Introduction
Support for right-wing and radical-right populist parties and candidates increased globally in recent years. These politicians advocate socially conservative policies, stringent immigration laws, and restrictions on minority rights. Previous research in the US context has attributed the electoral success of these right-wing leaders to their increasing support among white, male, and Christian populations, motivated by perceptions among these groups that their social status is under threat (Craig and Richeson Reference Craig and Richeson2014; Gidron and Hall Reference Gidron and Hall2017; Mutz Reference Mutz2018b; Parker Reference Parker2021; Parker and Lavine Reference Parker and Lavine2024).
Social status based on race and nationality has played an important role in that regard. Social status in this research is understood as the prestige and respect groups have in society, irreducible to class or economic conditions. Mutz (Reference Mutz2018b) argues that threat to the social status of whites and Americans is one of the main causes of Trump’s electoral support among white voters. These reactionary voters are not simply motivated by economic hardship or a belief in racial superiority but instead by a perception that their group is becoming less valued, i.e., losing social status. Although it can be challenging to disentangle these motivations, experimental studies have shown that whites see racial progress as a zero-sum game in which they are losing (Norton and Sommers Reference Norton and Sommers2011) and as a threat to their status (Valentino and Brader Reference Valentino and Brader2011; Wilkins and Kaiser Reference Wilkins and Kaiser2014; Wilkins et al. Reference Wilkins, Hirsch, Kaiser and Inkles2016).
Status threat based on nationality – that the US is losing prominence relative to other nations – has similar effects (Mutz Reference Mutz2018b). Arguably, Trump’s “Make American Great Again” (MAGA) and “America first” slogans speak directly to these status threat sentiments (Gidron and Hall Reference Gidron and Hall2017; Lamont et al. Reference Lamont, Park and Ayala-Hurtado2017; Mutz Reference Mutz2018b; Parker Reference Parker2021; Sides et al. Reference Sides, Tesler and Vavreck2018; Stewart and Willer Reference Stewart and Willer2022). Various studies show that national identity is a strong predictor of attitudes against out-groups (Brown Reference Brown2000; Citrin, Reingold, and Green Reference Citrin, Reingold and Green1990; Pettigrew Reference Pettigrew, Tuch and Martin1997) and right-wing support (Gidron and Hall Reference Gidron and Hall2017; Mutz Reference Mutz2018b), as well as being one of the main sources of positive social status in the United States (Lamont Reference Lamont2000).
In this article, we investigate three open questions. Combining a vignette and a conjoint experiment, we analyze (1) if the effect of status threat is due to a generalized or a restricted shift of white voters toward conservatism; (2) whether and how that reaction is constrained by party identification; and (3) if we can empirically attribute a causal link between status threat communication and voter decisions.
The first question essentially addresses the scope of the electoral effect of racial and national status threat. When white Americans are exposed to political communications highlighting a supposed decline in their group status, is the effect on their political attitudes a broad or a narrow one? We derive two contrasting hypotheses from theories concerning the psychological mechanisms behind individuals’ reactions to group status threat.
According to the motivated social cognition perspective (Jost et al. Reference Jost, Glaser, Kruglanski and Sulloway2003), political conservatism is a response to psychological needs to reduce uncertainty and preserve personal and group standing. Threats to the status quo boost conservatism because they activate those psychological needs (Jost et al. Reference Jost, Glaser, Kruglanski and Sulloway2003; Jost, Federico, and Napier Reference Jost, Federico and Napier2009). The electoral implication is that when threats to the social status of one’s racial or national group become salient, this should increase the electoral appeal of candidates who advocate for preserving the status quo. Importantly, nothing in this argument prevents threat in one dimension from affecting support for conservatism in another. Supporting candidates with conservative stances in various domains should help reduce uncertainty about the social environment, alleviating the psychological anxiety associated with threat salience and uncertainty (Jost et al. Reference Jost, Glaser, Kruglanski and Sulloway2003). Thus, this mechanism should not only amplify support for candidates whose conservatism is directly related to the source of the threat but also for those who hold conservative positions in unrelated areas. For instance, threats to the social status of whites should reduce support among whites for candidates in favor of affirmative action policies benefiting non-white groups (Norton and Sommers Reference Norton and Sommers2011). But opposing expansion of redistributive programs, LGBT rights, and abortion rights are also forms of preserving the status quo, so support for candidates taking these stances should also increase. This suggests that the scope of status threat’s effects is broad or generalized conservative support and can go beyond policies targeting the source of the threat.
