White Americans’ feelings of being “last place” are associated with anti-DEI attitudes, Trump support, and Trump vote during the 2024 U.S. presidential election

Introduction

Racial economic inequality in the United States is stark. In 2019, the average White family had about 13 times the median wealth of the average Black family (Bhutta et al., 2020). Likewise, the average White family has 6 times the wealth of the average Latinx/Hispanic family (Urban Institute, 2024). Thus, on average, White Americans are doing well economically compared to Black and Latinx Americans, who continue to face significant barriers to wealth accumulation stemming from historical and present-day racism (Derenoncourt et al., 2022; Gómez, 2022).

One important psychological consequence of racial wealth inequality is that it contributes to societal stereotypes that link race with socioeconomic status (SES). Americans, regardless of their own race/ethnicity, tend to assume White people are wealthy and that Black and Latinx people are poor (Brown-Iannuzzi et al., 2019; Zou & Cheryan, 2017). As a result of assuming that most White people are wealthy, many White Americans may feel like they are “falling behind” their racial group and not living up to the SES stereotype of White people (Cooley et al., 2021). This feeling of “falling behind” may be further compounded by between-group status comparisons—namely, White Americans’ beliefs about where they stand in the SES hierarchy relative to other racial/ethnic groups. In particular, some White Americans may feel like they are in, or close to, “last place”—worse off than all other racial/ethnic groups, including their own. This perceived position (i.e., being “last place”) within the racial/ethnic SES hierarchy may draw some White Americans to alt-right worldviews given that such ideology aims to elevate the status of White Americans (Hartzell, 2018). Of note, we use “alt-right ideology” and “alt-right” to reference belief systems and candidates who center White nationalism and thus support/endorse policies that favor “White interests” (Forscher & Kteily, 2020; Hawley, 2017).

In the present paper, we test how the context of high racial economic inequality in the U.S.—inequality that disproportionately benefits White Americans (vs. most other racial/ethnic groups)—may contribute to a sense of despondency among some White Americans. Further, we predict that White Americans who feel they are in “last place”—worse off compared to all other racial/ethnic groups, including their own—may be particularly likely to support alt-right ideology, DEI bans, and may have been most likely to vote for President Trump in the 2024 U.S. presidential election.

Within-Group Status Comparisons

Recent research suggests that, when forming a sense of their subjective status, White Americans most often compare their own status within their racial group (i.e., to other White people; Caluori et al., 2024; Cooley et al., 2021). Due to race/class stereotypes that associate wealth with White people (e.g., Brown-Iannuzzi et al., 2019), this may lead many White Americans to make upward status comparisons with others from their racial group; and, comparing oneself to others who are perceived as better off can negatively impact psychological well-being (Diel et al., 2025; Festinger, 1954). Consistent with this reasoning, prior work using representative quota samples of non-Hispanic, White Americans found that, on average, White Americans perceive themselves to be worse-off in the SES hierarchy than most other White Americans (Caluori et al., 2024; Cooley et al., 2021). Moreover, these status perceptions (controlling for the influence of objective status), predicted decreased positive emotions at a second future time point, which then predicted worse well-being at a third future time point (Caluori et al., 2024). These findings suggest that because White people are often stereotyped to be wealthy, the average White American, regardless of their objective status, may feel like they are falling behind—and this subjective experience has meaningful emotional consequences.

Between-Group Status Comparisons

Although White Americans may primarily compare their status to others within their own racial group, they are also likely to evaluate their status through between-group comparisons (Bobo & Hutchings, 1996; Brown-Iannuzzi et al., 2025; Cooley et al., 2025). Racially/ethnically minoritized groups (e.g., Black and Latinx Americans) are often stereotyped as poorer than most White Americans (e.g., Brown-Iannuzzi et al., 2019; Zou & Cheryan, 2017). Blumer’s Group Position Theory (1958) suggests that prejudice is exacerbated when White Americans feel their superior position is being challenged. Further, Bobo and Hutchings (1996) extend this model to suggest that the relative position of the group who is doing the challenging also influences the degree of intergroup threat. Extending from these models, we suggest that the current economic climate may lead some White Americans to feel as if their individual position in the economic hierarchy is challenged both by feeling they are “worse off” than other White people and that other racial/ethnic groups may be superseding them (or close to it) in status as a result of perceived racial progress (Kraus et al., 2019; Norton & Sommers, 2011). This perception of precarious positionality both within and between racial groups may activate fears of being in “last place” (i.e., last place aversion; Kuziemko et al., 2014). So, while prior work has documented that feelings of “falling behind” other White people has predictable emotional/well-being consequences for White Americans, we propose that the simultaneous perception that one is being passed by, or close to being passed by, all other racial groups (i.e., feeling both low within-group and low between-group status) is likely to have distinct consequences for political attitudes.

The Relationship Between Feeling in “Last Place” and Political Attitudes

For White Americans who feel like they are in “last place” (or close to it), their perceived low status conflicts with stereotypes that White people are wealthy, while intergroup competition is also perceived to be high. This conflict may produce a feeling among those White Americans that they are entitled to be wealthier (Kunst & Obaidi, 2020), that the current system is not attuned to their needs, and that they need a candidate who focuses on their plight as White people (Jardina, 2019; Major et al., 2018). Due to the simultaneous perception that other racial groups are passing them, or close to passing them, in status, these individuals are likely to see policies that aim to uplift racial minorities, such as DEI-related policies, to be personally harmful and unjust. As such, we anticipate that White Americans who feel like they are in “last place” (or close to it) will be particularly likely to support worldviews and ideology that align with the alt-right including White nationalism, anti-government beliefs (due to the perception that the government does not currently prioritize White Americans’ best interests), and support for policies that ban DEI initiatives (Hartzell, 2018).

Previous work lends evidence to this hypothesis (Cooley et al., 2024). Across two large representative quota samples, collected in spring/summer 2023, when non-Hispanic, White Americans were asked to rank the SES of their own racial group, other racial groups, and themselves as individuals, about 10% of each sample fell into a latent profile that reflected a tendency to feel worst-off, falling behind all racial groups (including their own; Cooley et al., 2024). On average, participants who fell in this “last place” profile, compared to the other White Americans in the sample, were the most likely to support alt-right ideology. These effects were robust to controlling for objective indicators of SES (income and education), as well as participants’ conservatism, age, and gender. Importantly, “last place” White Americans were not objectively the lowest status people in these samples. This suggests that White Americans’ subjective feelings about their position in the racial economic hierarchy are not simply a reflection of objective experiences of wealth or poverty. Instead, we reason that these perceptions may stem from the current sociopolitical climate around Whiteness in the U.S. In the present work, we replicate and extend upon these findings in the context of the 2024 U.S. presidential election, combined with a longitudinal design, to assess the novel outcomes of differential candidate support and voting behavior as the election nears.

“Last-Place” Political Rhetoric

When Barack Obama was elected President in 2008, many White working-class voters interpreted the election of the United States’ first Black President to be a de-prioritization of their needs and an empowerment of minority groups at their expense, deepening divides on both racial and party lines (Abramowitz & McCoy, 2019; Tesler, 2020). Likewise, the great recession during Obama’s presidency led some White Americans to scapegoat Black Americans for their own economic plight, putting into motion a narrative that people of color are helped by the federal government at White Americans’ expense (McKenzie, 2014). The narrative of White Americans being “left behind” has gained traction in recent years, as reflected by increasingly overt themes of economic displacement, perceived loss of status, and diminishing opportunities for White Americans (Hochschild, 2016; Jardina, 2019; McKenzie, 2014).

These “left behind” narratives were also prominent in the campaigning of current President Donald Trump during the 2016, 2020, and 2024 U.S. presidential elections (Goethals, 2018). Indeed, some have theorized that the populist messaging that Trump campaigned with in 2016 was internalized by some White voters as speaking to their racial group and contributed to Donald Trump’s success in the 2016 presidential election by acknowledging the status threat many White voters were feeling (Abramowitz & McCoy, 2019; Knowles & Tropp 2018; Mutz, 2018). If so, it follows that White Americans who are most drawn to Trump’s populist rhetoric in 2024 may be those who feel subjectively that they are in, or close to, “last place”— passed by, or close to being passed by, both stereotypically high-status groups (i.e., White and Asian Americans), and racially marginalized groups (e.g., Black and Hispanic Americans) within the SES hierarchy.

