Abstract
Although there is a tendency to think all forms of essentialism—the belief that characteristics are inherent and unchangeable—are similar, some theories suggest different foundations and outcomes. We investigated if belief systems about the stability of political ideology (trait essentialism) and the fundamental nature of partisans (social essentialism) predict prejudice in opposite ways and if they do so via differential relations with blame. Across six studies (N = 2,231), we found that the more people believe the trait of political ideology is fixed (trait essentialism), the more they think that Republicans and Democrats are inherently different (social essentialism). Crucially, despite this positive correlation, trait essentialism was negatively linked to partisan prejudice and social essentialism was positively linked. The essentialism to prejudice links were driven, in part, by differential associations with blame attributions. Media messaging robustly influenced both types of essentialist thinking, with implications for prejudice.
There is increasing antagonism between members of opposing political parties (Pew Research Center, 2022). The animosity Americans hold toward members of the political opposition is referred to in various ways by social scientists, including affective polarization, social polarization, partyism, partisan animosity, and partisan prejudice (Hartman et al., 2022; Iyengar et al., 2012; Lelkes & Westwood, 2017; Mason, 2018; Sunstein, 2015). Across approaches to measuring dislike for political opposition, evidence consistently points to a dramatic increase in partisan prejudice in the United States since the 1980s (Iyengar et al., 2012; Iyengar et al., 2019). For example, between 2016 and 2022 alone, there was a marked rise in the percentage of Americans who call members of the other political party closed-minded, dishonest, unintelligent, and lazy (Pew Research Center, 2022). Notably, the percentage of people who regard members of the other party as immoral skyrocketed from 35–47% to 63–72%.
Animosity toward rival partisans can result in denigrating and avoiding the opposition, showing preferences for co-partisans, and can have implications for attitudes and behaviors beyond the political sphere (Iyengar et al., 2019; Lelkes & Westwood, 2017). For example, political prejudice can influence factors from the friendships and romantic relationships people choose, to hiring-related decisions, to where people decide to live and shop (Gift & Gift, 2015; Huber & Malhorta, 2017; Iyengar & Krupenkin, 2018; Iyengar et al., 2019). Given the stark increase in partisan prejudice in recent years, and the potential broad implications, social scientists seek to understand factors that foster this prejudice—especially those that are potentially modifiable and can be leveraged to improve attitudes (Ahler & Sood, 2018; Iyengar et al., 2019; Levendusky, 2018; Levendusky & Malhotra, 2016). In this research, we add to the growing body of literature aiming to understand the roots of partisan prejudice by focusing on cognitive factors that have received little empirical investigation in this context but are theoretically relevant. Namely, drawing on mindset and attribution theory, we explore the role of belief systems about the nature of political ideology (trait essentialism) and partisans (social essentialism). We also investigate why these beliefs are linked to prejudice with a focus on blame attributions. Finally, we explore how beliefs may be influenced by media messaging regarding political ideology and partisanship.
Trait Essentialism
We situate our current research in two rich literatures exploring the effects of beliefs in the fixedness and stability of attributes: the mindset and essentialism literatures. Trait essentialist beliefs, also referred to as fixed mindsets (Ryazanov & Christenfeld, 2018), are intuitive beliefs regarding the immutable nature of personal traits and attributes that serve as potent frameworks guiding how people interpret the world around them (Dweck, 2000; Dweck et al., 1995). Mindset scholars distinguish fixed mindsets from growth mindsets, believing that individual characteristics are malleable (Dweck, 2000; Dweck et al., 1995; Dweck & Leggett, 1988). Although much of the early research focused on understanding how beliefs about the nature of traits impact self-regulation, motivation, and goal achievement (Burnette et al., 2013), it quickly became apparent that these beliefs also mattered for how people view other people (Molden & Dweck, 2006).
