Journal of Political & Military Sociology

The White Working Class, Authoritarianism, and Unions

Gregg Robinson

Grossmont College

This article explores the relationship between authoritarianism, union membership, and attitudes toward Donald Trump. Matched samples of both unionized and non-unionized white working-class persons in San Diego County were asked about their attitudes toward Trump and other candidates running for office in the 2016 presidential elections, along with questions that probed their levels of authoritarianism and ethnocentrism. The major finding of this research is that, while there were similar levels of authoritarianism in both the unionized and non-unionized samples, the social psychological tendency of authoritarianism only predicted support for Trump among the non-unionized sample. The reasons for this relationship are explored in more depth in this article.


Since Donald Trump’s presidential victory, the white working class (WWC) has become a major topic of interest. From Arlie Hochschild to Angus Deaton, from Justin Gest to J. D. Vance attention has focused on this shrinking but strategic group. Written off by some as a demographic irrelevancy or a racist backwater, the importance of this group in electing Trump sent reporters out to talk to the experts who have studied this class and the associated anger of this community. The question many in the media sought to answer was: why did members of this class vote as they did?

The answers that are given fall in two groups. On the one hand, there are those who argue that economic vulnerabilities (Deaton 2016; Leipziger 2016; Sides and Tesler 2016; for international comparisons see Cavaille, Gidron, and Hall 2016; Kriesi et al. 2012) drove white working-class anger and political choices. On the other hand, a second group argues that social psychological pathologies were the source of WWC attraction to authoritarian politicians like Trump (Blumgart 2016; Ehrenfreund 2016; Klinkner 2016; MacWilliams 2016; Meyerson 2016; Taub 2016; Tesler 2016; for an international perspective see Cornelis and Van Hiel 2015; Kaufman 2016; IPSOS/MORI 2016;). According to this view, economic problems only “activated” the psychological predispositions, such as authoritarianism and ethnocentrism (Merolla et al. 2011; Stenner 2005).

Of course, while economic and social psychological explanations are not mutually exclusive, the greatest support, both vocal and empirical, has come from the psychological approach (Drutman 2017; Mac Williams 2016; Meyerson 2016; Taub 2016; Tesler 2016). Following this line of research, this article focuses predominantly on the psychological argument, while concentrating on the importance of unions as mediators between social psychological pathologies and political outcomes. This approach draws on Putnam (2000) and particularly on Skocpol (2003, 2017; Skocpol and Williamson 2012) because of their concern over the disappearance of civic organizations as resource providers for both ideological formation and political mobilization. More specifically, it will be argued that the presence or absence of labor unions in the lives of voters in 2016 was a crucial ingredient in determining whether authoritarianism and ethnocentrism became the basis of the WWC’s support for Trump.

The present research is the continuation of work I began nearly a decade ago during the height of the 2008 foreclosure crisis when I examined the response of a WWC suburb in southern California to that crisis (Robinson 2013, 2014, 2015). While that study determined that many angry people were willing to blame banks and the corporate elite for their economic problems, no community groups with roots and resources in WWC communities existed to channel that anger into a movement.

Similarly, the current article also looks at similar samples of WWC southern Californians. I explore WWC attitudes toward the November 2016 presidential election. Specifically, matched samples of both unionized and nonunionized WWC members in San Diego County were asked about their attitudes toward Trump, as well as other presidential candidates, along with questions that probed their levels of authoritarianism and ethnocentrism. What I found was that, while there were similar levels of authoritarianism in both the unionized and nonunionized samples, this social psychological tendency only predicted support for Trump among the nonunionized sample. The reasons for this connection are explored in greater depth below.

Economic Explanations in the Literature

The parallels between recent political events in the world and those of the 1930s are both disconcerting and much commented upon in the media and scholarly literature (Judis 2016; Frank 2016). The recent economic meltdown that brought the world economic system to the brink of collapse in a manner similar to that of the Great Depression was followed by a paler version of the authoritarianisms of the 1930s. For many analysts, the key to understanding the growth in popularity of right-wing figures like Trump is to recognize the obvious connection to economic crises. Such scholars argue that the combination of globalization, followed by the 2008 economic crisis, produced certain pathologies, including anger and resentment, among the world’s working classes that led to support for authoritarian elites (Atkins 2016; Deaton 2016; Case and Deaton 2015; Edsall 2016; Friedman 2016; Guo 2016; Porter 2016; Mankiw 2017; Muro and Liu 2016).

Others like Thomas Frank (2016) and John Judis (2016) agree with those who maintain the economic origins of working-class anger, but they further stipulate that this anger was monopolized by the right wing because of the left’s abandonment of economic populism. It is argued, for example, that the Democratic party has turned away from what some saw as a sexist and racist WWC in order to focus on the needs of women, minorities, and gays (Klein 2017; Lilla 2017). Thus, when the economic crisis hit in 2008, the Democratic Party paid lip-service to the economic problems of the working classes, but its policies served the interests of Wall Street and of minorities. Thus, the forgotten WWC was ripe for the appeal of an authoritarian who attacked globalization and promised that manufacturing jobs would return..

Social Psychological Explanations

While such economic arguments have been offered to explain the WWC’s support of Trump, these arguments have been criticized by others for lacking specificity (MacWilliams 2016; Meyerson 2016; Serwer 2017). For example, some have maintained that the most economically destitute among the WWC were not those most likely to vote for Trump (Klinkner 2016; Serwer 2017; Tesler 2016). Instead, exit polls indicate that the determinative attitudes for Trump support were authoritarianism and ethnocentrism (Serwer 2017; MacWilliams 2016; Taub 2016). Specifically, these explanations of why these individuals voted for Trump highlight social psychological theories of authoritarianism and the closely connected concept of ethnocentrism. As early as 1950, Adorno and his team of researchers argued that personality disorders led individuals to be attracted to authoritarian leaders. Since then and over the last quarter century, Adorno’s research has been updated with researchers substituting a more cognitive psychological orientation for Adorno’s outdated Freudianism (e.g., Lakoff 2002, 2016), while others have focused on methodology (Altemeyer 1981; Hetherington and Weiler 2009; Stenner 2005). Specifically, scholars maintained that the central problem of Adorno’s traditional measures of authoritarianism was that contemporary publics could see through them. Recognizing that surveys were asking them about their racism and ethnocentrism, respondents lied (or manipulated their answers) to produce more socially acceptable responses. To get around this problem, a series of alternative questions and scales were developed that could predict authoritarianism (Feldman 2003; Stenner 2005). Four questions about child-rearing practices were found to be predictive of this tendency (Feldman 2003; MacWilliams 2016; Stenner 2005). Specifically, it was maintained that contrasting questions, such as whether a child should “have respect for elders or be encouraged to be independent” (more on these below), were clever ways of measuring authoritarianism without respondents realizing what was actually being studied (Hetherington and Weiler 2009; MacWilliams 2016; Stenner 2005).

Such contrasting questions and their measures have been employed by political scientists to explain Donald Trump’s popularity (MacWilliams 2016; Taub 2016). As was the case with Adorno’s study, when authoritarianism was associated with leaders like Hitler and Mussolini, the revised test for authoritarianism found similar support for Trump on the part of those with a psychological predisposition for authoritarianism.