Opposition to Diversity, Equity, and Inclusion Efforts

Notably, the “left behind” rhetoric has been reflected in policy relevant to diversity, equity and inclusion, in particular (DEI; Hochschild, 2016). President Trump’s 1776 commission/report during his first term (Executive Order, 2020), as well as his 2024 campaign platform (Wendling, 2025), foreshadowed his subsequent flurry of executive orders banning DEI efforts after his 2025 inauguration. Thus, anticipating large-scale DEI bans if Trump were to win the election, we also examined whether non-Hispanic, White Americans who feel the threat of being (or soon becoming) “last place” may also be those who are most supportive of DEI bans. We reasoned that for these White Americans, DEI policies may seem not only unnecessary, but personally harmful.

Relationships Between Feeling “Last Place” and Trump Support Over the 2024 U.S. Election

Finally, it is an open question as to how subjective perceptions of feeling “last place” (or close to it) may relate to political outcomes over time, and during times of political transitions, such as a U.S. presidential election. One possibility is that the relationship between feeling “last place” and support for alt-right ideology and candidates like Donald Trump becomes stronger as the election nears due to the increased salience of candidates’ political rhetoric as the election season peaks. However, it is also possible that in a climate of extreme political polarization, such as in the contemporary U.S., the behavior of political candidates (or other exogenous shocks) would have to be relatively extreme to shift belief systems appreciably (Axelrod et al., 2021). Such a possibility would predict a relative stability of the relationship between feeling “last place” and support for alt-right ideology, Donald Trump, and voting behavior as the election nears. Through a longitudinal design, over the span of several months and including timepoints both before and after the election, we tested these possibilities.

Overview of Longitudinal Study

The present study has three key hypotheses. First, we predicted that there would be a distinct group of White Americans who feel “last place,” and that such a perception, controlling for objective status, would predict the highest support for alt-right ideology and candidates with platforms that align with alt-right worldviews. This first prediction represents a replication of Cooley et al. (2024), while also extending those findings to a novel longitudinal design and during a consequential point in U.S. history (a presidential election). Second, testing a novel hypothesis, we predicted that these “last place” White Americans would be most supportive of policies that aim to ban DEI initiatives and would report the strongest support for President Trump leading up to the election, as well as the highest proportion of votes ultimately cast for President Trump. Finally, given our longitudinal design we further tested the novel question of whether these associations increased in strength as the election neared; or, conversely, whether they were relatively stable, perhaps due to strong political polarization undermining the influence of exogenous shocks (Axelrod et al., 2021).

To test these hypotheses, we conducted a five-wave longitudinal study, with three waves before the 2024 U.S. presidential election and two waves after the election, launched between September 4th, 2024 and November 20th, 2024.^[1] At each time point, non-Hispanic, White Americans living in the United States were asked to complete a survey consisting of a measure of within- and between-group subjective status (i.e., our key predictor; see Figure 1) as well as support for alt-right ideology; support for an unnamed, purportedly-local candidate with a platform that aligned with alt-right worldviews; support for DEI bans; and voting intentions and behavior in the 2024 U.S. presidential election. To assess patterns of responses to the within- and between-group measure of subjective status, and generate subjective status profiles—such as the hypothesized “last place” profile—we chose latent profile analysis (LPA) for several reasons. First, the underlying process of LPA is much like factor analysis, except instead of clustering items from a scale based on shared characteristics, individuals in a sample are clustered based on shared patterns of responding. Thus, LPA is a powerful, person-centered approach to data analysis that is particularly relevant to understanding holistic patterns of responses such as those captured by the within- and between-group subjective status measure of focus here. LPA has the additional benefit that it moves beyond the often-critiqued method of testing 2- and 3-way interactions between variables which requires exceedingly (and often prohibitively) large sample sizes to adequately detect effects, and these types of interactions are notoriously difficult to replicate (e.g., Giner-Sorolla et al., 2024). Finally, we were replicating and extending upon prior work that used this analysis strategy.

Methods

A link to Supplemental Materials, study materials, de-identified data, and analysis code can be accessed here: https://osf.io/uznha/. This study was not pre-registered; however, our hypotheses replicate and extend findings from an existing publication for which all studies were pre-registered (Cooley et al., 2024). All methods below were reviewed and approved by the lead authors’ institutional review board (proposal # ER-F24-03) to ensure adequate protection of participants.

Statistical Power

We worked with CloudResearch’s “Managed Research” platform to recruit a sample of non-Hispanic, White Americans with census-based representative quotas on age, gender, region of the country, and education. We over-sampled at wave 1 (N ~ 600) to account for attrition across waves. This larger sample size also accommodates current standards for statistical power for latent profile analysis (LPA) which we planned to apply to responses to our within- and between-group subjective status measure based on wave 1 responses. Current recommendations for LPA suggest we would need approximately 500 participants to detect the correct number of latent profiles (Nylund et al., 2007).

Next, we planned to use mixed-effects models for our longitudinal analyses to examine the effect of a time-invariant, continuous predictor (i.e., posterior probabilities of within- and between-group subjective status profile membership) on repeatedly-measured political outcome variables. The literature on power analyses for mixed-effects models suggested that our analysis would require approximately N = 200 for adequate power (.80), assuming a small/medium effect and a high intraclass correlation coefficient (i.e., ICC; Brysbaert & Stevens, 2018). That said, the required sample size that emerges when simulating power analyses for multilevel models is variable based on minor tweaks in researcher assumptions. Thus, our main strategy was to recruit the maximum number of participants that we could afford. This resulted in a goal to close the study with N = 300 White Americans who completed all 5 waves.

Participants

We retained all participants who completed at least waves 1 and 2 for our LPA analysis to generate subjective status profiles. This allowed us to both have a sufficient sample size for LPA and also ensured that we only retained participants who would be used in our longitudinal prediction models (i.e., participants who only completed wave 1 could not be included in longitudinal models, so determining their profile placement would be moot). Thus, LPA analyses were conducted on responses to the within- and between-group subjective status measure, as assessed at wave 1.^[2]

Next, we handled missing data on this measure via listwise deletion (N = 19) given our interest in the relative placement of all racial groups (and to mirror the strategy used to handle missing data on this measure in Cooley et al., 2024). Because we were interested in psychological processes that stem from being enmeshed in the U.S. cultural context, we removed N = 27 participants who were not born in the United States, as measured at wave 1 (again, mirroring Cooley et al. [2024]). Our final sample who completed at least both wave 1 and wave 2 of the study consisted of 506 non-Hispanic, White Americans who were born in the United States. They were, on average, 48.63 years of age (SD = 16.59; 279 women; 227 men) and had a median education of “some college, no degree (or associate’s degree),” a median income of 40,000-59,999 USD, and the following political affiliations: 28.5% Republican, 22.9% Independent, 46.6% Democrat, and 2.0% Other. Table 1 presents the descriptive statistics by wave on each of our representative quota variables.

Table 1

Descriptive Statistics of Representative Quota Variables by Wave

Participants: Longitudinal Analyses, Mixed-Effects Models

Because we only recruited participants at subsequent waves who had completed the prior wave, it was not possible for participants to skip one wave and then return to the study. Demographics of those who completed at least waves 1 and 2, and thus who were used in the LPA analysis conducted on wave 1 responses to our within- and between-group subjective status measure, are described above; those who continued to complete wave 3 included 449 respondents; those who also completed wave 4 included 377 respondents; and, finally, those who completed all 5 waves included 291 respondents. Our strategy for handling missing data due to attrition across waves is discussed in more detail in the “Analyses” section.

Procedure

Participants were contacted in early September 2024 to complete a longitudinal study through late November. If they agreed, they continued to be contacted again every 7 to 14 days. The first wave began with a robot and captcha check to ensure that respondents were human. If participants did not identify as non-Hispanic and White, they learned that they were not eligible for this study. These items were then followed by an attention check. If participants failed the check, they learned they were not eligible to continue with the study. If they passed all of these sections at wave 1, participants read an informed consent for the 5-wave study and completed a variety of demographic items to accommodate our quota-based sampling. If participants fit our desired quotas, they then continued into the study. Our key measures, broken down by wave of inclusion, are detailed below.

Measures

Key Items Included at All 5 Waves

Within- and Between-Group Subjective Status Measure. Each wave began with participants responding to the recently developed “Perceived Self-Group Hierarchy” (PSGH) measure of their perceived position within the racial economic hierarchy and the perceived position of a variety of racial groups, including their own (see Figure 1 and Brown-Iannuzzi et al., 2025 for scale development; see also Cooley et al., 2024). Participants were asked to consider their social status in the U.S. in terms of money, education, job prestige, and political power, as well as the relative status of “White people”, “Black people”, “Asian people” and “Latinx people.” Responses to this measure were analyzed via LPA to obtain posterior probabilities of classification into each of the within- and between-group subjective status profiles for each participant and these probabilities then served as predictors of the outcomes that follow.