Recent research has focused on the important linkage between trait essentialism and prejudicial attitudes. For example, research looking at perceptions of individuals with stigmatized traits, such as weight, shows that fixed mindsets, or trait essentialism, can reduce the extent to which people blame others for their devalued condition (e.g., Hoyt & Burnette, 2020). Decades of research into attribution theory demonstrates that the less people believe others can control their stigmatizing condition, the less prejudice they exhibit toward them (Crandall & Reser, 2005; Weiner, 1985; Weiner et al., 1988). Overall, in contexts ranging from weight to mental health to sexual orientation, trait essentialist beliefs can decrease stigma by reducing perceived responsibility, especially attributions related to controllability (Haslam & Kvaale, 2015; Hegarty, 2010; Hoyt & Burnette, 2020). When people deem others to have little control over their behavior, or have low levels of intentionality (Malle & Knobe, 1997), they are less likely to deem them responsible for their behavior (Alicke, 2000; Heider, 1958; Malle et al., 2014; Weiner, 1995). In sum, trait essentialist beliefs regarding stigmatized attributes are often associated with lower levels of prejudice. This is largely driven by a shared belief that people are not inherently personally responsible for, or cannot control, the trait.
Social Essentialism
Importantly, beliefs in the unchanging nature of traits associated with groups of people are often connected with broader perceptions about the nature of the groups that possess those traits. The more people deem personal attributes as fixed (trait essentialism), the more they think that these attributes are representative of a shared human nature (e.g., Haslam et al., 2004). That is, trait essentialism is positively associated with social essentialism. However, unlike trait essentialism, social essentialism is a strong predictor of greater prejudice, stereotyping, and discrimination (Bastian & Haslam, 2006; Haslam et al., 2002; Rothbart & Taylor, 1992; Ryazanov & Christenfeld, 2018; Yzerbyt et al., 2001). The more people perceive social groups as having a shared collective essence, the more inclined they are to attribute events around them to those group memberships (Yzerbyt & Rogier, 2001). And, the more social groups are seen as having an inherent essence, the more likely outgroup members are to attribute negative characteristics and intentions to them, resulting in prejudice (Agadullina & Lovakov, 2018). This association between social essentialist thinking and prejudice toward outgroup members is well established in scholarly work (Agadullina & Lovakov, 2018). In the domain of politics, the more people think of partisanship as a social identity, the more partisan prejudice they display (Rothschild et al., 2019). Thus, contrary to findings regarding trait essentialism, social essentialist thinking regarding outgroups is associated with greater levels of prejudice, driven at least in part by attributing negative characteristics and intentions to them.
In the current work, we explore if beliefs about the stability of political ideology (trait essentialism) and the fundamental nature of partisans (social essentialism) predict prejudice in opposite ways. Trait essentialism focuses on inherent, unchangeable characteristics, aligning with the view of political ideologies as deeply ingrained personal beliefs. Social essentialism pertains to the belief that certain groups have a shared, inherent nature, aligning with the view of political parties as monolithic entities. Thus, we conceptualize trait essentialism as beliefs about individual ideology and social essentialism as beliefs related to partisanship to capture the nuanced differences in how people perceive fixed traits versus group identities. However, we acknowledge the significant overlap between these constructs. For example, liberals and conservatives can also be seen as groups subject to social essentialism. Overall, we recognize that ideology and political party are highly correlated, and indeed we expect trait and social essentialism to be positively related to each other, yet differentially linked to prejudice.
We build upon distinct, yet converging, lines of research across different domains showing that there are contradictory effects of fixed beliefs about individual attributes relative to beliefs about groups of people with those attributes on prejudicial attitudes. For example, in the domain of mental illness, the mixed blessings model outlines how fixed biological and genetic explanations for mental disorders can both decrease stigma by attributing the illness to causes outside of the person’s control and increase stigma by increasing social essentialist beliefs about those with mental illness (Haslam & Kvaale, 2015). Additionally, in the mindset literature, the double-edged sword model outlines how believing in the fixed nature of stigmatized characteristics can paradoxically both strengthen and mitigate prejudice (Hoyt & Burnette, 2020). Fixed mindsets of traits, or trait essentialism, in stigmatizing contexts can predict less prejudice via decreased perceived responsibility for the problem, but simultaneously can predict more prejudice toward others via increasing social essentialist thinking about the group with the stigmatized traits (Hoyt & Burnette, 2020). These double-edged sword effects have been shown in a variety of contexts including weight (Hoyt et al., 2017), addiction (Burnette et al., 2024), poverty (Hoyt et al., 2023), criminality (Hoyt et al., 2022), and anorexia nervosa (Hoyt et al., 2024). Additionally, in the domain of sexual prejudice, believing that sexual identity is biologically determined and unchangeable (trait essentialism) predicts lower levels of blame and greater tolerance, whereas believing that sexual identity is a fundamental and informative category or social category (social essentialism) predicts greater sexual prejudice (Haslam & Levy, 2006; Haslam et al., 2002; Hegarty, 2010; Hegarty & Pratto, 2001; Hoyt et al., 2019). Likewise, we build on this latter work, to explore how beliefs in the unalterable nature of political ideology (trait essentialism) and the fundamental nature of partisans (social essentialism)—in this case Democrats and Republicans—relate to each other and also differentially predict partisan prejudice via differential links to blame attributions.