A connection between the recent economic problems among groups like the WWC and increased authoritarianism was explained by some social psychologists as the result of “activation” (Stenner 2005). That is, psychological predispositions, such as authoritarianism, become activated by economic events which provide stressors that motivate the hunts for scapegoats that are typical of authoritarians. Scholars argued that this tendency was particularly true for “middle level” authoritarians—people who were not strongly authoritarian but were authoritarian enough so that under situations of social stress (unemployment, economic change, cultural transition, etc), their dormant authoritarian psychological tendency would become activated (Stenner 2005).

Such explanations, however, leave cultural explanations underdeveloped. They evince very little understanding of how people with authoritarian tendencies are influenced by the pre-established beliefs through which culture channels such psychological orientations (e.g., the antisemitism of Germany or the racism of the US). These earlier individualistic approaches to authoritarianism meant that these explanations were less sensitive to the ways in which cultural traditions of racism, sexism, and ethnocentrism offered their adherents avenues for self expression. This shortcoming has led to the emergence of a second set of social psychological explanations.

Probably the most extensive literature attempting to explain the rise of right-wing populism, including Trump’s popularity, comes from analysts of culture. Within cultural explanations, there exists both a right- and left-wing approach to studying the association of cultural values with the rise of the extreme right among the WWC. The right-wing variant is found in the work of Charles Murray (1984, 2012) and J. D. Vance (2016) who see the economic deterioration of the WWC, not as a cause of cultural change, but as a consequence of them. In this regard, these scholars emphasize the emergence of values in the 1960s that undermined religious traditionalism, male responsibility, and the work ethic so important to economic success. For these analysts, the Trump vote grew out of this cultural pathology. By contrast, the left-wing variant cultural explanation focuses on racism and the reaction to the deterioration of white skin privilege (Beauchamp 2016; Meyerson 2016; Noden 2016; Palumbo-Liu 2016; Serwer 2017). According to this approach, the main reason for the rise of Trump’s support and other forms of authoritarianism is the loss of WWC privilege in the face of struggles by women and other minorities for equal justice.

A social psychological variant of this cultural theme that is more sympathetic to the WWC is found in the work of Arlie Hochschild (2016) and Justin Gest (2016). In their analyses, WWC people struggle to understand the changes that have destroyed both the jobs they depend on and the communities in which they live. Out of this economically based dislocation emerges a deep sense of marginality and crushed hope that generates ethnocentrism, which both Hochschild and Gest explain in social psychological terms. These scholars stress that cultural ethnocentrism is the result of a social psychological politics of nostalgia and resentment. WWC members yearn for a return of factories, the communities they supported, and the ethnic-sexual traditionalism of the 1950’s. This, in turn, leaves WWC people open to appeals of politicians like Trump who promise to return society to that more favorable by-gone era.

The Political Science of Civic Organizations

While these social psychological theories are convincing, they often fail to recognize the importance of groups that mediate between such psychological and cultural dynamics, on the one hand, and their political expression, on the other. As my previous research illustrated, the importance of organizational membership (or, more accurately, the lack of such membership) was key to understanding how the WWC community I studied reacted to the foreclosure crisis (Robinson 2013, 2015). Drawing upon the work of political scientists who have studied the decline of civic organizations, I offer here another perspective on the WWC.

Robert Putnam’s Bowling Alone (2000) voiced concern about the impact that the disappearance of civic organizations (social clubs, political organizations, trade unions, etc.) would have on democracy. In order to participate effectively in the democratic process citizens must learn how to make speeches, debate opponents, run for office, and join with like-minded people to express their desires. Organizations such as social clubs, unions, and sports leagues, have been some of the most effective groups within which such learning takes place. Membership in these organizations, however, has declined dramatically since the end of the Second World War, and with it democratic participation has waned (Putnam 2000).

Theda Skocpol (2003) has argued that this decline has disproportionately affected members of the working and lower classes. She maintains that, while the decline in membership in civic organizations affects all classes, the impact has been greatest on the working class because the middle classes have compensated for such civic isolation with increased participation in professional (e.g., AMA, ABA) and cultural (e.g., NOW, NAACP, Sierra Club) organizations. No such compensating organizations exist among the lower classes, leaving workers and the poor as the most civically isolated groups in society.

More recently, Skocpol’s analysis of Trump’s victory likewise focuses on organizational resources (Marshall 2016). In her recent study, Skocpol pointed to the organizational infrastructure provided by moneyed elites (especially the Koch brothers) at the same time that organizational resources among Democrats (especially those provided by unions) have declined. This, she argued, helped to account for the victory of Trump in states like Michigan and Ohio. By contrast, organized labor’s decline has affected politics in two areas. First, the resources to turn out WWC voters have been cut and, second, this decline marginalized organized labor’s role in formulating a message that might have successfully reached working-class voters, particularly in regards to deindustrialization, trade, and jobs—key issues that influenced the WWC vote in the upper Midwest.

Skocpol’s insights seem essential to understanding the WWC’s authoritarian turn in the 2016 presidential election. Specifically, how did union membership affect the political decisions of the WWC? Clearly, a deeper understanding of labor unions’ effects on politics is needed.

Unions and Politics

Union membership has long been noted as a significant determinant of political activities, particularly among working-class people. From the money that unions provide to candidates to the person-power that unions mobilize to get out the vote (GOTV), from the impact that unions have on the political attitudes of their members and families to the willingness of those members to participate in politics, this set of organizations has been one of the few channels of working-class political influence (Clark 2009, 71–89; Delaney and Masters 1991; Skocpol 2003). Like Skocpol, most scholarly attention has been devoted to money and GOTV (Kramer 2017; Schlozman 2013; Yates 2009), but some scholars add specificity to Skocpol’s point about unions and the WWC vote (Clark 2009, 71–89; Masters and Delaney 2008). In this regard, unions have often engaged in internal educational campaigns to inform their members about candidates and issues and to illustrate how these affect working-class interests (Clark 2009, 71–89; Delaney et al. 1990). Unions send literature to members, hold forums, push candidates to address union issues, and even hold coffee klatches and home visits to inform and mobilize the rank-and-file (Clark 2009, 71–89; Delaney et al. 1990). Some research indicates that those with the greatest union involvement are most likely to be affected by these educational campaigns (Clark 2009, 71–89; Delaney et al. 1990).

In this article, I use the above research on unions to explore how union membership mediates between authoritarianism/ethnocentrism and Trump support. I look at whether contact with unions and their educational campaigns affected attitudes toward the 2016 presidential race and, particularly, toward Trump. I examine whether the degree of union involvement by workers increases the effectiveness of union influence.

An important point of clarification is in order, however. I am not arguing that research on civic groups provides an alternative to research on the psychological roots of Trump’s political appeal. Instead, I attempt to specify how such psychological dynamics are mediated by groups like unions. Some political commentators and academics have spent far too much time arguing over what the single most important determinant of the Trump election was (Douthat 2017; Klinkner 2016; Serwer 2017; Tesler 2016), while the truth probably involves some combination of factors. Thus, while psychological and economic forces are important in determining WWC attitudes toward Trump, I aim to expand our understanding by adding to it the mediation by unions.


Drawing on the above discussion of the literature, I argue that there are two sets of measurable variables one could use to understand how unions affected WWC people’s social psychological attitudes toward Trump. The first originates in studies of authoritarianism—particularly the child-rearing scale. Specifically, I wanted to know if and how unionization mediated between this social psychological measure of authoritarianism and WWC attitudes toward Trump.