H1 Threats to the racial status of whites or the national status of Americans increase the electoral appeal of various candidates’ conservative positions among white American voters, including conservative positions not directly related to the source of the status threat.
However, it is also plausible that a social status threat increases support only for those with specific positions addressing the perceived source of the threat. If psychological existential and epistemic needs are situationally triggered by the salience of race-based threats, for instance, a conservatism that minimizes changes on unrelated issues like abortion or LGBT rights may be insufficient to address the anxiety created by the threat. If so, the electoral consequence of status threat is much more restricted, affecting only threat-related issues. The social identity theory (Tajfel et al. Reference Tajfel, Turner, Austin, Worchel, Hatch and Schultz1979; Tajfel and Turner Reference Tajfel, Turner, Worchel and Austin1986) and the sociofunctional threat-based hypothesis (Cottrell and Neuberg Reference Cottrell and Neuberg2005) provide further justification for this expectation of restricted conservative shift.
Under status threat, dominant groups will increase their support for conservative candidates that stand for preserving the status quo or for re-establishing stable hierarchies and conditions of a “glorious past,” representing a period of less insecure in-group dominance (Gidron and Hall Reference Gidron and Hall2017; Lamont et al. Reference Lamont, Park and Ayala-Hurtado2017; Mutz Reference Mutz2018b; Parker Reference Parker2021; Sides, Tesler, and Vavreck Reference Sides, Tesler and Vavreck2018; Stewart and Willer Reference Stewart and Willer2022). But this reaction should be restricted and favor only candidates whose policies somehow target the source of threat. For instance, a threat to the social status of whites should lead to stronger resistance to affirmative action policies benefiting non-whites than to other liberal policies such as abortion laws or LGBT rights. When the threat to positive social status is based on nationality, reactions should target policies that can reduce the status or relative power of the nation, such as restricting trade privileges granted to other countries.
The sociofunctional argument posited by Cottrell and Neuberg (Reference Cottrell and Neuberg2005) also suggests that status threats should elicit specific, rather than generalized, reactions. This framework asserts that different out-groups are perceived to pose qualitatively distinct threats to in-group members, which in turn evoke distinct emotional responses. Consequently, it is reasonable to anticipate specific reactions depending on the perceived source of the status threat.
H2 Threat to the racial status of whites increases the electoral appeal among white Americans of candidates’ conservative positions against racial inclusion, but not the appeal of other unrelated conservative positions.
H3 Threat to the national status of Americans increases the electoral appeal among white Americans of candidates’ conservative positions related to maintaining American culture and trade dominance, but not the appeal of other unrelated conservative positions.
We add that partisanship should also restrict the effect of status threat on support for conservative candidates. In the US context, it is well-known that partisanship plays a major and enduring role in voter decision-making (Bartels Reference Bartels2002; Boonen Reference Boonen2019; Hooghe and Boonen Reference Hooghe and Boonen2015; Jennings, Stoker, and Bowers Reference Jennings, Stoker and Bowers2009). Partisans often follow party lead when expressing policy support (Bisgaard and Slothuus Reference Bisgaard and Slothuus2018; Bullock Reference Bullock2011; Clifford, Leeper, and Rainey Reference Clifford, Leeper and Rainey2023) and choose co-partisan candidates unless that candidate takes dissonant positions on many politically salient issues simultaneously (Mummolo, Peterson, and Westwood Reference Mummolo, Peterson and Westwood2021). While there is some evidence that individuals will at least temporarily shift partisan identity in response to status threat (Craig and Richeson Reference Craig and Richeson2014), it is not clear whether they will go so far as to actually choose an out-party candidate.
