It’s About Hate: Approval of Donald Trump,Racism,Xenophobia and Support for Political Violence

Abstract

Is approval of Donald J. Trump associated with support for political violence? If so, what explains the link between Trump approval and political violence? Using an original, nationally representative survey of over 1,500 adults in the United States we produce two findings. First, individuals who express approval for Trump are also significantly more likely to endorse positive descriptors for the January 6, 2021 attack on the U.S. Capitol and are more likely express support for the use of political violence more broadly. Second, the effects of Trump approval on support for the use of political violence are mediated through racist and xenophobic attitudes. Trump supporters in the study disproportionately exhibit racist and xenophobic/anti-foreigner attitudes, and these attitudes are associated with a positive endorsement of both January 6 and the use of political violence.

Keywords

political violence January 6 Trump xenophobia racism

Are supporters of former U.S. President Donald J. Trump more likely to exhibit approval for political violence? If so, why might this be the case? We ask these questions because the Trump phenomenon – his campaign, his presidency, and the wider political culture he represents – has long been associated with violence and endorsement of violence. On the campaign trail, and then later when in office, Trump habitually employed norms-transgressing violent rhetoric and advocated the use of violence against critics and opponents (Cineas, 2021; Kelly, 2020). Researchers found that counties hosting Trump campaign rallies in the run-up to the 2016 U.S. Presidential election saw subsequent increases in hate-motivated incidents (Feinberg et al., 2019). Trump’s election in 2016 is empirically associated with a surge in hate crimes (Edwards & Rushin, 2018). Lipscomb (2017) reported that in the wake of the 2016 election, one-fourth of perpetrators of hate crimes in Florida indicated Trump and his policies and statements as motivators for their attacks. Nacos et al. (2020) found that violence and violent threats increased during Trump’s term in office, while Levine (2020) and Swaine and Adolphe (2019) identified over 50 cases of violence in which the perpetrators invoked Trump’s influence as an explanation or justification for their behaviors. The Trump years began and ended with two large-scale violent events associated with him and his political movement: the August 2017 “Unite the Right” rally in Charlottesville, Virginia that Trump refused to unequivocally condemn stating that there were, “good people on both sides; ” and the January 2021 attack on the U.S. Capitol which followed a political rally headlined by Trump urging participants to disrupt the certification of the 2020 U.S. Presidential election.¹

In this study, we examine whether approval of Donald Trump is empirically associated with endorsement of political violence and, if so, why this is the case. Specifically, we investigate whether individuals who approve of Trump are more likely to use positive words when describing the January 6, 2020 violent insurrection at U.S. Capitol and whether they are more likely to endorse the use of violence more generally to achieve political objectives or to send political messages. Using an original survey of over 1,500 adults in the U.S. we find that Trump approval is strongly associated with both of these measures of support for political violence.²

Moreover, we examine why Trump approval is linked with support for political violence. We argue that the relationship between Trump approval and political violence is mediated through increased racism and xenophobia. This is consistent with Pape’s (2021) research suggesting that individuals arrested for participation in the violence of the January 6 event may have been motivated by racial animus prompted by demographic changes in the United States. We find that individuals who approve of Trump are more likely to exhibit racist and xenophobic attitudes which facilitate their support for political violence. In the next section, we discuss the theoretical links between Trump support, racism and xenophobia and support for political violence.

Support for Political Violence in the United States

Recent public opinion polls show that public support for political violence is a salient issue in the United States. A December 2021 survey conducted by the Washington Post and University of Maryland found that 34% of Americans believed that taking violent action against the government is justified, while 62% said it was never justified. The survey showed a significant partisan divide. While 23% of Democrats agreed that violence against the government was justified, 40% of Republicans regarded political violence as justifiable (Balz et al., 2022). These patterns are roughly in line with previous survey findings (see, for example, Bright Line Watch, 2020; Cox, 2021; Diamond et al., 2020; Kalmoe & Mason, 2022). Diamond et al. (2020) note that the trend toward greater acceptance of political violence seems to be increasing in the United States. The American public also seems to think that political violence has become a part of the American political landscape. Agriesta (2021) found that 71% of surveyed Americans expected political violence to occur in the wake of future elections. In a poll on public civility, researchers found that the average American believed that the U.S. is on course to experience another civil war (Georgetown University Institute of Politics and Public Service, 2019). Finally, Kalmoe and Mason (2021) suggest that recent incidents of political violence in the U.S., such as the January 6, 2021 assault on the U.S. Capitol, may actually increase the American public’s tolerance or approval of political violence, suggesting that political violence and attitudes toward political violence are mutually reinforcing.

However, surprisingly few studies have investigated the factors that might contribute to Americans expressing support for political violence. There are a handful of notable exceptions. Using an experimental design, Kalmoe (2014) found that subjects were more likely to express support for political violence when exposed to violent rhetoric by politicians. Kalmoe and Mason (2019) determined that subjects exhibiting trait aggression and strong partisan identity and those that expect their party to be victorious in the next election are more likely to support political violence. Kalmoe and Mason (2022) also identify misogyny as a characteristic of respondents who are more likely to support political violence. Bartels (2020) found that political cynicism was associated with increased tolerance of political violence. However, crucially, Bartels also found that respondent level of ethnic antagonism – hostility toward non-White individuals – was a more substantive predictor of support for political violence. Finally, in a recently published study, Armaly et al. (2021) determined that individuals exhibiting “Christian Nationalism” – an ideology that merges Christian and nationalist American group identities – were more likely to tolerate the use of political violence, but that this effect was enhanced when interacted with individuals’ perceived victim status, conspiratorial thinking and White identity salience.

Moreover, no academic studies to our knowledge have considered the relationship between Trump approval and attitudes regarding political violence. This is an oversight given other polling results showing that Trump supporters may be more amenable to political violence. A December 2021 Washington Post/University of Maryland poll found that 43% of respondents who voted for Trump in the 2020 election believed violence against the government to be justified, as opposed to 40% of Republicans as a whole (Balz et al., 2022). Furthermore, polling by Bright Line Watch (2020) determined that willingness to endorse uncivil behavior, harassment and political violence are higher among Trump supporters relative to Trump critics, regardless of party affiliation.

Trump Approval, Racism/Xenophobia and Support for Political Violence

We argue that one important, but not singular, factor linking Trump approval and support for political violence is increased White racial animus and xenophobic attitudes. During his presidential campaign, and later when in office, Trump frequently made remarks denigrating Latinos, African Americans, Muslims, and immigrants (Lopez, 2020). As President, Trump inaugurated a partial ban on immigrants from certain Muslim countries and implemented harsh counter-immigration practices for Latin American migrants. He also disparaged the home countries of non-White immigrants, expressed a preference for immigrants from Northern European countries and castigated non-White members of Congress.

