The personality of a personality cult? Personality characteristics of Donald Trump's most loyal supporters

The unusually loyal supporters of Donald Trump are often described as a cult. How can we understand this extreme phenomenon in U.S. politics? We develop theoretical expectations and use the Big Five personality dimensions to investigate whether Trump's most loyal supporters share personality characteristics that might make them inclined to cult-like support. We find that (1) Trump

behavior that would have sunk most traditional politicians."This support appears detached from ideology or policy success.At least in part due to this core of personally loyal supporters, Trump has been able to turn the GOP into a party of what some term "explicit" White nationalism (Massey, 2021, p. 5), while the United States has shifted onto the list of potentially backsliding democracies (e.g., International IDEA, 2021).
While the concept of a personality cult is rarely discussed in the political science literature on U.S. politics, it has been frequently referenced in the news media to characterize Trump's appeal.For example, in June 2018, the Editorial Board of The New York Times noted "the striking degree to which President Trump has transformed the Republican Party from a political organization into a cult of personality."Figure 1 shows a considerable increase in the monthly rate of articles in The New York Times with at least one mention of the phrase "personality cult" or "cult of personality" (blue squares) and in the 6-monthly right-aligned moving average (blue line) from June 2015, when Trump formally announced his candidacy, reaching a peak in late 2020 and early 2021.A similar pattern emerges for articles mentioning "Trump" and "cult" in the same paragraph (red dots/line). 1  In this article, we use two studies to investigate whether individual personality characteristics can help us understand the extreme loyalty of Trump's "base" of supporters.Study 1 is an online survey of U.S. adults designed for this project.Study 2 is the 2016 American National Election Study (ANES).We employ the Big Five personality dimensions (Agreeableness, Conscientiousness, Extraversion, Neuroticism, and Openness), which have been consistently associated with political attitudes (Carney et al., 2008;Fortunato et al., 2018;Gerber et al., 2010Gerber et al., , 2012Gerber et al., , 2013;;Mondak, 2010;Mondak & Halperin, 2008).Gerber et al. (2010) show that personality traits rival factors such as income and education as correlates of political behavior.
A striking finding across these two studies conducted more than 4 years apart (September 2016 -January 2017 and April 2021) is that the most loyal Trump supporters have high scores on 1 See the online supporting information for search terms (Figure S1) and for data from the Wall Street Journal showing a somewhat different pattern (Figure S2).

