![[PDF]](/pb-assets/Icons/adobe-pdf-icon-1498649896243.png){kind=icon}
Abstract
Recently released data from the 2016 American National Election Study allow us to offer a multifaceted profile of white voters who voted for Donald J. Trump in the 2016 presidential election. We find that Trump’s supporters voted for him mainly because they share his prejudices, not because they’re financially stressed. It’s true, as exit polls showed, that voters without four-year college degrees were likelier than average to support Trump. But millions of these voters—who are often stereotyped as “the white working class”—opposed Trump because they oppose his prejudices. These prejudices, meanwhile, have a definite structure, which we argue should be called authoritarian: negatively, they target minorities and women; and positively, they favor domineering and intolerant leaders who are uninhibited about their biases. Multivariate logistic regression shows that, once we take these biases into account, demographic factors (age, education, etc.) lose their explanatory power. The electorate, in short, is deeply divided. Nearly 75% of Trump supporters count themselves among his enthusiastic supporters, and even “mild” Trump voters are much closer in their attitudes to Trump’s enthusiasts than they are to non-Trump voters. Polarization is profound, and may be growing.
Introduction
Picture a mass rally with banners flying and placards held high. A chant rises from the crowd: “We want a strong, determined leader who will crush evil and take us back to our true path.” Nearby another massive rally is underway, with a different chant: “We would rather have respectful children than independent children.”
Which of these two rallies would be more alarming to civil libertarians? Which would be likelier to inspire talk of authoritarianism and hate? Curiously, for decades, two major national surveys—the American National Election Study and the General Social Survey—have attempted to gauge the danger of public authoritarianism by asking respondents to choose between traits in children. But the “authoritarianism” revealed in this way is tame, even innocuous. The premise is that attitudes toward children’s traits divide respondents into two camps, those who gravitate toward “authoritarian submissiveness” and those who retain a spirit of democratic independence (Feldman, 2003; Stenner, 2005). But the concept of authoritarianism, as originally defined (Adorno et al., 1950), involved far more than submissiveness alone. Anger, hatred, and intolerance were all central to the original notion—and to the traumatized century from which it sprang. In this century, society remains very much in thrall to anger and hate; and yet, more often than not, scholars have looked elsewhere to understand threats to liberty and equality.
In 2011, with this concern in mind, we asked the American National Election Study (ANES) to include a small scale that taps aggressiveness as much as submissiveness. This small scale, which we drew from a tradition of research on “Right-Wing Authoritarianism” (Altemeyer, 1981, 1988, 1996), was distilled from a field test of the full RWA scale in 2005. The ANES accepted our proposal and included the five-item RWA scale in the 2012–2013 Election Study,1 where it proved to have exceptionally strong predictive power with respect to anti-Black and homophobic prejudice (Smith et al., 2014, 2015). In 2016, the ANES again included several of our RWA items, two of which now comprise what we call our “Domineering Leader” scale.
Our starting point is the hypothesis that prejudice is fueled more by aggressiveness than by submissiveness, and that it is accompanied by the wish for a domineering leader who will punish the “undeserving.” This wish is clearly authoritarian in the original sense, but we give the notion of authoritarianism a fresh spin. In contrast to most of the established theories, we posit that people with authoritarian tendencies follow domineering leaders less for the pleasure of submission than for the pleasure of forcing moral outsiders to submit. Vicarious participation in the domination and punishment of out-groups is a core part of the authoritarian wish to follow a domineering leader. Hence, to activate this wish, leaders must be punitive and intolerant. Authoritarianism is not the wish to follow any and every authority but, rather, the wish to support a strong and determined authority who will “crush evil and take us back to our true path.” Authorities who reject intolerance are anathema, and must be punished themselves.
Voting With Attitude
The American National Election Study is an extraordinary source of insight. Since 1948, the ANES has surveyed representative samples of eligible voters in every national election. In 2016, as before, this survey was conducted both face-to-face and online, before the election and after. Our Domineering Leader scale was included, and proved to be highly predictive of support for Donald Trump. So too did measures of prejudice.
On one level this is a familiar result. The construct of authoritarianism was originally conceived as part of an attempt to explain prejudice (Adorno et al., 1950), and many studies, over three-quarters of a century, have showed a strong and consistent statistical association between authoritarianism (aggressiveness and submissiveness) and many forms of bias, from ethnocentrism to misogyny and homophobia (Smith, 2006).2 The 2016 ANES data conform to that pattern, albeit with new wrinkles. Explaining that finding is the mission of this article. But beyond the specifics, there is also a broader implication. It appears that authoritarianism and prejudice are not simply associated but are, in fact, two sides of the same coin. “Authoritarian aggressiveness” takes aim at despised groups, and “authoritarian submissiveness” prompts support for leaders who target those same groups. In other words, authoritarianism is support for intolerant leaders, because they are intolerant.
What we found in the 2016 ANES data is that authoritarianism in this sense was strongly associated with support for Donald Trump. We also found that the public is deeply divided (McAdam and Kloos, 2014), and that, along with polarization over race (Abramowitz, 2013, 2016), there are profound divisions over tolerance and governance more generally. It is often said that this polarization is a class phenomenon; that “the white working class” in particular resists tolerance and tolerant governance. But we find that the effects of class are complex—that pro-Trump voters hail in large numbers from many strata and milieus—and that the effects of class are mediated, in the large majority of instances, through biases and other attitudes.
In a sequel to the present article, “Nativism, Populism, and the White Working Class” (in Critical Sociology, 2018, forthcoming) we address this issue in greater detail, offering a close look, in particular, at the ethnographic literature on the “broke and patriotic,” “the politics of resentment,” and the interactive effects of class and prejudice more generally.3 But suffice to say, for now, that Trump’s white base is more readily found among voters who want domineering and intolerant leaders than among voters of any particular class background. Whether rich or poor, young or old, male or female, college or non-college educated, white voters supported Trump in 2016 when they shared his prejudices, and very seldom otherwise.