Several studies demonstrate that these racist and xenophobic utterances and policies were key to garnering the support of Trump's electoral base. For example, Schaffner et al., 2016 found that racist and sexist attitudes were the most important predictors of voter support for Trump in 2016, eclipsing economic grievances and mistrust of politics and politicians. Trump’s explicitly bigoted appeals were a particularly crucial aspect of his popularity with less educated White voters in the 2016 race (Glick et al., 2002). Sides et al. (2019) demonstrate that Trump supporters in 2016 were animated by racism against African Americans, Islamophobia, anti-immigrant xenophobia and a fear of “demographic displacement” by non-White individuals. In a longitudinal study comparing the attitudes and voting behaviors of a cohort of respondents in 2011 and 2016, Mason et al. (2021) found that respondents who expressed racist and xenophobic attitudes in 2011 were significantly more likely to have voted for Trump in 2016, but not other Republicans running for office. These findings are consistent with research in other democracies showing that xenophobia – particularly anti-immigrant attitudes – drive electoral support for extreme right populist political parties (Cutts et al., 2011; Semyonov et al., 2006).

Group Conflict Theory, Racism, Xenophobia and Support for Political Violence

We expect Trump approval, racism and xenophobia and support for political violence to be interconnected and reinforcing. A small handful of studies suggests to us that this is a sound expectation. In a survey fielded by the American Enterprise Institute, respondents who agreed that White individuals face higher levels of discrimination than non-White individuals in contemporary America were more likely to also believe that the use of violence is needed to preserve the traditional American values (Cox, 2021). Cox further explains that a core contemporary belief across right-wing individuals that advocate violence is that the social, cultural and political dominance of White Christian males is under assault by growing demographic diversity and that Donald Trump is working to reverse this trend. As previously mentioned, Bartels (2020) found that ethnic antagonism, specifically beliefs that by Whites that non-White individuals are gaining greater access to political power and government resources, is a driver of pro-political violence attitudes. At the same time, research by Newman et al. (2021) demonstrates that racist and xenophobic speech by politicians emboldens bigoted individuals to voice and act upon their prejudices. This suggests that in addition to attracting supporters, Trump’s racism and xenophobia empowered and licensed supporters to enhance their bigoted attitudes and behaviors. Newman et al. (2021) refer to this as “The Trump Effect.”

Our expectation that the racist and xenophobic attitudes that are associated with Trump support reinforce endorsement of political violence is consistent with theories explaining social group divisions, particularly in the context of external threats or competition over resources. Brewer (1999) explains that when groups in society face an external threat or come into competition with one another, individuals who are members of those groups begin to exhibit closer affinity for other ingroup members (“ingroup love”) and greater hostility towards outgroup members (“outgroup hate”). Ingroup love and outgroup hate work in tandem with one another, and exacerbate the process of dehumanization of outgroup members, an important preliminary step towards tolerance of violence (Hogg et al., 2017; Kunst et al., 2018; McCauley and Mosklenko 2008). Indeed, research of religious group conflicts within countries finds that percept of threat and competition – beliefs that rival communities pose a threat or are gaining more relative power – increases support for violent extremism and terrorism (Kunst et al., 2018; Obaidi et al., 2018; Thomsen et al., 2014) or an individual’s propensity for aggression (Merrilees et al., 2013).

Hypotheses

This discussion leads us to the hypotheses we test in the study. First, given the centrality of violence and violent rhetoric within the Trump phenomenon, we expect individuals who approve of Trump to be more likely to support political violence:

H1: Individuals who express greater approval of Trump are more likely to express greater support for political violence.

Second, because Trump supporters are more likely to exhibit racist and xenophobic attitudes, given the aforementioned group conflict theory literature we expect racism and xenophobia to mediate the effect of Trump approval on support for political violence:

H2: The relationship between Trump approval and support for political violence is mediated through racist and xenophobic attitudes.

Note that our hypotheses argue for an association between Trump approval, racism and xenophobia and support for political violence rather than a unidirectional causal relationship. This is because our theory allows for Trump approval to both predict and to be reinforced by both racial or xenophobic animus and support for political violence. We briefly discuss the implications of this decision in the conclusion.

Research Design

We test these hypotheses with a survey instrument we designed and fielded on an online panel of 2,500 U.S. respondents managed by Lucid Theorem.³ The Lucid Theorem panel provides a representative sample that is regarded as suitable for research (Coppock & McClellan, 2019) and has been used in previous academic public opinion studies (see, for example, Cassese et al., 2020; Miller, 2020; Motta, 2021; Sylvester, 2021). We fielded the survey between September 6 and 16, 2021. We eliminated 141 respondents who either did not finish the survey, were not located in the United States or were using a VPN that obscured their location or who failed an attention check embedded in the questionnaire.⁴ This left 1,909 total respondents. The median respondent took 14.1 minutes to answer the survey.

Dependent Variables

There are two outcome variables in the study. The first measures respondents’ perceptions regarding the January 6, 2021 violence at the United States Capitol building in Washington, D.C. To operationalize this variable, we used two survey questions to ask respondents to select words they would use to describe both the January 6 event and the people who participated in the January 6 event. In each question, respondents were presented with a list of four positive and four negative words and asked to select as many or few as they liked.⁵ Respondents were also provided with an open-ended response option to enter their own word or words to describe the January 6 event and participants that we subsequently coded as positive or negative.⁶ Respondents were also provided the option of not selecting any terms or could state that they were not familiar with the January 6 occurrence.⁷ We then created a net-positive indicator of terms for the January 6 violence by subtracting the sum of all negative terms from the sum of all positive terms identified by respondents. This produced a measure ranging from −10, for a respondent who selected 10 negative words and no positive words to describe the January 6 event and participants including open-ended responses, to 8, indicating a respondent that used 8 positive words and no negative to describe January 6.

Overall, respondents had a negative view of January 6. The mean respondent used, on balance, two negative words to describe the event and participants of January 6 while the median respondent used three negative words, on balance. Only around 22.9% of respondents used one or more net positive words to describe January 6. A frequency distribution of the first dependent variable – net positive description of the January 6 event – is presented in Figure 1.

Figure 1.

Frequency distribution of first dependent variable: net positive words used to describe January 6, 2021 Violence.