The Big Five
In this section, we develop theoretical expectations for each of the Big Five personality dimensions and extreme political followership.We expect that cult members' extreme followership is characterized by unwavering loyalty to, and persistent agreement with, a leader.Cult members do not simply share the leader's policy positions or believe the leader can effectively achieve specific ends.They adopt the leader's direction and positions unreservedly and unquestioningly.We argue that adopting, practicing, and displaying such loyalty can be psychologically fulfilling in itself.Individuals feeling anxious and vulnerable, and seeking disciplined obedience and unquestioned leadership, are drawn to personality cults.As Walsh et al. (1995) suggest, people seeking others' acceptance, guidance, and support are more likely to join cults.A personality cult, Pittman (2017, pp. 540-42) writes, may provide a sense of belonging to an imagined extended family led by a supreme patriarch.
Cult members' view of their leader, Sundahl (2023) writes, is characterized by resilience of authority, as the leader's status is not contingent on political success; symbolic elevation of the leader above others in society; and religious parallels in the representations of the leader and associated practices.We discuss how we operationalize these three aspects of cult followership in the Research Design section.
In this section, we offer conjectures about how each of the Big Five might incline individuals to strong loyalty to a political leader.The Big Five are "big" as each summarizes several more specific personality characteristics.We therefore follow Soto and John (2009) in breaking down each dimension into two facets.While we discuss some facet-level expectations, our theorizing is mainly at the higher level of the Big Five because of limited literature on the facets and political behavior.
Agreeableness is associated with being affectionate, appreciative, kind, soft-hearted, and warm.Its facets are Altruism and Compliance. 2 We expect the most loyal Trump supporters to score low on this dimension.High Agreeableness has been connected to being less persuaded by weak arguments and more persuaded by strong arguments (Xia & Habashi, 2015).Agreeableness appears connected to critical attention rather than the dogmatic acceptance associated with cult followership.Cult members are ready to dogmatically defend the views of the cult and its leader against political opponents.
Gerber et al. ( 2011) find an association between Agreeableness and aversion to political conflict.Trump loyalists are often presented as open to conflict and confrontation.Other studies find Agreeableness is negatively associated with political interest and knowledge (Gerber et al., 2011a;Mondak, 2010;Mondak & Halperin, 2008).Low Agreeableness could therefore be connected to the firm political engagement of cult members (although dogmatic followership might be incompatible with political knowledge).
Conscientiousness is associated with being efficient, organized, reliable, responsible, and thorough.Its facets are Order and Self-Discipline.We expect the most loyal Trump supporters to get high scores on Conscientiousness, as extreme political followership might be based on an appreciation for the self-discipline required by a leader who demands unwavering loyalty.Conscientiousness is associated with obedience to others' demands (Mashiko, 2008), as well as intolerance for uncertainty (Zmigrod et al., 2018), which might explain the view of a leader as infallible.
Conscientiousness is further tied to inflexibility, consistent with dogmatic followership, and to a desire for achievement through conformance (McCrae et al., 1993).The strong extreme conformity of converts to cults is found to give a sense of independence, especially from parental control (Levine, 1984;Walsh et al., 1995) and "individual accomplishment" (Straus, 1979).This connection is also found in studies on dogmatism (Duckitt, 2009).Committed obedience requires Self-Discipline, but may also bring desired Order to an individual's life.
Fortunato et al. ( 2018) found a high level of Conscientiousness among Trump voters during the 2016 primaries.They consider this a function of supporters' perceptions of Trump himself as Conscientious, but add that Trump's behavior after taking office cannot be perceived as Conscientious.At least one systematic study of expert perceptions of Trump's personality found he was low in Conscientiousness (Nai & Maier, 2018).Thus, our expectations are not based on this homophily proposition.
Extraversion is associated with being active, assertive, energetic, outgoing, and outspoken.Its facets are Assertiveness and Activity.There is no obvious connection between Extraversion and cult followership.Obedience to authority, for example, is neither positively nor negatively associated with Extraversion (Mashiko, 2008).We discuss Extraversion in more detail below when considering potential confounding factors.
Neuroticism is associated with being anxious, moody, nervous, tense, touchy, and worried.Its facets are Anxiety and Depression.We expect high levels of Neuroticism among loyal Trump supporters, since anxiety and a tendency to worry could explain a desire for purpose and stability provided by a strong leader.Emotional vulnerability is connected to recruitment to cults 2 The Compliance facet differs from the obedience and dogmatism associated with cult followership.The survey items measuring Compliance concern being forgiving, not starting quarrels, and not finding fault with others.(Curtis & Curtis, 1993).Lofland and Stark (1965) find dissatisfaction with current life situation and conflicting motivations to make individuals predisposed to join cults.They feel vulnerable and seek approval and security in the cult.We therefore expect Trump loyalists to score highly especially on the Anxiety facet.
Psychologists have found a high prevalence of neurotic personality characteristics in adherents of religious cults, although these studies are few and based on convenience samples (Levine & Salter, 1976;Rousselet et al., 2017).Some find a connection between Neuroticism and obedience (Zeigler-Hill et al., 2013), while others do not (Mashiko, 2008;Rim, 1984).
Finally, Openness is associated with being curious, imaginative, insightful, intellectual, original, and widely interested.Its facets are Openness to Ideas and Openness to Artistic Expression, or "Aesthetics."We expect to find low levels of Openness, and especially of Openness to Ideas, among loyal Trump supporters, as extreme followership is contradictory to being reflective, imaginative, and receptive to new ideas and information.It conflicts with dogmatism (Costa & McCrae, 1992a).Mashiko (2008), however, finds no association between Openness and obedience to authority.