Figure 1

Subjective Within- and Between-Group Status Measure (Brown-Iannuzzi et al.,
2025; Cooley et al., 2024)

Alt-Right Beliefs. Participants’ support for ideology that aligns with the alt-right (Hawley, 2017) was assessed via two separate scales taken from Cooley et al. (2024). One scale was a 17-item scale created by Cooley et al. (2024) that includes items assessing the following attributes of alt-right ideology: perceptions of anti-White bias (e.g., “White people are generally under attack in the U.S”); anti-government beliefs (e.g., “The government threatens my personal rights.”); violent anti-government beliefs (e.g., “When the government isn’t working, violence is sometimes the answer”); and anti-immigrant beliefs (e.g., “Immigrants are contaminating the U.S. way of life”). Cronbach’s alpha for this scale across waves was high, varying from .95 to .96.

The next alt-right ideology scale was an adaptation of a 5-item scale used by Kamenowski et al. (2021), as assessed in Cooley et al. (2024). This scale recorded agreement to items like, “America is superior to other nations;” “White people should be the leaders in the U.S.;” and “Foreigners will never be real Americans no matter how much time they spend in the U.S.” All items for both scales were evaluated via seven-point Likert scales (1= strongly disagree, 7 = strongly agree). Cronbach’s alpha for this scale across waves was high, varying from .86 to .89.

Opinions on Right and Left Movements. Participants then rated their feelings about Right-Wing events or social movements (e.g., “January 6th storming of the U.S. Capitol”; 11 items) and Left-Wing events or social movements (e.g., “#BlackLivesMatter”; 8 items) on a seven-point Likert scale (1 = extremely negative, 7 = extremely positive; 8 items). These items were taken from Cooley et al. (2024) with the addition of assessing feelings toward “Project 2025” in waves 2-5. Cronbach’s alpha for both support of right-wing movements/events and support of left-wing movements/events varied from .89 to .96 across waves.

Fictional Candidates. Using previously published items (Cooley et al., 2024), participants next read about the platforms of two unnamed political candidates who were presented as “recent candidates for state-level political positions” and indicated their opposition or support for both candidates (1 = strongly oppose, 7 = strongly support). Candidate #1 was designed to have views that align with alt-right ideology, such as espousing policies that support white nationalism (i.e., voter identification requirements, strict immigration policies, and prevention of DEI education), and Candidate #2 was designed to have views similar to left-wing politicians and candidates (i.e., advocacy for government-run healthcare, higher taxes, and more wealth redistribution).

Attitudes Toward DEI. Finally, participants were informed about the University of Florida’s decision regarding the removal of DEI education with the following statement: “The University of Florida recently announced that it would be eliminating all DEI (Diversity, Equity, and Inclusion)-related positions. This move complies with a state law that bars state universities from using government funds for initiatives that promote DEI.” They were then asked to rate their agreement to three statements about DEI education (e.g., “I think more states and universities should enact policies that will eliminate DEI positions on college campuses”) on a sliding scale (0 =strongly disagree, 100 = strongly agree). Cronbach’s alpha for this scale across waves was high, varying from .97 to .99.

Key Wave-Specific Items

Waves 1-3 (Pre-Election).

Voting Intentions and Support for Candidates. At waves 1-3, participants were also asked to indicate their support of, or opposition to, each of the candidates running in the 2024 presidential election at the time of the study: Donald Trump, Kamala Harris, and Third-Party Candidates on 1 (strongly oppose) to 7 (strongly support) scales. They were then asked how likely they were to vote for each candidate as well as how likely they were to vote in the 2024 presidential election, versus abstain from voting, on a sliding scale (0 = not likely at all, 100 = extremely likely).

Wave 4 (Day-After-Election).

Cast Votes. The day after the election, participants were asked a yes-or-no question of whether they voted in the U.S. presidential election. If they answered “yes,” participants were given a follow-up question asking them to report if they voted for Donald Trump, Kamala Harris, or a third-party candidate. If they answered “no,” participants were asked why they did not vote in an open-ended question.

Exploratory Items Waves 1-5

Because of the expense of longitudinal recruitment, as well as the timing of this study at a unique moment in U.S. history, we included additional measures to assess a variety of distinct research questions that will be analyzed in separate manuscripts. These items are detailed in the Supplemental Materials.

Analyses

Latent Profile Analyses

To identify participants’ within and between-group subjective status profiles, we planned to conduct LPA with the following profile indicators: relative placement of the self, White people, Black people, Hispanic people, and Asian people. We used Mplus 8.10 and 8.11 with Mixture Add-On and model defaults such that variances were allowed to vary within profile but not between profiles and covariances were fixed to 0 both within and between profiles (Muthén & Muthén, 2017). All other model decisions were selected to follow prior work using LPA with this measure, which included the use of maximum likelihood estimation with robust standard errors (Cooley et al., 2024).

Criteria for Evaluating LPA Model Fit. Following recent guidelines for LPA (Ferguson et al., 2020), we began our analysis by comparing model results starting with a one-profile baseline model, which was compared to a two-profile model, which was then compared to a three-profile model and so on, until it was clear that the model with one fewer profile had better fit. To evaluate relative model fit we examined the log-likelihood value, Akaike’s Information Criterion (AIC), the Bayesian Information Criterion (BIC), and the Sample-Adjusted BIC (SABIC). For each of these statistics, better model fit is reflected by lower values; however, the magnitude of the change between each model is also important. For example, these indices may decrease between two models, but to such a small degree that the improvement in fit is not meaningful, in which case the more parsimonious model would be selected. Additionally, we examined the results of two tests that compare the fit of a given model with the fit of a model with one fewer profile: the Lo-Mendell Rubin (LMR) test and bootstrapped log likelihood test (BLRT). Both of these tests indicate a preference for the more parsimonious model when they do not reach statistical significance. Finally, we examined the degree of classification certainty, via model entropy. Entropy values range from 0 to 1 with values closer to 1 representing greater classification certainty, meaning that the placement of participants into profiles is expected to have less error. We also aimed to avoid solutions that yielded any profiles that accounted for less than 10% of the sample to follow general recommendations (Ferguson et al., 2020), as well as to replicate the decision-making strategy used in Cooley et al. (2024).

Longitudinal Models Using Profile to Predict Political Outcomes

After establishing participants’ probabilities of falling into each of the identified within- and between-group subjective status profiles via LPA, we conducted longitudinal mixed-effects models, specifying a random intercept for participant and a random slope of wave, with posterior probabilities of profile membership (i.e., probability of belonging to each profile), wave, and their interactions as predictors of each of our political outcomes. This approach, of using posterior probabilities of profile membership rather than categorical profile placement (i.e., placing participants into their highest probability profile), makes use of graded profile membership rather than assuming hard classification, thereby incorporating classification uncertainty into the model. Importantly, the variables used to define the latent profiles (perceptions of each group’s status and self-status) were distinct from the distal outcomes (support of alt-right ideology, DEI bans, and Trump), reducing concerns about overlap or bias in prediction. As such, parameter estimates reflect relationships with latent profiles while accounting for inevitable classification error. Including interactions with wave allowed us to test whether the relation between probabilities of profile membership and political outcomes changed over time. Our models nested observations within participant and specified a random slope for the effect of wave (coded from 1-5).

Missing Data in Longitudinal Models. Given that our LPA only used data for participants who completed at least waves 1 and 2, there were not any participants in our dataset who completed wave 1, but not wave 2. However, several variables had missing values at wave 3, wave 4, and wave 5. These included alt-right beliefs, opinions on right and left movements, support for fictional candidates, and attitudes toward DEI (missingness at wave 3 = 11%, wave 4 = 25%, wave 5 = 42%), as well as support for Trump and likelihood of voting for Trump (missingness at wave 3 = 11%; these items were not measured after wave 3). We used multiple logistic regression to assess how missingness on these variables was related to observed data, including probabilities of profile assignment, support for Trump at wave 1 (to account for political lean), alt-right beliefs at wave 1 (as measured by Cooley et al., 2024, to ensure missingness on other DVs was not related to a key variable of interest), age, education, income, and gender (see Supplement for full analyses). Missingness on these variables was only significantly associated with age at all waves, with greater missingness among younger participants. This suggests that data may be missing at random (MAR). To proceed with our planned mixed-effects model analyses, data were multiply imputed using the Mice package in R (van Buuren & Groothuis-Oudshoorn, 2011). Models using pairwise deletion did not produce results that differed meaningfully from models using imputed data, and results using pairwise deletion are reported in the Supplemental Materials.

