowest to highest observed values of each independent variable, holding all others at their means (King, Tomz, and Wittenberg 2000). We present full regression tables for all models in online appendix C.

We begin in figure 2 with immigration attitudes, displaying the predicted probability of supporting Trump separately for white, Black, Latino, and Asian respondents. As the figures clearly illustrate, immigration attitudes are powerfully predictive of support for Trump. Moving immigration attitudes from their most liberal to conservative values is associated with 62.5 percentage point increase in support for Trump for white Americans (95% CI: [61.7,63.2]), 17.8 percentage points for Black Americans (95% CI: [15.9,19.8]), 49.8 percentage points for Latinos ([47.6,52.0]), and 41 percentage points for Asian Americans (95% CI: [37.4,44.5]).

Figure 2 Immigration attitudes and Trump support

Notes: Predicted probability of supporting Trump moving from most liberal (0) to most conservative (1) immigration policy attitudes, holding all other values at their means. 95% confidence intervals. Data from Nationscape Survey (2019–2021). Full regression tables appear in online appendix table C1.

We turn next to figure 3, where we display the results of our racial resentment models. Similar to figure 2, we find a strong relationship. Moving racial resentment from its lowest to highest values is associated with an increase in Trump support of 51.7 percentage points (95% CI: [50.8,52.6]) for white Americans, 9.6 percentage points (95% CI: [8.4,10.9]) for Black Americans, 30.8 percentage points (95% CI: [29.2,32.5]) for Latinos, and 29.6 percentage points (95% CI: [26.7,32.5]) for Asian Americans.

Figure 3 Racial resentment and Trump support

Notes: Predicted probability of supporting Trump moving from lowest (0) to highest (1) levels of racial resentment, holding all other values at their means. 95% confidence intervals. Data from Nationscape Survey (2019–2021). Full regression tables appear in online appendix table C1.

For figure 4, we regress Trump support on our other measure of prejudice: white ethnocentrism. Similar to previous analyses, moving white ethnocentrism from its lowest to highest values is associated with an increase in Trump support of 48.2 (95% CI: [46.2,50.3]), 14.8 (95% CI: [12.3,17.5]), 37.6 (95% CI: [34.3,40.8]), and 33.0 (95% CI: [27.5,38.4]) percentage points for whites, Blacks, Latinos, and Asian Americans, respectively.

Figure 4 White ethnocentrism and Trump support

Notes: Predicted probability supporting Trump moving from lowest (0) to highest (1) levels of White ethnocentrism, holding all other values at their means. 95% confidence intervals. Data from Nationscape Survey (2019–2021). Full regression tables appear in online appendix table C2.

Finally, in figure 5, we show almost identical patterns. Moving sexism from its lowest to highest values is associated with increases, across the board, in support for Trump from 8.3 percentage points (95% CI: [6.9,9.8]) for Black Americans to 33 percentage points (95% CI: [31.6,34.4]) for white Americans. Latinos and Asians once again fall in the middle with increases of 20.7 percentage point (95% CI: [16.7,24.8]) and 17.9 percentage point (95% CI: [15.7,20.1]) increases for Asian and Latino respondents, respectively.

Figure 5 Sexism and Trump support

Note: Predicted probability supporting Trump moving from lowest (0) to highest (1) levels of sexism, holding all other values at their means. 95% confidence intervals. Data from Nationscape Survey (2019–2021). Full regression tables appear in online appendix table C2.