Proposed Model
First, we expect the two types of essentialism to be positively correlated—yet we predict that trait essentialism will be associated with lower levels of prejudice, whereas social essentialism will be associated with greater levels of prejudice. Second, building on the theoretical work in the mindset, essentialism, and attribution theory literatures, in the current research we test the mechanisms linking each type of essentialism to prejudice (see Figure 1). Specifically, building from the work showing that trait essentialist beliefs can decrease stigma by reducing perceived responsibility, especially attributions related to controllability (Haslam & Kvaale, 2015; Hegarty, 2010; Hoyt & Burnette, 2020), we explore whether trait essentialism is linked to lower prejudice via beliefs that opposite partisans are not as responsible for their political ideology and not attributing malicious intent to them—both of these types of blame are closely tied to the individual. Additionally, building from the work showing that social essentialism beliefs can lead people to attribute negative events, characteristics, and intentions to outgroup membership (Agadullina & Lovakov, 2018; Yzerbyt & Rogier, 2001), we test if social essentialism is linked to greater prejudice via beliefs that opposite partisans are to blame for the political strife, meaning the extent to which people hold the opposing political party responsible for political turmoil and division. We focus on strife generally, rather than a specific issue like inequality or inflation, to avoid complexities and noise associated with party differences in the importance placed on different social issues.
Proposed Model of the Mechanisms Linking Trait Essentialism and Social Essentialism to Partisan Prejudice.
Intervening on Essentialist Beliefs Through Media Messaging
Finally, our third aim is to investigate an important factor that might influence these beliefs: media messaging. In several domains, from weight to anxiety to addiction, media articles can effectively alter people’s beliefs regarding the changeability or static nature of attributes (Burnette et al., 2019; Hoyt, Burnette, et al., 2019; Hoyt et al., 2023). For example, disease messaging related to obesity can foster stronger trait essentialism (Hoyt et al., 2017). In the context of political ideology, we propose that media messaging can influence essentialism beliefs, with the primary goal of exploring whether media messaging can decrease partisan prejudice through messaging that either increases trait essentialism or decreases social essentialism. To alter trait essentialist thinking, we focus on a message about the genetics of political ideology. Given the strong link between genetic explanations and trait essentialism (Ryazanov & Christenfeld, 2018), media messaging on genes shaping political views likely serves to foster more trait essentialism regarding political ideology, which in turn, may reduce individual responsibility and intent attributions as well as prejudice (Dawes & Weinschenk, 2020; Edsall, 2022; Junger, 2019).
To intervene on the other lever linking essentialist thought to partisan prejudice, we examined messaging designed to decrease social essentialism. Given the positive relationship between trait and social essentialism, we also endeavored to offset any inadvertent reduction in trait essentialist beliefs in our social essentialism reduction message. Finally, in our last study, we sought to more stringently test our theoretical model by addressing the limitations of our initial approach, which aimed to reduce prejudice by either increasing trait essentialism or decreasing social essentialism. Specifically, we compared a message designed to decrease social essentialism with one designed to decrease trait essentialism. In sum, in the current work, we seek to directly manipulate not only trait essentialism about individual ideology, but also beliefs about the underlying essence of partisans, and we do so by using naturalistic media articles.
Research Overview
In this work, we extend the research on potentially modifiable cognitive factors that predict partisan prejudice by focusing on essentialism beliefs and corresponding attributions linking these beliefs to prejudice. Specifically, we aim to extend the literature on how trait and social essentialism predict animus to the context of partisan prejudice. In this research, we have three principal aims, and we outline which aims we test across the six different studies. First, our primary aim is to show that while trait essentialism and social essentialism are positively related due to their shared belief in the static and unchanging nature of things, they predict prejudice in opposite ways. Specifically, trait essentialism beliefs predict lower levels of prejudice and social essentialism beliefs predict greater levels of prejudice. As this is a primary aim, we examine these relations in all six studies.