The second dependent variable measures respondents’ support for the use of political violence. Research by Westwood et al. (2022) demonstrates that previous public opinion studies elevate support for political violence among Americans because, in part, respondents have different interpretations of what political violence is. Following Westwood et al. (2022) we constructed our second dependent variable using a range of questions about respondents’ support for several different types of politically-motivated violent or destructive behaviors, including attacks on property, to capture all possible interpretations. This dependent variable is an additive index based on five survey questions asking respondents if they believe the use of violence is acceptable or is necessary to achieve political goals, to send a message to politicians and to solve problems in the country.⁸ We modeled three of these on questions developed by Kalmoe (2014) in his study of trait aggression and support for political violence. Responses to these questions were combined into an index ranging from 5–25,⁹ where 5 indicates rejection of political violence across the board and 25 indicates the highest level of support for political violence across the board.

A plurality of respondents, around 31.1%, had a score of 5 on the political violence index, suggesting that they unequivocally do not support political violence across all questions. Around 10% had a score of 15 or higher, suggesting that they were neutral to supportive of the use of political violence. The median score for respondents was 8. Figure 2 presents a frequency distribution of the second dependent variable, respondent support for political violence.

Figure 2.

Frequency distribution of second dependent variable: support for political violence.

Independent Variable

The main independent variable of the study is approval of former U.S. President Donald Trump. To measure Trump approval, we asked respondents if they approved or disapproved of the job that President Trump did when he was in office.¹⁰ A plurality, around 39.55%, strongly disapproved. Around 10.6% moderately disapproved. 10.3% neither approved nor disapproved. 18.9% and 20.7% of respondents either moderately or strongly approved. Overall, in the sample, 50.1% registered some level of disapproval while around 39.6% registered some level of approval for the performance of Trump while he was in office. We also asked respondents about their approval of current President Joe Biden and use this question in our analysis to provide a contrasting picture of the impact of Presidential performance attitudes on January 6 and political violence. Again, a plurality of respondents registered strong disapproval of Biden, around 32.1%, while 12.3% moderately disapproved. 13.6% neither approved nor disapproved while around 25% registered moderate approval and 17.1% strong approval. It is important to note that Trump approval and partisan identity are quite correlated but are not identical. Around 30.7% of respondents who either moderately or strongly approved of Trump did not identify as Republicans.¹¹

Presidential approval is the metric that is most frequently used to operationalize public support for presidents (Newport & Saad, 2021). In the survey we sought to capture respondents’ support/approval of both Trump and Biden and because Trump is a former president, we phrased the question as retrospective approval of his presidency. We have several reasons to believe Trump approval to be indicative of current attitudes about Trump. Research by Panagopoulos (2012) uses current and historical public opinion data to demonstrate that contemporary post-presidency approval ratings of former presidents are strong, positive predictors of mean and final approval of presidents when they were in office. This is particularly true of recent presidents. Public opinion data shows that the difference between current/post-presidential approval ratings of George Bush Sr., Bill Clinton and George W. Bush differed by only 3.3 points on average. We therefore expect that sentiments on the Trump presidency to not have changed much since his time in office, particularly given that only 8 months elapsed between the survey and the end of his administration. Moreover, after his presidency Trump remains very salient in contemporary American politics. He is a key opinion leader for many Americans, particularly Republicans and conservatives, and continues to both fundraise and hold political rallies ahead of a widely expected 2024 presidential run. He is highly sought after for campaign endorsements by Republican office seekers.

That said, Trump was a highly controversial president and remains a contentious political figure today. We account for the possibility that some respondents who may have supported Trump previously have changed their minds about him, possibly in response to the turbulent aftermath of the 2020 campaign. Therefore, in all models we control for whether a respondent is a former Trump supporter. This is operationalized as a dichotomous variable coded 1 for respondents who reported voting for Trump in the 2020 presidential election but who at the time of the survey moderately or strongly disapproved of his performance as president. Very few respondents fit into this category; only around 2.8%. Most respondents who expressed approval of the Trump presidency in the survey also reported voting for him in 2020, suggesting that he has retained his supporters in the post-presidential era.

Mediator

The mediating variable of our study measures respondents’ racist and xenophobic attitudes. We used two questions to assess racism,¹² with phrasing derived from Brigham (1993), and four to assess xenophobia or negative attitudes towards foreigners and immigrants from Saucier (2013). We combined responses to these into an additive index (α = .78) that ranges from 6–30, where higher scores indicated higher levels of racism/xenophobia. The median score for respondents was 16.

Controls

In the estimations, we also hold constant an array of demographic and attitudinal factors that may affect respondents’ attitudes toward the January 6^th event and the acceptability of political violence. These include the respondent’s age, gender, household income, employment status, education level, partisan affiliation, political ideology, racial identification, religion, religiosity, degree of engagement in political activities, media consumption and region of residence with the U.S. Following Kalmoe (2014) and Westwood et al. (2022), we also control for respondent trait aggression, using a measure derived from Bryant and Smith (2001). Control variables were derived both from Lucid Theorem demographic data on respondents as well as questions we crafted.

In the sample, respondent age ranges from 19 to 98. The median age was 47. Around 47% of respondents were male. The median household income for respondents was between $45,000 and $49,999 per year. Around 11.6% of respondents reported that they were unemployed and looking for work. The median respondent reported completing “some college” but did not have a degree, however around 42.5% reported having at least a Bachelor’s degree. Around 44.8% of respondents identified at Democrats, while 34.5% identified as Republicans. The remainder indicated that they were independents or had no party affiliation. Around 33.4% of respondents indicated that they were politically conservative overall while around 32.2 indicated they were liberal and 34.3 that they were moderate. Around 70.9% of respondents identified their race as White, non-Hispanic while around 10% identified as Black, non-Hispanic, 5.5% as Asian or Pacific Islander, 1.2% as Native American and 11.9% as Hispanic or Latino/Latinx.

63.6% of respondents indicated that they were Christian or a member of a Christian religious sect. To assess religiosity, we used three questions that asked respondents’ self-declared religiosity,¹³ frequency of attendance of religious services¹⁴ and prayer.¹⁵ We combined these to create an index ranging from zero to 12, where a zero indicates that the respondent identified themselves as not religious or spiritual and 12 indicates that respondent is both religious and regularly prays and attends worship services. Around 17.8% of respondents indicated that they were not religious at all. Around 6.2% indicated that they were religious or spiritual, but not observant. The remainder were religiously observant on some level. The median religiosity score for respondents was 6. Because politically conservative Christians are an important base for Trump, we interacted conservative political ideology with the Christian variable.

