AUTHORITARIANISM AND POLITICAL VIOLENCE: ISSUES WITH THEORY AND MEASUREMENT

There are mixed results in the literature regarding the link between the authoritarian personality and support for political violence. Whereas trait aggression is associated with support for political violence, authoritarian personality—measured using the Feldman and Stenner (Reference Feldman and Stenner1997) parenting measures—is not correlated with trait aggression (Kalmoe Reference Kalmoe2014). Kalmoe and Mason (Reference Kalmoe and Mason2022, 164–67) also found that trait aggression significantly predicts partisan political violence; although they do not measure authoritarianism per se, they described those possessing the traits correlated with support for violence as authoritarians. However, their partisan political-violence scale includes measures that gauge support of violence against out-group political officials and, more abstractly, the use of violence to achieve political goals. Their political-violence scale in part (likely by design) is a dimension of authoritarianism—that is, aggression toward the out-group (see the online appendix). In other words, supporting political violence is what makes one an authoritarian.

Using support for violence to measure right-wing authoritarianism (RWA) is common practice, but this complicates efforts to demonstrate that authoritarians support violence. Faragó, Kende, and Krekó (Reference Faragó, Kende and Krekó2019), for instance, leveraged items from the traditional Altemeyer RWA scale, to predict aggression toward out-groups. Although it is unclear which RWA items Faragó and colleagues used, the traditional RWA scale includes items that implicitly and explicitly gauge support for aggression toward out-groups (Altemeyer Reference Altemeyer1981). Conversely, Armaly and Enders (Reference Armaly and Enders2022) found that authoritarianism is one of six variables that predict support for political violence. In this case, support for political violence is predicted by authoritarianism rather than being a measure of it. Armaly and Ender’s (Reference Armaly and Enders2022) authoritarian measure is a relatively new instrument, developed by Heller et al. (Reference Heller, Decker, Schmalbach, Beutel, Fegert, Brähler and Zenger2020), but it consists of standard elements of RWA—including support for submission to traditional authorities and consequences for troublemakers in society.Footnote ²

This issue is not new—previous studies pointed out that measures of authoritarianism (namely, the Altemeyer RWA scale) have incorporated attitudes and behaviors that theories predict should be downstream of authoritarianism as a predisposition, such as prejudice and intolerance toward out-groups (Feldman Reference Feldman2003). Violence against the government is not an exception; the RWA scale scores those who agree that it is better to “trust the judgment of proper authorities in government and religion than to listen to the noisy rabble-rousers in our society who are trying to create doubt in people’s minds” as more authoritarian (Altemeyer Reference Altemeyer1981). Perhaps “proper” is the operative word, but the RWA scale also places those who reject “challenging our government” and who “do what the authorities tell us” as more authoritarian. It is difficult to reconcile the unprecedented assault on the US Capitol—the historical seat of Congress—with the traditionalism and submissiveness that the RWA scale associates with authoritarianism (Perlstein Reference Perlstein2021).

This problem is not limited to the news media labeling insurrectionists as authoritarians; the link between RWA and support for violence against the government is explicit in academic literature. Research links those who score high on RWA with the belief that “the people convicted for their role in the violent January 6 attacks on the US Capitol are really patriots who are being held hostage by the government,” and “American patriots may have to resort to violence in order to save our country” (Public Religion Research Institute 2024). Armaly and Enders (Reference Armaly and Enders2022), whose RWA-inspired authoritarianism measure predicts support for political violence, measure support for political violence using respondents’ agreement with three statements on 1–5 Likert scales. Notably, one of the statements is support for political violence toward government: “Violence is sometimes an acceptable way for Americans to express their disagreement with the government.” Another statement is more ambiguous but similarly invokes the notion that violence like that of January 6 is permissible: “It is acceptable to use violence in advancing political goals these days.”