It is, of course, fundamentally important to understand why some social strata are likelier than others to hold these prejudices, and we will address those issues in our sequel to this article. But here the most relevant point is that voting in 2016 was more immediately and decisively associated with attitudes than with demographics.
We established this point statistically by multivariate logistic regression. The virtue of this procedure is that it allows us to assess the relative influence of many variables at once. Here, we examine 17 variables in all—five demographics (gender, education, age, marital status, and income) and a dozen attitudes. Our intent is to explain two outcomes: first, what divides Trump from non-Trump voters; and second, what divides strong from mild Trump voters. With that goal in mind, we examined the effects of all 17 of our variables on the voting decisions of the universe of white voters for which the 2016 ANES provides complete data—1883 voters in all. Of that total, Trump received 51.99% of the votes (i.e., 979),4 and 716 of his supporters (73.14%) voted for him enthusiastically.5
In the account that follows, we present our findings about Trump voters step-by-step, point-by-point. And for a complete picture of our findings, readers can consult our full statistical tables, including our regression tables, at the website of the Association for Critical Sociology, http://criticalsociology.org/the-anger-games-who-voted-for-donald-trump-in-the-2016-election-and-why/
Divided by Worldviews
Demography is not destiny. Many explanations of the 2016 election focus narrowly on demographic variables, but our research shows the limitations of that one-sided approach. The most influential explanation of this kind revolves around “the white working class”—an explanation which, tellingly, does not hold up well under close scrutiny.
A massive pre-election study (Rothwell and Diego-Rosell, 2016) showed that Trump supporters are relatively unlikely to be jobless or to work part time. They have above-average wages, low exposure to immigrant workers, and under 10% of them work in production.6 Of the total of white adults without bachelor’s degrees (135 million) in the United States, only 18.5 million are production workers.7 In the 2016 primaries, meanwhile, Trump’s voters averaged just three months less education than other Republican voters. They were better educated than the general public, and their incomes were $16,740 above the medians in their states (Manza and Crowley, 2017: 14 and 23–24). In the general election, moreover, Trump outpolled Clinton by a larger margin among voters with annual incomes from $70,000 to $120,000 than he did among any other income group.8
Our ANES findings add nuance to this account. Trump voters were not only reasonably secure economically; they felt reasonably secure. This was one of the few areas in which the attitudes of Trump and non-Trump voters did not differ significantly. To the extent that voters were financially insecure, they were insecure across party lines.
Almost every other attitude, however, does distinguish Trump from non-Trump voters. The larger story of the 2016 election is that attitudes came to the fore and eclipsed demographics. Of course, this is not to deny that education, gender, age, marital status and income matter greatly. Social statuses interact with attitudes in innumerable significant ways. But our ANES research shows that attitudes were the main dividing lines between Trump voters and other voters. The decisive reason that white, male, older and less educated voters were disproportionately pro-Trump is that they shared his prejudices and wanted domineering, aggressive leaders more often than other voters did. Why these prejudices and preferences are unevenly distributed remains to be explained. But what we know now is that these attitudes are found across the demographic spectrum, and that wherever they appear, they prompt support for Donald Trump.
We reached this conclusion by several steps. We began by examining demographics alone. We then included a single attitude; then two attitudes; and then a dozen. That exercise helped us better understand pro-Trump voting. We then performed the same exercise to see what differentiates enthusiastic Trump voters from his milder supporters.
In Tables 1a–1d, we look at all three groups of interest: strong Trump voters, mild Trump voters, and non-Trump voters. We see immediately that the demographic story is complex. With respect to gender, marital status and age, strong and mild Trump voters are much closer to each other than they are to non-Trump voters. But the same is not true for income and education.
|
Table 1a. Differences in white voters’ support for Trump in the 2016 election (N = 1883).
|
Table 1b. Prejudices among white voters in the 2016 election (N = 1883).
|
Table 1c. White voters’ attitudes toward authority in the 2016 election (N = 1883)
|
Table 1d. Other key attitudes among white voters in the 2016 election (N = 1883)
Mild Trump voters are much more affluent than strong Trump voters, with incomes just $1,000 below the median annual income of non-Trump voters.
When we turn to attitudes, however, we find that Trump voters close ranks. Strong and mild Trump voters are divided by a chasm from non-Trump voters. In Table 1b, we see just how sharply they differ with respect to five major prejudices.
In all cases, prejudices clearly divide Trump voters from non-Trump voters. The same is true when we inspect attitudes towards children and leaders.
Both of these measures sharply divide Trump from non-Trump voters.
We see again that Trump voters differ significantly from non-Trump voters in every case.9 Calculations of odds ratios show, further, that each 1-point increase in support for domineering leaders corresponds to a 16.2% increase in the chance of voting for Trump. Each 1-point increase in prejudice against women corresponds to a 23.5% increase in the likelihood of voting for Trump. For immigrants the equivalent figure is 22% and for African Americans the figure is 19.4%. Suffice to say that each of these attitudes matters greatly. In fact, the smallest difference between Trump and non-Trump voters with respect to attitudes is still bigger than the biggest difference within the Trump camp.10
Respectful Children and Domineering Leaders
When Bob Altemeyer of the University of Manitoba introduced the Right-Wing Authoritarianism (RWA) scale in the 1980s, the study of authoritarianism had been out of fashion for many years. Largely due to his efforts (Altemeyer, 1981, 1988, 1996), survey research on authoritarianism revived in the 1990s.11 But until the ANES adopted our short RWA scale in 2012, these surveys were restricted mainly to small convenience samples. Hence, what we are now learning about authoritarian attitudes in the ANES surveys is new and unique.
In 2016, we distilled our concern for Domineering Leaders into a scale with these two items: (1) “Our country will be great if we honor the ways of our forefathers, do what the authorities tell us to do, and get rid of the ‘rotten apples’ who are ruining everything”; and (2) “What our country really needs is a strong, determined leader who will crush evil and take us back to our true path.”