Due to the strong party identification in the US context, we expect that party attachment will be the main driver of partisans’ candidate evaluation, mitigating the effect of status threat. For instance, Republican (Democratic) voters will be more likely to choose a candidate endorsed by the Republican (Democratic) party regardless of threat or policy factors. While we expect at least a restricted increase in support for conservatives (H2 and H3 ), we expect it within partisan lines. For instance, Democratic voters will support Democratic candidates but will prefer more conservative Democrats over more liberal Democrats on threat-relevant policies when status threat becomes salient.
H4 Party attachment is the main driver of partisans’ candidate evaluation and selection, mitigating the effect of policy positions of the candidate, even when status threat becomes salient.
Finally, we anticipate a stronger conservative shift among Democratic voters, rather than Republican ones, because of a ceiling effect (Brown, Rucker, and Richeson Reference Brown, Rucker and Richeson2021; Nail et al. Reference Nail, McGregor, Drinkwater, Steele and Thompson2009). If Republican voters are already more likely to support conservative candidates at the baseline, any increase in conservatism due to status threat will be difficult to detect. Additionally, if Republicans are already high in status threat (perhaps due to co-partisan elite rhetoric), it may be more difficult to detect differences between Republicans in a control condition and those exposed to status threat.
H5 The effect of status threat on voters’ candidate preference is stronger for Democratic than Republican voters.
Research design
Experimental design
We investigate the hypotheses using an online survey that combines a vignette and a conjoint experiment. This allows us to examine how the presence and type of status threat affects participants’ preference for candidates with randomly assigned liberal and conservative positions on different policies.
Participants were randomly assigned to either a status reassuring condition or one of three possible status threat conditions: white status threat, American status threat, or a combined condition. The full vignettes, and all other survey materials, are included in the online supplement.
Economic status based on income conditions and social status based on respect and esteem in society are related, but one cannot be reduced to the other (Ganzeboom, De Graaf, and Treiman Reference Ganzeboom, De Graaf and Treiman1992; Ridgeway Reference Ridgeway2013; Sedikides and Guinote Reference Sedikides and Guinote2017). As our focus is on the effect of social status, we expose everyone, including those in the status reassuring condition, to negative information about the US economy, including inflation and the decline in household incomes. If differences are found between treatment groups, they can be attributed to the parts of the vignette related to social status - rather than simply the presence of threat.
For the status reassuring condition, this baseline economic information was followed by a reassurance that American society largely remains the same. Participants in the racial status threat condition, however, were presented with information about the declining status of white Americans, while participants in the national status threat condition received information about a decline in the US’ global status. Participants in the combined condition received both sets of information.
After reading the assigned vignette, subjects evaluated candidates in a forced-choice paired-profile conjoint experiment. Each subject chose the candidate they would vote for between two profiles presented on the same screen and rated each candidate on a 7-point scale, repeating this task six times.
The profiles presented statements about candidates’ positions on affirmative action, abortion rights, LGBT rights, immigration, income redistribution, and trade policy with China. Each issue position was randomized independently and with equal probability. For instance, a candidate could be liberal on abortion policy but conservative on affirmative action. In order to increase the realism of the task and reduce the number of times any participant saw the exact same position statement attributed to different candidates, the liberal or conservative position in each issue was sampled from a list of different statements grouped by the ideological stance they represent. Our aim is not to evaluate the effect of each specific statement; rather, we seek to understand the overall influence of status threat on voters’ general inclination toward conservatism across various policy dimensions. We thus employed an indicator variable (1=conservative; 0=liberal) for each policy topic to capture whether the sampled statement was liberal or conservative in nature. The full list of statements and a validation analysis of their ideological leanings using a large-language model can be found in the online supplement. For each task, issue order was randomized, but the order for both candidates being compared was the same. Table 1 shows an example of two profiles.
Conjoint table example
We randomized two sets of conjoint profiles, one with and one without information about candidates’ party affiliation. For the latter, the information “Independent” appeared under party affiliation to avoid guessing by participants. The main dependent variable is the respondents’ candidate choice.
After completing the candidate conjoints, participants were asked about perceived status shifts and self-reported status threat anxiety. This allows us to measure whether participants are aware of the specific type of status shift and whether their anxiety is increased by it, without potentially creating a compound treatment on the conjoint outcome.