Results

LPA Results

Replication: Identifying Within- and Between-Group Subjective Status Profiles

The LPA results for iterative model fit led us to test 1 to 5 profile models; the results of these tests appear in Table 2. Using the decision-making strategies defined in the Analyses section, and the same strategies used in Cooley et al. (2024), there was some uncertainty as to whether the 3-profile or 4-profile model was the best fit, especially when combining theory, prior findings (i.e., Cooley et al., 2024 settled on a 3-profile model), a preference for parsimony, and the data itself. That said, from a purely data-driven perspective, the 4-profile model best fit the data. To address this ambiguity, we analyzed the data both ways, with the 4-profile model results appearing in the main manuscript and the 3-profile results, analyzed identically to Cooley et al., 2024, in Supplement.^[3]

The indicator means for the selected 4-profile model appear in Table 3 and a visual of these means, mapped onto the subjective status measure of interest, appear in Figure 2.

Table 2

Comparative Model Fit Statistics for 1 to 5 Profile Models

Description of 4-Profile Model Results

For comparison purposes, we will start by noting that prior work identified three profiles named based on where the self was placed in the hierarchy: a “second place” profile, a “third place” profile, and a “last place” profile (see Cooley et al., 2024). The findings below are similar, but given our selection of the 4-profile model for the present data, the “third-place” profile observed in prior work seems to have split into two versions: one with groups relatively compact, perhaps reflective of a perceived tight status competition between groups, and one with groups relatively dispersed. We describe these profiles in more detail below.

The “second place (tied)” profile appears in Figure 2, top panel, and accounts for 17% of our sample. In this profile, participants see the self as behind the perceived high status of White people (as we see across all profiles). That said, in this profile, participants see the self as tied for second place in status with a group stereotypically assumed to be high status—Asian Americans—and ahead of two groups stereotyped to be low status, Black and Hispanic Americans (Zou & Cheryan, 2017). This profile is very similar to the “second place” profile from Cooley et al., 2024.

The “third place-compact” profile appears in Figure 2, second from the top panel, and accounts for 51% of our sample. In this profile, participants see the self as falling behind relatively high-status White and Asian people, but ahead of relatively poor Black and Latino people who are perceived as not differentiable in status. That said, as compared to the profile described next, the perceived status of the self is behind, but closer to, White and Asian people; and ahead of, but closer to, both Black and Hispanic people. For these reasons, we named this profile “third place-compact” as the placement of all racial groups, as related to the self, is relatively condensed—perhaps reflecting a perceived tight status race. This profile most strongly mirrors the single “third-place” profile that emerged in Cooley et al., 2024.

A new profile, varying a bit from prior work, is the “third place-dispersed” profile, which appears in Figure 2, third panel from the top, and accounts for 17% of our sample. This profile is quite similar to the prior profile, with the most notable difference being that the perceived status of the self is further behind White and Asian people, and the perceived status of Black and Hispanic people is further behind the self. In short, the rankings of the self and all racial groups are the same as the prior profile, but everyone is spread further apart within this profile, perhaps reflecting perceptions of a more dispersed status race, and thus relatively lower intergroup status threat.

Finally, a “last place” profile emerged (see bottom panel of Figure 2), accounting for 15% of the sample. This profile replicates the key profile of theoretical interest from Cooley et al. (2024), except that instead of perceiving the self as falling significantly behind all other racial groups included on the scale as in prior work, this profile now reflects seeing the self as tied for last place with a racial/ethnic minority with a long history of experiencing racial discrimination in the U.S.: Black Americans. Replicating prior work, it is notable that White Americans in this profile also perceive that there is a tight race among all racial groups at the top with all points clustering relatively close together and toward the top of the scale. As a reminder, this is the profile we theorized would be predictive of the highest alt-right extremism (replicating prior work), most support for DEI bans (new outcome), the most Trump support (new outcome), and that would have the highest proportion of people within the profile ultimately voting for Trump (new outcome).

Table 3

Summary of Indicator Means for Selected 4-Profile Model

Figure 2

Visual Depiction of 4-Profile Model Results

Longitudinal Results

Stability of Responses to Subjective Status Measure

Initially, when designing our study, we had hoped to examine potential changes in profile membership over time. Indeed, this was the reason we included the within- and between-group subjective status measure at each wave. However, given the complexity and expense of recruitment for longitudinal designs, we were not able to overrecruit enough at wave 1 to maintain an adequate sample size for stable profile classification (N = 500) over all 5 time points. Nylund and colleagues (2007) use simulations to show that LPA results produce accurate findings consistently when N > 500, but that at lower sample sizes, fit statistics like BIC and BLRT (as we use here) become less reliable for determining the correct number of profiles. For these reasons, we focused on profile membership as determined by our largest set of cohesive data (wave 1), while further only including participants who would also be included in our longitudinal prediction models (i.e., those who completed at least wave 1 and wave 2). To further justify this decision of treating profile probabilities as a relatively stable individual difference over the timespan of our study, we also conducted some additional statistical tests which are described next.

One way to assess response stability to our within- and between-group subjective status measure is to assess the stability of placement of each indicator (i.e., the self, White people, Black people, Asian people, and Latinx people) across all 5 waves. To assess this, we calculated an intraclass correlation coefficient (ICC) for each profile indicator across waves 1-5. The ICC reflected high stability in responses to each profile indicator: self-placement ICC = .82, 95% CI [.79, .85]; White-people-placement ICC = .66, 95% CI [.62, .71]; Black-people-placement ICC = .71, 95% CI [.67, .75]; Latinx-people-placement ICC = .67, 95% CI [.62, .71]; and Asian people placement ICC = .73, 95% CI [.69, .77]. While there is some variability in stability—with the placement of the “self” being most stable, and placement of “White people” as least stable—the indicators used to generate the latent profiles demonstrated moderate-to-good temporal stability overall (ICCs ranged from .66 to .82). This suggests that the constructs used to define profiles are relatively stable individual differences, although we revisit this point in the general discussion.

Replication and Extension to Longitudinal Context of 2024 U.S. Presidential Election: Alt-Right Outcomes

Next, we tested whether we generally replicated the finding that subjective perceptions of falling in “last place (tied)” predicts general support for alt-right ideology and the highest support for an unnamed, purportedly-local candidate with a platform that aligns with alt-right worldviews.^[4] Both of these outcomes were taken directly from prior work (Cooley et al., 2024), and thus represent a direct replication; however, they simultaneously extend prior work to a novel longitudinal design during a particularly consequential moment in U.S. history, taking place 1.5-2 years after data collection for Cooley et al. (2024). Due to the longitudinal nature of these data, we were further interested in whether these effects became stronger as the election neared.

To test these hypotheses, as described in detail in the Analyses section, we ran longitudinal mixed-effects models with posterior probabilities of profile membership, wave, and their interactions as predictors of each of our three alt-right outcomes. Because the four posterior probabilities of membership in profiles 1, 2, 3, and 4 sum to 1, they are perfectly collinear. To avoid multicollinearity, we include only the probabilities of membership in profiles 1, 2, and 3 in our models, using profile 4, “last place (tied),” as the omitted reference category. We selected profile 4 as the reference because a higher probability of being placed in that profile was hypothesized to predict the highest levels of support of all measured political outcomes.

To aid interpretation, we mean-centered continuous predictors (i.e., profile probabilities, income, education, and age). As a result, the intercept represents the predicted value of the dependent variable when all predictors are at their mean. Likewise, the coefficients for the profile probabilities (see Table 4) indicate how the outcome changes as the probability of belonging to each profile increases, with the probability of belonging in the “last place (tied)” profile decreasing accordingly. For example, a negative and statistically significant coefficient for the probability of belonging in the “second place (tied)” profile when predicting support for an alt-right candidate would indicate that as the probability of belonging to this profile increases relative to the “last place (tied)” profile, support for an alt-right candidate decreases.