Social dominance orientation and right-wing authoritarianism
In addition to these personality dimensions, we measure social dominance orientation (SDO) and right-wing authoritarianism (RWA).These are commonly categorized as values3 or social attitudes (Duckitt, 2001;Pratto et al., 2006), since they are more context-dependent and affected by social and political influences than the personality dimensions (Caspi et al., 2005;Costa & McCrae, 1992b;Gosling et al., 2003).RWA is an intragroup phenomenon concerning relationships between individuals of the same group.Duckitt et al. (2010) convincingly identify three elements of RWA: obedience (Conservatism), moral conformity (Traditionalism), and strict social control (Authoritarianism).SDO, on the other hand, is an intergroup phenomenon concerning relationships between ingroups and outgroups.It expresses a desire for one group's superiority over another.
We expect loyal Trump supporters to score highly on both SDO and RWA, as others have found (Dean & Altemeyer, 2019; we discuss related literature in Section 7 of the online supporting information).A key concern for our study is that RWA and SDO are related to the Big Five.Openness is negatively associated with RWA, and both Openness and Agreeableness are negatively associated with SDO (Akrami & Ekehammar, 2006).Perceived threat, which can be tied to Neuroticism, is related to authoritarianism (Feldmann & Stenner, 1997).Extraversion has been associated with both RWA (Ekehammar et al., 2004;Huddy & Del Ponte, 2020) and SDO (Caprara & Vecchione, 2013;Hofstetter, 2023).
We therefore include both RWA (all three elements) and SDO as controls in our analysis to distinguish the association of these social attitudes from more fundamental personality characteristics.
Trump is often portrayed as a populist politician.However, not all populist leaders receive such personal loyalty from their voters (Fierman, 2021).There is evidence for some association between the Big Five and populism.Agreeableness is negatively associated with support for populist parties (Ackermann et al., 2018;Bakker et al., 2016Bakker et al., , 2021)), while Conscientiousness may be positively related to such support (Ackermann et al., 2018;Aichholzer et al., 2018).Extraversion may also be related to support for populists (Ackermann et al., 2018) or simply strong partisanship (Gerber et al., 2010).
It will therefore be important for us to show that any association between loyalty to Trump and the Big Five is not confounded by either Conservative or Liberal ideology, including ideological self-identification or identification with a political party, or with factors common among supporters of populist leaders in general.

R E SEA RC H DE SIGN Data 4
We designed a survey specifically for this study and fielded it online using Lucid Marketplace among 1038 U.S. residents in April 2021 (Study 1).We used quotas for age, sex, race, region, and education based on the U.S. census (full details in the online supporting information).We subsequently became aware of a short-form measure of the Big Five in the 2016 ANES (Study 2), enabling us to make reasonable, although not exact, comparisons for many of our analyses.
The timing of Study 1 allows us to assess the loyalty of Trump supporters following his election loss in November 2020 and the end of his presidency in January 2021.The survey has three sections, containing items for measuring, first, the Big Five, second, SDO and RWA, and third, Trump's presidency and the 2020 election and its aftermath.We thus follow Gerber et al.'s (2011b, p. 283) recommendation to place the personality section before the politics section to reduce the potential for biased estimates of personality and outcomes.
Positions along the personality dimensions are indicated by respondents' attitudes toward a list of statements, presented in randomized order, using a 5-point Likert scale ranging from "strongly disagree" to "strongly agree."In the political science literature, short 10-item personality batteries, and especially Gosling et al.'s (2003) Ten-Item Personality Inventory (TIPI), are standardly used.However, these have disadvantages in terms of reliability and validity (John et al., 2008, pp. 137-38) and may underestimate the relation between personality and political ideology (Bakker & Lelkes, 2018).Longer inventories also allow for assessment of the facets of each personality dimension. 5 We therefore use John et al.'s (1991) 44-item Big Five Inventory (BFI), with nine items for Agreeableness, nine for Conscientiousness, eight for Extraversion, eight for Neuroticism, and 10 for Openness.The BFI has higher internal consistency, convergence, and discriminant validity than other long-form Big Five measurements (John et al., 2008, pp. 131-38).We follow Soto and John's (2009, p. 89) advice and adjust the BFI using an "acquiescence scale" to correct for potential bias introduced by pro-and con-trait items.We also follow their procedure for coding the two facets of each Big Five dimension.
Study 2 uses the TIPI but reveals a similar connection between the Big Five and loyal followers of Trump to that found in Study 1. Study 2 places the TIPI battery after most of the politics-related questions in the survey, but this would not make it more likely to yield results 5 We discuss short-and long-form personality measures further in the online supporting information.