For similar reasons I also chose to study the impact of ethnocentrism on WWC feelings towards the 2016 election. Asking respondents about their attitudes toward various ethnic outgroups (Latinos, Muslims, immigrants, etc.) was easy to measure, and these attitudes were also found to have been important in predicting orientations toward Trump in previous research (Klinkner 2016; Serwer 2017; Tesler 2016). Moreover, this set of measures also has the added benefit of providing access to comparable national data (American National Election Study) that has studied the impact of ethnocentrism on Trump support in an early 2016 survey. This comparison with a larger national survey also allowed me to differentiate union effects from the effects of the distinct socio-cultural environment of San Diego.

California is a “Fair Share” state, meaning that every worker in an organized workplace must pay union fees. Thus there is little or no difference in the cost between being a union member and a non-member. This, in combination with union efforts to enroll members, means that the vast majority of workers at unionized shops are formal members, which reduces the problem of self-selection into union membership. This also means that union participation in California ranges from the strongly committed who join voluntarily to the totally indifferent who merely pay the union fees they are required to pay. This, in turn, provided the opportunity to study the impact of the degree of union commitment on authoritarianism and ethnocentrism.

In the 2016 presidential election, polling data indicated that California would go strongly for Clinton. This, in combination with southern Californians’ more accepting attitudes toward immigration and minority groups, led me to predict that the San Diego general WWC sample would be less supportive of Trump than the American National Election sample (ANES) but still more likely to vote for him than for Clinton. This lower level of support in the San Diego WWC community would likewise be the result of lower levels of ethnocentrism. Finally, I also hypothesized that the best predictors of general San Diego WWC attitudes toward Trump would be the social psychological factors of authoritarianism and ethnic group hostility. Since I expected there would be less ethnocentrism in the San Diego sample, I thought the best predictor of its attitudes toward Trump would be the authoritarianism scale.

More specifically, the central hypotheses generated were:

1. Most unions in San Diego have large Latino memberships. This, in combination with union educational efforts, should reduce the willingness of union members to express ethnocentrism in contrast to both the general San Diego WWC and national (ANES) samples.

2. I was not certain whether there would be differences in measured authoritarianism between the San Diego general WWC sample and the union sample given the disguised nature of this measure. I estimated, however, that there would be no statistically significant differences between the two samples (the ANES did not ask these questions, so I could not compare it to the national sample).

3. More importantly, I believed that both ethnocentrism and authoritarianism would be less important in determining union members’ attitudes toward Trump than among either the general WWC San Diego sample or the general sample.

4. Finally and most importantly, I also hypothesized that those most involved in unions would have the lowest level of association between either authoritarianism or ethnocentrism, on the one hand, and support for Trump, on the other.


The study of these issues involved a survey of roughly 200 WWC general voters in San Diego and a similar survey of nearly 200 WWC members of trade unions in the same community. The telephone survey (conducted August–September 2016) consisted of a random sample of both landlines and cell phones drawn from voter rolls (obtained from a commercial voter file company that supplemented the Registrar of Voters names with additional demographic information).

An initial list of 2500 phone numbers of white Anglo voters who had less than a bachelor’s degree (the definition of WWC used in this study) and another 2500 phone numbers of similar WWC union members who, according to the voter file company, had these same characteristics. These two samples were matched for age, income, and zip code. Given previous experiences of conducting surveys using this data source, it was expected that less than ten percent of the names would be available and appropriate for interviews. In some of these cases the information provided by the voter file company regarding ethnicity, education, location, or union membership was inaccurate so those respondents were eliminated from the final pool. Thus, of the initial list of 2500 general WWC names, only 1846 were functional (due to wrong numbers, disconnected lines, no answer, etc.); this group was called on average eight times with only 942 live contacts made (largely due to answering machines). Of those, 740 respondents refused to participate, with 212 agreeing to be interviewed. While this response rate is generally seen as low, it is increasingly common in telephone surveys (Babbie 2007; Singleton and Straits 2005). The union sample had a similar, though somewhat lower, response rate: 1778 functional numbers, 811 live contacts, and 164 participants. After examining the sample more closely, I discovered that the voter file company’s list included a large number of retired union members. To compensate, a second sample was drawn with a list of 500 younger union members. This produced an additional group of 30 completed interviews, giving a final total of 194 union members.

With such a response rate, an obvious concern is non-response bias. That is, a risk exists that those with an orientation correlated to the areas of interest under consideration might be those most predisposed to either respond or refuse to do so. To check for non-response bias a number of evaluations were made that have been suggested in the literature (Babbie 2007; Groves 2006; Keeter et al. 2000; Singleton and Straits 2005). Late responders were compared to early responders (respondents reached after six to eight calls were compared to those called one or two times) in regards to their attitudes toward Trump, Muslims, and Latinos, as well as tendencies toward authoritarianism. The assumption was that late responders would be similar to those not responding at all. That is, those who were hardest to contact were likely to be similar to those not reached at all. This test found no measurable bias.

Also considered was the possibility that people who were most progressive in the first place were those most likely to be willing to be interviewed (due to hostility toward Trump, especially among the union members). This would be an important source of bias because it would put into question many of the conclusions of this research. The voter subsample provided a means to check for this bias. Specifically, if it were true that progressive respondents were more likely to be those participating in the survey, then one would expect a higher participation rate among registered Democrats than among registered Republicans (since the former were more likely to be progressive in the first place). I checked the samples against the voter sampling frame and found the opposite to be the case: There were more Republicans and Independent respondents in the voter rolls probably because of the tendency for older respondents to participate in both samples. While this was reassuring, regarding the possibility of selection bias among progressive respondents, it created a need to weight the sample in order to compensate for age and political party membership.

One last qualification should be made here. Because the San Diego samples were taken from voter rolls, it is likely that respondents were somewhat more politically engaged than members of the general population who did not bother to register to vote. While this fact should be kept in mind when comparing the national sample to the San Diego sample, it is less likely to affect comparisons between the two San Diego samples since both were drawn from voter rolls.


Dependent Variables

In order to have a national point of comparison, most of the questions in this survey were taken from the “Political Thermometer” section of the 2016 ANES. Respondents were asked to evaluate people and organizations on a 100-point scale ranging from 0 (feeling very cold or antagonistic) to 50 (feeling neutral) to 100 (feeling extremely warm or supportive). Questions asked in this manner included those about the major three candidates in the 2016 election (Trump, Clinton, and Sanders). A simple question about the respondents’ likelihood of voting for the two major candidates was also asked (“If the election were held today, who would you be most likely to vote for president?”). Respondents could choose from the following responses: Trump, Clinton, or Someone Else.

Independent Variables

The demographic variables used in this study included age, gender (dummy coded with male as the reference category in the multivariate analysis), and income (with categories ranging from less than $30,000 to more than $100,000). Ideological issues were operationalized by asking two questions. First, “No matter what your registration, do you think of yourself as closer to the values of the Republican Party or the Democratic Party?” For this question, responses were coded as either Democratic, Neither, or Republican. Second, respondents were also asked, “Would you consider yourself politically to be liberal, middle of the road, or conservative?” Responses were coded as strongly liberal, somewhat liberal, middle of the road, somewhat conservative, or very conservative.

