Since the 2016 election, I have reviewed nearly every academic article containing the name “Donald Trump.” This huge literature has plenty of disagreements—but the dominant findings are clear: attitudes about race, gender, and cultural change played outsized roles in the 2016 Republican primaries and general election, with economic circumstances playing a limited role.

Media coverage of this research has often framed it as a victory for those who argue “Trump supporters are racist and sexist” over those who argue “Trump supporters are left behind economically.” Concerned scholars and commentators have criticized this research as hopelessly biased and have often tried to revive the economic story. But another strain of conservative explanation for Trump’s support—one that is focused on aversion to “political correctness”—turns out to be quite close to the racial and cultural explanations.

Beneath the divide lurks a consensus: Many people dislike group-based claims of structural disadvantage and the norms obligating their public recognition. Those voters saw Trump as their champion. The 2016 election produced greater candidate and voter division around the celebration of diversity and accepted explanations for group disparities. Trump and Clinton extended the influence of these factors, but they had long been rising in importance in dividing the American parties.

Naming and Shaming

This hidden consensus is obscured due to semantics. The names of common scales of survey questions used to predict Trump support tend to irk conservatives:

  • “Racial resentment,” an aspect of “symbolic racism,” is measured by asking for agreement or disagreement with statements like “Generations of slavery and discrimination have created conditions that make it difficult for blacks to work their way out of the lower class” or “Irish, Italian, Jewish, and many other minorities overcame prejudice and worked their way up. Blacks should do the same without any special favors.”
  • “Hostile sexism” is measured with agreement or disagreement with statements like “when women lose to men in a fair competition, they typically complain about being discriminated against” or “women are too easily offended.” The related “modern sexism” scale taps similar attitudes.
  • “Authoritarianism” is measured with questions asking respondents to choose between pairs of parenting values such as “independence” vs. “respect for elders” or “self-reliance” vs. “obedience.”

Thus, in a crude popularization, respondents who favor obedient children and individual-level explanations for economic disadvantage could easily be labeled racist, sexist authoritarians. The questions would no doubt be labeled and interpreted differently if academia (and the media) were composed of more conservatives and fewer liberals.

Yet these question scales are reliable and valid. Scholars have worked for decades to differentiate them from general ideological views. They help predict lots of attitudes, including support for Trump. These attitudes are far more explanatory than direct measures of bias, though traditional racism and sexism (and white identity) also explain some votes. Studies finding influential perceptions of white or male discrimination are actually based on relative comparisons with disadvantaged groups.

Scholars have a tendency to imply that the liberal ends of these scales are normal and look for deviation from them, such as by measuring “denial of discrimination.” But one could equivalently restate common findings in the other direction. People who reflexively believe claims of discrimination because they perceive structural barriers for disadvantaged groups were more likely to support Hillary Clinton, as were those favoring permissive parenting.

Structural Discrimination Perceptions

Liberal pundits and scholars see evidence of structural group disadvantages and long-term effects from discrimination. Failure to acknowledge discrimination thus seems to blame the victim based on negative stereotypes. Since public opinion is notoriously inconsistent and contradictory, however, we cannot impute opinions beyond the endorsed sentiments.

The influential attitudes are not simple manifestations of racism and sexism. Political sophistication leads to lower scores on the racial resentment scale because sophisticates tend to make global attributions regarding social outcomes, allowing them to associate anything with broader sociopolitical causes. White and black Americans, moreover, understand the racial resentment scale differently, and many minorities score high on it. Only overt racism predicts whether whites discriminate and whether they have racially biased evaluations of others. Racial resentment does not predict either.

The liberal side of the scales may also represent biased thinking. Liberals perceive more racism and sexism than racial minorities and women say they experience. Experiments show that liberals perceive tests where men or whites perform better as less credible than equivalent tests showing women or minorities doing better, even though conservatives rate them equally credible. Liberals are thus predisposed to believe discrimination is the cause of disadvantaged group disparities.