4
A replication package is available on Goldsmith's Dataverse page at: https://doi.org/10.7910/DVN/6F0RDX.consistent with Study 1.The TIPI is also asked in the postelection survey, so questions from the preelection portion of the ANES are unlikely to affect it.
In Study 1, to measure RWA, we employ Duckitt et al.'s (2010) battery, which includes six items for each RWA dimension.For SDO, we use Pratto et al.'s (1994) 16-item battery on attitudes toward (in)equality between groups and social dominance.These employ the same 5-point Likert scales.
In Study 2, we construct measures for the three elements of RWA and for SDO based on the available questions, attempting to reflect the measures used in Study 1 as closely as possible.We give these variables slightly different names from those in Study 1 to indicate they are not based on identical survey items (see Table S1 in the online supporting information for details).We use eight items to construct an Authoritarian index, five items for a Tradition index, and self-placement on a 7-point Liberal-Conservative scale to measure Conservative ideology.We use three items to construct a Dominance index.
The political part of the Study 1 survey consists of 18 items about Trump during and after his presidency, the legitimacy of the 2020 presidential election and Trump's claim that it was stolen, as well as the storming of the Capitol and Trump's role in this event.The items are largely adopted from surveys conducted by Axios-Ipsos ( 2021) and Washington Post-ABC (2021).
To measure loyalty to Trump, we seek indicators of broad and dogmatic loyalty, as discussed above.These identify individuals who tend to strongly support Trump based on who he is, rather than those who have high, but not perfect, levels of agreement based on Trump's policies or specific achievements.
In Study 1, we selected seven items to measure extreme loyalty to Trump, our main dependent variable, which we label "Trump Followers."Strong Trump followers (1) voted for Trump in 2020, (2) "strongly approve" of Trump's performance as president, (3) "strongly approve" of his handling of the coronavirus, (4) believe Trump will go down in history as an "above average" or "outstanding" president, 6 (5) believe Trump acted "responsibly" after the 2020 election, (6) believe Republican leaders "should follow Trump's leadership," and ( 7) "strongly agree" with the view that mainstream media has treated Trump unfairly.
We consider these responses consistent with Sundahl's (2023) three parameters of political personality cults.Elevation is apparent in strong approval of Trump's performance while in office, and in historical legacy better than most or outstanding; resilience is seen in continued faith in Trump's leadership after his election loss; and religious parallels appear in the view of Trump's infallibility and persecution suggested by belief in his responsible behavior after the 2020 election and unfair treatment by the news media.
To measure loyalty to Trump in Study 2, which obviously occurred in a different context, we use five items.We code respondents as Trump Followers if they (1) voted for or intended to vote for Trump in 2016, (2) had a Strong Preference for Trump tied to their vote or intended vote, (3) found Nothing to Dislike about Trump, (4) rated Trump at 90 or above on a 0-100 feeling thermometer, and (5) responded that Trump makes them feel Hopeful "most of the time" or "always."Coding this variable based on evidence collected in the preelection wave of the survey where possible (items 3-5) helps it conform to Sundahl's resilience element: These are not influenced by knowledge of Trump's success in the election.
While we consider these "all-in" indicators of extreme followership theoretically appropriate for identifying cult-like followers of Trump, we also examine the robustness of our results using different indicators which allow more equivocal views of Trump.
In addition, to identify the broader pool of individuals at least moderately positively disposed toward Trump, we created a Trump Index based on the seven (Study 1) or five (Study 2) variables.This is a continuous 0-1 index of support for Trump in each study, which gives equal weight to pro-Trump responses to each item.For our main analysis, we focus on the subset of respondents who score above the mean on each index (.39 in Study 1 and .32 in Study 2; see the online supporting information for details and further descriptive statistics).This reflects the assumption that only respondents positively disposed toward Trump are likely to become very loyal followers.However, our results are very similar if we use the full sample for each study (see the online supporting information).
All items used in the analysis for Study 1 and Study 2 are standardized on a 0-1 scale for ease of comparison.Summary statistics, including for several control variables, are presented in Table 1.These include the population-weighted means for Study 2, which we use in the analyses.Weighting the Study 1 data based on age and gender categories from the U.S. Census leads to some changes in average values, including bringing the percentage of older respondents closer to that of Study 2. However, we present unweighted results for Study 1 because it was not designed to be a probability sample (Till & Matei, 2017). 7Results change little with the weighted sample.We focus on results robust across both the unweighted Study 1 and the weighted Study 2 samples.
For Study 1, just under 10% (9.8%) of our respondents in the full sample were strong Trump Followers, comprising 33% of Republicans, 5% of Independents, and 0.6% of Democrats.For Study 2, they are just under 6% (5.8%), consisting of 13% of Republicans, 3% of Independents, and 0.8% of Democrats.There are many reasons to expect some differences in the proportion of respondents exhibiting extreme loyalty to Trump, including the 4-year period between the studies and differences in sampling.However, our Trump Index yields proportions more similar (12% and 11%, respectively for Studies 1 and 2) to strong Trump supporters at or above a threshold of 0.9 on the index.
There are other notable differences across the datasets (Table 1).Our online-only sample for Study 1 skews younger and less Republican.Among the Big Five, the Study 1 sample is less Conscientious and less Open.It also has higher scores on Authoritarianism, Traditionalism, and Social Dominance.