Aligned with Cooley et al. (2024), we additionally controlled for age, objective status (i.e., income and education) and gender (1: man; -1: else) so that we could isolate the effects of subjective status. Each of our control variables were initially also allowed to interact with wave (so we could control for their effects over time), as were profile probabilities, so we could assess differential relationships between profile probabilities and political outcomes as the election neared. Notably, however, for none of our outcomes was the effect of profile moderated by wave of the study. For this reason, the results of simplified models without interaction terms with wave are reported below and the more complex models with interactions with wave are reported in Supplemental Materials. We report results separately for each of our 5 political outcomes next.

Alt-Right Ideology (Two Scales). Results when predicting support for alt-right ideology were partially consistent with hypotheses and partially replicated Cooley et al. (2024; see Table 4). Support for alt-right ideology (as measured by both scales) decreased significantly as the probability of belonging to the “second place (tied)” profile increased, relative to the “last place (tied)” profile, bKamenowski = -0.80, p <.001; bCooley= -0.73, p = .004. Support for alt-right ideology (as measured by both scales) also decreased as the probability of belonging to the “third place–compact” profile (bKamenowski = -0.12, p = .559; bCooley = -0.20, p = .353) or “third place–dispersed” profile (bKamenowski = -0.33, p = .182; bCooley = -0.34, p = .205) increased, relative to the “last place (tied)” profile, but these effects were not statistically significant.

To better understand the differences between profiles in these outcomes, we also computed marginal means for each profile. Because posterior probabilities range from 0 to 1, we estimated the expected outcome as if a participant had a probability of 1 of belonging to each respective profile (with the remaining probabilities set to 0). This allows us to obtain model-implied means for each profile without imposing any classification rule. Conceptually, this approach is similar to probing effects of a continuous predictor at meaningful values (e.g., 0 and 1), but in this case we are using the posterior probabilities generated by the latent profile model. We repeated this process for this outcome and all outcomes that follow.

As can be seen in Tables 5 and 6, when predicting support for alt-right ideology via either scale, the “last place (tied)” profile had the highest estimated marginal means in a rank-order sense, relative to the other profiles. Statistically, both the “last place (tied)” profile and the “third place-compact” profile had significantly higher estimated marginal means than the “second place (tied)” profile, but did not differ significantly from the “third place-dispersed” profile.

Table 4

Effect of Profile Probabilities on Alt-Right Outcomes, Dropping Non-Significant Interaction with Wave

Fictional Alt-Right Candidate. Support for an unnamed political candidate whose views aligned with alt-right ideology decreased significantly as the probability of belonging to the “second place (tied)” profile increased, b = -1.39, p < .001, or as the probability of belonging to the “third place–dispersed” profile increased, b = -0.99, p = .016 (see Table 4), relative to the “last place (tied)” profile. Support for this candidate also decreased marginally as the probability of belonging to the “third place–compact” profile increased relative to the “last place (tied)” profile, b = -0.58, p = .087.

As can be seen in Tables 5 and 6, estimated marginal means for each profile show that the “last place (tied)” profile has the highest support, in a rank-order sense, for the candidate whose views align with alt-right ideology. Statistically, this profile has significantly higher support for this candidate than both the “second place (tied) and “third place-dispersed” profiles, although support expressed by the “last place (tied)” profile does not significantly differ from support expressed by the “third place-compact” profile.

Table 5

Estimated Marginal Means for Models Predicting Alt-Right Outcomes from Profile Probabilities

Table 6

Pairwise Contrasts of Estimated Marginal Means for Alt-Right Outcomes with Effect Size Estimates

Extension: DEI Ban Support and Trump-Relevant Outcomes

DEI Ban Support. Next, we tested our novel hypothesis that non-Hispanic, White Americans who had a higher probability of being in the “last place (tied)” profile would report the greatest support for DEI bans—bans that aligned with the campaign platform of Trump (and ultimately were reflected during President Trump’s immediate executive orders post-inauguration).

The results of the mixed-effects model predicting DEI ban support can be seen in Table 7. Support for DEI bans decreased significantly as the probability membership in the “second place (tied)” profile increased, b = -25.07, p < .001, or as the probability of membership in “third place–dispersed” profile increased, b = -14.07, p = .040, relative to the probability of membership in the “last place (tied)” profile. Support for DEI bans also decreased as the probability of membership in the “third place-compact” profile increased relative to the probability of membership in the “last place (tied)” profile, b = -8.78, p = .122, but this effect was not statistically significant.

Of note, as can be seen in the estimated marginal means reported in Tables 8 and 9, the “last place (tied)” profile is the only profile that has significantly higher DEI ban support than both the “second place (tied) and “third place-dispersed” profiles, although it does not significantly differ from the “third place-compact” profile.

Trump Support and Trump Voting Likelihood (Pre-Election). Next, we tested our novel hypotheses that non-Hispanic, White Americans who had a higher probability of being in the “last place (tied)” profile would report the greatest support for Trump (pre-election), the greatest likelihood of voting for Trump (pre-election), and would be the most likely to actually vote for Trump (reported the day after the election). Again, we were also interested in whether the former two effects became stronger as the election neared (i.e., moderation by wave).

When predicting Trump-related outcomes, as with all other outcomes, there was no moderation by wave, indicating that the relationship between posterior probabilities of profile placement and both Trump support and Trump voting likelihood did not change from September 4th up until the election (i.e., waves 1-3; see Supplemental Materials for those non-significant interactions with wave). Thus, we report the results of our simplified models, without interactions with wave, below.

As can be seen in Table 7, and generally consistent with hypotheses, greater probability of membership in the “second place (tied)” or “third place–dispersed” profiles—relative to the “last place (tied)” profile—predicted significantly lower support for Trump (b = –1.03, p = .003; b = –0.87, p = .016, respectively) and lower likelihood of voting for him (b = –22.16, p = .002; b = –20.38, p = .007, respectively). Greater probability of membership in the “third place–compact” profile was also associated with lower Trump support (b = –0.45, p = .133) and lower likelihood of voting for him (b = –8.72, p = .164), but these effects did not reach statistical significance.

Table 7

Effect of Profile Probabilities on Trump-Related Outcomes, Dropping Non-Significant Interaction with Wave

Of note, as can be seen in the estimated marginal means in Tables 8 and 9, as when predicting support for DEI bans, the “last place (tied)” profile shows the highest support for Trump and the highest reported likelihood of voting for him, in a rank-order sense, relative to the other profiles. Statistically, it is the only profile that has significantly higher support for Trump, and higher reported likelihood of voting for Trump, than both the “second place (tied) and “third place-dispersed” profiles. Although, the “last place (tied)” profile and “third place-compact” profiles do not significantly differ from one another for either outcome.

Table 8

Estimated Marginal Means for Models Predicting Trump-Related Outcomes from Profile Probabilities

Table 9

Pairwise Contrasts of Estimated Marginal Means for Trump-Related Outcomes with Effect Size Estimates

Wave 4 Outcome: Actual Vote for Trump (Day After Election)

Finally, we predicted self-reported voting behavior, as measured at wave 4, completed the day after the election. As a reminder N = 377 participants completed wave 4. Because actual vote was only measured at one wave, this analysis is not longitudinal in nature and we simply used profile probabilities and our control variables to predict whether participants voted for Trump (1) versus another candidate/no vote (0; N = 35 reported that they did not vote) via logistic regression. In particular, a logistic regression model was estimated to predict the likelihood of voting for Trump from latent profile membership probabilities, income, education, age, and gender. The results of this analysis mirrored the results for our other Trump-relevant outcomes (see Table 10).

As shown in Table 10, a greater probability of membership in the “third place–dispersed” profile (OR = 0.46, p < .001) or the “second place (tied)” profile (OR = 0.63, p = .011) was associated with substantially and significantly lower odds of voting for Trump relative to the “last place (tied)” (omitted) profile. By contrast, a greater probability of membership in the “third place–compact” profile, relative to the probability of membership in the “last place (tied)” profile, was not associated with different odds of voting for Trump (OR = 1.11, p = .483). Thus, the highest odds of voting for Trump are associated with a greater probability of membership in either the “last place (tied)” or “third place–compact” profiles (which do not differ statistically), while membership in the “second place (tied)” and “third place–dispersed” profiles predict lower odds.

Table 10

Logistic Regression Predicting Trump Vote in 2024 Presidential Election by Posterior Probabilities of Profile Placement

Discussion

Overall, our results generally supported our three hypotheses, with some minor deviations. First, LPA combined with responses to our measure of within- and between-group subjective status (Brown-Iannuzzi et al., 2025) revealed the theorized “last place” subjective status profile that reflected a group of non-Hispanic, White Americans who feel they are “falling behind” the perceived higher status of their own racial group as well as Asian Americans. However, in contrast to prior work, this “last place (tied)” profile further reflected seeing the self as tied in status with Black Americans (and nearly so with Hispanic Americans), rather than passed by them in status. We have several reflections on this general replication and slight change from the data collected 1.5-2 years prior.