These items have often proven their effectiveness. But until now the power of these items at the level of national politics has been largely conjectural. Twenty-three years ago, Altemeyer said that if the public’s views about “crushing evil” matched those of his student samples, “tens of millions of North Americans” would agree that strong determined leaders should be given a free hand to crush evil (1994: 134). Now, we can confirm the truth of that guess. Nearly 63 million voters pulled the lever for Donald Trump in 2016, and they agreed with unusual fervor and frequency that evil and “rotten apples” should be smashed. That attitude sharply divided them from other voters.
Authoritarianism, in short, is a dividing line. But that fact has been obscured until now by the conventional wisdom about authoritarianism – namely, that authoritarianism is best measured by what we prefer to call the Child Trait scale. This scale asks respondents to choose between four pairs of traits in children: “(1) Independence or respect for elders? (2) Curiosity or good manners? (3) Obedience or self-reliance? (4) Being considerate or well-behaved?”
The ANES regards the underlined choices as classically authoritarian. Sharing that premise, a political scientist (Wood, 2017) has already announced that authoritarianism had little effect on the 2016 election.12 But that conclusion is undermined by the responses to our Domineering Leader items. When we originally proposed this scale to the ANES, we hypothesized that the wish for a leader who would “get rid” of troublemakers would sway right-wingers more than the wish for respectful and mannerly children. That thesis now has empirical support.
The Child Trait scale has real strengths. In 2012, it was modestly but significantly predictive of resentment toward African Americans. RWA was stronger, but the Child Trait scale showed genuine independent strength (Smith et al., 2014). The same has not proven to be true, however, with respect to Trump voting.
In the tables to follow, we display data on Trump voters in the columns that we have labeled Trial 1 and Trial 2. In Trial 1, we look only at gender, income, marital status, age, and education. What we learn here is that, when we leave attitudes aside, education is strongly associated with Trump voting and marital status has good-sized effects. (Age is also meaningfully associated with voting for Trump. Since age is counted year-after-year, small numbers can add up, as they do in this case.)
This proves that voters without bachelor’s degrees—and, to a lesser extent, older and married voters—were disproportionately represented among Trump voters.13
In Table 2b, we distinguish strong from mild Trump voters by the same method, and what we see closely resembles the pattern we observed in Table 2a.
|
Table 2a. Variables relevant to support for Donald Trump among white voters (N = 1883).
|
Table 2b. Variables relevant to strength of support among white Trump voters (N = 979).
This tells us that education is strongly associated with enthusiasm for Trump. The effect in Table 2b isn’t quite as strong or significant as it was in Table 2a, but we can say with confidence that strong Trump voters are even likelier than mild supporters to lack a college degree. We also see that the odds of voting enthusiastically for Trump decline modestly, but significantly, as income increases. These results confirm what we inferred earlier from the raw numbers—that mild Trump voters resemble non-Trump voters with respect to income and education.
Let’s return now to the second trials reported in Tables 2a and 2b. The point of these trials was to see whether preferences for certain childhood traits inspired pro-Trump voting. At first glance this seems to be true. When we factor in only Child Trait attitudes, we obtain a strong and significant coefficient (.277***) and the coefficient for education declines considerably in both strength and significance (from .834*** to .416**).14 The same pattern appears in Table 2b with respect to enthusiastic Trump voting, though to a lesser extent.
If we were to stop here, we might conclude that preferences regarding children’s traits had significantly influenced the outcome of the 2016 election. But in Tables 2c and 2d, we test that hypothesis by the simple expedient of examining one additional variable—attitudes toward domineering leaders. The result is that this variable, which we regard as authoritarian in the classical sense, now takes center stage. Child Trait preferences retain their statistical significance but they lose much of their predictive power—and demographics fade into the background.15
|
Table 2c. Variables relevant to support for Donald Trump among white voters (N = 1883).
|
Table 2d. Variables relevant to strength of support among white Trump voters (N = 979).
Support for Domineering Leaders
Ahead, we add a third column, Trial 3, which shows what happens when we place the Domineering Leader scale alongside the Child Trait scale. We look first at Trump voting, and then at enthusiastic Trump voting.
Plainly, the effect is huge. As we see in Trial 3, the inclusion of just two attitude variables steeply reduces the statistical significance of all demographics except marriage, and even this now has weak reliability. Education, which initially lost much of its strength when Child Trait attitudes were included, now falls to virtual inconsequence. Attitudes toward children and leaders both yield significant and highly reliable results, and the wish for domineering leaders now eclipses every other variable.
We see a similar story ahead. In Trial 4, from the universe of 979 white voters who voted for Trump in 2016, we see the influence of our variables on the chance that they voted for him enthusiastically.
What we see here is striking. The wish for a domineering leader now figures as the main dividing line between enthusiastic and less enthusiastic Trump voters, with undiminished statistical significance. Child trait preferences, in contrast, fall out of the picture.16
Later, when we factor in other attitudes, the larger story becomes more nuanced. But the essential points remain valid—that enthusiastic Trump voters are also enthusiastic about domineering leaders, and that they are not especially enthusiastic about respectful children.
Table 3a shows, in another way, that voters without bachelor’s degrees tend to support Trump because they want an authoritarian leader. Few observers, at this point, would doubt that Donald J. Trump sees the presidency as an imperial office. Table 3a suggests that his supporters agree, and that, among white voters in general, agreement of this kind is far more widespread among less educated voters than among college-educated voters.
|
Table 3a. Domineering leaders and education. Dispersion of attitude scores toward leaders by education, white voters only (N=1883).
We see here that the wish for a domineering leader attracts less educated voters more strongly than college-educated voters at every level. In the upper quartile of the distribution, in particular, less educated voters express a very high level of enthusiasm for strong leaders who will silence troublemakers and crush evil. Similar enthusiasm is found in the upper quartile of college-educated voters, but with an intensity no greater than that found at the 50th percentile among less educated voters.