Second, we explore the mechanisms, testing if trait essentialism predicts lower levels of prejudice through less blame, whereas social essentialism predicts greater prejudice through greater blame. Specifically, in Study 1, we first examine the social essentialism to political strife blame to prejudice link outlined in the bottom part of Figure 1. In Studies 3, 5 and 6, we evaluate the full mediation model with both types of essentialism and all blame attributions, predicting prejudice. These studies use both experimental designs and simple cross-sectional approaches.
Third, we focus on how different media messages, with ecological validity, impact essentialism beliefs, blame attributions, and subsequent attitudes toward members of the opposite political party. We manipulate beliefs in different ways across the studies. Namely, in Study 2, we focus on increasing trait essentialism, which should help to mitigate prejudice, relative to a control. In Study 3, we aim to decrease social essentialism and compare this with genetic messaging. In Study 4, we build on this and include a control group. Finally, in Study 6, we try to parse theoretical ideas by seeking to weaken both trait and social essentialism independently, and we include a control (in Study 5 we focus on aim 2 above looking more closely at the mechanisms using self-reports of essentialism, rather than manipulations). Beyond these primary goals, we also explore if effects are similar or if they differ for Democrats and Republicans, though this is purely exploratory with no a priori theoretical predictions.
Study 1
In our first study, we focused on our primary aim and started to examine the mediating mechanisms by investigating the blame link between social essentialism and prejudice.
Methods
Participants and Procedure
We sought a minimum sample of 250. Analyses reveal this is roughly the sample size needed to have .90 power to detect small effects (0.20; α = .05; n = 255; Faul et al., 2007), and 250 offers stable estimates in correlational research (Schönbrodt & Perugini, 2013). We oversampled to ensure data quality and to maximize power. Thus, we recruited 358 participants from the United States from CloudResearch’s Approved Group Participants (Hauser et al., 2022) compensating them $1.25 for participation. Using custom prescreening, we solicited half Republicans and half Democrats to help ensure our sample represented both major political parties.
Across studies, we used three techniques to help ensure data quality and screen for careless responding (Curran, 2016). Namely, we incorporated a bot check, time spent answering questions, and attention checks. Participants were included in the final sample if they passed the bot check, spent at least 2 seconds per item (Huang et al., 2012; Wood et el., 2017), and answered the attention checks correctly. The attention check items asked participants to identify what second-year high school students are called in the United States and to respond to the item “I see myself as someone who did not read this statement.” People failed the checks if they did not answer “sophomore” to the first, or if they selected “agree” or “strongly agree” for the second. In this study, 346 people remained after data screening. Of these, 21 indicated they were Independents and were removed, leaving a final sample of 325 (53.2% Democrat, 46.8% Republican; mean age = 42.55 years, age range = 22–82; 177 women, 147 men, and eight not indicating their gender; 25 Black, 12 East Asian, 263 White, 18 Latino/a/x, four Middle Eastern or North African, seven Native American, zero Hawaiian or other Pacific Islander, six South Asian, three bi/multiracial, one other race or ethnicity).
Participants gave consent before completing a measure of their political party identity and political ideology. Next, participants responded to measures of trait essentialism, 1 social essentialism, and blame. Finally, they responded to a series of measures of partisan prejudice before responding to demographic questions. Our data and materials for all studies are available on the Open Science Framework (OSF; https://osf.io/bj8z4/). We report all manipulations, measures, and exclusions in these studies. The Institutional Review Board (IRB) approved all procedures.
Measures
Participants responded to items using a seven-point scale from strongly disagree to strongly agree unless noted otherwise.
Political Party
We assessed participants’ political party and ideology with well-established measures (Levendusky, 2017). Participants were asked to indicate if they generally consider themselves a Democrat, Republican, or Independent. Those responding as Independent were asked if they think of themselves as closer to the Democratic or Republican party (however, we excluded these participants and did not use this information). We also asked participants if they consider themselves a strong or not very strong partisan. We focused on partisanship in this research as our primary outcome measures of prejudice were focused on opposite partisans. Although we focus on partisanship, we did ask participants to identify their political views from extremely liberal to extremely conservative. As expected, political party identification and ideology were highly correlated across studies. 2
Trait Essentialism
We assessed trait essentialist thinking about political ideology with eight items: five modified from traditional mindset scales and three items modified from trait essentialism scales (Hegarty & Pratto, 2001; Hoyt, Morgenroth, et al., 2019). Participants indicated their agreement to items such as “Political orientation is caused by biological factors such as genes and hormones” and “Your political orientation is something that you can’t change.” Higher values represent stronger trait essentialism.