A number of questions taken from the General Social Survey (2004) asked about civic organizational involvement. Only two regarding union membership are reported here, however. First, union respondents were asked about their degree of union involvement, “How would you describe your involvement in your union? Would you say that you were: Very active, Fairly active, Not that active, Not at all active.” Finally, respondents were also asked, “How frequently do you read information from your union sent to you by way of newsletters, emails, or other forms of communication? Frequently, Occasionally, Seldom, Never.”

In addition to questions measuring respondents’ civic organization involvement, questions involving race and ethnocentrism were also taken from the thermometer section of the ANES and asked respondents to describe their feelings about Muslims and Latinos. A question about attitudes toward immigrants was also created based on the same 100-point scale. It is not reported here both because it was not used in the ANES survey and because it was unrelated to any of my dependent variables.

Finally, the tendency toward authoritarianism was evaluated using the child-rearing scale mentioned previously developed by Stenner (2005). There are four sets of questions asking respondents to choose between two alternatives: whether they believe it is more important for a child to have “independence or respect for elders,” “obedience or self-reliance,” “curiosity or good manners,” and “being considerate or well-behaved.” Following Stenner, I valued each of the authoritarian responses (respect, obedience, good manners, and well-behaved) as “1” and the others as “0,” combined the scores, and divided by four. This produced a scale with a minimum of 0, a maximum of 1, a mean of .47, and an alpha of .6. The usual cut off for an acceptable alpha is .7 (DeVellis 2012), but previous research has reported similar coefficients and still found the scale to be a useful predictor of political choice and social attitudes (Stenner 2005).

Because the ANES did not ask the child-rearing questions, I looked for a question in that survey that was associated with authoritarianism. One of the questions from the thermometer section of the ANES that involved attitudes toward the police was found to significantly correlate with the child-rearing measure of authoritarianism (Pearson R=.12, p=.001; and Spearman’s Rho=.17, p=.000). While this correlation is not overwhelmingly strong, I report it here so as to have a national point of comparison.

It is worth saying a few words here about the southern California community from which these samples were drawn. California, in general, and San Diego, in particular, have seen major demographic changes over the last half century. From a largely white, conservative, and Republican region, they have become disproportionately Latino, immigrant, and Democratic. Thus, this is a community in which the ethnic changes so much associated with the cultural/racism explanation for the WWC’s support for Trump are most advanced. Moreover, because the ethnic composition of southern California is where the country as a whole is heading in the next half century, this study may provide a glimpse of future nation-wide ethnic relations. Indeed, the 2014 Census estimated that by 2020 the majority of children in the United States will be of Latino background, and by 2040 the country will follow California in becoming a majority minority nation (Colby and Ortman 2015; U.S. Census Bureau 2014).


The data analysis begins by comparing the San Diego samples to the ANES national sample in order to compare how the San Diego community compares to the larger population (see Table 1).

Table 1.  Feeling Thermometer: (0–100, 0 = Very Cold; 50 = Neutral; 100 = Very Warm)

Means for Politicians, Police, and Minorities

White Working Class (No College Degree)


Trump (Standard Error)

Clinton (Standard Error)

Sanders (Standard Error)

Police (Standard Error)

Latinos (Standard Error)

Muslims (Standard Error)

Authoritarianism (Standard Error)

N = 397








San Diego Non-Union Mean
N = 188








San Diego Union Members
N = 177








*From American National Election Study: Pilot January 2016 (SPSS weighted)

Table 1 highlights a number of striking similarities and differences between the groups. Several important similarities between the samples stand out. In all three, politicians were held in relatively low regard, with none breaking the 50/neutral barrier. It is also interesting that Bernie Sanders was seen more favorably than Trump in all three samples. Clinton, by contrast, was the least positively evaluated candidate in both the national and San Diego general samples (but not in the sample of union members). Finally, of the groups evaluated in all three samples, the police were the most positively evaluated, with Muslims receiving the most negative evaluations.

Table 1 also illustrates points of divergence between the three samples. For example, San Diego respondents were much more positive about both Latinos and Muslims than respondents at the national level. In both cases there was over a 15-point difference in attitudes towards Latinos and Muslims between the ANES and San Diego samples. Regular contact with large numbers of people from both groups (San Diego not only has a large Latino population but a significant Muslim one as well, particularly in the working-class areas from where our sample was drawn) seems to have dampened ethnic hostility (or, at least, the willingness to express it). It is also striking that there was very little difference between the two San Diego samples in regards to attitudes toward the police, Latinos, Muslims, or authoritarianism. These last points should be stressed since they contradict one of my initial hypotheses. Indeed, union membership seems to have made no difference in regards to tendencies toward authoritarianism or ethnocentrism, at least as measured by these questions. The “liberalizing” impact of union membership that I had predicted was not supported by this data.

By contrast, the liberalizing effect of union membership was evident in attitudes toward Trump and Clinton. The union sample evinced more negative attitudes toward the former and more positive ones towards the latter than either of the other samples. This was particularly striking in the evaluation of Trump: Union members differed from both other WWC San Diegans and ANES respondents by at least 8 points (statistically significant). For union members, Trump was the least positively evaluated politician on this list, while for the other two samples he was close to being the most positively evaluated. It is also important to emphasize the significance of union impact on evaluations of Clinton. She was by far the most negatively evaluated of the politicians by the non-union samples, but not by union members.

Attitudes toward Trump and Clinton are explored in more depth in Table 2, where sample members were asked which of these candidates they would vote for in the November election. What is striking about this table is that, unlike Table 1, the differences in intent to vote for Trump were nearly indistinguishable between the national, non-union, and unionized samples (within the margin of error for union versus non-union San Diego samples). On the other hand, the intent to vote for Clinton by unionized members was dramatically greater than was found among the other two samples. In addition, this intent among unionized respondents was also greater than the positive feelings about her found among these same respondents in Table 1. There is obviously a group of Trump supporters that does not vary much from sample to sample. What changes with WWC unionized voters is that there were far fewer third-party voters among them—that is, people who were either less ideologically aligned or more dismayed with the two main candidates. Among unionized voters this ambiguity was resolved: Clinton was their candidate. While the popularity of Clinton among unionized voters in Table 1 was hardly enthusiastic, many of these less-than-enthusiastic unionized voters still indicated they would vote for her (see Table 2). Union contact seems to have resolved this ambiguity in Clinton’s favor. They put aside their concerns and decided to cast their vote for her.

Table 2.  If the Election Were Held Today, Who Would You Vote For?
White Working-Class Voters



San Diego Non-Union

San Diego Union













Don’t Know




Not Vote








Table 3 supports this interpretation of unionization’s impact on ambivalent voters as the effects of unionization on the choice of candidates are examined more closely. Here, unionized members’ choices are compared on the basis of voters’ degree of involvement in unions. Recall that one of my original hypotheses was that those most closely tied to unions would be most likely to vote for Clinton rather than Trump. The data in Table 3 seems to reject this hypothesis. There is no statistically significant association between level of union involvement and presidential choice in this data. Specifically, no statistically significant differences exist in voter choice between “very active,” “fairly active,” and “not that active” union members. This same lack of connection also applies to those who were most likely to read union information (data not presented). That is, there was no association between likelihood of reading union information and political choice.

Table 3.  Degree of Union Involvement and Candidate Choice


Very Active

Fairly Active

Not That Active

Not At All Active
















Don’t Know









There is, however, one qualification to this finding. If the comparison had been one between “Not at All Active” and all other union members, the difference would have been statistically significant. It should be kept in mind that over three-quarters of the non-active union members were inactive due to retirement. They represented individuals who had lost contact with their unions because they were no longer in the workforce. It should also be pointed out that these respondents were obviously older than current union members, which would have predisposed them to more conservative political choices.