We also find many college-educated voters who disavow the wish for an autocratic leader. Their median score (3.7) reflects more resistance to domineering leadership than support; the score at the 25th percentile reflects even stronger resistance.
At this point, skeptical readers might suspect that the Domineering Leader scale is so strongly associated with support for Donald Trump because the scale items so closely resemble his campaign rhetoric. But the evidence suggests, rather, that Trump owes his success to the fact that he tailored his rhetoric to the wishes of Republican voters. At the outset of the Republican primaries there were 17 candidates. None of the others sounded anything like Donald J. Trump. Not one promised to stifle dissent, crush evil, build walls, or ban Muslim immigrants with anything resembling Trump’s hubris. Republican voters, choosing between a paradigm-shifting candidate and a cast of extras, chose the bully in the china shop. The wish that fired that choice was a preference for a domineering and intolerant leader who would put their prejudices into practice.
The strength of the Domineering Leader items is not a new discovery. Altemeyer and others have successfully used these items for decades, and our analysis of the 2012 ANES data showed that these items were strongly associated with homophobia and resentment toward African Americans.17 Overall, attitudes in 2012 proved to be much stronger than demographics in predicting prejudice, and our 2012 five-item RWA scale demonstrated high strength and significance (Smith et al., 2015). A 1-point uptick in five-item RWA scale scores corresponded to 27.1% less support for same-sex marriages and 23.3% less support for same-sex adoptions, and similar results obtained for racial resentment.18
Our Domineering Leader scale, which has a powerful correlation of 0.891 with the five-item 2012 RWA scale, is similarly effective in predicting prejudice.19
Getting Rid of Rotten Apples and Crushing Evil
The Child Trait scale is intended to measure submissiveness. The Domineering Leader scale measures something different—the wish for a strong leader who will force others to submit. The premise is that evil is afoot; that money, the media and government authority—and even “politically correct” moral authority—have been usurped by undeserving interlopers. The desire for a domineering leader is the desire to see this evil crushed.
At issue here is not simply a preference for strong leadership. Asked whether Hillary Clinton “provides strong leadership,” 71.0% of the overall ANES sample agreed, compared to 62.7% who said the same thing about Trump. But when asked whether she “speaks her mind,” respondents agreed only 32.5% of the time, compared to a whopping 81.6% for Trump. Nor were they in doubt about what Trump thinks. Strength of opinion, and the willingness to express opinions strongly, evidently carried more weight than the perception of strength per se.
It is often assumed that voters who prefer authoritarian leaders simply want to submit. But often, what obsevers construe as submissiveness can be better understood as supportiveness; and the support that is offered is generally conditional. Only leaders who pledge zero tolerance for the undeserving are supported, and only as long as they live up to their promises. Authorities who tolerate usurpers are regarded as usurpers themselves.
Arlie Russell Hochschild (2016a) offers an incisive portrait of this way of thinking in her recent ethnography of Tea Party supporters in Louisiana. From interviews with 40 subjects, she distilled a vignette that she believes tells the “deep story” beneath the details. When her interview subjects were shown this vignette, they generally said, “Yes, that’s it exactly”; “You read my mind.”
Hochschild (2016b) calls this vignette “The Line-Cutters.” Tea Party supporters, she says, feel that people are “cutting in line” ahead of them—African Americans, career women, immigrants, and refugees—and to add insult to injury, they’re asked “to feel sorry” for the line-cutters, even when they “see President Barack Hussein Obama waving [them] forward. He’s on their side. In fact, isn’t he a line-cutter too?” Obama and his liberal allies “have removed the shame from taking. The government has become an instrument for redistributing your money to the underserving. It’s not your government anymore; it’s theirs.”20
This vignette captures the spirit of many survey statements that Trump voters endorse with enthusiasm. In 2006, researchers found that a short RWA scale that featured our “rotten apples” item explained significant variance in support for the war in Iraq.21 Two recent online studies found that short RWA scales correlate strongly with enthusiasm for Trump,22 and one of those studies found a strong association between RWA and this statement: “It takes a macho guy like Trump, who doesn’t let anyone push him around, to be President of the United States.”23 Another study showed that 72% of Trump supporters agree with the following statement: “Because things have gotten so far off track in this country, we need a leader who is willing to break some rules if that’s what it takes to set things right.”24 Interestingly, considerably lower levels of agreement with this statement were found among voters whom pundits often associate with Trump—white voters with high school degrees (62%), voters who fear being victims of terrorist attacks (57%), Republicans in general (57%), and voters with only some college (51%).25
Exit polls, meanwhile, showed that pro-Trump voters overwhelmingly wanted a president who would shake things up. Nearly 40% of the 24,558 voters who were polled in this way said that what they want most in a president is the ability to “bring change.” This answer was given nearly twice as often as the second most common response, experience (22%)—and so many Trump voters said “brings change” (85% of all Trump voters) that they accounted for nearly 32% of the entire electorate.26
Barack Obama had promised change, too. But what he proposed was not, in the apt words of Christopher Parker and Matt Barreto, change they could believe in.27
Attitudes Matter and Prejudices Matter Most
Respect for rule-breaking leaders who refuse to tolerate “line-cutting” often coincides with scorn for leaders who play by the rules and reward the allegedly undeserving—the lazy grasshoppers of Aesop’s fable, who cut in line ahead of the hard-working ants. It seems, in short, that the wish for a domineering leader is effectively a prejudice against other kinds of leaders; and that prejudice, in turn, often coincides with other familiar prejudices, as we see from the 2016 ANES.
We saw earlier that less educated white voters register consistently high scores on Domineering Leader items. Now, in the following table, we see a very similar pattern with respect to prejudices against women and minorities.