We also control for respondents’ political engagement and media consumption patterns as these might affect attitudes toward January 6 and political violence. Our expectation is that politically disengaged respondents who do not follow news are less likely to have strong, motivating attitudes about the January 6 event and contentious political behaviors, like political violence. To do this, we created a dichotomous measure coded 1 for respondents who reported that they voted in the 2020 U.S. Presidential election and who reported that they had contacted an elected official, urged another person to vote or had participated in a political meeting or protest at least once in the past 3 years. Around 37.1% of respondents had voted and reported being politically engaged. We also included a question measuring how often respondents reported following the news or current events.¹⁶ Around 53.8% of respondents reported that the followed the news at least once a day to multiple times per day. We also examined where respondents get their news and created two dichotomous measures each coded 1 for respondents reporting that they get most of their news from right-wing news sources¹⁷ and from social media.¹⁸

We also control for respondents’ level of trait aggression. This is crucial as individuals who are more aggressive might have distinct attitudes about January 6 and the acceptability of the use of political violence. To measure trait aggression, we used four questions also used by Kalmoe (2014) who derived them from Bryant and Smith (2001). These questions four questions are combined to create an additive index that ranges from 4, indicating no level of trait aggression, to 20, indicating high trait aggression.¹⁹ The median respondent scored 10 on this additive scale.

Finally, to account for possible regional variation within the sample, we included U.S. region dichotomous indicators for respondents from the U.S. South, West and Midwest. The Northeast is the reference category. Around 39.1% of our respondents hailed from Southern states. Around 18.4% were from the West and around 20.5% were from the Midwest.

Descriptive statistics for all variables used in the analyses are summarized in Table 1.

Table 1.

Descriptive Statistics.

Variable	Obs.	Mean	Std. Dev	Min	Max
Positive description of Jan. 6, 2021	1602	−1.998127	3.830985	−10	8
Support political violence	1889	9.266278	4.348745	5	25
Approval of Trump	1909	2.706129	1.618508	1	5
Former Trump supporter	1909	.0282871	.1658352	0	1
Approval of BIDEN	1909	2.826611	1.520235	1	5
Racism, Xenophobia	1884	16.17357	4.930114	6	30
Age	1909	47.30854	16.8913	18	98
Gender (Male=1)	1909	.4698795	.4992227	0	1
Household income	1862	9.341031	6.920383	1	24
Unemployed	1909	.1162913	.3206579	0	1
Education level	1894	4.568638	1.91948	1	8
Democrat	1909	.4484023	.4974609	0	1
Republican	1909	.3452069	.4755603	0	1
Conservative	1909	4.01781	1.734983	1	7
White, non-Hispanic	1909	.7092719	.4542173	0	1
African American, non-Hispanic	1909	.1000524	.3001485	0	1
Asian, Pacific islander	1909	.0550026	.2280451	0	1
Native American	1909	.0115244	.1067592	0	1
Hispanic	1909	.1178628	.32253	0	1
Christian	1909	.635935	.4812931	0	1
Religious	1568	7.591199	2.958008	2	12
Political engagement	1909	.3708748	.4831656	0	1
News frequency	1909	5.170246	1.671639	1	7
Social media	1908	.1954927	.3966834	0	1
Right-Wing media	1909	.0471451	.2120047	0	1
Aggression	1909	10.47197	3.255197	4	20
South	1909	.3907805	.4880532	0	1
West	1909	.1838659	.3874763	0	1
Midwest	1909	.2058669	.4044397	0	1

Results

To test the hypotheses, we conduct two sets of analyses.²⁰ For the first set of analyses, we examine the effects of respondent Trump approval on the two dependent variables. The results of these analyses are summarized in Table 2.

Table 2.

Effects of Trump Approval on Attitudes Toward January 6 and Political Violence.