Thus, the impact of union membership on political choice was more of a nominal than an ordinal relationship—a milieu effect. That is, it was not just the activists who were affected by the lobbying efforts of their union for a particular candidate, but those who had any association with their union as well. In short, it appears that there was something about having a union present in the workplace that helped convince and mobilize ambivalent non-activists and non-joiners. While I will speculate more about this in the conclusion, the importance of this connection should be noted here.

Multivariate Analysis

Table 4 regresses attitudes toward Trump on various predictor variables for the national, San Diego non-union, and unionized samples. The first three columns of this table have the same variables regressed on the dependent variable to preserve comparability with the ANES national sample. Because the ANES did not ask the child-oriented authoritarian scale questions, the last two columns of Table 4 repeat the analysis for the San Diego samples with this authoritarianism scale added.

Table 4.  Regression of Trump Thermometer Attitudes on Various Variables

B (Std Error) / R2

San Diego Non-Union
B (Std Error) / R2

San Diego Union Members
B (Std Error) / R2

San Diego Non-Union
B (Std Error) / R2

San Diego Union Members
B (Std Error) / R2








.15 (1.0)

–.08 (.09)

.20 (.15)

.16 (.16)

.27 (.18)


–3.4 (3.3)

–5.31 (3.42)

–4.9 (6.11)

–3.3 (6.02)

–3.78 (6.92)


–.03 (.06)

–1.3 (1.2)

–.12 (.33)

–.28 (1.64)

–.19 (.31)








–.40 (.06)/.19***

–.23 (.08)/.02**

–.11 (.14)

–.18 (.08)/.01*

–.12 (.20)


–.07 (.08)

–.19 (.13)

 .03 (.20)

.11 (.19)

.12 (.24)








–.19 (.06)/.02**

–.19 (.12)

–.30 (.12)/.09***

–.01 (.12)

–.40 (.14)/.10***


–.03 (.07)

–.18 (.08)

.19 (.13)

–.02 (.12)

.22 (.16)


2.90 (1.0)

11.02 (2.04)/.06**

7.63 (3.1)/.04*

 5.34 (3.9)

3.25 (5.02)

Party Orientation

3.00 (.90)/.09***

17.92 (2.66)/.42***

21.44 (3.84)/.44***

24.4 (2.7)/.45***

19.6 (5.2)/.45***








.30 (.06)/.04*

.37 (.12)/.02*

.19 (.13)

.30 (.13)/.02**

.17 (.15)





25.3 (7.8)/.08***

11.8 (8.30)

Total R2












* p≤0.5; **p≤.01; ***p≤.001

For the national sample, attitudes toward Clinton (negative), the police (positive), and party orientation (Republicans positive) predicted feelings about Trump, but attitudes toward Muslims were overwhelmingly most important. Over half of the R2 in this national equation was the result of this variable alone. True to the predictions of the cultural approach mentioned previously, voters nationwide seem to be driven by hostility towards an out-group. Second in importance were factors related to political beliefs: Party orientation was the second most important variable explaining attitudes toward Trump, which in combination with attitudes toward Clinton—another political variable—accounted for nearly a third of the R2 in this equation. Finally, the only authoritarianism-related question asked of ANES sample members—attitudes toward the police—was also significantly involved in Trump evaluations, though it contributed only a limited amount of R2.

In contrast to the ANES sample, for most southern Californian WWC respondents, party orientation, rather than hostility to Muslims, drove their feelings about Trump. Four-fifths of the R2 in this equation was accounted for by party orientation, while less than 10% originated from hostility towards Muslims. Voting for a candidate affiliated with the Republican party was the most important factor predicting the WWC vote in San Diego. But even this striking finding minimizes the impact of political factors for this group. The second most powerful predictor of feelings about Trump was another political variable, ideology, with conservatism predicting about ten percent of R2. Combined, these two political factors accounted for ninety percent of the explanatory power of this data.

The importance of partisanship, however, should not distract us from recognizing the significance of the diminution of hostility toward Muslims as a predictor. While this factor was important, the coefficient was cut roughly in half and the size of the R2 was reduced by a factor of more than nine in comparison to the national data. Thus, for general San Diego WWC respondents, the atmosphere of southern California seems to have reduced the importance of ethnocentrism in accounting for their feelings about Trump in comparison to WWC voters in the rest of the country.

The only other important factor in this equation was the authoritarian-linked attitudes toward the police. The coefficient is roughly comparable to that of the national sample, though the R2 contributed is cut in half. This measure of authoritarianism is important both nationwide and among San Diegan WWC respondents since, as I indicated above, much research has shown that authoritarianism is a significant determinant of attitudes toward Donald Trump.

Turning to the union member data, we see that that sample is nearly identical to the non-union sample in terms of the importance of political variables. Party orientation and ideology follow the same pattern found in the non-union sample, with the former accounting for nearly three quarters of the R2 and the latter nearly ten percent. If the connection between attitudes toward Clinton and feelings about Trump are considered “political,” then all of the R2 in this table was determined by these political factors.

There are, however, two major contrasts between the national and general San Diego samples, on the one hand, and the unionized sample, on the other. First, and most importantly, attitudes toward Muslims were not significant predictors of the dependent variable for union members. By contrast, this was the most important determinant of attitudes toward Trump among members of the ANES national sample, and it was an important factor among the general San Diego sample. Thus, the ethnocentrism that has been so important in most analyses of attitudes toward Trump was unimportant in this union data.

A detailed examination of the relationship between union membership, hostility toward Muslims, and the Trump vote (data not presented) indicates that the impact of the former was through its effect on those that held middle level hostility toward this group. Union membership did not dramatically affect the likelihood of voting for Trump among either those strongly hostile to Muslims (they overwhelmingly voted for Trump whether members or not of unions) or those highly tolerant toward Muslims (who likely overwhelmingly voted against him). Instead, it was upon those who were in the middle of the scale that unions had their greatest impact. This group represented a plurality of the respondents—about 45% of each of the samples in this area.

The second contrast is that attitudes toward the police was also not a significant predictor for unionists. This is particularly important for this discussion given the centrality of authoritarianism to this research. Where such an authoritarian attitude toward the police was significantly related to attitudes in both the national and general San Diego samples, this attitude failed to obtain statistical significance for unionists. In spite of having similar attitudes toward the police in comparison to their fellow WWC San Diegans (see Table 1), this concern failed to impact unionists’ feelings toward Trump.

Let us consider authoritarianism in more depth. In the absence of comparative national data (since the ANES did not ask the child-rearing questions), the comparison with the general San Diego WWC community is instructive. With the addition of this variable to the model, a number of things change. First, the importance of ideology disappears as a predictor for both general and unionized samples, suggesting that authoritarianism was behind the explanatory power of this variable. While the change in the relationship between feelings toward Trump and attitudes towards Muslims is not as dramatic, it follows the same pattern as that of ideology: There is a diminution of the explanatory power of anti-Muslim attitudes in the data (although it still remains significant).