Less educated voters show more negativity toward every stigmatized group at every percentile. This matches the distribution we see displayed, in Table 3c, with respect to domineering leaders and African Americans:
|
Table 3b. Prejudices and education. Dispersion of scores on selected by education, white voters only (N=1883).
|
Table 3c. Authoritarianism and prejudice. Dispersion by education of key attitude scores, white voters only (N=1883).
The parallels here are almost uncanny. We see something similar, but with finer and more complete detail, when we examine all 17 variables at once.
Here, once all attitude variables are taken into account, not one demographic variable continues to show statistical strength, with the slight and insignificant exception of marital status. Child Trait preferences also fade into insignificance, and financial worries and non-fundamentalist religiosity play miniature roles.28
In all, eight attitudes predict Trump support: conservative identification; support for domineering leaders; fundamentalism; prejudice against immigrants, African Americans, Muslims, and women; and pessimism about the economy. These attitudes were revealed by responses to scales new and old. Among them were the well-known “racial resentment” scale, which explores attitudes toward African Americans, and similar scales compiled from newer items concerning immigrants and Muslims.
One of the small innovations in our statistical analysis was to merge four “hostile sexism” items, which appeared in the 2016 ANES survey for the first time, with three “modern sexism” items that had been used previously. Factor analysis showed that these items cohere, which enables us to probe sentiments toward women with a seven-item scale that predicts Trump voting with substantial strength and significance.29
Overall, what we see is that a spectrum of attitudes inspired pro-Trump voting, and that many of these attitudes are particularly common among older, less educated, and male voters. Central among these attitudes is the wish for domineering presidential action against line-cutters and rotten apples.
One apparent paradox in Table 4a deserves mention. Readers might wonder how Trump voters can be so pessimistic about the economy and yet show only average concern about their personal finances. The answer, it seems, lies in the well-known finding that pessimism about the economy reflects partisan biases more than personal concerns. Voters who dislike incumbent presidents tend to judge the economy harshly. Further, as Michael Tesler (2016) has shown for both 2012 and 2016, racial resentment plays an especially large role in negative views of the economy. A test with ANES data in 2012 and 2016 showed that pessimism about employment was strongly and significantly associated with racial resentment (.369*** and .416*** respectively). When Tesler controlled for party loyalty and ideology, he found that racial resentment alone accounted for a nearly 40% chasm between the economic views of racial liberals and conservatives.30
|
Table 4a. Trump voting, white voters only (N = 1883).
Similarly, Rothwell and Diego-Rosell (2016) found that negative views of the economy are common among older and other white voters not because they are suffering disproportionately economically but as a function of their politics.31
Table 4b gives us the big picture for enthusiastic Trump voters in the same way that Table 4a revealed the big picture for the Trump voters per se.
|
Table 4b. Variables relevant to strength of support among white Trump voters (N = 979).
We see here that five attitudes distinguish enthusiastic Trump voters from others. Four of the eight attitudes that distinguish Trump from non-Trump voters also distinguish strong from mild Trump voters: pessimism about the economy, support for domineering leaders, Christian fundamentalism, and anti-immigrant sentiment. And the difference makes a difference: the higher the score, the greater the chance of a pro-Trump vote. But it is also striking that conservative identification and negativity toward women do not differentiate strong from mild Trump voters.
It is also striking to note that ordinary racial resentment does not appear to divide enthusiastic from milder Trump supporters. Rather, a new anti-minority scale, which we constructed from items about perceived discrimination against whites, emerges as a pivotal dividing line.32 Strong Trump partisans are significantly more likely than others to allege anti-white discrimination. This reflects a heightened and, in tone, embittered complaint about what they see as line-cutting.33 The defiant wish for a domineering and impolitic leader, which is strongest among Trump’s most fervent supporters, coalesces here with the wish for a reversal of what his base perceives as an inverted moral and racial order.34
What Next?
Trump’s presidency is off to a rocky start, and some of his less dedicated supporters have begun to peel away. Pollsters have noted that several categories of Trump voters are showing less enthusiasm for him, including independents. But his core constituency remains loyal, so much so that pundits now routinely wonder: What, if anything, could Trump do that would alienate his base?
The implication of the evidence presented in this article is that Trump could lose the loyalty of his base if he were to lose his credibility as a domineering leader who insults and thwarts “line-cutters.”35 A failure of strength would resonate. But even that would prove only that Donald J. Trump himself was not the harsh and effective leader they had imagined. Trumpism could survive without Trump.
What, then, is Trumpism? Many liberals have hoped that 2016 was an aberration, that many middle-of-the-road voters cast their ballots for Trump without really sharing his views. Proponents of the “white working class” thesis, in particular, say that many Rust Belt factory workers, left behind by globalization, voted for Trump in the hope that he was sincere about his populist rhetoric.36 If that were true, then winning these voters back to the liberal fold would be fairly straightforward, since Trump’s fidelity to Wall Street would soon disillusion them. Other liberals have hoped that a good many voters, in the sluggish aftermath of the 2008 recession, were simply fearful about the uncertain future and willing to gamble on a risk-taker. Still others have hoped that hostility to Clinton and the Democratic leadership was the key to Trump’s appeal, and that this hostility can be overcome by well-crafted and appealing initiatives.
These scenarios strike us as unrealistic. Most Trump voters cast their ballots for him with their eyes open, not despite his prejudices but because of them. Their partisanship, whether positive (toward Trump and the Republicans) or negative (against Clinton and the Democrats), is intense. This partisanship is anchored in anger and resentment among mild as well as strong Trump voters.37 Anger, not fear, was the emotional key to the Tea Party (Banks, 2014),38 and that seems to be true for Trumpism as well. If so, the challenge for progressives is greater than many people have imagined. Hostility to minorities and women cannot be wished away; nor can the wish for domineering leaders. The anger games are far from over.