	[1]	[2]	[3]	[4]	[4a]	[5]	[6]	[7]	[8]
	Positive Description ofJan 6, 2021	Support Political Violence	Positive Description of Jan 6, 2021	Support Political Violence	Support Political Violence – binary^a	Positive Description of Jan 6, 2021	Support Political Violence	Positive Description of Jan 6, 2021	Support Political Violence
Approval of Trump	1.534***	0.157***	1.241***	0.274***	0.231***			1.237***	0.270***
Approval of Trump	(0.044)	(0.025)	(0.064)	(0.038)	(0.046)			(0.066)	(0.048)
Former trump supporters^b			0.441	0.278	0.031			0.413	0.257
Former trump supporters^b			(0.450)	(0.274)	(0.343)			(0.541)	(0.366)
Approval of Biden						−0.585***	0.040
Approval of Biden						(0.077)	(0.041)
Democrat			−1.066***	0.010	−0.172	−1.110***	−0.190	−1.089***	0.014
Democrat			(0.215)	(0.122)	(0.152)	(0.248)	(0.128)	(0.251)	(0.129)
Republican			−0.634*	−0.565***	−0.632***	0.446*	−0.181	−0.613*	−0.563***
Republican			(0.247)	(0.140)	(0.172)	(0.262)	(0.135)	(0.240)	(0.143)
Conservative			0.328***	0.077	0.075	0.496***	0.137**	0.325***	0.082
Conservative			(0.081)	(0.048)	(0.058)	(0.088)	(0.047)	(0.088)	(0.054)
Christian			0.445	0.250	0.159	0.668	0.275	0.469	0.268
Christian			(0.387)	(0.236)	(0.285)	(0.426)	(0.238)	(0.346)	(0.229)
Conservative* Christian			−0.051	−0.064	−0.053	−0.143	−0.078	−0.061	−0.068
Conservative* Christian			(0.091)	(0.055)	(0.066)	(0.101)	(0.055)	(0.079)	(0.058)
Religious			0.007	0.007	−0.008	0.099***	0.022	0.009	0.006
Religious			(0.025)	(0.014)	(0.017)	(0.027)	(0.014)	(0.025)	(0.014)
Age			−0.007	−0.026***	−0.026***	−0.016**	−0.028***	−0.007	−0.026***
Age			(0.005)	(0.003)	(0.004)	(0.006)	(0.003)	(0.007)	(0.005)
Gender (Male = 1)			0.028	0.417***	0.429***	0.237	0.443***	0.018	0.418***
Gender (Male = 1)			(0.149)	(0.088)	(0.107)	(0.164)	(0.088)	(0.122)	(0.111)
Household income			0.021*	−0.018*	−0.017*	0.015	−0.019**	0.022	−0.018*
Household income			(0.012)	(0.007)	(0.009)	(0.013)	(0.007)	(0.013)	(0.008)
Unemployed			−0.312	−0.258*	−0.290*	−0.551*	−0.257*	−0.331	−0.257**
Unemployed			(0.242)	(0.134)	(0.167)	(0.267)	(0.135)	(0.222)	(0.096)
Education level			−0.060	0.012	0.019	−0.087*	0.009	−0.059	0.012
Education level			(0.043)	(0.025)	(0.031)	(0.047)	(0.025)	(0.042)	(0.025)
White, non-Hispanic			−0.211	−0.847**	−0.782*	−0.357	−0.835**	−0.214	−0.822*
White, non-Hispanic			(0.523)	(0.288)	(0.373)	(0.576)	(0.291)	(0.556)	(0.330)
African Am, non-Hispanic			0.145	−0.640*	−0.394	−0.120	−0.698*	0.144	−0.627*
African Am, non-Hispanic			(0.565)	(0.311)	(0.404)	(0.622)	(0.315)	(0.595)	(0.313)
Asian, Pacific islander			0.170	−0.372	−0.337	−0.132	−0.414	0.154	−0.365
Asian, Pacific islander			(0.551)	(0.318)	(0.406)	(0.606)	(0.320)	(0.529)	(0.333)
Native American			−0.500	−0.756*	−0.395	−0.314	−0.772*	−0.503	−0.702
Native American			(0.781)	(0.451)	(0.544)	(0.859)	(0.446)	(0.639)	(0.534)
Hispanic			0.148	−0.587*	−0.337	−0.103	−0.575*	0.154	−0.568*
Hispanic			(0.527)	(0.296)	(0.381)	(0.580)	(0.298)	(0.466)	(0.294)
Political engagement			0.220	0.361***	0.274*	0.204	0.328***	0.218*	0.354**
Political engagement			(0.157)	(0.094)	(0.115)	(0.173)	(0.094)	(0.109)	(0.104)
News frequency			−0.136**	−0.087**	−0.112**	−0.131*	−0.099***	−0.138*	−0.086**
News frequency			(0.051)	(0.029)	(0.035)	(0.056)	(0.029)	(0.061)	(0.032)
Right-Wing news			0.805*	−0.020	0.160	1.404***	0.226	0.818*	−0.023
Right-Wing news			(0.355)	(0.208)	(0.252)	(0.390)	(0.206)	(0.336)	(0.198)
Social media			−0.040	−0.034	−0.040	0.144	−0.025	−0.033	−0.048
Social media			(0.203)	(0.115)	(0.139)	(0.224)	(0.115)	(0.215)	(0.120)
Aggression			0.029	0.117***	0.107***	0.075**	0.123***	0.028	0.114***
Aggression			(0.023)	(0.014)	(0.017)	(0.026)	(0.014)	(0.021)	(0.016)
South			0.087	−0.124	−0.030	0.220	−0.076	0.090	−0.125
South			(0.194)	(0.115)	(0.141)	(0.213)	(0.115)	(0.142)	(0.134)
West			0.576*	0.086	0.090	0.765**	0.120	0.578**	0.082
West			(0.232)	(0.138)	(0.171)	(0.255)	(0.138)	(0.196)	(0.187)
Midwest			−0.036	−0.066	−0.104	0.204	−0.003	−0.028	−0.068
Midwest			(0.224)	(0.132)	(0.162)	(0.247)	(0.131)	(0.182)	(0.160)
Constant^c	−6.015***	−0.386***	-5.310***	−1.168**	0.530	−1.816*	−1.478***	−5.259***	−1.180**
Constant^c	(0.137)	(0.082)	(0.759)	(0.431)	(0.541)	(0.848)	(0.441)	(0.631)	(0.416)
Obs.	1602	1889	1553	1829	1829	1553	1829	1553	1829
F/LR χ²	1200.29***	38.57***	52.26***	448.72***	333.16***	34.19***	397.58***	50.04***	32.43***
R-Squared	0.4286	0.0042	0.4710	0.0508	0.1317	0.3589	0.0450	0.4694	0.2251
Model	OLS regression	Ordered logit	OLS regression	Ordered logit	Logit	OLS regression	Ordered logit	OLS regression	Ordered logit
Survey Weights Applied?	No	No	No	No	No	No	No	Yes	Yes

*p ≤ .1 **p ≤ .01 ***p ≤ .000.

Descriptions of dependent variables.

• Positive Description of Jan. 6, 2021 = Respondent used more positive terms to describe January 6, 2021 U.S. Capitol event and participants.

• Support Political Violence = Respondent expresses support/tolerance for use of political violence (believes violence is necessary to achieve political objectives, believes some politicians need to be “scared straight” through violence, believes worst politicians should ‘get a brick through window,’ believes problems with government should be ‘handled with bullets,’ believes should use violence to express political feelings.

^aDependent variable truncated into binary measure where 1 = above median support for political violence.

^bDichotomous variable coded 1 for respondents who reported voting for Trump in 2020 but who moderately or strongly disapproved of Trump’s performance as president when surveyed.

^cOnly first constant cut-point reported to save space in models 2, 4, 6, 8.

Because the dependent variables are a continuous measure with a normal distribution (positive description of January 6) and an ordinal scale measure (support for political violence) in our analyses we use ordinary least squares and ordered logistical regression analyses respectively for these analyses. To ease substantive interpretation of the impact of Trump approval on support for political violence, we also conduct a logistical regression analysis on a dichotomous version of the dependent variable that is coded 1 for respondents expressing an above median level of support for political violence (model 4a). Because we have a number of control variables, we also conduct naïve estimations that only include the independent variable as a check (see models 1 and 2). Though the sample itself is representative of the U.S. adult public, as a check we also built survey weights²¹ and ran the estimations applying these weights (see models 7 and 8). These reproduce the main findings of the study.

For the second set of analyses, we conduct tests of mediation using a statistical package produced by Hicks and Tingley (2011).²² The results of these tests are summarized in Figures 3 and 4. In these tests, we examine whether and to what extent racist and xenophobic attitudes mediate the effect of Trump approval on respondents’ positive assessment of January 6^th and support for political violence.

Figure 3.

Results, racism and xenophobia as mediator between trump approval and January 6.

Figure 4.

Results, racism and xenophobia as mediator between Trump approval and support for political violence.

The results presented Table 2 broadly support our first hypothesis. Across these tests, respondents who express higher levels of approval for Trump are more likely to use more positive words to describe the January 6 event and participants and are also more likely to express support for the use of political violence. To guard against the possibility of inflated standard errors due to multicollinearity with the control variables and predictors, models 1 and 2 present the results of the naïve estimations, sans controls. In both of these estimations, Trump approval positively predicts January 6 and political violence support, a finding we reproduce in all other models.