In sum, there is robust evidence that all of our measures of support for the status hierarchy—immigration attitudes, prejudice, and sexism—are uniquely and powerfully predictive of support for Donald Trump. ³⁶ While all of these attitudes are held at similar levels across racial groups, their associations with support for Trump vary by racial group. Consistent with theory, the strength of associations roughly corresponds to each groups positioning within the racial hierarchy with Asian Americans and Latinos between whites at the top and Black Americans at the bottom. ³⁷

Next, we run a series of tests to probe the robustness of our empirical results. First, readers might be concerned with our measure of support for Trump. We find identical results using Trump approval rather than support for Trump over Biden in a head-to-head match up (online appendix tables C3 and C4). We also show that results do not appear to be sensitive to including each IV in separate regressions versus all together in a single regression (online appendix table C5). Second, to test whether our results are unique to the Nationscape survey, we replicate our analyses with two additional surveys that have complementary advantages. In online appendix tables C6 and C7, we replicate our findings using the 2020 Cooperative Election Study (CES), and find substantively identical results. Readers might also be concerned that neither the NS nor the CES use culturally-sensitive sampling procedures (Barreto, Reny, and Wilcox-Archuleta 2017). Using the 2020 Collaborative Multiracial Post-Election Survey (CMPS), a multiracial and multilingual post-election survey that over-samples Black, Latino, and Asian American respondents, we again replicate our analyses and find substantively identical results (onlline appendix tables C9 and C10). These replications suggest that our findings are not simply an artifact of our choice of survey. Our results are robust to various surveys with different field dates, sample sizes, sampling techniques, and questionnaires.

Our theory suggests that Trump was unique in activating this existing support for the status hierarchy in the mass public. First, if this is true, we should see that the relationship between these attitudes and support for Trump is stronger among those who pay more attention to politics, or “receive the message,” relative to those who pay less attention (Zaller 1992). In figure 6, we display our same models as in figures 2–5, but this time interacting each key independent variable with a dummy for attention to politics (1=“follow what’s going on in government most of the time”, 0=“some of the time” to “hardly at all”) and displaying changes in predicted probabilities with 95% confidence intervals. Across the board, the association between each independent variable and support for Trump is consistently stronger among those who pay attention to politics, suggesting that respondents are linking their predisposition to political choices, as expected and consistent with theory.

Figure 6 Attention, status hierarchy, and support for Trump

Note: Change in predicted probability supporting Trump moving from lowest (0) to highest (1) levels of each IV for high and low political interest respondents, holding all other values at their means. 95% confidence intervals. Data from Nationscape Survey (2019–2021). Full regression tables appear in online appendix tables C11 and C12.

Second, readers might be concerned that respondents are “learning” their attitudes from Trump rather than having their attitudes activated by Trump’s rhetoric, suggesting a different causal model. While we suspect that these attitudes—which are rooted in group-based antagonisms and are strongly rehearsed and highly crystallized—are unlikely to change substantially in the face of elite rhetoric (Tesler 2015), we estimate a model predicting support for Trump in 2016 as a function of 2011 attitudes using the Democracy Fund Voter Study Group and YouGov Views of the Electorate Research (VOTER) Survey panel dataset (N=5,567). While the racial and ethnic subgroups are smaller in this survey (the Asian American sample is too small to analyze), and we do not have perfectly corresponding measures of support for the status hierarchy, this approach leverages the temporal nature of panel data to rule out concerns of reverse causality (Lenz 2012). These results, which largely replicate, are included in online appendix tables C17, C18, and C19.

Third, we should see a stronger association between each independent variable and support for Trump relative to other Republican figures whose rhetoric and actions are not as clearly advocating reinforcing the status hierarchy. In figure 7, we run the same models as with figures 2–5 but include models estimating support for i) the generic Republican congressional ballot and ii) self-reported support for Romney in past voting behavior. Again, we find that moving each independent variable from its minimum to its maximum values is associated with a much stronger change in support for Trump than for other Republican figures, consistent with our theory and with a story of activation. We replicate this analysis using the VOTER panel data. In online appendix tables C17, C18, and C19, we also show that support for the status hierarchy measured in 2011 is more predictive of support for Trump in 2016 than Romney in 2012 among the same respondents, particularly for white and Latino respondents. ³⁸ We do not find consistent results for Black respondents between these two elections, however, which may be partly due to Obama’s presence on the ballot in the 2012 election and his ability to turn out a much broader swath of Black voters than other Democratic candidates before or since (Parker 2016). We also find in these models that other factors, like ideology and partisanship, are not being uniquely activated in 2016 relative to 2012 among racial and ethnic minority voters, which could provide an alternative explanation for our findings.