Acknowledgements
We presented an earlier version of this article to the Self & Society Symposium at the annual meeting of the American Sociological Association (ASA) in Montreal on August 11, 2017. We thank the participants in that symposium for their input and we thank Carmen Russell of the ASA for his efforts to publicize this article. We also owe debts of gratitude to David Fasenfest for his meticulous editorial attention; to our anonymous peer reviewers for their excellent suggestions; to Ian McGregor, Nicholas Valentino, Eric Kaufmann, and Daniel Luc Sullivan for sharing information with us; and to the University of Kansas, which gave David Smith a sabbatical in spring 2017 and, in 2014, supported our research on the 2012 election with a Summer Research Grant. Finally, we warmly thank the co-authors of our several conference papers on that earlier research (Shane Willson, Danny Alvord, and Mick McWilliams).
Statistical Tables
Many of the tables within this article present our findings in abbreviated form. To see these findings in full, see “The Anger Games” link at the Association for Critical Sociology website, http://criticalsociology.org/the-anger-games-who-voted-for-donald-trump-in-the-2016-election-and-why/
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Notes
1.
The five-item RWA scale that the American National Election Study included in the 2012–2013 survey consists of these five items: (1) There is no “ONE right way” to live life; everybody has to create their own way. (2) Our country needs free thinkers who will have the courage to defy traditional ways, even if this upsets many people. (3) Our country will be great if we honor the ways of our forefathers, do what the authorities tell us to do, and get rid of the “rotten apples” who are ruining everything. (4) What our country really needs is a strong, determined leader who will crush evil and take us back to our true path. (5) The “old-fashioned ways” and “old-fashioned values” still show the best way of life.
2.
The authoritarian syndrome is often said to consist of aggressiveness, submissiveness, and conventionalism, but we think the latter is better understood as a form of submissiveness (to rules as well as to rulers). In the original formulation by Adorno et al. (1950: 228), the word authoritarian was restricted to aggressiveness and submissiveness.
3.
We drew these phrases from the titles of recent books by Francesco Duina (2017) and Katherine Cramer (2016).
4.
These figures do not exactly match the exit polls, but they tell essentially the same story.
5.
We determined strength of support by means of ANES item V162035—POST: Preference strong for Pres cand for whom R voted.
6.
Jonathan Rothwell and Pablo Diego-Rosell (2016: 1, 14, 15, 18 and passim) examined zip code-level data. They found that, with respect to voters’ attitudes in 2016, employment was the 70th leading factor and holding a production job was 94th. More important than economic insecurity was the fact of living in white enclaves with thin populations, high levels of reliance on social security income, poor health (diabetes but not obesity), high but not drug-related mortality rates in middle age, and relatively low education levels.
7.
Another 17 million are small business owners. See the census figures in Moody (2017). It is also relevant (Bouie, 2016) that the non-college working class is now just 58% white.
8.
Trump’s second largest margin came among voters with incomes between $50,000 and $70,000 (Griffin and Teixeira, 2017). See also the related analysis by Mike Davis (2017).
9.
Since the small differences in pocketbook worries between Trump from non-Trump voters are not statistically significant we omit the raw numbers.
10.
All differences reported ahead are statistically significant, and most (including all Trump vs non-Trump differences) are significant at the p < .01 level.
11.
See, for an overview of the field, the many articles on RWA and related subjects that Political Psychology has published in the past two decades.
12.
Wood (2017), reporting ANES data, found that Trump voters had slightly lower Child Trait scores than other Republican voters in recent elections, but scored higher on racial resentment. A shift from the 50th to the 75th percentile on the Child Trait scale increased the chance of voting for Trump by 3%; the same shift on the racial resentment scale yielded roughly a 20% increase. See also MacWilliams (2016, 2017).
13.
Here the ANES data confirm one of the main results from the 2016 exit polls; namely, that less educated voters were disproportionately pro-Trump. Our findings clarify that point by showing that less educated white voters tended to support Trump because they tended to share his prejudices to an unusually high degree.
14.
In counterpoint, we see that the effect of Child Trait attitudes on marital status appears minimal.
15.
Our data do not permit clear insight into rural voters, but we now have the Survey of Rural America (2017), which shows that rural voters supported Trump at a level (54%) akin to the level among white voters, and that many of them are mild supporters. When asked whether immigrants strengthen or burden the US, 48% replied “strengthen” and 42% said “burden.” Roughly half said that discrimination against minorities is the main racial problem and the other half singled out alleged discrimination against whites. On the other hand, twice as many rural respondents (64%) say that irresponsible people get government help compared to those who say that needy people fail to get government help (32%).
16.
Stanley Feldman, one of the leading proponents of the view that “authoritarianism,” as measured by the Child Trait scale, consists mainly of submissiveness, focuses in a new paper (2017) on the punitiveness and intolerance that accompanies the wish for obedient children.
17.
The 2012 ANES survey obtained a representative national sample by multistage area-probability sampling. Of that sample a subset was recontacted via the internet in 2013. That subset replied to new items, including five we drew from Altemeyer’s 20-item RWA scale (two of which form our Domineering Leader items). Since we probed this data with an interest in bias, we restricted our analysis to white subjects for whom we had non-missing values for the core items about African Americans and to respondents who identified as heterosexual for the homophobia items.
18.
In 2012–2013 the ANES also included a four-item version of the Social Dominance Orientation scale that we had proposed. This scale worked well in our 2012-2013 analyses but was not included in the 2016 ANES.
19.
Thanks to Dr. Brock Ternes for running these numbers for us.
20.
A longer version of this story appears in Hochschild (2016a: 137–140). Here she adds “overpaid public sector workers” to the list of supposed line-cutters. It bears recalling, in this connection, that Trump voters have above-average income and employment and, also, above-average levels of disability assistance. Line cutting, it seems, is in the eye of the beholder.
21.