Models 3 and 4, and model 4a, contain all controls and are the main models of the first analysis. In model 3, Trump approval is found to predict the use of positive words to describe the January 6 event and participants. Substantively, for each ordinal level increase in Trump approval – from strongly disapprove to moderately disapprove, to neither approve nor disapprove, to approve to strongly approve – respondents used on average between 1.2 and 1.5 more positive words to describe the January 6 event and participants.²³ This is graphically represented in Figure 5.²⁴ In Figure 5, respondents who strongly or somewhat disapproved of Trump used a net of 4 and 3 negative words respectively to describe January 6, well below the median level for all respondents. Respondents who somewhat approved of Trump used about an equal number of negative and positive words to describe January 6. Respondents who strongly approved of Trump expressed one net positive term to describe January 6. Trump approvers, on average, use between 3 and 4 more positive terms, net, than Trump skeptics when depicting the January 6 events.

Figure 5.

Trump approval and positive description of Jan 6, 2021, Marginal Effects. Notes: Marginal effects simulations of effect of one-unit increase of independent variable on dependent variable. Based on Model 3 in main analysis. All covariates included in model.

In model 4, Trump approval positively predicts support for political violence. We find the same results in model 4a when we truncate the dependent variable into a dichotomous measure coded 1 for respondents exhibiting a higher than median level of support for political violence. Using model 4a, which simplifies presentation of the substantive results, we produced a first difference marginal effects simulation and graphed the results in Figure 6.

Figure 6.

Trump approval and support for political violence, marginal effects (using binary measure of political violence). Notes: Marginal effects simulations of effect of one-unit increase of independent variable on dependent variable. Dependent variable is binary, coded 1 for respondents expressing above median support for political violence. Based on Model 4a in main analysis. All covariates included in model.

In Figure 6, Trump skeptics are shown to express below median support for political violence while Trump approvers express above median support. Respondents who either somewhat or strongly disapproved of Trump had a .39 to .43 probability of expressing an above-median level of support for political violence. However, respondents who either somewhat or strongly approved of Trump had around a .54 to almost .58 probability of supporting political violence above the median level in the sample.

In models 5 and 6, we reran the estimations using Biden approval instead. In both of these estimations, Biden approval does not predict a positive assessment of January 6 or support for political violence. Indeed, respondents who approved of Biden used a net of .60 more negative words than positive words to describe January 6, on average. Biden supporters were no more or less likely to support political violence.

Finally, in models 7 and 8 we reran the main estimations using our survey weights. These reproduce, very closely, the findings of the main models. Trump approval positive predicts both positive descriptions of January 6 and support for political violence among respondents.

Findings for Controls

Several the controls are significant in the models, but not consistently so. The control for former Trump supporters is not significant in any model. However, when it is run without Trump approval, the main independent variable in the study, it becomes a significant negative predictor of positive description of January 6, but not for support for political violence.²⁵ This suggests that the small number of respondents who were former Trump supporters but now express reservations do not have a positive view of January 6 and its participants. This is a potentially interesting finding that might merit future investigation.²⁶ Conservative political ideology is significant and positive, but not consistently across models, while Republican party identification is mostly negatively signed and not significant. We suspect that this puzzling finding is possibly the result of multicollinearity. Trump approval, conservative ideology and Republican identification are all strong correlates of one another.²⁷ When each of these are run by themselves in the models, with the other controls, they are generally significant positive predictors.²⁸ This suggests that collinearity errors are suppressing significance and/or changing the signs of coefficients.

Younger and male subjects are more likely to positively describe January 6 and support political violence, though not consistently across all models, while education level is only significant, and negative, in one model. Respondents from higher income households were less likely to positively support political violence but were generally no more or less likely to describe January 6 in positive terms. However, unemployed respondents generally exhibited less support for political violence and January 6. The former finding is consistent with the literature linking economic deprivation with political violence while the later finding is consistent with other public opinion literature demonstrating that economically vulnerable people are less likely to express support for political violence (Blair et al., 2013). Racial or ethnic identification is not a consistent predictor across the models. Most of the respondent racial identification categories are negative signed but are not consistently significant. There is no indication that respondents of a particular race or ethnicity are more likely to positively describe January 6 or support political violence. Political engagement and trait aggression are significant and positive in most models, with some inconsistencies. Patterns of media consumption are decidedly mixed. Respondents exhibiting higher levels of news consumption are less likely to positively describe January 6 and endorse political violence. Respondents who rely upon right-wing media for their news are more likely to use positive terms to describe January 6 but are not more likely to support political violence. This is intuitive as right-wing media has played an important role in shaping a positive narrative about January 6 while also painting left-wing instances of political violence in a negative light. However, social media consumption is not significant across the board. Finally, Christian identification, the interaction between Christian and conservative, and the region dummies are mostly not significant in the estimations.

Results of Mediation Test

To test the second hypothesis, we evaluated racist and xenophobic attitudes as a potential mediator. The results of these tests are summarized in Figures 3 and 4. Results suggest that, as hypothesized, racism and xenophobia significantly mediate the relationship between Trump approval and attitudes about January 6 and political violence. The results show that respondents who approved of Trump were more likely to exhibit racist and xenophobic attitudes,²⁹ and those attitudes were statistically associated with a propensity to use positive words to describe the January 6 event and participants and to regard political violence as acceptable.³⁰ Around 24.5% of the effect of Trump approval on attitudes about January 6 is mediated through increased racism and xenophobia. Around 64.3% of the effect of Trump approval on support for political violence is mediated through racism and xenophobia. This provides support for hypothesis 2. However, the results provide evidence of partial mediation only. This suggests that other potential unobserved mediators are important components of the impact of Trump approval on both attitudes about January 6 and support for political violence.

Conclusion

In summary, we find that individuals who approve of former President Trump are more likely to endorse political violence, and to positively assess the events and participants of the January 6 event. We also find that Trump approvers are motivated by racial animus and mistrust and hatred towards immigrants and foreigners, and that these attitudes, in part, may make them more accepting of political violence. Our findings are robust to the inclusion of a number of significant controls and to different model specifications. They strongly suggest that an important component of the relationship between Trump approval and the normalization of political violence is, indeed, hate. Understanding the attitudinal drivers of support for political violence is crucial. Though other research shows that a vast majority of respondents who express support or tolerance for political violence do not engage in actual violent behavior themselves, Kalmoe (2014) points to a wide literature demonstrating that attitudes toward violence help to facilitate willingness to act on those attitudes.