Social Essentialism
Participants responded to a four-item scale of social essentialism modified from Hoyt, Morgenroth, et al.’s (2019) measure of social essentialism of sexual orientation. Participants responded to items such as “Members from the same political party (e.g., conservatives), are very similar to one another and have many things in common” and “Knowing that someone belongs to a particular political party tells us a lot about that person’s character.” Higher values represent stronger social essentialist beliefs.
Political Strife Blame
In this study, we assessed the extent to which people blame members of the opposite party for the political divide with the following three items: “[OppPartisans] are to blame for the turmoil in politics,” “[OppPartisans] are responsible for today’s political problems,” and “The political divide today is the fault of the [OppPartisans].” Higher numbers represent greater blame attributions.
Partisan Prejudice
We incorporated four different assessments of prejudice, including affective and cognitive components. The three scale-based measures, negative affect, negative trait evaluations, and the disliking thermometer, were highly correlated across studies and generally showed the same effects. Thus, we combined for ease of presentation, and we present disaggregated results in the supplemental materials at OSF.
Negative Affect
Using the measures from Osmundsen et al. (2021), we asked participants to indicate how strongly they feel both negative (angry, frustrated, afraid) and positive (hopeful, enthusiastic, proud) emotions when they think about Democrats and Republicans. We examined affective responses to their opposing political party members. We recoded positive items so higher numbers represent more negative affect.
Negative Trait Evaluations
Using another well-established measure of partisan polarization, we asked participants to indicate how well eight traits apply to the opposing party using a five-point scale, ranging from “not well at all” to “extremely well” (Garrett et al., 2014; Levendusky, 2018). The traits included: American, intelligent, honest, open-minded, generous, hypocritical, selfish, and mean. We recoded the positive traits so that higher numbers represent more negative trait evaluations, and thus more negative attitudes.
Disliking Thermometer
Participants responded to a single-item feeling-thermometer question from the American National Election Study (ANES), a popular approach to assessing affective polarization (Iyengar et al., 2012; Levendusky, 2017). They were asked to rate how favorable and warm they feel toward both Democrats and Republicans on a thermometer graded from 0 to 100. We examined responses to their opposing political party members. Values were recoded such that higher scores indicate less liking and warmth.
List of Likes and Dislikes
Finally, we asked participants to list anything that they like and dislike about the opposing party (Levendusky & Malhorta, 2016; Levendusky, 2017). They were given the opportunity to list up to six of each. We summed the number of likes and dislikes they entered. Each unique like or dislike listed was counted as 1, except for mentions of the opposite partisans being human or being people, which were counted as 0.5. A small number of participants (n = 6) listed more than 6 dislikes; we capped their total number at 6.5. The range of scores of likes was 0 to 6 and dislikes was 0 to 6.5. For analyses, we focus on the number of dislikes given our focus on partisan prejudice.
Results
See Table 1 for means, standard deviations, and correlations between scales.
Means, Standard Deviations, Reliability, and Bivariate Correlations.
Note: + = p < .06; * = p < .05; ** = p < .01; *** = p < .001; Political party: 1= Democrat, 2 = Republican; TE-Inc = Trait Essentialism Increase Message; SE-Red = Social Essentialism Reduction Message; TE-Red = Trait Essentialism Reduction Message; Cont = No Message Control.
Aim 1: Essentialism and Prejudice
We started by exploring our primary questions. Is trait essentialism of political ideology positively correlated with social essentialism, and do these two belief systems predict partisan prejudice in opposite directions? First, trait essentialism and social essentialism were significantly, positively, and moderately correlated with each other (see Table 1). Second, we found that stronger endorsement of trait essentialism negatively predicted the prejudice composite and the number of dislikes regarding the opposite party. Additionally, as expected, social essentialism showed the opposite pattern; greater endorsement of social essentialist beliefs positively predicted prejudice and the number of dislikes regarding members of the opposite party.