Skitka et al. (2006: 378–379). Besides the “rotten apples” item, Skitka’s team used RWA items that say the country must “smash the perversions eating away at our moral fiber”; “put tough leaders in power” to “silence troublemakers”; and seek “a mighty leader who will … destroy the radical new ways and sinfulness.”
22.
Choma and Hanoch (2016); Pettigrew (2017); and Van Assche and Pettigrew (2016).
24.
See Vasilopoulos et al. (2015) about the wish for a domineering leader in France in the wake of the terror attack on the satirical journal Charlie Hebdo.
25.
Jones et al. (2016: 23).
26.
See the exit polls reported by CBS online.
28.
Irreligiosity, in contrast, predicted anti-Trump views (Rothwell and Diego-Rosell, 2016).
29.
For insight into the role of sexism in the election, see Wayne et al. (2016) and Valentino et al. (2017). For a cogent analysis of the dimensionality of gender bias, see Begany and Milburn (2002).
30.
Tesler (2016: 56–57).
31.
Rothwell and Diego-Rosell (2016: 3). They cite Carol Graham, whose most recent study (2017) appeared subsequently.
32.
Pessimism about the economy divides Trump from non-Trump voters much more than it distinguishes strong from mild Trump supporters. This is understandable in the light of Tesler’s finding about the association between economic pessimism and racial resentment—which we also detected in the 2012 ANES data (Smith et al., 2015).
33.
One of the anti-immigrant items strongly suggests line-cutting as well—namely, that immigrants “take jobs away from people already here.”
34.
Martin Gilens (1999) offers an incisive account of the roots and branches of this worldview.
35.
Karina Korostelina (2017) shows effectively how important it is to Trump’s base to see him insult his opponents.
36.
This view underpins many of the contributions to Democrats and the White Working Class (Greenberg et al., 2017). See also the earlier work by Andrew Levison (2013).
37.
The strong vs mild distinction could be studied in many ways. Emily Ekins (2017) used proprietary longitudinal data to conduct a latent class analysis. She found that Trump voters subdivide into four large clusters and one smaller cluster. That finding would be worth pursuing further, along with other relevant findings (e.g., the fact that suburban voters in 2016 fell midway between rural and urban voters in the Survey of Rural America).
38.
This conclusion, which many others have reached as well (see, e.g., Vasilopoulos et al., 2015), is consistent with the findings of “approach/avoidance” research, which shows that enraged campaigners surge ahead, while fearful citizens, in contrast, prefer reticence and retreat.
References
|
Abramowitz, A (2013) The Polarized Public. Saddleback, NJ: Pearson. Google Scholar | |
|
Abramowitz, A (2016) Partisan nation. Conference paper, State of the Parties Conference, Bliss Institute, University of Akron, November. Available (accessed 4 September 2017) at: http://www.uakron.edu/dotAsset/e14a54b3-82e0-416a-aead-f386fc1a83f9.pdf Google Scholar | |
|
Adorno, T, Sanford, R, Frenkel-Brunswik, E. (1950) The Authoritarian Personality. New York, NY: Harper & Bros. Google Scholar | |
|
Altemeyer, B (1981) Right-Wing Authoritarianism. Winnipeg: University of Manitoba Press. Google Scholar | |
|
Altemeyer, B (1988) Enemies of Freedom. San Francisco, CA: Jossey-Bass. Google Scholar | |
|
Altemeyer, B (1994) Reducing prejudice in right-wing authoritarians. In: Zanna, M, Olson, J (eds) The Psychology of Prejudice, Vol. 7. Hillsdale, NJ: Erlbaum, 131–148. Google Scholar | |
|
Altemeyer, B (1996) The Authoritarian Specter. Cambridge, MA: Harvard University Press. Google Scholar | |
|
Banks, A (2014) Anger and Racial Politics. New York, NY: Cambridge University Press. Google Scholar | Crossref | |
|
Begany, J, Milburn, M (2002) Psychological predictors of sexual harassment. Psychology of Men & Masculinity 3(2): 119–126. Google Scholar | Crossref | |
|
Bouie, J (2016) What pundits keep getting wrong about Donald Trump and the working class. Slate, 5 May. Available (accessed 19 August 2017) at: http://www.slate.com/articles/news_and_politics/politics/2016/05/what_pundits_keep_getting_wrong_about_donald_trump_and_the_working_class.html Google Scholar | |
|
Choma, B, Hanoch, Y (2016) Cognitive ability and authoritarianism: Understanding support for Trump and Clinton. Personality and Individual Differences 106: 287–291. Google Scholar | Crossref | ISI | |
|
Cramer, K (2016) The Politics of Resentment: Rural Consciousness in Wisconsin and the Rise of Scott Walker. Chicago, IL: University of Chicago Press. Google Scholar | Crossref | |
|
Davis, M (2017) The Great God Trump and the White Working Class. Jacobin, 7 February. Available (accessed 11 October 2017) at: https://www.jacobinmag.com/2017/02/the-great-god-trump-and-the-white-working-class Google Scholar | |
|
Duina, F (2017) Broke and Patriotic: Why Poor Americans Love Their Country. Stanford, CA: Stanford University Press. Google Scholar | |
|
Ekins, E (2017) The Five Types of Trump Voters. Democracy Fund Voter Study Group, June. Available (accessed 21 June 2017) at: https://www.voterstudygroup.org/reports/2016-elections/the-five-types-trump-voters Google Scholar | |
|
Feldman, S (2003) Enforcing social conformity. Political Psychology 24(1): 41–74. Google Scholar | Crossref | ISI | |
|
Feldman, S (2017) Authoritarianism, threat, and intolerance. In: Borgida, E, Federico, C, Miller, J (eds) At the Forefront of Political Psychology. New York, NY: Routledge, in press. Available (accessed 26 August 2017) at: http://stanleyfeldman.site44.com/Feldman_Authoritarianism,%20Threat,%20and%20Intolerance.pdf Google Scholar | |