First, we think it is noteworthy that a very similar “last place” subjective status profile emerged at all. This replication, even years later, suggests that the feeling among some White Americans that they are “last place,” even if tied for it, has been a persistent undercurrent in U.S. politics. Second, while the present sample of non-Hispanic, White Americans did not rate their status as significantly below Black Americans, they did indicate that they saw their status as undifferentiated from this racial group with a long history of experiencing racism in the United States (Derenoncourt et al., 2022; Gόmez, 2022). Finally, just as in prior work, participants in the “last place (tied)” profile are not only perceiving themselves to be in last place, but they are also rating the status of all groups, including themselves, the highest on the scale as compared to the other two profiles. Thus, these “last place (tied)” White Americans do not seem to see everyone as suffering low status; instead, they seem to perceive a tight status race at the top of the SES hierarchy. It would be important for future work to continue to explore the factors that lead to these perceptions for some White Americans, but not for others.

Our results using posterior probabilities of profile placement to predict alt-right and Trump-related outcomes also generally supported our hypotheses with some caveats. Overall, the pattern across profiles, for all 5 continuous political outcomes, can be understood as a ranking from highest to lowest predicted support: last place (tied) > third place–compact > third place–dispersed > second place (tied). While these patterns replicated across all outcomes, the specific contrasts that reached statistical significance varied somewhat. Most notably, across all continuous outcomes, a higher probability of being in the “last place (tied)” profile, relative to the “third place-compact” profile, was associated with higher values on alt-right and Trump-specific outcomes, but not to a statistically significant degree. This marks a notable difference from Cooley et al., 2024 in which the contrasts between the “last place” profile and the “third place” profile (which mirrored the “third place-compact” profile here) were in the same direction as observed here, but statistically significantly different for all alt-right outcomes assessed in that work.

One reason for this deviation from prior findings may be that a higher proportion of White Americans—i.e., not only the “last place (tied)” White Americans, but also the “third place-compact” White Americans—may have been animated by perceived status threat in the time surrounding the 2024 presidential election. Indeed, increased perceptions of status threat are associated with a variety of sociocultural factors that were present in the 2024 election season such as increased conservatism (Craig & Richeson, 2014) and increased radicalization (Pfundmair & Mahr, 2022). Additionally, it is important to note that both “last place (tied)” and “third-place-compact” profiles share the perception that one’s own status is quite close to two groups who are often seen as low status (i.e., Hispanic and Black Americans). Thus, this perception of being either tied with, or close to be being tied with, the perceived low status of Hispanic and Black Americans may be an important shared factor driving associations with our political outcomes. That said, we should also note that for all Trump-relevant, continuous outcomes—namely, DEI ban support (i.e., policies associated with Trump), Trump support pre-election, and Trump voting likelihood pre-election— the “last place (tied)” profile was the only profile that reported statistically significantly higher values on these outcomes than either the “last place-dispersed” or “second place (tied)” profiles. So, while the “last place (tied)” profile and “third place-compact” profile did not significantly differ from one another, the “last place (tied)” profile still stood out as potentially most differentiated (in the high direction) on these outcomes as compared to others in the sample.

An important contribution of the present data is that they are the first to show that perception of one’s personal position in the racial economic hierarchy, controlling for objective status, is predictive of support for Trump and actual voting behaviors, as well as support for a consequential set of policies, DEI bans, which ended up marking Trump’s first Executive Orders after being inaugurated in January 2026 (The White House, 2025). As such, these findings shed light on one possible psychological mechanism behind DEI opposition—namely, if DEI aims to “lift up” minoritized racial groups like Black and Hispanic Americans, such policies may seem profoundly unfair and personally harmful to White Americans who feel they are in a tight status competition with both of these groups (as reflected in both the “last place [tied]” and “third place-compact” profiles).

Another important and novel contribution of the present manuscript is the longitudinal nature of the study. This design allowed us to assess whether higher posterior probabilities of falling in the “last place (tied)” profile predicted even higher alt-right ideology, DEI ban support, and Trump support as the election neared, perhaps due to the accumulation of campaign rhetoric aimed at increasing status threat (e.g., rhetoric that demonizes outgroups; Mutz, 2018; Savin & Treisman, 2024). However, in contrast to this possibility, the association between probabilities of membership in each profile, and our outcomes of interest, were relatively stable over time. The general absence of an effect of time could be due to our study spanning a relatively short timeframe (i.e., approximately 3 months). Indeed, research suggests that partisan identities are relatively stable over time (Kinder & Kalmoe, 2017). Additionally, within a highly polarized political climate, as we have in the U.S., voters may be more likely to stick with their established partisan identities, making these types of effects relatively stable over time (Abramowitz & Webster, 2018) and relatively unaffected by exogenous shocks (Axelrod et al., 2021).

Although our theoretical perspective, and associated choice of data analysis strategy (i.e., LPA), centered our interest in the joint influence of both within- and between-group perceptions of relative status, some may wonder whether both of these factors drive associations with our political outcomes equally, or if one factor drives that association more than the other. While it is difficult to parse out the relative contribution of intragroup versus intergroup perceptions given our measure and analytical approach, a couple of nuances to our findings can speak, at least indirectly, to this question. First, the profile with the largest gap between “self” and “White people”—reflecting the most extreme feelings of within-group deprivation, was not the “last place (tied)” nor the “third place-compact” profile. This nuance of our findings suggests that within-group comparisons alone are not the driving force behind the association of subjective status profiles and support/voting for Trump.

Relatedly, a variety of work highlights that intergroup comparisons may be especially likely to stoke prejudice and threat, which could suggest that between-group comparisons may be the more powerful factor in predicting the political outcomes of focus here (e.g., Cooley et al., 2018; Cooley et al., 2019; Wildschut & Insko, 2007). Indeed, in terms of between-group comparisons, both the “last place (tied)” and “third place-compact” profiles share the quality of all groups being perceived as relatively close together in status, consistent with high perceived intergroup competition. Thus, the perception of a tight intergroup hierarchy, or perhaps perceived proximity to two disadvantaged racial groups (i.e., Hispanic and Black Americans), may be a factor underlying the association between these two profiles and higher support for the outcomes we measured here.

That said, future research should continue to explore the exact psychological processes associated with these distinct perceptions of personal status within the racial economic hierarchy that may mediate/drive differential political attitudes among non-Hispanic, White Americans. For example, future work could utilize mixed-methods to first identify which profile participants have the highest probability of falling into via large quantitative analysis, and then follow-up with a subset of participants from each profile for qualitative interviews to better understand the psychology, background, and experiences that lead to these divergent perceptions of personal position within the racial economic hierarchy (Syed & Westberg, 2025).

Limitations and Future Directions

A limitation of the current study is its reliance on correlational data and our inability, due to power concerns, to assess possible changes in profile placement over time,^[5] which prevents us from making definitive claims about causality and the directionality of effects. While we hypothesized that the “last place (tied)” status profile would contribute to support and voting for Trump, it is equally plausible that engagement with Trump-related movements (e.g., MAGA), or rhetoric, as well as an affinity for Trump, may cultivate or intensify feelings of being last place. Future research should aim to recruit a larger sample at wave 1 in order to attain a minimum of 500 participants at each time point. Such a process would enable researchers to conduct latent profile analysis (LPA) at each time point on responses to the within- and between-group subjective status measure so that potential changes over time in posterior probabilities of profile membership, could be assessed as they relate to sociopolitical experiences. With this revised methodology, researchers could then use cross-lagged panel models to assess whether profile probabilities predict subsequent support for alt-right ideologies/Trump, or conversely, whether support for alt-right ideologies/Trump predicts subsequent changes in profile probabilities.

Relatedly, future studies should examine how profile membership may shift in response to salient social or political events. A wide range of political, social, economic, and other factors are likely to shape perceptions of one’s personal status within the racial economic hierarchy. Investigating the impact of such racialized societal events on the subsequent qualities of White Americans’ within- and between-group subjective status profiles could offer valuable insight into the broader socio-political conditions that give rise to perceptions of personal relative deprivation and/or perceptions of a tight racial economic hierarchy.