This joint experimental design has many advantages. First, although many observational studies have demonstrated the association between status threat and policy attitudes or candidate support (Gidron and Hall Reference Gidron and Hall2017; Mutz Reference Mutz2018b; Parker Reference Parker2021; Parker and Lavine Reference Parker and Lavine2024), experimental studies on the effect of status threat on voters’ choice between candidates remain rare. This limits researchers’ ability to draw causal conclusions due to possible omitted confounders and the dependency of the results on modeling choices. This has led to controversies about inferences from those analyses and the role of other factors, such as material self-interest (Morgan Reference Morgan2018; Mutz Reference Mutz2018a). Our design allows us to leverage an internally valid causal interpretation of the connection between status threat communication and voters’ choice of conservative candidates, while also mitigating the dependency of the conclusion on modeling choices.
Second, some previous studies have used indirect measures of social status concern, such as social dominance orientation (Gidron and Hall Reference Gidron and Hall2017; Mutz Reference Mutz2018b; Parker Reference Parker2021). These indirect measures are good approximations, but they can add noise to the estimation and do not fully capture the intended concept of status threat. Instead, we directly manipulate perceptions of status threat by randomly assigning a vignette which closely captures the concept of group social status.
Finally, previous studies have used voters’ policy attitudes or actual or declared vote share for the candidates as outcome variables (Gidron and Hall Reference Gidron and Hall2017; Mutz Reference Mutz2018b; Parker Reference Parker2021). One important limitation in this case is that real candidates appear to voters as a bundle of features, including party affiliation and multiple policy positions. When those studies find an association between measures of status threat and vote for a candidate, we cannot be sure that voters were attracted by a status-related policy position adopted by the candidates instead of some unrelated feature, such as the candidate’s charisma, prior political history, or any other policy positions. Although our analysis is not designed or intended to adjudicate among competing explanations for real-world candidate selection, our design improves the instrumentalization of the concept of status threat and grants a causal interpretation of the connection between status threat and the electoral appeal of candidates’ conservative positions on different policy dimensions. We discuss some limitations of our design in the concluding section.
Statistical procedures
The parameter of interest is the Conditional Average Marginal Causal Effect (CAMCE). camce. The estimation is conducted using linear probability models (LPMs) of respondents’ candidate choice on the interaction between candidates’ issue positions and the treatment groups. We estimate the models using pooled data as well as various data subsets categorized by voters’ partisanship and candidates party affiliation.
We use the no-party profile to test hypotheses H1 to H3 . If hypothesis H1 is correct, candidates’ conservative positions on all issues should have a stronger positive impact on their electoral support among participants exposed to the status threat conditions. If the hypotheses H2 and H3 are correct, we should observe an increase only for positions connected to the threat.
To test H4 , we focus on the profiles containing candidate partisanship. If H4 is supported, a Democratic voter exposed to status threat should support more conservative Democratic candidates over more liberal Democrat candidates. When the choice is between a Democrat and a Republican candidate, however, they should prefer the Democrat regardless of the candidates’ issue positions. We should see a similar reaction among Republican voters.
Lastly, we conduct a statistical test of differences in regression coefficients to assess if the differences in interactive effects across subsamples are significantly different from zero. If H5 is correct, the impact of status threat should be stronger among Democrats than among Republicans. In the online supplement, we present the results of a series of robustness checks, which include exclusion of speeders, reanalyzing the data using models that incorporate random effects by policy statement, as well as logistic models. The online supplement also include details about out attention checks.
Data collection
Data was collected using an online survey of white US-resident adults. We collected 2,042 complete interviews, with 1,533 cases randomly selected to receive information about candidates’ party affiliation. Details of the sample size calculation are in the online supplement.
Participants were recruited via the online survey platform Prolific (Palan and Schitter Reference Palan and Schitter2018). Participation was available to those who self-identified as non-Hispanic white, born in the US, and 18 years old or older. Participants received payment based on the median length of the survey. We imposed a soft quota by partisanship, seeking to have a sample with 50% Democratic voters and 50% Republican voters. Ultimately, 1,029 respondents were Democratic and 1,013 were Republican. Independents were excluded to increase the sample size of partisans to test H4 and H5 .