The problem is not limited to only the RWA scale. Other, arguably more valid measures of authoritarianism—such as the Feldman and Stenner (Reference Feldman and Stenner1997) parenting items—associate support for obedience and “good” behavior and manners with authoritarianism. Hibbing (Reference Hibbing2022) explicitly identified the issue of asserting authoritarians as the source of violence against government, noting that the activities of the January 6 insurrectionists—destroying iconic national property, aggression toward law enforcement, and threats made to the then–Vice President of the United States—cannot reasonably be qualified as reflecting traditionalism and submission to authority.

IS THE RELATIONSHIP CONDITIONAL?

There are clear inherent contradictions in an authoritarian supporting violence against the government, both conceptually and in measurement. However, this does not mean that we would never see a relationship between authoritarianism and political violence—rather, the nature of the relationship may depend on the questions we use and when we ask them.

Because the child-rearing measure of authoritarianism does not incorporate measures of violence and is still recognized as a valid exogenous measure of authoritarianism (Engelhardt, Feldman, and Hetherington Reference Engelhardt, Feldman and Hetherington2021), it is an appropriate way to measure the authoritarian personality. The measure better captures the conceptual tradeoff between autonomy and social cohesion that is the basis for attitudes indexed by RWA. Moreover, it avoids an overlap between the independent and dependent variables, given that RWA operationalizations often include violence in statements.

However, whether this measure predicts support for political violence should depend on two factors: (1) does support for political violence explicitly mention the government; and (2) who is running the government at the time the question is asked. Another study found that the degree of endorsement of political violence depends on the clarity of targets (Clifford, Lopez, and Lothamer forthcoming). We therefore might expect that the effect of authoritarianism on violence will be conditional, depending on whether authoritarians perceive those who are in charge of the government to be legitimate. In the contemporary American context, and according to research that connects support for Republicans with the child-rearing measure of authoritarianism, the relationship might reverse depending on which political party controls the White House (Hetherington and Weiler Reference Hetherington and Weiler2018; MacWilliams Reference MacWilliams2016—but see Luttig Reference Luttig2021 for a rebuttal to this claim).

What about when measures of political violence do not specify the government? One issue with many measures of support for political violence is that they ask about general support without offering context, leaving a respondent to infer what “violence” means and, ostensibly, at whom or what the violence will be aimed (Westwood et al. Reference Westwood, Grimmer, Tyler and Nall2022). We suspect that if the effect of authoritarianism on violence against the government is conditional, it will not reliably predict support for generic political violence either because some authoritarian partisans may infer that this means violence against the in-group controlling the White House. Thus, even the use of an exogenous measure of authoritarianism will not result in a consistent relationship with support for political violence.

SECURITARIAN PERSONALITY

If not authoritarian, then how might we classify those who consistently support violence against the government or otherwise? Hibbing (Reference Hibbing2020, Reference Hibbing2022) developed an alternative concept to describe supporters of “strong leaders” who do not exhibit obedience to authority: “securitarians.” According to Hibbing (Reference Hibbing2020, Reference Hibbing2022), securitarians are motivated to achieve insider-friendly policies, including the strengthening of national defenses, restricting immigration, and protecting the culture from what they perceive as moral decay. Hibbing (Reference Hibbing2020) measures the securitarian personality through agreement with the following statements: “Projecting weakness is just about the worst thing a person could do”; “I think a good deal about the security of my family and my country”; and “‘Being prepared’ to face threats is the best motto for living one’s life.” Hibbing’s (Reference Hibbing2020) securitarian worldview statements also include: “A central goal for our country should be to become so strong that outsiders will realize it does not make sense to attack us”; “If we are not vigilant, we will quickly be victimized by criminals, immigrants, and by the power of foreign countries”; and “Just about the worst thing for a country is to be perceived as weak.” If this account is accurate—that securitarianism better predicts the types of attitudes and behaviors associated with authoritarianism—then we should expect to see specific attitudes and behaviors associated with personal security and the projection of strength to predict support for political violence.