There are a couple of important limitations to our study that suggest avenues for future research. First, an important limitation of our study, which we regard as a preliminary analysis of the relationship between Trump support and tolerance of political violence, is that it does not test for a unidirectional causal relationship. As a political figure, Trump may inspire his supporters to view political violence as more acceptable. However, Trump supporters may be attracted to him because they are more tolerant of political violence, a priori. Future research should investigate the directionality of the relationship between Trump approval and support for political violence, perhaps using an experimental design.

Second, the results show that Trump approval and racist and xenophobic attitudes are important, but are only part of a complex and multifaceted story of attitudes toward January 6 and political violence. This is illustrated by the relatively low r-squared values for the models in Table 2 and the finding of partial rather than full mediation in Figures 5 and 6. Future research could test Trump approval against other predictors of political violence, mentioned in our literature review, such as misogyny, political cynicism and various types of elite communications. Rival mediators, such as erosion of trust in political elites and institutions, denigration of democratic values and affective political polarization, may also help to explain the impact of Trump approval on attitudes toward political violence. These should also be investigated, particularly as the January 6 events become further contested and shaped by America’s political culture wars.

Finally, there is the possibility that factors that we tested, or failed to test to avoid overfitting the models, interact with Trump approval in important ways that could add nuance to its relationship with support for political violence. Perhaps certain types of Trump approvers or Trump fans with particular types of attitudes, experiences and proclivities are most likely to favorably depict January 6 and endorse the use of political violence. Future work could investigate these possibilities by examining intersectional political identities and attitudes more systematically (e.g., Whites, males, Christian Republicans) or by using latent class analysis to identify subgroups of supporters who may be most supportive of political violence.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the McCourtney Institute.

ORCID iD

James Piazza

Notes

Author Biographies

James A. Piazza is Liberal Arts Professor of Political Science at The Pennsylvania State University.

Natalia Van Doren is a Ph.D. candidate in psychology at The Pennsylvania State University and is currently completing her residency in clinical psychology at the Veterans Administration in Palo Alto, CA.

References

Agriesta

(2021). CNN Poll: Most Americans think election results could lead to violence in the coming years. CNN, 12(March). https://www.cnn.com/2021/03/12/politics/cnn-poll-political-divisions

Armaly

M. T.

Buckley

D. T.

Enders

A. M.

(2021). Christian nationalism and political violence: Victimhood, racial identity, conspiracy and support for the Capitol attacks. Political Behavior, 44(2), 937–960. https://doi.org/10.1007/s11109-021-09758-y

Balz

Clement

Guskin

(2022). Republicans and Democrats divided over Jan. 6 insurrection and Trump’s culpability, Post-UMD poll finds. The Washington Post. January 1.

Bartels

L. M.

(2020). Ethnic antagonism erodes Republicans’ commitment to democracy. Proceedings of the National Academy of Sciences, 117(37), 22752–22759. https://doi.org/10.1073/pnas.2007747117

Blair

Christine Fair

Malhotra

Shapiro

J. N.

(2013). Poverty and support for militant politics: Evidence from Pakistan. American Journal of Political Science, 57(1), 30–48. https://doi.org/10.1111/j.1540-5907.2012.00604.x

Brewer

M. B.

(1999). The psychology of prejudice: Ingroup love or outgroup hate? Journal of Social Issues, 55(3), 429–444. https://doi.org/10.1111/0022-4537.00126

Brigham

J. C.

(1993). College students’ racial attitudes. Journal of Applied Social Psychology, 23(23), 1933–1967. https://doi.org/10.1111/j.1559-1816.1993.tb01074.x

Bright Line Watch (2020). American democracy on the eve of the 2020 election. Bright line Watch October 2020 surveys. http://brightlinewatch.org/american-democracy-on-the-eve-of-the-2020-election/

Bryant

F. B.

Smith

B. D.

(2001). Refining the architecture of aggression: A measurement model for the buss–perry aggression questionnaire. Journal of Research in Personality, 35(2), 138–167. https://doi.org/10.1006/jrpe.2000.2302

10.

Cassese

E. C.

Farhart

C. E.

Miller

J. M.

(2020). Gender differences in COVID-19 conspiracy theory beliefs. Politics & Gender, 16(4), 1009–1018. https://doi.org/10.1017/S1743923X20000409

11.

Cineas

(2021). Donald Trump is the accelerant: A comprehensive timeline of Trump encouraging hate groups and political violence. Vox. January 9.

12.

Clayton

Davis

N. T.

Nyhan

Porter

Ryan

T. J.

Wood

T. J.

(2021). Elite rhetoric can undermine democratic norms. Proceedings of the National Academy of Science, 118(23), 1073. https://doi.org/10.1073/pnas.2024125118

13.

Coppock

McClellan

(2019). Validating the demographic: Political, psychological, and experimental results obtained from a new source of online survey respondents. https://doi.org/10.1177/2053168018822174

14.

Cox

(2021). After the ballots are counted: Conspiracies, political violence, and American exceptionalism. Survey Center on American Life. American Enterprise Institute. January. https://www.americansurveycenter.org/research/after-the-ballots-are-counted-conspiracies-political-violence-and-american-exceptionalism/

15.

Cutts

Ford

Goodwin

M. J.

(2011). Anti-immigrant, politically disaffected or still racist after all? Examining the attitudinal drivers of extreme right support in britain in the 2009 European elections. European Journal of Political Research, 50(3), 418–440. https://doi.org/10.1111/j.1475-6765.2010.01936.x

16.

Diamond

Drutman

Lindberg

Kalmoe

Mason

(2020). Americans increasingly believe violence is justified if the other side wins.

17.

Edwards

G. S.

Rushin

(2018). The effect of President Trump's election on hate crimes. Available at SSRN 3102652 .

18.

Feinberg

Branton

Martinez-Ebers

(2019). The Trump effect: How 2016 campaign rallies explain spikes in hate. Texas AM University Commerce mimeo.

19.

Georgetown University Institute of Politics and Public Service (2019). New Poll: Voters find political divisions so bad, believe U.S. is two-third of the way to “Edge of a Civil War”. October 23. https://politics.georgetown.edu/2019/10/23/new-poll-voters-find-political-divisions-so-bad-believe-u-s-is-two-thirds-of-the-way-to-edge-of-a-civil-war/

20.

Glick

Lameiras

Castro

Y. R.

(2002). Education and Catholic religiosity as predictors of hostile and benevolent sexism toward women and men. Sex Roles, 47(9), 433–441. https://doi.org/10.1023/A:1021696209949

21.