The experiment was preregistered at OSF,Footnote 1 approved by an Institutional Review Board, and fully adheres to principles of ethical research. Data and scripts for replication are available at Harvard Dataverse (Ferrari and Smith Reference Ferrari and Smith2025).
Data analysis
Tables with descriptive statistics, covariate balance, conjoint profile and task order effects checks, and full regression results are in the online supplement. Overall, we find little support for any of our hypotheses regarding status threat. Rather, the primary drivers of voter decision-making is partisanship and policy.
Figure 1 shows the frequency that candidates with different party affiliations and policy positions were selected by partisans in the status reassuring condition. Results are very similar for the other treatment conditions (see online supplement). The frequencies suggest that partisanship was the most relevant feature. Republican participants chose Republican candidates 66% of the time, while Democratic participants chose Democratic candidates 63% of the time. These frequencies were higher than the effect of any other policy position, except for Democratic voters selecting candidates based on abortion. Policy positions were also influential, however, and their influence didn’t completely disappear when candidates’ party was presented – as can be seen by comparing the left and right panels of Figure 1.
Percentage of time (y-axis) voters who identify with the Democratic or Republican party (color code) chose candidates with conservative (top panels) or liberal positions (bottom panels) in various policy areas (x-axis). The left panels show cases in which the candidate’s profile included party affiliation. For instance, Republican voters selected candidates who were conservative toward abortion 58% of the time when the candidate’s profiles didn’t include party affiliation (top right panel). This figure includes only cases in the status reassuring condition.
Figure 2 shows results estimated using non-partisan candidates’ profiles only, testing hypotheses H1 to H3. Candidates’ policy positions significantly affected participants’ candidate choice. In the baseline status reassuring condition, taking a conservative position on any one of those issues reduces support among Democratic voters, but increases support among Republican voters. However, there is no consistent evidence that any status threat condition increased the appeal of any particular position. In other words, we observe neither a generalized nor a restricted conservative shift among partisan voters.
Point estimates (shapes) and 95% confidence intervals (bars) capturing the causal effect (x-axis) of status threat (shapes and colors) on voters’ support for candidates due to their conservative policy position (y-axis). Estimates based on linear probability models using non-partisan candidate pairs (column panels). Row panels show subsamples by voters’ partisanship. SE clustered by subject.
Figure 3 is similar to Figure 2 in all regards, except that we included the partisan profiles only in order to text H4. Status threat didn’t produce any substantial conservative shift whatsoever among partisans, but policy positions and partisanship have significant effects. Being affiliated with the Republican party reduces support for an otherwise all-liberal candidate by 30 percentage points (p.p.) among Democratic voters, but increases support by 36 p.p. among Republican voters. The only other effect that rivals the partisanship effect is being conservative on abortion among Democratic voters.
Point estimates (shapes) and 95% confidence intervals (bars) capturing the causal effect (x-axis) of status threat (shapes and colors) on voters’ support for candidates due to their conservative policy position (y-axis). Estimates based on linear probability models using partisan candidate pairs (column panels). Row panels show subsamples by voters’ partisanship. SE clustered by subject.
Finally, Figure 4 shows the predicted probability of voting for a candidate under different status threat conditions as a function of the number of conservative policy positions taken by the candidate. Partisans tolerate more ideological dissidence from in-party candidates, but do not do so for out-party candidates. When partisanship is less relevant (Democrat vs Democrat, Independent vs Independent, or Republican vs Republican), Democratic voters accept up to two conservative positions before their likelihood of selecting a candidate drops below 50%. However, when choosing between candidates from different parties, Democrats become more tolerant for in-party candidates, accepting up to three or four conservative positions, and less tolerant for out-party candidates, accepting only zero or one conservative position. The results for Republicans are similar, and neither is significantly impacted by status threat.
Predicted probability of voting for the candidate as a function of the number of conservative positions taken on different issues (x-axis) by status threat exposure (shapes), voters’ partisanship (line types), and party profile of the candidates competing (panels).