Another caveat to interpreting the present findings is the way in which we measured both within- and between-group subjective status, as well as our political outcomes. For example, research indicates that the type of scale used—here we used a recently developed 0 to 100 sliding scale on which groups and the self were placed (see Brown-Iannuzzi et al., 2025)—influences how people report inequality (Eriksson & Simpson, 2012). Likewise, it is interesting to note that although White Americans in the “last place (tied)” and “third place-compact” profiles reported the highest support for alt-right ideology and Trump-related outcomes, their means were still closer to the middle of the scales versus the extremes. This could reflect social desirability concerns that seem to lead polls to underestimate degree of support for candidates like Trump (Enns et al., 2017) or it could reflect that the beliefs of the average alt-right and Trump supporter are not as extreme as some may assume them to be (Westfall et al., 2015).

Finally, this study focused exclusively on the subjective status profiles of non-Hispanic, White Americans because of their group’s current advantaged position in the racial status hierarchy. This focus limits the generalizability of the findings to other racial and ethnic groups. Future research should examine how participants’ own race and ethnicity influences the types of subjective status profiles that emerge, especially given recent work that indicates these perceptions are quite distinct among Hispanic Americans, as compared to those observed for the non-Hispanic, White Americans sampled here (Cooley et al., 2025). Thus, it is likely that one’s own racial or ethnic identity may shape how individuals position themselves relative to both their own racial group and other racial groups—with unique predictive effects on political attitudes.

Conclusion

As economic inequality, and racial economic inequality, persist in the U.S., more White Americans may come to feel “left behind” within their own racial group while also perceiving rising threat from other racial groups. This dual sense of marginalization, both within and between groups, may amplify White Americans’ angst that they are “falling behind” with implications for U.S. political perspectives.

Conflicts of Interest

The authors declare no competing interests.

Acknowledgements

This work was support by a grant from the Russell Sage Foundation #2204-38042.

Data Availability Statement

Study materials, de-identified data, and analysis code are available at https://osf.io/uznha/.

Supplementary Materials

The supplementary materials can be found here.

Author Contributions

All authors designed the study. E.C. and J.B. worked with Cloud Research to collect the data. E.C., J.B., N.C., A.K., F.B., and W.C. analyzed the data. E.C. and N.C. wrote up the results. A.K. and F.B. drafted the introduction, methods, and discussion which was then edited by E.C. as their faculty mentor for several rounds before sharing with the rest of the research team, all of whom provided critical revisions. The full team then reviewed the updated draft of the manuscript in a second round of review. All authors then approved submission.

There were also three rounds of revisions in response to editor and reviewer comments during peer review. E.C. completed the first draft of each of these extensive revisions and adjusted data analyses in the ways requested. N.C. then reviewed those revisions and completed additional data analyses as needed. W.C. then checked the revised data analysis code. All other authors then reviewed each new draft of the manuscript before resubmission.

Endnotes

[1] Wave 1: opened Sept 4th, closed Sept 19th; wave 2: opened Sept. 30th, closed October 7th; wave 3: opened Oct. 16th, closed October 21st; wave 4: opened Nov. 6th, closed Nov. 7th; wave 5: opened Nov. 20th, closed Nov. 21st ↩

[2] As a reviewer noted, it would be very interesting to examine whether and why profile placement may change over time. However, given that several analyses supported the idea that responses to our within- and between-group subjective status measure were stable over time, and because sample size dropped below recommended sample sizes for stable profile placement in LPA (i.e., N = 500) for waves 3-5, we felt it was most reasonable to treat this variable as a stable individual difference, as measured initially at wave 1. See Supplemental Materials for more information. ↩

[3] The 3-profile model, combined with a classify-analyze approach used in Cooley et al., 2024, yields results that are highly comparable to prior findings. Likewise, Supplemental Materials includes comparisons of the highest probability profile placement for each participant in the 3-profile model vs. the 4-profile model. Perhaps most relevant to the current theory, participants placed in the “last place (tied)” profile in the 3-profile model are the only participants for whom no one was reclassified into a different profile in the 4-profile model (see Supplemental Figure 2S). This suggests that most of the uncertainty in classification was driven by difficulty sorting participants into the other three profiles, but that individuals were sorted into the “last place (tied)” profile with relative certainty. ↩

[4] We also measured support for alt-right-aligned events and social movements; however, we adjusted this measure from Cooley et al. (2024) across waves, as we added Project 2025 after wave 1. Because this is not a direct replication due to this change, these analyses appear in R code, but not here. ↩

[5] Although there was good stability in the placement of the self over time in the present data, there was only moderate stability in placements of the racial groups. ↩

References

Abramowitz, A., & McCoy, J. (2019). United states: Racial resentment, negative partisanship, and polarization in Trump’s America. The ANNALS of the American Academy of Political and Social Science, 681(1), 137–156. https://doi.org/10.1177/0002716218811309

Abramowitz, A., & Webster, S. W. (2018). Negative partisanship: Why Americans dislike parties but behave like rabid partisans. Political Psychology, 39(S1), 119–135. https://doi.org/10.1111/pops.12479

Axelrod, R., Daymude, J. J., & Forrest, S. (2021). Preventing extreme polarization of political attitudes. Proceedings of the National Academy of Sciences, 118(50), e2102139118. https://doi.org/10.1073/pnas.2102139118

Bhutta, N., Chang, A. C., Dettling, L. J., & Hsu, J. W. (2020, September 28). “Disparities in wealth by race and ethnicity in the 2019 Survey of Consumer Finances,” FEDS Notes. Washington: Board of Governors of the Federal Reserve System, September 28, 2020, https://doi.org/10.17016/2380-7172.2797‌

Blumer, H. (1958). Race prejudice as a sense of group position. The Pacific Sociological Review, 1(1), 3–7. https://doi.org/10.2307/1388607

Bobo, L., & Hutchings, V. L. (1996). Perceptions of racial group competition: Extending Blumer’s theory of group position to a multiracial social context. American Sociological Review, 61(6), 951–972. https://doi.org/10.2307/2096302

Brown-Iannuzzi, J. L., Cooley, E., McKee, S. E., & Hyden, C. (2019). Wealthy Whites and poor Blacks: Implicit associations between racial groups and wealth predict explicit opposition toward helping the poor. Journal of Experimental Social Psychology, 82, 26 34. https://doi.org/10.1016/j.jesp.2018.11.006‌

Brown-Iannuzzi, J. L., Cooley, E., Vlasak, D., Lisnek, J. A., Lei, R. F., Yeager, C., & Elacqua, N. (2025). Using a novel “Perceived Self-Group Hierarchy” measure to predict White Americans’ health via feelings of “falling behind.” Social Science & Medicine, 374. https://doi.org/10.1016/j.socscimed.2025.118061.

Brysbaert, M., & Stevens, M. (2018). Power analysis and effect size in mixed effects models: A tutorial. Journal of Cognition, 1(1), Article 9. https://doi.org/10.5334/joc.10

Caluori, N., Cooley, E., Brown-Iannuzzi, J. L., Klein, E., Lei, R. F., Cipolli, W., & Philbrook, L. E. (2024). Perceptions of falling behind “most White people”: Within-group status comparisons predict fewer positive emotions and worse health over time among White (but not Black) Americans. Psychological Science, 35(2), 175-190. https://doi.org/10.1177/09567976231221546

Cooley, E., Brown-Iannuzzi, J. L., Brown, C. S., & Polikoff, J. (2018). Black groups accentuate hypodescent by activating threats to the racial hierarchy. Social Psychological and Personality Science, 9(4), 411-418. https://doi.org/10.1177/1948550617708014

Cooley, E., Brown-Iannuzzi, J. L., Caluori, N., Elacqua, N., & Cipolli, W. (2024). White Americans who perceive themselves to be “last place” in the racial status hierarchy are most drawn to alt-right extremism. Communications psychology, 2(1), 102. https://doi.org/10.1038/s44271-024-00154-w

Cooley, E., Brown-Iannuzzi, J. L., Lei, R. F., Cipolli, W. III, & Philbrook, L. E. (2021). The policy implications of feeling relatively low versus high status within a privileged group. Journal of Experimental Psychology: General, 150(11), 2346–2361. https://doi.org/10.1037/xge0001051

Cooley, E., Hester, N., Cipolli, W., Rivera, L. I., Abrams, K., Pagan, J., Sommers, S. R., & Payne, K. (2019). Racial Biases in Officers’ Decisions to Frisk Are Amplified for Black People Stopped Among Groups Leading to Similar Biases in Searches, Arrests, and Use of Force. Social Psychological and Personality Science, 11(6), 761-769. https://doi.org/10.1177/1948550619876638

Cooley, E., Jeong, H. Y., Pacheco Ramirez, E., & Guha, M. (2025). Perceiving oneself near the top of a tight status hierarchy predicts alt-right ideology, Trump support, and opposition to DEI among Hispanic/Latine Americans. Manuscript under review for the Russell Sage Foundation issue “Inequality in America: Beliefs, Attitudes, and Actions.”