Seeing these attitudes as more than simple bias may help understand Trump support across genders and races. Although women were less likely to support Trump than prior Republican candidates, “hostile sexism” and “racial resentment” had similar effects on male and female voters. Trump gained minority vote share (compared to Mitt Romney, who had to face Barack Obama) and some Obama-Trump voters were racial minorities. But minority vote choice was also partially driven by attitudes toward diversity and value change. Racial and gender attitudes are related to broader cultural views (such as agreement that “the American way of life is threatened”) that are widely subscribed to by Americans across social groups.

Evidence of economic effects on Trump support has been based on relating voting trends and economic performance across geographic areas, such as counties. But these findings may still be explained by cultural views. Shocks from Chinese imports, for instance, drive negative attitudes towards minorities more than attitudes toward free trade. Geographically, high Latino-growth areas were more likely to move toward Trump. Voters were especially sensitive and emotionally responsive to claims of an impending American white minority. The voters most enthusiastic about Trump early in the campaign were those highest in “racial resentment.” And even Bernie Sanders supporters who defected to Trump tended to stand out for their low perceptions of discrimination rather than for their economic views.

Reconciling Explanations Based on Political Correctness

Research on “political correctness” advances a similar cultural story with a conservative spin. Asking about statements that might be offensive to particular groups increased support for Trump. His supporters were more fearful about restrictive communication norms. Beliefs that political norms around offensive speech silence important discussions and prevent people from sharing their views are widespread, particularly among conservatives. Many conservatives say they cannot discuss topics like gay rights, race, gender, or foreign policy for fear of being called racist or sexist. Opposition to political correctness thus incorporates aversion to norms toward discrimination claims. When voters begin to question society’s norms, they can see candidates (even those who lie regularly) as more authentic truth tellers when they subvert those norms.

Questions tied to political correctness could tap the same underlying views as those explicitly about racial and gender discrimination. For example, conservatives might favor statements like “Police officers and soldiers are too worried about offending minorities to do their jobs effectively” or to disagree that “Men are responsible for making sure their statements to women are not perceived as offensive or inappropriate.” The findings so far do not imply that the specific questions asked are the only or best way to tap the underlying important attitudes. My own work finds that racial attitudes are related to broader sentiments about celebrations of diversity and the undermining of traditional American values.

Associations between all of these views and Trump support should not be taken as unidirectional and causal. There has been no increase in “racial resentment” for 30 years. It is instead increasingly associated with most other political attitudes, increasing among Republicans and decreasing among Democrats. Politically-aware partisans are most likely to sort into their majority party viewpoint on racial attitudes. Panel studies show reciprocal causation, with prior partisanship driving racial views more than the reverse. In this election, like others, the vast majority of partisans supported their own party’s candidate—and many adopted the views of their party’s standard-bearer as their own.

Vote choice predictions used for other elections were still substantially improved by adding racial attitude measures. Even those sentiments measured before Trump’s rise predicted future attitudes toward him. But there have now been more changes in racial attitudes among Democrats than Republicans, suggesting that reactions to Democratic elite messages and the campaign context were also driving views.

The Racial Context of Trump’s Rise

When racial and gender views are invoked to explain Trump support, the campaign context usually referenced is Trump’s own statements—and there is no shortage of material. He first rose to political prominence by championing the “birther” issue (the idea that Obama might be ineligible for the presidency because he was born in Kenya). He began his campaign by claiming that Mexico was sending “rapists” to enter the country illegally. He later attacked a federal judge of Mexican heritage as inherently biased, the textbook definition of racism. On gender, he called Clinton a “nasty woman” and was caught on an old Access Hollywood video bragging about sexual assault. According to the common view, he made the subtext of prior campaigns the text and turned dog whistles into blowhorns. Running after the first racial minority president and against the first female nominee made his statements particularly salient.

But it is not clear that Trump’s direct statements were responsible for activating voters’ cultural views. His negative statements about minority groups were recognized by voters—but not positively. In open-ended responses, “racist” was the number one negative thing said about Trump even among Republicans. And a surprisingly high proportion of Trump voters said they did not like him personally, often citing his language.

