White Voter Support for Trump

Two frameworks are useful in trying to explain the surprising 2016 election, and even the surprising nomination, of Donald Trump. One framework focuses on the macro level, covering the larger circumstances and conditions that permitted if not facilitated Trump’s election. The other framework focuses on the micro level, on the specific variables that resulted in Trump’s election. One can justly see Trump’s election as a perfect storm, a combination of many things that permitted his razor edge victory, about 70,000 votes total in Michigan, Pennsylvania, and Wisconsin.

American voters have gone to the polls five times in this young century to elect a president. In 40 percent of the cases the person who did not receive the most votes became president. Given the closeness of the election, we may never reach consensus about why Trump won the election. This post takes a micro view by discussing research on white support for Trump undertaken by Carl Stempel, a sociologist at the California State University, East Bay.

ANES Survey Data

Using surveys conducted by the 2016 American National Election Studies Stempel attempts to explain support for Trump among white voters. “Support for Trump” combines votes for Trump, voted enthusiastically for Trump, and a “feeling thermometer”* towards Trump. Each of these is weighted equally. Stempel controls for background/demographic data: age, gender, income, education, attends religious services, geographical region, country of origin, and employment status. He finds that older people, males, those in the middle four deciles of income ($40,000 – $100,000), lower middle education (high school through some college), southerners, U.S. born, and “homemaker” like Trump more. However, he finds that none of the demographic/background variables are strongly associated with Trump. Together, they explain about 11% of the variation in Trump support.

Before getting into the variables that most explain Trump support it is important to note that almost all the variables are inter-related. However, each of the variables is controlled for all the other ones. The final list of explanatory variables are either the ones that are quite different than the other explanatory variables (that is, they capture a different source of Trump’s appeal) or are more strongly associated with Trump support than the variables that are not significant) or the explanatory variables fall into both camps.

In the analysis below the explanatory variables are ranked from higher standardized beta coefficients to lower standardized beta coefficients. These are estimates resulting from a regression analysis that has been standardized. The beta coefficients compare the strength of the effect of each independent variable to the dependent variable – the higher the beta value, the stronger the effect.

Variables with statistical significance of p < .01 or less

The two explanatory variables that most strongly influence by far white support for Trump deal with immigration and declining economic conditions.

Immigration. The strongest variable for Trump support concerns restricting immigration and deporting unauthorized immigrants. In other words, “pro-Trump people really want to restrict immigration a lot…”)

Interestingly, these Trump supporters tend not to see immigrants as threatening the nation’s economy or culture. Stempel says that this may mean that Trump supporters are somewhat politically correct on issues of race and immigration. He explains this oddity by suggesting that overall Trump supporters are “disidentifying with his blatant racism and xenophobia and want to position their resentments as principled and fair, in response to being victimized and excluded or to authorities’ failure to enforce laws.” Again, this is how the survey respondents answered the survey questions. It does not necessarily mean this is what they believe. In any event, the responses to the survey questions dealing with whether immigrants threaten the economy or the country’s culture are negatively related to support for Trump. The answers to these questions fall just above p < .05 significance at .059.

Perception of a Declining U.S. Economy. The second most influential variable explaining support for Trump deal with economic issues, specifically the perception that the U.S. economy is declining, and declining especially since 2008. This variable does not measure levels of personal economic problems (worse off than last year, worry over money problems, family member or close friend lost a job). These personal economic problems are positively but only weakly related to support for Trump. Overall, personal economic problems are not an important factor in explaining Trump support.

The perception that the U.S. economy is declining also “captures the squeeze on modestly educated middle- and working class whites who feel they (or their children) are losing their opportunity at the American dream…and who have seen better jobs going away to globalization, robots, machines, and computers.” Stempel suggests that the strength of this variable may be caused by “Fox News’ constant talk about how Obama is ruining or stalling the economy, echoed by Trump on the campaign trail….” and “…affirmative action beneficiaries like the Obamas.”

The next pair of variables show high influence on Trump supporters. One is authoritarianism and the other is anti-gay or transgender policies.