Craig, M. A., & Richeson, J. A. (2014). On the precipice of a “majority-minority” America: Perceived status threat from the racial demographic shift affects White Americans’ political ideology. Psychological Science, 25(6), 1189–1197. https://doi.org/10.1177/0956797614527113

Derenoncourt, E., Kim, C. H., Kuhn, M., Schularick, M., (2022). Wealth of two nations: The U.S. racial wealth gap, 1860–2020. Q. J. Econ. 59. https://doi.org/10.1093/qje/qjad044

Diel, K., Hofmann, W., Grelle, S., Boecker, L., & Friese, M. (2025). Prepare to compare: Effects of an intervention involving upward and downward social comparisons on goal pursuit in daily life. Personality and Social Psychology Bulletin, 51(9), 1523-1537. https://doi.org/10.1177/01461672231219378

Enns, P. K., Lagodny, J., & Schuldt, J. P. (2017). Understanding the 2016 US presidential polls: The importance of hidden Trump supporters. Statistics, Politics and Policy, 8(1), 41-63. https://doi.org/10.1515/spp-2017-0003

Eriksson, K., & Simpson, B. (2012). What do Americans know about inequality? It depends on how you ask them. Judgment and Decision Making, 7(6), 741-745. https://doi.org/10.1017/S1930297500003284

Exec. Order No. 13,958, 3 C. F. R. 215 (2020). https://trumpwhitehouse.archives.gov/presidential-actions/executive-order-establishing-presidents-advisory-1776-commission/

Ferguson, S. L., Moore, E. W. G., & Hull, D. M. (2020). Finding latent groups in observed data: A primer on latent profile analysis in Mplus for applied researchers. International Journal of Behavioral Development, 44(5), 458-468. https://doi.org/10.1177/0165025419881721

Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117–140. https://doi.org/10.1177/001872675400700202

Forscher, P. S., & Kteily, N. S. (2020). A psychological profile of the alt-right. Perspectives on Psychological Science, 15(1), 90–116. https://doi.org/10.1177/1745691619868208

Giner-Sorolla, R., Montoya, A. K., Reifman, A., Carpenter, T., Lewis, N. A., Aberson, C. L., Bostyn, D. H., Conrique, B. G., Ng, B. W., Schoemann, A. M., & Soderberg, C. (2024). Power to Detect What? Considerations for Planning and Evaluating Sample Size. Personality and Social Psychology Review, 28(3), 276-301. https://doi.org/10.1177/10888683241228328

Goethals, G. R. (2018). Donald Trump, perceptions of justice, and populism. Leadership, 14(5), 513-523. https://doi.org/10.1177/1742715018793741

Gómez, L. E. (2022). Inventing Latinos: A new story of American racism. The New Press.

Hartzell, S. L. (2018). Alt-white: Conceptualizing the ‘alt-right’ as a rhetorical bridge between white nationalism and mainstream public discourse. Journal of Contemporary Rhetoric, 8(1/2), 6-25.

Hawley, G. (2017). Making sense of the alt-right. Columbia University Press.

Hochschild, A. R. (2016). Strangers in their own land: Anger and mourning on the American right. New York, NY: The New Press.

Jardina, A. (2019). White identity politics. Cambridge University Press.

Kamenowski, M., Manzoni, P., Haymoz, S., Isenhardt, A., Jacot, C., & Baier, D. (2021). Religion as an influencing factor of right-wing, left-wing and Islamist extremism. Findings of a Swiss youth study. PLoS ONE, 16(6), e0252851. https://doi.org/10.1371/journal.pone.0252851

Kinder, D. R., & Kalmoe, N. P. (2017). Neither liberal nor conservative: Ideological innocence in the American public. The University of Chicago Press.

Knowles, E. D., & Tropp, L. R. (2018). The racial and economic context of Trump support: Evidence for threat, identity, and contact effects in the 2016 presidential election. Social Psychological and Personality Science, 9(3), 275-284. https://doi.org/10.1177/1948550618759326

Kraus, M. W., Onyeador, I. N., Daumeyer, N. M., Rucker, J. M., & Richeson, J. A. (2019). The misperception of racial economic inequality. Perspectives on psychological science: a journal of the Association for Psychological Science, 14(6), 899–921. https://doi.org/10.1177/1745691619863049

Kunst, J. R., & Obaidi, M. (2020). Understanding violent extremism in the 21st Century: The (re)emerging role of relative deprivation. Current Opinion in Psychology, 35, 55–59. https://doi.org/10.1016/j.copsyc.2020.03.010

Kuziemko, I., Buell, R. W., Reich, T., & Norton, M. I. (2014). “Last-place aversion”: Evidence and redistributive implications. The Quarterly Journal of Economics, 129(1), 105–149. https://doi.org/10.1093/qje/qjt035

Major, B., Blodorn, A., & Major Blascovich, G. (2018). The threat of increasing diversity: Why many White Americans support Trump in the 2016 presidential election. Group Processes & Intergroup Relations, 21(6), 931–940. https://doi.org/10.1177/1368430216677304

McKenzie, B. D. (2014). Political perceptions in the Obama era: Diverse opinions of the Great Recession and its aftermath among Whites, Latinos, and Blacks. Political Research Quarterly, 67(4), 823–836. https://doi.org/10.1177/1065912914541702

Muthén, L. K., & Muthén, B. O. (2017). Mplus User’s Guide (8th ed.). Los Angeles, CA: Muthén & Muthén.

Mutz, D. C. (2018). Status threat, not economic hardship, explains the 2016 presidential vote. Proceedings of the National Academy of Sciences, 115(19). https://doi.org/10.1073/pnas.1718155115

Norton, M. I., & Sommers, S. R. (2011). Whites see racism as a zero-sum game that they are now losing. Perspectives on Psychological Science, 6(3), 215–218. https://doi.org/10.1177/1745691611406922

Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling: A Multidisciplinary Journal, 14(4), 535–569. https://doi.org/10.1080/10705510701575396

Pfundmair, M., & Mahr, L. A. M. (2022). Regaining power: How feelings of exclusion during COVID-19 are associated with radicalism among critics of containment policies. Frontiers in Psychology, 13, 952760. https://doi.org/10.3389/fpsyg.2022.952760

Savin, N., & Treisman, D. (2024). Donald Trump’s words (No. w32665). National Bureau of Economic Research.

Syed, M., & Westberg, D. W. (2025). Mixed-methods research in Psychology: Rationales and research designs. Advances in Methods and Practices in Psychological Science, 8(2), 25152459251343919. https://doi.org/10.1177/25152459251343919

Tesler, M. (2020). Post-racial or most-racial? Race and politics in the Obama era. University of Chicago Press.

Urban Institute. (2024, April 25). Nine charts about wealth inequality in America. https://apps.urban.org/features/wealth-inequality-charts/

van Buuren, S., & Groothuis-Oudshoorn, K. (2011). mice: Multivariate imputation by chained equations in R. Journal of Statistical Software, 45(3), 1–67. https://doi.org/10.18637/jss.v045.i03

Wendling, M. (2025, February 13). What is project 2025? The right-wing wish list for Trump’s second term. BBC News. https://www.bbc.com/news/articles/c977njnvq2do

Westfall, J., Van Boven L., Chambers J. R., & Judd C. M. (2015). Perceiving political polarization in the United States: Party identity strength and attitude extremity exacerbate the perceived partisan divide. Perspectives on Psychological Science. 10(2), 145-58.https://doi.org/10.1177/1745691615569849

Wildschut, T., & Insko, C. A. (2007). Explanations of interindividual–intergroup discontinuity: A review of the evidence. European Review of Social Psychology, 18(1), 175-211. https://doi.org/10.1080/10463280701676543

The White House. (2025, January 20). Ending Radical And Wasteful Government DEI Programs And Preferencing. https://www.whitehouse.gov/presidential-actions/2025/01/ending-radical-and-wasteful-government-dei-programs-and-preferencing/

Zou, L. X., & Cheryan, S. (2017). Two axes of subordination: A new model of racial position. Journal of Personality and Social Psychology, 112(5), 696–717. https://doi.org/10.1037/pspa0000080