Next, we explored the relative strength of the association between these two belief systems and prejudice by conducting ordinary least squares regressions with both types of essentialism predicting partisan prejudice. Both trait essentialism (ß= −.31, p < .001, 95% CI = −0.30, −0.15) and social essentialism (ß= .39, p < .001, 95% CI = 0.24, 0.41) simultaneously and robustly predicted the prejudice composite in opposite directions. Overall, in this model 18% of the variance in partisan prejudice was explained by trait and social essentialism (R2 = .18). Similarly, both trait essentialism (ß= −.22, p < .001, 95% CI= −0.57, −0.19) and social essentialism (ß= .26, p < .001, 95% CI = 0.32, 0.76) simultaneously and robustly predicted the number of dislikes in opposite directions. Additionally, 9% of the variance in number of dislikes for opposite party was attributed to these two belief systems (R2 = .09).
Aim 2: Mediating Mechanism of Political Strife Blame
Next, we examined the role of political strife blame in the relationship between social essentialism and partisan prejudice—the bottom portion of the overall model outlined in Figure 1. The political strife blame measure was significantly associated only with social essentialism and not trait essentialism (see Table 2). Additionally, political strife blame was positively associated with the prejudice composite. Finally, to test for mediation, we used Hayes’ (2018) PROCESS Model 4 to compute bootstrap-based confidence intervals (95%) for the estimate of the indirect effect of social essentialism on prejudice through blame. There was a significant indirect effect of essentialism on both the prejudice composite and number of dislikes.
Study 1: Indirect Effects of Social Essentialism on Measures of Partisan Prejudice through Blame.
Exploratory Question: Moderation by Political Party
We then explored whether these relationships between the two belief systems and the measures of partisan prejudice differed across the two parties. Specifically, we conducted moderation analyses using PROCESS Model 1, with the belief system and party identification (1 = Democrat, 2 = Republican) predicting each outcome measure. We found a significant interaction between trait essentialism and political party on the prejudice composite (see Table 3), but not number of dislikes (p = .181). This interaction revealed that the negative association between trait essentialism and partisan prejudice was significant only for Democrats but not Republicans 3 (see Figure 2). There were no significant interactions between social essentialism and party on either measure (ps > .534).
Study 1: Conditional Effects of Trait Essentialism on Partisan Prejudice for Democrats and Republicans.
Study 1: The Association Between Trait Essentialism and Partisan Prejudice as a Function of Political Party.
Study 1 Discussion
In sum, findings supported our primary predictions and extended past work showing contradictory effects of essentialism beliefs to the domain of political prejudice (e.g., Haslam & Kvaale, 2015). Namely, the more people believed that the trait of political ideology is fixed, the more they believed that conservatives and liberals are inherently different types of people. Critically, despite this positive link between the two types of essentialism, there were differentiating links to prejudice. Whereas trait essentialism of ideology predicted less partisan prejudice, social essentialism beliefs about the nature of partisans predicted greater prejudice. Regression analyses revealed that these effects were robust and independent of one another.
Examining the mechanism for social essentialism, we found that social essentialist beliefs predicted partisan prejudice through blaming opposite partisans for the political strife in society. Next, exploratory analyses revealed that the link between greater endorsement of trait essentialism and decreased levels of partisan prejudice was found only for Democrats and not Republicans on the composite partisan prejudice measure, but not the number of dislikes measure. There were no interactions between party affiliation and social essentialism on prejudice.
Study 2
Having established the basic relationships between trait essentialism, social essentialism, and partisan prejudice, we turned our focus in this study to our third aim: how does messaging about political ideology impact essentialism and subsequent prejudice? In Study 2, we focused on trait essentialism, including a message that outlined the genetic underpinnings of political orientation. Namely, we tested the prediction that, relative to a no-message control, a genetics of ideology message will increase trait essentialism and, in turn, decrease prejudice.