Methods
To assess how personality and social attitudes are associated with Trump Followers, we use logistic regression models for our main dependent variable, which is coded "1" when all seven (Study 1) or five (Study 2) responses indicate loyalty to Trump, and "0" otherwise.For the main independent variables, the Big Five, we use the standard indices described above.The full results are provided in Tables S3a-S6 in the online supporting information.We present marginaleffect graphs in the main analysis, showing change in probabilities for each variable moving from the 5th to 95th percentile value, when all other covariates are held at their means. 8Our references to statistical significance also relate to these marginal effects, unless otherwise noted.
While personality is the most fundamental characteristic we measure, we include controls for Conservatism, GOP identification, RWA and SDO, as well as other potential confounders.Gerber et al. (2012), similarly, use a series of attitudinal and demographic controls to assess potential relationships between political attitudes, political partisanship, and the Big Five.The measurement of each variable is described in Table S1 in the online supporting information.Table S2 provides descriptive statistics for the unweighted full samples. 7Weighting would not correct for any selection bias occurring within weighted categories for a purely online sample.

8
We restrict the sample for estimating marginal effects to observations within the 5th to 95th percentile rangeacross all variables in order to estimate the marginal effects on the same respondents for each variable.
Among our controls are five demographic characteristics.Support for Trump is often shown to be higher among voters who are White, male, older, religious, and less educated (e.g., Morgan & Lee, 2018;Tyson & Maniam, 2016).Race and gender are potentially as fundamental to shaping world view as personality.Personality may also change with the experience and self-awareness that comes with age or education.Religiosity 9 may be a fundamental factor shaping world view if produced by childhood socialization.Further, since White Republican men are typically seen as among the largest and most consistent group of Trump supporters (Igielnik et al., 2021), we include an interaction of these categories (Table S6 in the online supporting information).Important studies explaining Trump's 2016 election victory have pointed to the key role of voters adhering to White identity (Grimmer et al., 2023;Sides et al., 2018).We therefore check the robustness of our findings to a control for this available in Study 2 (Table S9).We discuss controls for characteristics of populist voters below (see also Tables S10 and S11).9 An ordinal indicator of self-identification as "an atheist," "not a religious person," or "a religious person" in Study 1 or an indicator of frequency of church attendance in Study 2.

R E SU LT S Trump Followers and the Big Five
Our initial models show the association between strong loyalty to Trump and the Big Five and the effect of controlling for GOP identification and Conservatism (Figure 2 and Tables S3a-S4b in the online supporting information 10 ).The models in Figure 2 are restricted to the subset of respondents who score above the mean on the Trump Index (n = 475 in Study 1 and n = 1378 in Study 2).This analysis can thus establish whether, among all Trump supporters, there is a subset distinct in terms of degree of support and personality characteristics.Figure S3 (see the online supporting information) presents very similar results for the full samples.Without controls, Studies 1 and 2 show a positive association between and Trump Followers and a negative association between Openness and Trump Followers (left panels, Figure 2).There is also a significant positive association with Extraversion in Study 2 (bottom-left panel).But when the controls for GOP identification and Conservatism are introduced (right panels), the relationship between Openness and Trump Followers shrinks in magnitude and becomes statistically insignificant in both datasets.GOP identification in both studies, unsurprisingly, is strongly associated with followership of Trump, as is Conservative ideology in Study 2. But Conscientiousness remains positively related to Trump Followers in both studies.Extraversion is no longer significantly associated with Trump Followers in Study 2 (p = .198)with these controls. 1110 In all figures, whisker bars represent the two-tailed 95% confidence intervals.
11 Study 2, the ANES, employs a complex sampling design.In addition to the sampling weights (pweights), variables to identify the strata and primary sampling unit (PSU) should be applied.However, using the reduced sample of respondents scoring above the mean on the Trump Index causes an error due to a stratum with a single sampling unit.We therefore only employ the pweights.However, our results are very similar using the full sample, including the strata and PSU adjustments (see the online supporting information).