One of the most important findings (not only in this table, but in the entire study) is in regards to authoritarianism. Authoritarianism measures were the second-most important determinant of attitudes toward Trump among the general San Diego population. However, they had no impact among union members. Authoritarianism accounted for 15% of the R2 among general San Diego WWC respondents. By contrast, this set of authoritarian factors had no predictive power whatsoever among unionists. Despite the centrality of authoritarianism in predicting the outcome of this last election in other research, union members were apparently immune to its impact. Although they shared similar tendencies toward authoritarianism as other WWC San Diegans (see Table 1), somehow union membership inoculated them against it influencing their feelings about Donald Trump.

Similarly to ethnocentrism (attitudes toward Muslims), the relationship between union membership, authoritarianism, and the Trump vote (data not presented) shows that the impact of the former was through its effect on those in the middle level on this scale. Union membership did not dramatically affect the likelihood of voting for Trump among either strong authoritarians or strong non-authoritarians. Instead, it was on those who were in the middle of the scale that unions had their greatest impact. They also comprised the largest number of union respondents, with nearly three-fourths of these union members falling in the middle of this scale.

Union members’ attitudes toward Clinton were also much more important in determining feelings toward Trump than among respondents in the non-union San Diego sample. Whereas this variable had no impact upon the general WWC San Diegan population’s attitudes toward Trump, it contributed roughly one-fifth of R2 among union members. Union membership seems to have convinced WWC people that Clinton served their interests in a way that diminished their attraction to Trump. This was no small accomplishment given the general lack of enthusiasm for her even among union members.


Most of my expectations about the impact of union membership on attitudes toward Trump were supported, although some of the specifics turned out to be more complicated than predicted.

Hypothesis 1: My expectation that ethnocentrism among unionized sample members would be lower than among either the national or San Diego general WWC samples was only partially supported. While there was a difference in ethnocentrism among union members and the national sample, no such difference existed between the former group and other WWC San Diegans. Both San Diego samples were significantly less likely to express ethnocentrism toward either Muslims or Latinos than was the national sample. Union membership, however, played little to no role in this expressed tolerance. Instead, the key factor appears to have been the more diverse environment of southern California.

Hypothesis 2: Authoritarianism was no lower among WWC union members than among the general population of WWC San Diegans, confirming my speculation at the outset of this study.

Hypothesis 3: Neither ethnocentrism nor authoritarianism had any impact on union member WWC attitudes toward Trump, which was not true for the general WWC sample of San Diego. This was a key finding, and one which was particularly interesting, given the nearly identical degree of authoritarianism among both the general and unionized samples. It appears that union membership somehow short-circuited the connection that so many have found between psychological authoritarianism and political orientation.

Hypothesis 4: Unexpectedly, there was no relationship between degree of union involvement and attitudes toward Trump or the election more generally. This relationship turned out to be simpler than I had expected: It was all or nothing. Any contact at all with a union, rather than the degree of union involvement, had the predicted effect on political attitudes. Neither the self-described degree of involvement in the union nor the degree to which the respondent was exposed to union communications had any association with attitudes toward Trump.


When I was in the union, I’d get crap from them [about elections], and I guess I would read it. I had a friend, John, who was real into that stuff and he was hot on this and that, and so I went along with him. But I wasn’t that into it. I guess I kinda just went along cuz he was a drinkin’ buddy. But that was as close as I ever got to joinin’ any political thing.

This excerpt from an interview I conducted with a former union carpenter for my analysis of the impact of the foreclosure crisis on a WWC suburb in San Diego (Robinson 2013, 153), summarizes the findings of this article. Specifically, the interviewee displayed the ethnocentrism that social psychologists have found to be at the heart of the Trump appeal. However, while the worker was in a union, union membership had minimal impact on his political choices. Like so many members of the WWC, he was not anchored to an ideological approach while in the union. Indeed, in the interview, he admitted something he now found shameful: in the old days he used to vote Democratic. However, as he stressed, it was not because he really believed in any of the party “propaganda,” but because he did not know any better. He described himself as having been “naive” and overly influenced by the people with which he associated. Once he lost his union ties, he came “to his senses” and started voting “right” (in the dual sense of the word).

In short, the words of this interviewee, as well as the present investigation, reveal two things. First, psychology is not destiny; and, second and most importantly, studies of right-wing populism among the WWC have not paid enough attention to the role played by declining union membership. In this discussion I will draw out the implications of these two conclusions. First, I will address what the findings of my research mean for the study of the WWC and its tendency toward authoritarianism; and, second, I will address the the wider implications for American politics generally.

First, the importance of mediation by group membership on psychological predispositions must be stressed here. Previous research has noted this connection in regards to both African Americans and Latinos (Stenner 2005; Hetherington and Weiler 2009), two groups for whom authoritarianism is high but which has little impact on their political choices (they both vote overwhelmingly Democratic). This finding, which is very counterintuitive, has not been given the attention it deserves in the scholarly literature. The research presented in this article helps draw attention to such findings by widening the evidence base to include union membership. Groups like unions can either help neutralize the resentment associated with authoritarianism or channel it in other directions. Psychological predispositions do not automatically lead to political decisions; much depends on how they are mediated by groups and social networks. Indeed, even the most significant psychological traits can be successfully redirected or arrested given the right organizational environment. The fact that San Diegan WWC general and union samples are nearly identical in authoritarianism, but that support for Trump is only found to be the case for the former group, underlines the importance of such union mediation.

Additionally, the data also indicated that for both ethnocentrism (particularly anti-Muslim attitudes) and authoritarianism, union membership had an important impact on those in the middle of these scales. Those with strong feelings on either end of the scale were unaffected by union membership. Rather, it was members of the majority who were neither strongly Muslim-phobic nor authoritarian who were most open to the pressures that unions were able to wield. In short, it is this group of middle level authoritarians that has often been described as being “activated” by events in the economy or community (Hetherington and Weiler 2009; Stenner 2005). Despite this, however, the literature on “activation” has not taken the mediation by groups, such as unions, seriously enough. Despite the overwhelming attention paid to economic events and social-psychological orientations, such factors do not sufficiently explain political events, such as the rise of Trump. Thus, the literature on the decline of civic organizations is vital to understanding the present political conjuncture as well as the impact of authoritarianism and ethnocentrism on this conjuncture.

The current study also helps us clarify how unionization affects working-class political orientations. The fact that unionization affects non-activist political choices and minimizes the impact of social psychological predispositions, like authoritarianism and ethnocentrism, among those with both middle-level authoritarianism and almost any involvement in their union helps us better understand the mechanisms through which unions impact working-class political choices. Many people do not have strong political orientations or ideological commitments but instead have general emotional orientations toward politics. How these orientations are expressed in candidate choices or party support greatly depends on the social networks with which people associate. More specifically, while many of the union members in this research could have voted for Trump or a third party, the presence of a union served to orient them toward a candidate (Clinton) about whom they may not have been intrinsically enthusiastic. This makes sense: Committed Republican- or Democratic-oriented members of the WWC are unlikely to be swayed by even the most powerful of mediating organizations, while those without strong ideological orientations are, by contrast, those most likely to be affected, helping to explain why the degree of union involvement had little impact on political choice among members of this study’s unionized sample. In some ways those most affected by a union were those least involved in it because they were most often the people with the weakest political orientations.