Hicks

Tingley

(2011). Causal mediation analysis. The Stata Journal, 11(4), 605–619. https://doi.org/10.1177/1536867X1201100

22.

Hogg

M. A.

Abrams

Brewer

M. B.

(2017). Social identity: The role of self in group processes and intergroup relations. Group Processes & Intergroup Relations, 20(5), 570–581. https://doi.org/10.1177/1368430217690909

23.

Imai

Keele

Tingley

(2010a). A general approach to causal mediation analysis. Psychological Methods, 15(4), 309. https://doi.org/10.1037/a0020761

24.

Imai

Keele

Tingley

Yamamoto

(2010b). Causal mediation analysis using R Advances in social science research using R (pp. 129–154). Springer.

25.

Kalmoe

N. P.

(2014). Fueling the fire: Violent metaphors, trait aggression, and support for political violence. Political Communication, 31(4), 545–563. https://doi.org/10.1080/10584609.2013.852642

26.

Kalmoe

N. P.

Mason

(2019, January). Lethal mass partisanship: Prevalence, correlates, and electoral contingencies. National Capital Area Political Science Association American Politics Meeting.

27.

Kalmoe

N. P.

Mason

(2021). Is political violence justified? Monkey cage. Washington Post. January 11. https://www.washingtonpost.com/politics/2021/01/11/what-you-need-know-about-how-many-americans-condone-political-violence-why/

28.

Kalmoe

N. P.

Mason

(2022). Radical American partisanship: Mapping violent hostility; its causes, and the consequences for democracy. University of Chicago Press.

29.

Kelly

C. R.

(2020). Donald J. Trump and the rhetoric of ressentiment. Quarterly Journal of Speech, 106(1), 2–24. https://doi.org/10.1080/00335630.2019.1698756

30.

Kunst

J. R.

Boos

Kimel

S. Y.

Obaidi

Shani

Thomsen

(2018). Engaging in extreme activism in support of others’ political struggles: The role of politically motivated fusion with out-groups. PloS one, 13(1), Article e0190639. https://doi.org/10.1371/journal.pone.0190639

31.

Levine

(2020). ‘No Blame?’ ABC News Finds 54 cases invoking ‘Trump’ in connection with violence, threats, alleged assaults. ABC News. May 30.

32.

Lipscomb

(2017). Hate-fueled attacks rattle Florida after Trump’s election. Miami New Times.

33.

Lopez

(2020). Donald Trump’s long history of racism, from the 1970s to 2020. Vox. August 13.

34.

Mason

Wronski

Kane

J. V.

(2021). Activating animus: The uniquely social roots of Trump support. American Political Science Review, 115(4), 1508–1516. https://doi.org/10.1017/S0003055421000563

35.

Merrilees

C. E.

Cairns

Taylor

L. K.

Goeke‐Morey

M. C.

Shirlow

Cummings

E. M.

(2013). Social identity and youth aggressive and delinquent behaviors in a context of political violence. Political Psychology, 34(5), 695–711. https://doi.org/10.1111/pops.12030

36.

Miller

J. M.

(2020). Psychological, political, and situational factors combine to boost COVID-19 conspiracy theory beliefs. Canadian Journal of Political Science/Revue canadienne de science politique, 53(2), 327–334. https://doi.org/10.1017/S000842392000058X

37.

Motta

(2021). Republicans, not Democrats, are more likely to endorse anti-vaccine misinformation. American Politics Research. 1532673X211022639.

38.

Nacos

B. L.

Shapiro

R. Y.

Bloch-Elkon

(2020). Donald Trump. Perspectives on Terrorism, 14(5), 2–25. https://www-jstor-org-443.web.bisu.edu.cn/stable/10.2307/26940036

39.

Newman

Merolla

J. L.

Shah

Lemi

D. C.

Collingwood

Ramakrishnan

S. K.

(2021). The Trump effect: An experimental investigation of the emboldening effect of racially inflammatory elite communication. British Journal of Political Science, 51(3), 1138–1159. https://doi.org/10.1017/S0007123419000590

40.

Newport

Saad

(2021). Presidential job approval. Public Opinion Quarterly, 85(1), 223–241. https://doi.org/10.1093/poq/nfaa061

41.

Obaidi

Bergh

Sidanius

Thomsen

(2018). The mistreatment of my people: Victimization by proxy and behavioral intentions to commit violence among Muslims in Denmark. Political Psychology, 39(3), 577–593. https://doi.org/10.1111/pops.12435

42.

Panagopoulos

(2012). Ex-Presidential approval: Retrospective evaluations of presidential performance. Presidential Studies Quarterly, 42(4), 719–729. https://doi.org/10.1111/j.1741-5705.2012.04014.x

43.

Saucier

(2013). Isms dimensions: Toward a more comprehensive and integrative model of belief-system components. Journal of Personality and Social Psychology, 104(5), 921. https://doi.org/10.1037/a0031968

44.

Schaffner

B. F.

MacWilliams

Nteta

(2016). Explaining white polarization in the 2016 vote for president: The sobering role of racism and sexism. In Conference on the US Elections, (pp. 8–9).

45.

Semyonov

Raijman

Gorodzeisky

(2006). The rise of anti-foreigner sentiment in European societies, 1988-2000. American sociological review, 71(3), 426–449. https://doi.org/10.1177/000312240607100304

46.

Sides

Tesler

Vavreck

(2019). Identity crisis. Princeton University Press.

47.

Swaine

Adolphe

(2019). Violence in the name of Trump (August 28). The Guardian.

48.

Sylvester

S. M.

(2021). COVID-19 and motivated reasoning: The influence of knowledge on COVID‐Related Policy and Health Behavior. Social Science Quarterly, 102(5), 2341–2359. https://doi.org/10.1111/ssqu.12989

49.

Thomsen

Obaidl

Sheehy-Skeffington

J. A.

Kteily

Sidanius

(2014). Individual differences in relational motives interact with the political context to produce terrorism and terrorism-support. Behavioral and brain sciences.

50.

Weinschenk

A. C.

Panagopoulos

van der Linden

(2021). Democratic norms, social projection, and false consensus in the 2020 US presidential election. Journal of Political Marketing, 20(3-4), 255–268. https://doi.org/10.1080/15377857.2021.1939568

51.

Westwood

S. J.

Grimmer

Tyler

Nall

(2022). Reply to Kalmoe and Mason: The pitfalls of using surveys to measure low-prevalence attitudes and behavior. Proceedings of the National Academy of Sciences, 119(32), Article e2207584119. https://doi.org/10.1073/pnas.220758411