Authoritarian. The authoritarian variable deals with wanting a strong-man leader. This variable measures three items (1) the country would be great by getting rid of rotten apples, (2) country needs a strong leader to take us back to the true path, and (3) strong leader is good for the U.S. even if he bends rules to get things done. Stempel notes that this line of questioning is closely related to questions created by academics “wanting to understand fascism after WWII…”

Several questions relate to parenting (those wanting strict obedience vs. those wanting to foster independence and curiosity). Self-directed parenting is negatively related to pro-Trump support but is not significant.

Anti-Gay/Transgender Policies. Positions on gay marriage, adoption by gay parents, the “right” to discriminate against gays, and transgender bathroom policies/laws (reversed) all are strongly associated with being pro-Trump. These variables are more important than variables dealing with general traditional moral values (concern about moral decline, the need for more traditional family values, and newer lifestyles are breaking down society, as well as the reverse scored questions dealing with being positive about adapting to a changing world and respecting other minorities).

Stempel suggests that the anti-gay, anti-transgender significance “reflects strong discourses of ‘religious freedom’ and ‘they’re shaming us for upholding our traditional (Christian) morals’ that provide justification or cover for anti-gay policy positions, but not so much for anti-gay prejudice.

Four variable show moderate influence for pro-Trump support. These include anti-Muslim prejudice, anti-free trade, government should foster more economic quality (reversed), and resentment against Blacks.

Anti-Muslim Prejudice.  Anti-Muslim prejudice relates to feeling negatively towards Muslims, thinking Muslims are violent, and thinking Muslims are unpatriotic. Stempel notes that anti-Muslim prejudice is stronger or more freely expressed than prejudice against any racial group. He suggests that whites “apparently feel justified and freer to hate on Muslims.”

Anti-Free Trade. Anti-free trade is robustly associated with being pro-Trump. Stempel suggests this is related to the highly influential variable of the perception of a decline in the U.S. economy.

The Government Should Foster More Economic Equality (reverse scored). Although Trump supporters are anti-free trade Stempel importantly notes that other aspects of what might be called economic populism do not equate with pro-Trump support. Wanting the government to do more to address economic equality is “negatively and fairly strongly associated with being pro-Trump.” Stempel goes on to say that “Trump’s various efforts to lower taxes on the rich, even if at the expense of support for the poor, the elderly, and those in poor health, fit his real base.”

Resentment Against Blacks. Pro-Trump supporters resent Blacks. More specifically, they resent Blacks for not working their way up on their own and for blaming their problems on racism. White pro-Trump supporters also resent Blacks for their over-reliance on government and lack of a hard work ethic. On the other hand, anti-Black prejudice, such as the feeling thermometer and seeing Blacks as lazy or as violent, is negatively associated with support for Trump. At the same time, anti-Asian American prejudice is negatively associated with Trump support.

Stempel draws this conclusion regarding support for Trump and Black prejudice: “To be clear, pro-Trumpers have higher levels of immigrant threat and blatant anti-Black prejudice. What this analysis shows is that more politically correct expressions of racial and anti-immigrant resentment are the core drivers of Trumpism, not blatant prejudice.” Stempel also makes the point that “anti-Black resentment correlates very strongly with support for anti-unauthorized immigrant policies.”

Next in degree of influence on being pro-Trump are the two variables of opposition to Syrian refugees to the United States and pro-Christian fundamentalist.

Anti-Syrian Refugee Policy. Anti-Syrian refugee policy is fairly strongly and robustly associated with support for Trump.

Pro-Christian Fundamentalist. A pro-fundamentalist feeling thermometer is robustly and positively associated with support for Trump. This variable also involved concern about moral decline. Stempel suggests that this “may be partly because many liberal and moderate whites feel comfortable/justified in being prejudiced towards fundamentalist Christians.”

The least influential, but still statistically significant at the p < .01 level are these two variables: high political cynicism and the belief that people should get ahead without government support. Stempel doesn’t discuss these variables. It appears the negative relationship of political cynicism is entirely related to the variable measure of “most politicians are trustworthy.” In other words, supporters of Trump think most politicians are untrustworthy.

Variables with significance at the p <.05 level.