F I G U R E 2
Studies 1 and 2: Personality, party, and conservatism.Marginal Effect plots for logistic regression models.The sample is all respondents scoring above the mean value on the Trump Index (Study 1: mean = 0.39, n = 475; Study 2: mean = 0.32, n = 1378 (with GOP and Conservative controls)).Marginal effects are calculated for movement from the 5th to 95th percentile values for each variable (within this sample), while holding all other variables at their means.
These results are robust to two basic choices about the relevant sets of respondents for our concept: who we consider a strong Trump follower and how wide we cast the net for a sample of possible Trump loyalists.Figure 3 (left panels) shows our analysis for a different coding of the outcome variable.For each study, we categorize strong Trump support as a score greater than 0.9 on the Trump Index.This yields roughly equal proportions in each study (12% in Study 1 and 11% in Study 2; in the online supporting information (Figures S7 and S8, Table S8), we show multinomial logit results on the full sample using four categories: Strong Opposition, Moderate Opposition, Moderate Support, and Strong Support).Conscientiousness remains statistically significant in both studies and has the largest magnitude of the Big Five, although Extraversion is also significantly associated with this less stringent measure of strong Trump in Study 2 (Tables S7a-S7b in the online supporting information).
Analysis based on the sample of only Republican respondents (Study 1) and Republican Primary Voters (Study 2) is also shown in Figure 3 (right panels).Conscientiousness is the only personality dimension showing a significant association with Trump Followers, consistent with our results in Figure 2. In Table S16 in the online supporting information, we show similar patterns in Study 1 among self-identified "Trump Republicans," those who strongly support Trump in 2024, and in Study 2 for Trump voters in 2016.However, among self-identified 2020 Trump voters in Study 1, Conscientiousness does not produce a significant coefficient associated with being a strong Trump Follower, although it remains positive and of the greatest magnitude among the Big Five (Marginal effect = 0.27, p = .22).
We prefer to use the sample of U.S. adults above average on our Trump Index because, as an unconventional Republican leader, Trump appears to attract strong support from non-Republicans, and these "swing voters" can be an important factor in elections.For similar reasons, we believe results on the full sample are also relevant.These are also highly consistent with our main findings (see Table S7b in the online supporting information).
F I G U R E 3 Studies 1 and 2: Alternative dependent variable and GOP/primary respondents only.Marginal Effect plots for logistic regression models.The sample for the left panels is all respondents scoring above the mean on the Trump Index (Study 1: mean = 0.39, n = 475; Study 2: mean = 0.32, n = 1378).The sample for the right panels is all self-identified Republicans (Study 1, n = 264) or all Republican Primary voters (Study 2, n = 670), and the dependent variable is Trump Follower.Marginal effects are calculated for movement from the 5th to 95th percentile values for each variable (within the sample), while holding all other variables at their means.
When we examine a wide range of potential confounding factors including interaction terms for gender, race, and party identification, for RWA and SDO (Study 1), and for the corresponding three "Right" variables and Dominance (Study 2) (Figure S4 and Tables S5  and S6 in the online supporting information), this relationship between Conscientiousness and Trump Followers remains significant and of roughly stable magnitude in each study, increasing the chance of showing strong loyalty to Trump by about 10 to 15% in Study 1 and about 22 to 29% in Study 2, for a movement from the 5th to the 95th percentile values on the Conscientiousness index.In Study 1, this is somewhat lower than the impact of GOP identification or Conservatism.In Study 2, it is somewhat higher than the impact of GOP identification but substantially lower than that of Conservative ideology.
In the online supporting information, we also show that our findings are robust to inclusion of variables for White identity (e.g., Grimmer et al., 2023;Sides et al., 2018), populism-related controls for gender, income, education, and their interaction in Study 1 and a Populist Voter variable in Study 2 (Spruyt et al., 2016), an alternative Left-Right ideology variable, using a Generalized Structural Equation Modeling (GSEM), and using multiple imputation for missing data on the full sample in Study 2 (p = .082for Conscientiousness) (Figures S9-S16 and Tables S9-S14 in the online supporting information).