DATA AND MEASURES

We leveraged data from an original survey (N=3,103) fielded in September 2018 (Nielsen et al. Reference Nielsen, Kelley, Ellison, Johnson and Gervais2024), as well as data from the 2016, 2020, and 2024 ANES (Gervais et al. Reference Gervais, Dye, Acevedo, Ellison and Kelley2025). The original survey used a sample provided by the vendor Qualtrics, which recruited the nationally representative panel of randomly selected US adults with an oversample of gun owners using a specialized recruitment campaign. Qualtrics partners with online panel providers to supply a network of diverse, quality respondents. Each sample from the panel base is proportioned to the general population and then randomized before the survey is deployed. The models we employed use weighted data that (1) correct for the oversampling of gun owners; and (2) more closely match population estimates on key characteristics including gender, race, age, and region. All participants were compensated at rates that were fair in both local and global contexts and no groups were affected differentially.

Leveraging the 2018 original survey and the ANES data had several advantages. First, all four surveys included the Feldman and Stenner (Reference Feldman and Stenner1997) child-rearing measure of authoritarianism. (Details on the measure are included in the online appendix.) Second, we were able to assess whether support for political violence was conditional on who controlled the government: Democrats in the White House in 2016 (Barack Obama) and 2024 (Joe Biden) and a Republican in 2018 and 2020 (Donald Trump). Third, we could determine if explicitly mentioning support for political violence against the government made a difference. Because the 2018 and 2020 surveys both took place during Trump’s two presidencies, the 2018 survey included a measure of “support for political violence” that explicitly referenced the government. The survey asked respondents: “Do you think that it is ever justified for citizens to take violent action against the government, or is it never justified?” Respondents could choose “never” (0) or “justified” (1). The ANES question for 2020 (as well as for 2016 and 2024) gauging support for political violence does not mention violence against the government. Instead, it asks: “How much do you feel it is justified for people to use violence to pursue their political goals in this country?” with answers ranging from 1 (not at all) to 5 (a great deal).

Fourth, although neither of the datasets contained any of the statements that Hibbing (Reference Hibbing2020) used to measure the securitarian personality or worldview, they did include those that measured gun ownership. Hibbing (Reference Hibbing2020, 69–71, 140) found that securitarianism is associated with gun ownership and several pro-gun positions. This is consistent with preparing for outsider threats and “defending the castle,” so to speak (Hibbing Reference Hibbing2020, 69). Thus, we could determine whether behavior and attitudes that are downstream from the securitarian personality and worldview indeed were associated with support for political violence. For the 2018 survey analyses, we gauged gun ownership with a question that asked: “Do you personally own any guns (NOT including air guns, such as paintball, BB, or pellet guns)?” (1=yes, 2=no). For the ANES analyses, we used a gun-access measure (1=access to gun, 0=no access).Footnote ³ We also included several demographic controls, including age, education, female identity (gender), and nonwhite identity (race). We also believed it was relevant to control for political ideology and party identity. In several robustness-check models (see the online appendix), we also included other control variables, such as income, region, marital status, combat military service, and government trust. (Details about these variables are included in the online appendix.)

REPUBLICAN-CONTROL MODELS (2018 AND 2020)

We first assessed the degree to which authoritarian and securitarian traits predict attitudes about political violence during the years when a Republican, Donald Trump, controlled the White House (2018 and 2020). The 2018 survey asked whether political violence was ever justified, resulting in a binary outcome, whereas the 2020 survey used a multipoint scale to capture respondents’ level of agreement with political violence. To ensure comparability across surveys, we ran a linear model across both measures.Footnote ⁴ Table 1 displays the results from the linear regression models predicting support for political violence. In 2018 (model 1), authoritarianism had a negative relationship with the belief that violent action against the government is never justified. Substantively, a one-unit increase on the authoritarian index was associated with a 6.4-percentage-point decrease in the probability that a respondent supported violence against the government.Footnote ⁵

Table 1 Authoritarianism, Gun Access, and Support for Political Violence with Republican Control of the White House (2018 Survey and 2020 ANES)

Notes: Robust standard errors are in parentheses. **p<0.05, ***p< 0.01.