In paid advertising, it was the Clinton campaign that repeatedly raised these issues and endlessly replayed Trump’s statements. That made their ad campaign a vast historical outlier compared to prior elections; Clinton talked a lot less about policy issues and a lot less positively overall. Clinton raised the salience of norms about off-limits race and gender discourse, believing it would help her win votes (but may have also activated views of political correctness). Clinton talked far less about class, discounting “the rich vs. the middle class” message that has been a Democratic staple for generations. As a result, class attitudes had no effect in 2016, even though they had been dominant in 2012.

Far more often than his explicit racial statements, Trump incorporated rhetoric combining conservative sentiments with symbols that invoke racial attitudes. He mentioned “illegal” and “criminal” more than prior campaigns and exceeded Richard Nixon levels of “law and order” rhetoric, which had been effective in the past at marrying racial attitudes with broader ideas about liberalism. Meanwhile, seeing an advantage over Sanders with black voters in the Democratic primary, Clinton toured the country with the “mothers of the movement” who had lost children to police violence. Opinions of police use of force were related to Trump and Clinton support early in the campaign and law enforcement became an important Trump constituency, boosting Republicans.

Trump took advantage of a moment of rising racial conflict. As he began to campaign in 2015, there had been a large upsurge in attention to the Black Lives Matter movement, protests of police violence, and campus protests of discrimination. The Baltimore Freddie Gray protests and riots before his announcement, and the Ferguson anniversary protests after it, stimulated widespread media attention and public interest. A San Francisco shooting death that summer was used to (erroneously) blame “illegal immigrants” for rising crime in some cities. Later in 2015, a campus hunger strike at the University of Missouri stimulated similar racial protests at other universities. That December, fourteen people were killed in San Bernardino, California, igniting a debate about immigrant radicalization and neighbors’ fear of stereotyping.

Conservative media covered the escalating series of events continuously and sensationally, connecting them with a crisis atmosphere and rising minority group demands. Conservatives linked their views of universities and cities as coddling protesters and criminals, in their minds both stifling viewpoints and promoting disorder. Trump took advantage of the backlash against perceived new demands for cultural reorganization to redress discrimination.

Trump voters thus perceived rising crime alongside demands to limit police actions that hurt minorities, rising terrorism alongside norms against singling out Muslims, and declining opportunities for men alongside expectations to avoid mistreating women. Clinton voters saw rising diversity and increasing openness to people of all types being threatened by a backward-looking and shame-worthy candidate. Both perceptions were responses to the central messages of the candidates and the context of the campaign.

Reconciling Cultural Views of 2016

A lot of energy has been invested in understanding Trump support, even though he got fewer votes than Clinton nationally and won mostly the same voters as prior Republican candidates. But history rides on such contingencies as the decisions of a small share of voters in the upper Midwest. The popular interpretation of elections is also critical for future politics. That makes the expanding academic field of Trump studies important to get right.

I am not an expert on racial or gender attitudes, only a close reader of the literature who is more sensitive to conservative complaints about its popularization (given underrepresentation of conservatives in academia). Interested readers should consult great recent research from experts. Racial and gender attitudes are uniquely important, and not simply manifestations of general conservatism.

The long-term economic fortunes of rural areas and the relative economic standing of the white working class may, of course, still influence the development of the racial and cultural attitudes that were influential in 2016. But neither views of voters’ individual or community economic performance nor voters’ expectations about their economic prospects seem to have had much influence in 2016.

Beneath the mocking of “economic anxiety” as an explanation for Trump support lies copious evidence that attitudes toward group disadvantage and diversification were especially important in 2016 and had been rising in importance in prior elections. Conservatives may not like survey scales with pejorative names, but they should stop insisting that economic factors were more important than racial and cultural attitudes. Liberals, in turn, should stop assuming that Trump’s racist and sexist remarks directly won over racist and sexist voters. The messages sent by both campaigns as well as the news surrounding the election raised the salience of attitudes toward discrimination claims and norms surrounding political discourse. Recognition of these patterns can enable shared understanding of the 2016 election across the ideological divide.

Photo Credit: James McNellis from Washington, DC, United States [CC BY 2.0 (https://creativecommons.org/licenses/by/2.0)]