Facets of conscientiousness
In this section, we contrast how the Self-Discipline and Order facets of Conscientiousness are associated with Trump Followers and Conservatism and further examine the nature of the relationship between the facets and Trump Followers.Using Study 1, we code a binary variable for Conservatism to allow comparison across logit models, taking values of 1 if a respondent is in the top 10% on the Conservatism scale, otherwise 0.12 We do not use Study 2 for this analysis because the TIPI has only one survey item corresponding to each facet and is not recommended for such use (Gosling et al., 2003, pp. 523-24).
Self-Discipline is significantly related to Trump Followers, while Order is associated with Conservatism (Figure 4, Table S15a in the online supporting information), consistent with Xu et al. (2021).Altruism is positively and significantly associated with while Openness to Ideas is negatively associated and significant.Contrary to our conjecture regarding Neuroticism, counter-vailing tendencies of its facets do not emerge in association with strong Trump Followers.Across the Big Five, Self-Discipline is significantly associated with Trump Followers at p < .05,as is Active (Extraversion).Assertive (Extraversion) (p = .07)and Aesthetics (Openness) (p = .09)also approach standard significance.We show this model without a control for Conservatism to provide a direct comparison to the model using Conservative as a dependent variable.However, when the control for Conservatism is reintroduced, only Self-Discipline remains statistically significant (see Table S15a).These results indicate substantive differences, increasing our confidence that our results regarding Conscientiousness and Trump Followers are not an artifact of Conservatives' support for Trump.
When we assess the two-way interactions of several facets based on our theoretical discussion, few clear conditional relationships emerge with Trump Followers.For example, we find no statistically significant interaction effects between Anxiety (Neuroticism) and Compliance (Agreeableness), nor between Assertiveness (Agreeableness) and Compliance.We do find some evidence of an interaction between Self-Discipline and Compliance, at the highest values of Self-Discipline. 13he distribution of Conscientiousness, Order, and Self-Discipline scores for Trump loyalists and other groups of respondents further illustrates the distinctiveness of Trump loyalists.Specifically, Figure 5 presents kernel density plots for All respondents, those Above the Mean on the Trump Index, Conservatives, GOP identifiers, and non-GOP identifiers (Democrats or Independents) using Study 1.In all instances, Trump Loyalists show substantially greater density for high values of Conscientiousness and Self-Discipline than others in each category.The differences are least pronounced, but still apparent, when the comparison is among Conservatives.The pattern is less pronounced for the Order facet.
We can analyze the facets of Conscientiousness further to provide greater precision in the evidence supporting the association between Self-Discipline and cult-like followership.Specifically, only Self-Discipline is associated with all three alternative multi-item measures (described in Section 9 of the online supporting information) of strongest belief in Trump's Leadership, Trump's Status, and belief that Trump won the 2020 election (Figure 6).On the other hand, Self-Discipline is not associated with strongest Job Approval for Trump while president.We interpret this as consistent with the concept of a personality cult in that the leader is in a position of elevation and resilience.Resilience means followership does not depend on specific accomplishments (Job Approval), while elevation is reflected in exceptionally high belief in Status and Leadership.A religious element might be seen in belief in the 2020 Election Win, reflecting faith in Trump as both truthful and infallible.