Figure 7 Status hierarchy and support for other Republicans

Note: Change in predicted probability supporting Romney, Congressional Republicans, and Trump moving from lowest (0) to highest (1) levels of each IV, holding all other values at their means. 95% confidence intervals. Data from Nationscape Survey (2019–2021). Full regression tables appear in online appendix tables C13, C14, C15, and C16.

Discussion and Conclusion

Many media and political elites were flummoxed by Donald Trump’s surprisingly high support among Black, Latino, and Asian American voters in 2016 and 2020 relative to previous elections. Much of the extant research suggests that Trump’s anti-immigrant, racist, and sexist rhetoric, actions, and policy positions should have mobilized many non-white voters for the Democratic candidate. Indeed, some research suggested that they did. But this work ignores the fact that non-negligible segments of the Black, Latino, and Asian American populations in the United States hold beliefs that reinforce and maintain current status hierarchies even if those hierarchies are actively harmful to their groups.

We propose that this non-white support for traditional status hierarchies can be activated by entrepreneurial politicians much in the same way that it is activated in the white population. We test this theory using multiple surveys that have large samples of white, Black, Latino, and Asian American respondents but that were all fielded in different time periods, with different samples, and with different levels of cultural competence. Across the board, we find robust support for our theory. Not only do Black, Latino, and Asian Americans support the status hierarchy, multiply defined, but this support is strongly predictive of support for Donald Trump.

This article makes several contributions to the literature. First, it expands upon theories of white political behavior to show how support for traditional status hierarchies is not unique to white Americans and can be activated by political rhetoric and actions. Second, while most studies of voting theorize about and test theories on a single racial or ethnic group individually, we use surveys that are sufficiently large to allow us to comparatively and precisely test our theory with multiple groups and assess the robustness of our results to diverse sampling and methodological strategies. Finally, rather than focus on a single psychological construct and its effect on candidate support, we expand the scope of our study to attempt to understand how various related predispositions predict support for candidates who openly embrace a preservation of America’s traditional status hierarchies.

Our findings help us to better understand and contextualize Donald Trump’s surprising political success with racial and ethnic minority voters. It suggests that the GOP’s shift from dog-whistle racial appeals to more overt targeting of various immigrant and other racial, ethnic, and religious minority groups may not harm the party’s national prospects the way we might have expected prior to 2016.

While our results appear to be robust, our study has several limitations and opens up additional avenues of research for scholars interested in non-white voting behavior. First, while our research approach is consistent with other studies of voting behavior in recent elections (Sides, Vavreck, and Tesler 2018; Sides, Tausanovitch, and Vavreck 2022) and attempts to deal with issues of endogeneity, our use of cross-sectional public opinion surveys prevents us from firmly establishing causal effects. A recent overview of research on racial priming (Valenzuela and Reny 2022) reveals that very little experimental work has examined how these predispositions might be activated in the non-white population by elite rhetoric or the conditions under which activation might occur, suggesting space for experimental work on the priming of predispositions in non-white communities. Second, our study relies on measures of prejudice, like racial resentment, that were developed to theoretically only apply to white Americans and have poorer face validity for other populations like Black Americans. While we think that some of our other measures, like white ethnocentrism, partly overcome these issues, there is ample space to develop different measures of general out-group prejudice that might have better construct validity across groups. Finally, while it is beyond the scope of this study, we do not have evidence of the motivations underlying support for the status hierarchy among non-white Americans. While some researchers are increasingly investigating inter-group solidarity (Pérez and Kuo 2021) and animus within marginalized communities (see Pérez, Robertson, and Vicuña 2023), much more remains to be done.