On the other hand, the importance of partisanship in predicting political attitudes found in the San Diego data should give us pause. A significant portion of the WWC has become more partisan and moved towards an increasingly right-wing Republican party. It is unlikely that this particular group will move away from the far right. The insularity and self-reinforcing nature of modern politics, as well as the media environment that surrounds it, makes such a move unlikely. Indeed, while this move towards the right was originally thought to be the case for the WWC of the south (Bartels 2008; Achen and Bartels 2016), the tendency also appears to have crept northward. However, the phenomenon represents only a portion of the WWC, and as this study implies, such a move to the right is not inevitable. However, the likelihood of this occurring should be taken seriously. Thus, while many individuals in WWC communities could be attracted to a progressive agenda (as the popularity of Sanders indicates), time is not on the side of progressives.

Much has already been published (and much more will undoubtedly be written) about whether economic, cultural, or social psychological factors were the most important explanatory variables in Trump’s success with the WWC. The current research underlines the complexity of this discussion but adds to it the importance of mediation by unions and, by implication, other civic organizations. The bulk of the evidence, including some presented here, indicates that authoritarianism and ethnocentrism were particularly important in predicting who among the WWC would support Trump. However, these social psychological tendencies are not new. Why, then, is it the case that they are only now having an impact on politics? The concept of “activation” (Hetherington and Weiler 2009; Stenner 2005) is an only partially satisfying explanation. As I have argued, activation has recognized the importance of both economic events and psychological predispositions; however, it has not fully appreciated the role played by civic organizations (such as unions) on the Trump phenomenon. In short, the decline of unions, in conjunction with the economic and cultural shifts experienced, may help us understand why these psychological attitudes had the impact they did when they did. Thus, it is important to emphasize the significance of union mediation as a solution to growing authoritarianism. Importantly, we may not have to solve the problem of authoritarianism and ethnocentrism in order to sever their connection with political outcomes. To minimize the impact of such psychological tendencies, organizations, like unions, may be able to play a central role.

This leads to a wider and more speculative final point about US politics. Much of the debate within progressive circles has been about the direction the Democratic party should take in response to Trump’s election (Frank 2005, 2016; Kuttner 2016; Lakoff 2016; Levitz 2016; Skocpol 2017). Should the party wait for demographic changes that would cement its coalition of millennials, minorities, women, and the upper middle class or should it return to its working-class roots? Indeed, the debate rages between those who accuse Democrats of abandoning its base in the traditional working class in pursuit of Wall Street money (Frank 2016; Noden 2016; Kuttner 2016) and those who see a betrayal of minorities and women in any attempt to woo the WWC (Beauchamp 2016; Palumbo-Liu 2016; Serwer 2017). However a potential solution to this debate might be to bring a portion of the WWC into this coalition with minority groups. I contend that the road to such a coalition runs through the labor movement.

In conclusion, I do not believe that the data presented here apply only to southern California. While culturally, southern California is clearly more tolerant of minorities than many other parts of the country, this is only currently the case. As stated earlier, California, and especially southern California, is leading the way in the United States’ cultural transformation. The growth in the Latino population has turned a largely Republican state—one that gave us Nixon, Reagan, and Proposition 13—into one of the bluest states in the country. A similar Latinization is well underway across the country. In short, southern California is not so much different from the rest of the country but is, instead, leading it.

For this reason, the data presented here on the white working class—the group most often cited in Trump’s election—indicate that change is possible. With the country’s growing ethnic diversity, the ability to change the views of even a small minority of WWC voters could have a large impact. Progressives need not wait for demographic change to turn the tide in their favor; instead unionization may offer an answer. In short, progressives can minimize the harm by stopping the hemorrhaging of unions. Once accomplished, unions could then find ways to expand their organization in WWC communities. While it is not yet time to write off the WWC as a progressive voting bloc, the clock is ticking.


Achen, Christopher, and Larry Bartels. 2016. Democracy for Realists. Princeton, NJ: Princeton University Press.

Adorno, T. W., Else Frenkel Brunswick, Daniel J. Levinson, and R. Nevitt Sanford. 1950. The Authoritarian Personality. New York: Norton.

Altemeyer, Bob. 1981. Right-Wing Authoritarianism. Winnipeg: University of Manitoba Press.

Atkins, David. 2016. “Racism Alone Does Not Explain Trump Support “ American Prospect. November 4, 2016. Accessed April 5, 2018.

Babbie, Earl. 2007. The Practice of Social Research. 11th ed. Belmont, CA: Wadsworth Publishing.

Bartels, Larry M. 2008. Unequal Democracy. New York: Russell Sage.

Beauchamp, Zack. 2016. “White Riot: How Racism and Immigration Gave us Trump, Brexit, and a Whole New Kind of Politics.” Vox, June 20, 2016. Accessed April 5, 2018.

Blumgart, Jake. 2016: “Q & A: Is Fascism Back?” American Prospect. July 7, 2016. Accessed April 5, 2018.

Case, Anne, and Angus Deaton. 2015. “Rising Morbidity and Mortality in Midlife Among White Non-Hispanic Americans in the 21st Century.” Proceedings of the National Academy of Science 112 (49): 15078–15083.

Cavaille, Charlotte, Noam Gidron, Peter A. Hall. 2016. “Trumpism as a Transatlantic Phenomenon.” American Prospect. March 8, 2016. Accessed April 5, 2018.

Clark, Paul F. 2009. Building More Effective Unions. Ithaca, NY: Cornell University Press.

Colby, Sandra L., and Jennifer M. Ortman 2015. “Projections of the Size and Composition of the U.S. Population: 2014 to 2060 Population Estimates and Projections “ U.S. Census Current Population Reports.

Cornelis, Ilse, and Alain Van Hiel. 2015. “Extreme-Right Voting in Western Europe: The Role of Social-Cultural and Anti-egalitarian Attitudes.” Political Psychology 36 (6): 749–760.

Deaton, Angus. 2016.”Botched Globalization Led to Trump and Brexit.” Inquirer, September 10, 2016. Accessed April 5, 2018.

Delaney, John, and Marick Masters. 1991. “Unions and Political Action.” In The State of the Union, edited by George Strauss, Daniel G. Gallapher, and Jack Fioritio, 313–346. Madison, WI: Industrial Relations Research Association.

Delaney, John, Marick Masters, and Susan Schwochau. 1990. “Union Membership and Voting for COPE Endorsed Candidates.” Industrial and Labor Relations Review 43 (5): 621–635.

Douthat, Ross. 2017. “Race and Class and What Happened in 2016.” New York Times. November 29, 2017. Accessed April 6, 2018.

DeVellis, Robert F. 2012. Scale Development: Theory and Applications. Los Angeles: Sage Publications.

Drutman, Lee. 2017 “Political Divisions in 2016 and Beyond.” Voter Study Group. June 2017. Accessed April 6, 2018.

Edsall, Thomas. 2016. “Why Trump Now?” New York Times. March 1, 2016. Accessed April 6, 2018.

Ehrenfreund, Max. 2016 “A Strange but Accurate Predictor of Whether Someone Supports Donald Trump.” Washington Post. February 1, 2016. Retreived April 6, 2018.

Frank, Thomas. 2005. What’s the Matter with Kansas. New York: Holt.

———. 2016. Listen Liberal. New York: Metropolitan Books.

Friedman, Benjamin M. 2016. “On the Election III.” New York Review of Books, November 10, 2016. Accessed April 6, 2018.

Groves, Robert. M. 2006. “Nonresponse Rates and Nonresponse Bias in Household Surveys.” Public Opinion Quarterly 70 (5): 646–675.