In the online supplement, we demonstrate that the overall lack of effect of status threat exposure cannot be explained by respondents ignoring the message. We show evidence that the vast majority understood the message entirely or partially, and were able to recall it even after completing the candidate choice tasks. When the analysis is repeated only with respondents who answered the manipulation check question 100% correct, the results remain the same. We also show that subjects reacted to the treatments by changing their perceptions of the social status of whites and Americans, and also increased in status anxiety when exposed to a status threat condition. In sum, the overall picture that emerged is that, among partisans, policy position and partisanship dominate candidate choice. Exposure to the status threat communication has little impact on these choices, even when participants read and remembered these communications, and even when their status anxiety increases in response.
Conclusion
Recent electoral successes of right-wing populist candidates have generated debates and controversies about the role of status threat in producing a conservative shift among white voters (Craig and Richeson Reference Craig and Richeson2014; Gidron and Hall Reference Gidron and Hall2017; Jost et al. Reference Jost, Glaser, Kruglanski and Sulloway2003; Mutz Reference Mutz2018b; Morgan Reference Morgan2018; Parker Reference Parker2021; Parker and Lavine Reference Parker and Lavine2024). This study investigated whether exposure to messages emphasizing status threat can elicit the hypothesized conservative shift effect among white partisans in the United States, and whether this effect is confined to threat-related issues and the selection of in-party candidates only. Our findings indicate minimal effects from status threat exposure, alongside substantial influences from partisanship and candidates’ issue positions. Why did status threat not produce the anticipated effect?
One possibility is that threats to the esteem and prestige of social groups may hold less significance for white partisans when making candidate selections, particularly when partisanship and policy issues are at play. It is plausible that voters prioritize policies for reasons that are not closely tied to groups’ social status.
Another reason might be that the null results stem from certain limitations inherent to any single experimental study. First, our conjoint experiment was not designed to completely mirror the selection dilemmas faced by voters in primary or general elections. Achieving this would necessitate randomizing a vast number of features, including candidate appearance, experience, and fundraising, among others. Second, the salience of threats to racial, national, and other identities in the current political discourse in the United States may have already prompted white voters to react and internalize the effects of such threats in their behavior prior to the experiment. This pre-exposure is known to lead to an underestimation of causal effects, potentially producing null results (Druckman and Leeper Reference Druckman and Leeper2012). Third, the lack of observed effects might result from the intensity and frequency of exposure to status threat messages; stronger and more consistently reiterated messages could potentially yield the theorized conservative shift. However, it is also important to note that participants did attend to our messages, shifting their perceptions of social status of white and Americans, increasing in status anxiety, and recalling the content of their assigned status threat treatment even after having completed all six candidate conjoint decisions. It is not the case that the status threat messages produced no effects at all, only that they fail to have any impact on candidate choice compared to the much stronger effects of candidates’ policy positions and partisanship.
In sum, our study suggests that status threat may be inconsequential, or that while it does matter in other contexts, individuals may not exhibit heightened sensitivity to this type of threat in the context of candidate choice. Alternatively, it is possible that status threat was once influential, but its reactive effect has since then already been incorporated into voters’ preferences in the current political context, such that additional exposure does not lead to further shifts. In either case, our study serves to highlight the enduring importance of policy positions and especially partisanship for voter decision-making. When choosing between candidates, voters appear to care most about partisanship, attending to policy positions either when that position is especially important to them (such as Democratic voters attending to abortion positions) or when an in-party candidate substantially deviates from their ideology (i.e. a Republican taking multiple liberal positions). Status threat seems to be unable to derail this political decision-making process.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/XPS.2025.10006
Data availability
Support for this research was provided by the University of California, Riverside and the Volgenau Fellowship. The data, code, and any additional materials required to replicate all analyses in this article are available at the Journal of Experimental Political Science Dataverse within the Harvard Dataverse Network, at: doi: 10.7910/DVN/AUMLXM (Ferrari and Smith Reference Ferrari and Smith2025).
Competing interests
The authors report none. The views expressed in this article are those of the authors and do not reflect the official policies or positions of the US Naval Academy, Department of Defense, or any other government agency.
Ethics statement
This research was reviewed and approved by the United States Naval Academy Institutional Review Board (USNA.2024.0007-IR-EM2-A). We fully adhere to APSA’s Principles and Guidance for Human Subjects Research. Further information is available in the Supplemental Appendix under Ethics.
Open access