Several variables are significant at the p < .05 level. Three that have not been mentioned in passing above include use of military force to fight ISIS or to solve international problems, wanting smaller government, and social dominance orientation (reversed). This latter variable suggests that Trump supporters are opposed to the belief that society should make sure everyone has equal opportunity and to the notion that if people were treated more fairly there would be fewer problems. Trump supporters support the belief that we would be better off if we worried less about equality and that it is not a problem if some have more chance in life.

Unimportant Variables

Stempel identifies several variables that are not associated with support for Trump. Many of these variables drop out of significance because they their influence are carried by variables that have higher standardized Beta coefficients. Here are variables that are not robustly associated with white support for trump:

• Core conservative beliefs about government, especially the importance of reducing deficits and being against government regulation.
• Fear of terrorist attack
• Pro-white identity
• Anti-feminism and gender traditionalism
• Dislike for Obama

Stempel makes an important point about pro-Trump white supporters dislike for Obama.

“Trump supporters really dislike Obama. If I add an anti-Obama variable to the final equation, it becomes by far the strongest explanatory variable…. Adding anti-Obama to just the demographic variables, it explains a whopping additional 49% of variation in Trump support! I did not report this regression model to you because while it is important to know how much pro-Trump is anti-Obama, it muddies our understanding of the influence of all the policy and targeted variables, because it overwhelms the explanatory power of many of those variables… By adding the anti-Obama variable last, after all of the other explanatory variables, we can get some answer to the question of what motivates anti-Obama feelings by looking at which of the other variables decline most in explanatory power. The winners are anti-Black resentment, anti-gay/transgender policies/laws, anti-Muslim prejudice, government should foster greater equality, and wants smaller government.”

Comment

Looking back from 2019 onto these findings from 2016 I find few if any that are fully surprising. The strength of the anti-immigrant feeling is somewhat surprising. Viewing the immigration issue both statistically and from the perspective of mainstream media attention, immigration was not major deal in 2015. Yet Trump was able to take what may have been a latent or implicit fear held by some whites and manufacture it into a crisis for many whites.

Trump took what was at most a legitimate concern and transformed it into an imminent threat. The same can be said about the economy: he transformed a legitimate concern into an imminent threat for many whites, especially regarding his claims about the pernicious damage of free trade. Trump’s attacks on Obama with his emphasis that Obama was not born in the U.S. and his strong hints that he was a Muslim left many whites feeling angry about Obama and heightened their resentment against both Blacks and Muslims.

Trump’s repeated claims about him being the only person who could fix these disasters made him appeal to those who thought America needed a strong-man leader, even if it meant bypassing norms or at times breaking the law. The antipathy to political elites of any party, but especially the Democratic Party, led Trump to promise to “drain the swamp” and behave differently from established, traditional politicians who were to blame for the country’s fall. Only he could make America great again.

Trump has governed largely with the same modus operandi of creating crises where none really existed and then claiming he would (and sometimes claiming he did) fix the problem, eliminated the crisis. The creating or enhancing a crisis has occurred with immigration and the border wall, North Korea and missiles, China and trade, and Iran and nuclear fuel. Perhaps It has also occurred with Maduro and Venezuela. In each of these cases, Trump, in my opinion, has made things worse and, overall, may be taken advantage of by foreign leaders. All these issues have yet to play out fully. However, it is difficult to see how he can make things better, other than through using his selling/propaganda skills to convince his base that he has won and is making American great again.

Finally, it is likely that the desire for a strong-man and for some significant degree of authoritarianism may fit with Trump’s white base. Yet, the conservative current in the Republican Party, at least since Reagan if not before, has largely centered on the unfettered market as well as cultural conservatism. The Republican Party/Trump tax cuts and the Trump administration’s extensive deregulation push has satisfied the corporate/wealth heart of the Republican Party. Thus, Trump’s political strength rests in a white, perhaps evangelical-led, authoritarianism coupled with significant payoffs to the interests of corporations and their senior executives and shareholders.

[*The feeling thermometer has people place themselves on a scale of 0 to 100, where 0 is most negative and 100 is most positive and 50 is neutral.]