Guo, Jeff. 2016. “Death Predicts Whether People Vote for Donald Trump.” Washington Post. March 4, 2016. Accessed April 6, 2018.

Feldman, Stanley. 2003. “Enforcing Social Conformity: A Theory of Authoritarianism.” Political Psychology 24 (1): 41–74.

Gest, Justin. 2016. The New Minority. New York: Oxford University Press.

Hetherington, Marc J., and Jonathan D. Weiler. 2009. Authoritarianism and Polarization in American Politics. New York: Cambridge University Press.

Hochschild. Arlie. 2016. Strangers in their Own Land. New York: The New Press.

IPSOS/MORI. 2016. “Immigration is Now the Top Issue for Voters in the EU Referendum.” June 16, 2016. Accessed April 6, 2018.

Judis, John B. 2016. The Populist Explosion: How the Great Recession Transformed American and European Politics. New York: Columbia Global Reports.

Kaufman, Eric. 2016. “Brexit Voters: NOT the Left Behind.” Fabian Society. June 24, 2016. Accessed April 6, 2018.

Keeter, Scott, Caroline Miller, Andrew Kohut, Robert Groves, and Stanley Presser. 2000. “Consequences of Reducing Nonresponse in a National Telephone Survey. Public Opinion Quarterly 64 (2): 125–148.

Klein, Ezra. 2017. “What Steve Bannon Gets Right About Democrats—and Wrong about Trump.” Vox, April 18, 2017. Accessed April 6, 2018.

Klinkner, Philip. 2016. “The Easiest Way to Guess if Someone Supports Trump? Ask if Obama is a Muslim.” Vox. June 2, 2016. Accessed April 6, 2018.

Kramer, Curtlyn. 2017. “Vital Stats: The Widening Gap between Corporate and Labor PAC Spending.” The Brooking Institute. Accessed April 6, 2018.

Kriesi, Hanspeter, Edgar Grande, and Martin Dolezal. 2012. Political Conflict in Western Europe. New York: Cambridge University Press.

Kuttner, Robert. 2016. “The Hidden Injuries of Class, Race, and Culture.” The American Prospect, October 3, 2016. Accessed April 6, 2018.

Lakoff, George. 2002. Moral Politics. Chicago: University of Chicago Press.

———. 2016. “Understanding Trump.” George Lakoff. July 23, 2016. Accessed April 6, 2018.

Leipziger, Danny. 2016. “Make Globalisation more Inclusive or Suffer the Consequences.” Vox, December 8, 2016. Accessed April 6, 2018.

Levitz, Eric. 2016. “Reminder: Clinton Did Almost Nothing to Court Whites in the Rust Belt.” New York Magazine, November 15, 2016. Accessed April 6, 2018.

Lilla, Mark. 2017. The Once and Future Liberal. New York: Harper Collins.

MacWilliams, Matthew. 2016. “The Best Predictor of Trump Support Isn’t Income, Education, or Age. It’s Authoritarianism.” Vox, February 23, 2016. Accessed April 6, 2018.

Mankiw, N. Gregory. 2017. “What the President Could Learn from Professional Economists.” New York Times, March 10, 2017. Accessed April 6, 2018.

Masters, Marick F., and John T. Delaney. 2008. “Organized Labor’s Political Scorecard.” In What Do Unions Do? A Twenty Year Perspective, edited by James T. Bennett and Bruce E. Kaufman, 492–519. New Brunswick, N.J. Transaction Publishers.

Meyerson, Harold. 2016. “Trump’s Appeal to the Forgotten Man.” American Prospect. August 4, 2016. Accessed April 6, 2018.

Merolla, Jennifer L., Jennifer M. Ramos, and Elizabeth J. Zechmeister. 2011. “Authoritarianism, Need for Closure, and Conditions of Threat.” In Extremism and the Psychology of Uncertainty, edited by Michael A. Hogg and Danielle L. Blaylock, 212–226. Hoboken, NJ: Wiley-Blackwell.

Muro, Mark, and Sifan Liu. 2016. “Another Clinton Trump Divide: High Output American vs. Low Output America” November 29, 2016. Accessed April 6, 2018.

Murray, Charles. 1984. Losing Ground. New York: Basic Books.

———. 2012. Coming Apart. New York: Crown Publishing.

Noden, Kirk. 2016. “Why Do White Working-Class People Vote Against their Interests? They Don’t.” The Nation, November 17, 2016. Accessed April 6, 2018.

Palumbo-Liu, David. 2016. “Progressives Should Not Cave to Anemic Liberals in the Identity Politics Debate.” Vox. December 7, 2016. Accessed April 6, 2018.

Porter, Eduardo. 2016. “Where Were Trump’s Votes? Where the Jobs Weren’t.” New York Times. December 13, 2016. Accessed April 6, 2018.

Putnam, Robert. D. 2000. Bowling Alone: The Collapse and Revival of American Community. New York: Simon and Schuster.

Robinson, J Gregg, 2013. “The White Working-Class and the Foreclosure Crisis: Tracing the Roots of a Failed Movement in Southern California.” Sociological Perspectives 56 (1): 131–159.

———. 2014. “Political Cynicism and the Foreclosure Crisis.” Social Justice 40(3): 99 –118.

———. 2015. “The Left and the Foreclosure Crisis: Roots, Resources, and Identity.” Critical Sociology 43 (2): 181–198.

Schlozman, Daniel. 2013. “The Alliance of U.S. Labor Unions and the Democratic Party.” Scholars Strategy Network. October 24, 2013. Accessed April 6, 2018.

Skocpol, Theda. 2003. Diminished Democracy: From Membership to Management in American Civic Life. Norman: University of Oklahoma Press.

Skocpol, Theda. 2017 “A Guide to Rebuilding the Democratic Party from the Ground Up.” Vox. January 5, 2017. Accessed April 6, 2018.

Marshall, Josh. 2016. “Skocpol Responds to Judis.” Talking Points Memo, November 11, 2016. Accessed April 6, 2018.

Skocpol, Theda, and Vanessa Williamson. 2012. The Tea Party and the Remaking of Republican Conservatism. New York: Oxford University Press.

Singleton, Royce, and Bruce Straits. 2005. Approaches to Social Research. 4th edition. New York: Oxford University Press.

Sides, John, and Michael Tesler. 2016. “How Political Science Helps Explain the Rise of Trump (Part 3): It’s the Economy, Stupid.” Washington Post, March 4, 2016. Accessed April 6, 2018.

Stenner, Karen. 2005. The Authoritarian Dynamic. New York: Cambridge University Press.

Serwer, Adam. 2017. “The Nationalist’s Delusion.” The Atlantic. November 20, 2017. Accessed April 6, 2018.

Taub, Amanda. 2016. “The Rise of American Authoritarianism.” Vox. March 1, 2016. Accessed April 6, 2018.

Tesler, Michael. 2016. “Trump is the First Modern Republican to Win the Nomination Based on Racial Prejudice,” August 1, 2016. Accessed April 6, 2018.

U.S. Census Bureau. 2014. “U.S. Population Projections: 2014–2060.” December 10, 2014. Accessed April 12, 2018.

Vance, J. D. 2016. Hillbilly Elegy: A Memoir of a Family and Culture in Crisis. New York: Harper Collins.

Yates, Michael D. 2009. Why Unions Matter. New York: Monthly Review Press.