|
Gilens, M (1999) Why Americans Hate Welfare. Chicago, IL: The University of Chicago Press. Google Scholar | Crossref | |
|
Graham, C (2017) Happiness for All? Princeton, NJ: Princeton University Press. Google Scholar | |
|
Greenberg, S. (2017) Democrats and the White Working Class. Lexington, KY: Democratic Strategist Press, in collaboration with The American Prospect and The White Working Class Roundtable. Google Scholar | |
|
Griffin, R, Teixeira, R (2017) The Story of Trump’s Appeal. Democracy Fund Voter Study Group, June. Available (accessed 21 June 2017) at: https://www.voterstudygroup.org/reports/2016-elections/story-of-trumps-appeal Google Scholar | |
|
Hochschild, A (2016a) I spent 5 years with some of Trump’s biggest fans. Mother Jones, September–October. Available (accessed 19 August 2017) at: http://www.motherjones.com/politics/2016/08/trump-white-blue-collar-supporters/ Google Scholar | |
|
Hochschild, A (2016b) Strangers in Their Own Land. New York, NY: The New Press. Google Scholar | |
|
Jones, R, Cox, D, Dionne, E. (2016) How Immigration and Concerns About Cultural Changes are Shaping the 2016 Election. Washington, DC: Public Religion Research Institute, 23 June. Available (accessed 14 May 2017) at: https://www.prri.org/research/prri-brookings-poll-immigration-economy-trade-terrorism-presidential-race/ Google Scholar | |
|
Korostelina, K (2017) Trump Effect. New York, NY: Routledge. Google Scholar | |
|
Levison, A (2013) The White Working Class Today. Lexington, KY: Democratic Strategist Press. Google Scholar | |
|
MacWilliams, M (2016) The one weird trait that predicts whether you’re a Trump supporter. Politico, 17 January. Available (accessed 30 November 2016) at: http://www.politico.com/magazine/story/2016/01/donald-trump-2016-authoritarian-213533 Google Scholar | |
|
MacWilliams, M (2017) Intolerant and afraid: Authoritarians rise to Trump’s call. In: Fitzduff, M (ed.) Why Irrational Politics Appeals. New York, NY: Praeger, 121–137. Google Scholar | |
|
Manza, J, Crowley, N (2017) Working class hero? The Forum 15(1): 3–28. Google Scholar | Crossref | |
|
McAdam, D, Kloos, K (2014) Deeply Divided. New York, NY: Oxford University Press. Google Scholar | |
|
Moody, K (2017) Who put Trump in the White House? Against the Current, 186, January–February. Available (accessed 31 July 2017) at: https://www.solidarity-us.org/node/4859 Google Scholar | |
|
Parker, C, Barreto, M (2014) Change They Can’t Believe In. Princeton, NJ: Princeton University Press. Google Scholar | |
|
Pettigrew, T (2017) Social psychological perspectives on Trump supporters. Journal of Social and Political Psychology 5(1): 107–116. Google Scholar | Crossref | |
|
Rothwell, J, Diego-Rosell, P (2016) Explaining nationalist political views. Gallup Working Paper, 6 November. Available (accessed 6 May 2017) at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2822059 Google Scholar | |
|
Skitka, L, Bauman, C, Aramovich, N. (2006) Confrontational and preventative policy responses to terrorism. Basic and Applied Social Psychology 28(4): 375–384. Google Scholar | Crossref | ISI | |
|
Smith, D (2006) Authority fetishism and the Manichaean vision. In: Kalekin-Fishman D and Langman L, The Evolution of Alienation. Lanham, MD: Rowman & Littlefield, 91–114. Google Scholar | |
|
Smith, D, Hanley, E, McWilliams, R (2014) Personality and ideology. Conference paper, International Society for Political Psychology, July 5, Rome, Italy. Google Scholar | |
|
Smith, D, Hanley, E, Willson, S. (2015) Authoritarianism, social dominance, and generalized prejudice. American National Election Studies. Available (accessed 14 June 2015) at: http://www.electionstudies.org/onlinecommons/2016TimeSeries/Authoritarianism.pdf2015 Google Scholar | |
|
Stenner, K (2005) The Authoritarian Dynamic. New York, NY: Cambridge University Press. Google Scholar | Crossref | |
|
Survey of Rural America (2017) The Washington Post/Kaiser Family Foundation, June. Available (accessed 18 June 2017) at: http://ssrs.com/survey-rural-america/ Google Scholar | |
|
Tesler, M (2016) Post-Racial or Most Racial? Chicago, IL: The University of Chicago Press. Google Scholar | Crossref | |
|
Valentino, N, Wayne, C, Oceno, M (2017) Mobilizing sexism. Conference paper, American Political Science Association. Available (accessed 4 May 2017) at: https://drive.google.com/file/d/0B4H8rPzkMGIZSHlOMEMzenFuVlU/view Google Scholar | |
|
Van Assche, J, Pettigrew, T (2016) Comparative American and European far-right voters. Unpublished paper. Ghent University. Google Scholar | |
|
Vasilopoulos, P, Marcus, G, Foucault, M (2015) Emotional responses to the Charlie Hebdo attacks (21 November). Available (accessed 29 June 2016) at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2693952 Google Scholar | |
|
Wayne, C, Valentino, N, Oceno, M (2016) How sexism drives support for Donald Trump. The Washington Post, 23 October. Available (accessed 4 May 2017) at: https://www.washingtonpost.com/news/monkey-cage/wp/2016/10/23/how-sexism-drives-support-for-donald-trump/?utm_term=.3b0842858aa9 Google Scholar | |
|
Wood, T (2017) Racism motivated Trump voters more than authoritarianism. The Washington Post, 17 April. Available (accessed 27 April 2017) at: https://www.washingtonpost.com/news/monkey-cage/wp/2017/04/17/racism-motivated-trump-voters-more-than-authoritarianism-or-income-inequality/?utm_term=.2efc69ae6622 Google Scholar |