CONC LU DI NG DI SC US SION
The American Psychological Association (2022) defines the phenomenon of a personality cult as "exaggerated devotion to a charismatic political, religious, or other leader, often fomented by authoritarian figures or regimes as a means of maintaining their power."This reflects the concept's historical association with authoritarian regimes using state-controlled F I G U R E 4 Trump followers, conservatism, and facets of the Big Five.Marginal Effect plots for logistic regression models.The sample is all respondents scoring above the mean value on the Trump Index (Study 1: mean = 0.39, n = 475).Marginal effects are calculated for movement from the 5th to 95th percentile values for each variable (within the sample), while holding all other variables at their means.media and education for mythmaking around a leader, while stifling criticism (Plamper, 2012: p. xvii).It is therefore surprising to see it frequently used about supporters of a recent U.S. president (Figure 1).
There has been little analysis of whether certain types of voters in democracies might be inclined to such extreme loyalty to a political leader.Research in political science has been limited to a focus on elite motivations for, and political uses of, political personality cults F I G U R E 6 Aspects of Trump followers and facets of the Five 1).The sample is those scoring Above Average on the Trump Index (mean = 0.39, n = 475).Each dependent variable is an index as described in Section 9 of the online supporting information.(Crabtree et al., 2020;Gill, 1984).Most major works on the topic have been produced by historians, focusing on specific cases and leaders (Kershaw, 1987;Plamper, 2012).To the best of our knowledge, no study to date focuses on the characteristics of individual citizens associated with political personality cults.The topic is important in any context to better understand the nature, extent, and limits of mass engagement, and especially important in democracies, where citizens are free to choose whether to "buy into" or reject the cult.Other democratic leaders, including India's Narendra Modi, have also recently been described as leading political personality cults (e.g., Guha, 2022).
Our analysis suggests some U.S. voters show an extreme, cult-like loyalty to Trump rather than to a party or policies.Why might citizens in a democratic system show such unquestioning loyalty to an individual leader?
During the 2016 election, Ekins (2017) found that among Trump voters, 31% were "staunch conservatives," 25% "free marketeers," 20% "American preservationist," 19% "antielite," and 5% "disengaged."However, we contend that, for the most loyal supporters, Trump's appeal is based more on psychological needs met by his leadership style than on policy preferences or ideology.During the 2016 Republican convention, Trump famously claimed "I am your voice" and "I alone can fix it."Such claims indicate the sort of leader-follower relationship typical of personality cults.Sundahl's (2023) qualitative description of Trump's self-presentation and imagery conforms to her categories of elevation, resilience, and religious parallels.To assess whether such idealizing and loyalty to a leader might be connected to personality, we proposed five conjectures, expecting a positive association between two of the Big Five (Conscientiousness and Neuroticism) and the strongest levels of Trump support, a negative association for two (Agreeableness and Openness), and no association for one (Extraversion).Of these, only our expectations regarding Conscientiousness and Extraversion found empirical support.Thus, we do not claim to have fully confirmed our personality-based theory of extreme political followership.But the article provides a solid foundation for future theorizing and empirical work in a new but potentially crucial area of research, and we encourage further tests in new data and for different cases.
We provide robust evidence that Conscientiousness in particular distinguishes Trump's most loyal followers from broader groups of his supporters, including those with an overall above-average inclination toward him, other Republicans, and Conservatives.We also provide evidence that the cult-like aspects of Trump's Leadership are core elements perceived by followers sharing the characteristics of a Conscientious personality, and the Self-Discipline facet specifically.These relate most directly to elevation and resilience.
Trump's call for loyalty and claim that he "alone can fix it," we argue, appeal to a desire for discipline by giving a cause individuals can (and must) fully commit to.Trump demands commitment well beyond that implied by typical politicians in democratic systems.He denigrates not only ideological and electoral opponents but also those who might share power or authority within his ideological and political cohort.He is a jealous political leader and does not brook divided or shared loyalty.Citizens who sense this and are inclined to high levels of Self-Discipline, we argue, find the allure of Trump's leadership appealing because it meets a basic need making them susceptible to personalistic, loyalty-demanding leaders.
While our survey did not directly probe the religious-parallels aspect of extreme followership (Sundahl's study was published after we fielded the survey), we identify elements of infallibility, such as belief in Trump's so-called "big lie" of the stolen 2020 election.This aspect of extreme followership in politics should be explored in future research, but it seems plausible that Trump's most loyal supporters perceive his leadership with religious parallels.For example, Onishi (2021) writes that "MAGAism [Make-America-Great-Again-ism] is buttressed by religious narratives and imagery, and its gospel is spread through houses of worship every Sunday.For some evangelicals, Mr. Trump is a divinely ordained savior uniquely ordained to save the nation from ruin."Sundahl (2023, pp. 448-49) notes prominent figures in Trump's circle reference him in religious terms, including a Cabinet Secretary ("the chosen one"), a former campaign manager ("only God could deliver such a savior to our nation"), and a former Congresswoman ("a … godly, biblical president").
To help understand U.S. politics from 2016 onwards, and particularly the unusual and disruptive role of Donald Trump, we point to Trump's self-presentation as a savior.In 2023, after declaring his presidential candidacy, he said: "In 2016, I declared, 'I am your voice…'.Today, I add: I am your warrior.I am your justice.And for those who have been wronged and betrayed, I am your retribution" (Haberman & Goldmacher, 2023).Our results suggest that Trump's attraction extends beyond policy and the politics of Conservatism or Populism.While the results across our five conjectures were mixed, the robust findings across two studies for Conscientiousness, and for the Self-Discipline facet in Study 1, are not easily dismissed and worth further investigation.We contend that, for his most committed followers, the attraction is personality-based -both in terms of Trump's self-presentation to citizens and in terms of the personality characteristics making some citizens attracted to such leadership.Trump's appeal appears to fit Sundahl's (2023) three characteristics of a personality cult.The phenomenon of a political personality cult may have arrived in full force in U.S. democracy -and could potentially be its undoing.

AC K NOW L E DGM E N T S
Previous versions of this article were presented at the School of Politics & International Relations, Australian National University, and in the POLIS seminar series, Deakin University.We thank all seminar participants for their feedback.We also thank Paul Kenny and Alex Nai for detailed feedback on the manuscript.We thank Katja Curtin for excellent research assistance.

F U N DI NG I N FOR M AT ION
Australian Research Council Future Fellowship, grant FT140100763 (Goldsmith).Austrian Science Fund, grant ESP447 (Moen).

F
I G U R E 5 Kernel density plots (Study 1): Trump followers compared with all other respondents in each group (All Respondents, Above the Mean on the Trump Index, Conservatives, Republicans, and Non-Republicans).