Authoritarianism also was negative in the 2020 results but did not reach conventional levels of statistical significance (model 2). The inconsistency in effects across the 2018 and 2020 surveys likely was due to differences in how each survey measured support for political violence. The 2018 survey specified violence against the government, meaning to the extent that a relationship between authoritarianism and Republican identity exists, the measure likely captured support for violence against a Republican-controlled government. Conversely, the 2020 survey asked about using violence as a means to pursue political goals. Whereas some Republican authoritarians may have interpreted this as referring to the government, others may have seen it as violence directed toward out-groups. The ambiguity in the target of violence likely obscured any consistent relationship between authoritarianism and support for political violence.

Whereas authoritarianism showed no positive association with political violence in either the 2018 or 2020 survey, behavior downstream of securitarian personalities consistently predicted support for political violence regardless of how the question was framed. Models 3 and 4 in table 1 reveal that gun ownership (2018) and gun access (2020) both exerted positive effects on support for political violence.Footnote ⁶ In model 3, having access to a gun was associated with an 8.9-percentage-point increase in the probability of endorsing violence against the government. When applying the broader political violence measure in model 4, gun access was associated with a 0.16-unit increase in support for political violence relative to respondents without access, which represents a 12% increase relative to the weighted sample mean of 1.32. Although a 0.16-unit increase represents only 4% of the full range of the 1–5 scale, the distribution is highly skewed toward rejection of violence, with more than 82% of respondents selecting the lowest value (i.e., “not at all”). In this context, even modest shifts away from the overwhelming consensus represent meaningful behavioral differentiation. The consistency in the effect of securitarian-related behaviors indicates that securitarian attitudes operate independently of political context. Whereas authoritarians’ endorsement for violence was dependent on whether they were thinking about out-groups or their own political allies, securitarians supported political violence regardless of the political target evoked by the question.

DEMOCRATIC-CONTROL MODELS (2016 AND 2024)

Next, we assessed whether authoritarianism had the same level of effect on political violence in years when a Democratic leader was president. We used ANES surveys under two different Democratic leaders to ensure that the effect was not driven by attitudes specific to Obama or Biden. Linear regression models that predict support for political violence are shown in table 2. Contrary to the Republican-controlled years, authoritarianism was consistently associated with support for political violence with both coefficients in the 2016 and 2024 models, attaining conventional levels of statistical significance. In 2016 (model 1), a one-unit increase in authoritarianism was associated with a 0.032-percentage-point increase in support for political violence; in 2024 (model 2), the effect was slightly larger at 0.040 percentage points. Substantively, with a mean level of support for political violence of 1.30 in 2016 and 1.34 in 2024, every one-unit increase in political violence represents approximately 2.5% to 3.1% of the average level of support for political violence. The positive effect contrasts with the negative or nonsignificant effects observed during Republican control, suggesting that authoritarian attitudes toward political violence are highly contextual and depend on which party holds power.

Table 2 Authoritarianism, Gun Access, and Support for Political Violence with Democratic Control of the White House (2016 and 2024 ANES)

Notes: Robust standard errors are in parentheses. **p<0.05, *** p<0.01.

As in the Republican models, securitarian behavior was associated positively with attitudes toward political violence.Footnote ⁷ Models 3 (2016) and 4 (2024) present evidence that gun access was a consistent predictor of political violence support across both years of Democratic leadership. Substantively, respondents with access to a gun supported political violence approximately 0.34 units more in both 2016 and 2024, with almost identical effect sizes (0.335 and 0.354, respectively). A 0.34-percentage-point increase associated with gun access represents a 26% increase over the 2016 political violence mean of 1.30 and a similarly sized shift in 2024. This indicates that those with gun access are markedly more supportive of political violence despite widespread rejection of political violence by the broader population.

These effects indicate that behaviors associated with the securitarian personality maintain their predictive power irrespective of which party controls the presidency. Moreover, authoritarianism was no longer significant at conventional levels in 2016 (model 3), when gun access was included. The stability of the gun-access effect across both Republican and Democratic administrations indicates that securitarian attitudes operate independently of political context. Whereas authoritarians’ endorsement for violence was conditional on who holds power (and was not consistently significant), securitarians consistently endorsed violence regardless of the political leadership. This suggests that individuals who consistently support political violence may be classified better as securitarians than authoritarians.