Methods
Participants and Procedure
We sought a minimum sample needed to have .90 power to detect small to medium effects of analyses with two groups (.35; α = .05; n = 282; Faul et al., 2007). We recruited 359 participants from the United States from CloudResearch’s Approved Group participants (Hauser et al., 2022), compensating them $1.25 for participation. We used custom prescreening to help ensure roughly equal representation of both major political parties. We used the same techniques for data screening as in Study 1. In this study 339 people remained after data screening, and 17 were Independents, leaving a final sample of 322 (52.2% Democrat, 47.8% Republican; mean age = 41.66 years, age range = 19–72; 167 women, 151 men, two non-binary, and one transgender; 26 Black, 17 East Asian, 256 White, 29 Latino/a/x, zero Middle Eastern or North African, zero Native American, one Hawaiian or other Pacific Islander, three South Asian, three bi/multiracial, one other race or ethnicity).
After giving consent, participants completed a measure of their political party identity and political ideology before being randomly assigned to read an article discussing the role of genes in political ideology (n = 159) or to a no-treatment control (n = 163). Next, participants responded to a manipulation check (i.e., a measure of genetic attributions) before responding to measures of trait and social essentialism. Finally, participants responded to a series of measures of partisan prejudice before responding to demographic questions.
Article on the Genetics of Political Ideology
We asked half of the participants to read a shortened version of an article from the New York Times (Edsall, 2022) entitled How Much Do Your Genes Shape Your Politics? The article briefly chronicles mounting research that finds evidence that political attitudes and behaviors are at least partially heritable. After reading the article, participants were asked to briefly indicate how they would describe the main message of the article to a friend. The other half of the participants were in the no-treatment control condition and just completed the surveys. We recognize one limitation of this methodological approach is the discrepancy in survey length between the experimental and control conditions.
Measures
Participants completed the same measures as in Study 1 except we did not assess blame. 4
Manipulation Check
We assessed whether we adequately manipulated their beliefs in the genetic underpinnings of political ideology using a measure modified from Schneider et al. (2018). We presented participants with a series of eight behaviors, orientations, and conditions (e.g., height, being addicted to drugs) and asked them to indicate, using a percentage, the degree to which they believed it to be shaped by genetics, by the social environment, and by personal free will. For each item, the three percentages had to add up to 100%. We examined participants’ responses to the final item: being politically conservative/liberal.
Results
See Table 1 for means, standard deviations, and correlations between scales.
Aim 1: Essentialism and Prejudice
We started by looking at the basic correlations among the variables of interest. Once again, trait essentialism and social essentialism were significantly, positively correlated with each other (see Table 1). Next, we found that trait essentialism was negatively correlated with partisan prejudice on both the composite and dislike measures and social essentialism was positively correlated with both measures of prejudice.
Aim 3: Media Messages and Beliefs
To investigate the manipulation check, we conducted a MANOVA with all three attributions related to political orientation (i.e., genetics, environment, and free will) as the dependent variables and experimental condition as the between-subjects factor. Tests of between-subjects effects revealed that participants in the genetic message condition reported greater genetic attributions (M = 17.04; SD = 18.10) and fewer free will/choice attributions (M = 47.38; SD = 22.97) relative to those in the control condition (genetics: M = 6.92; SD = 13.91; F (1, 320) = 31.76, p < .001; partial η2 = .09; 95% CI = −13.66, −6.59; free will/choice: M = 57.45; SD = 25.23; F(1, 320) = 14.00, p < .001; partial η2 = .04; 95% CI = 4.78, 15.37). There was no effect of condition on attributions to the environment (p = .981).
Next, we examined the effect of experimental condition on both beliefs systems and on the measures of partisan prejudice. First, we conducted a MANOVA with trait essentialism and social essentialism as the dependent variables and experimental condition as the between-subjects factor. Participants in the genetic message condition reported greater endorsement of trait essentialism of political ideology (M = 2.55; SD = 1.13) relative to those in the control condition (M = 2.27; SD = 1.05; F (1, 320) = 5.29, p = .022; partial η2 = .02; 95% CI = −0.52, −0.04). However, there was no significant effect of experimental condition on social essentialism (F(1, 320) = 1.17, p = .281; partial η2 = .00; 95% CI = −0.36, 0.10). Next, a MANOVA on both partisan prejudice measures revealed no significant between-subjects effects (ps > .355). 5
We then examined whether the experimental condition had an indirect effect on partisan prejudice through its effect on trait essentialism using PROCESS Model 4. As can be seen in Table 4, reading an article about the genetic underpinnings of political ideology indirectly predicted lower levels of both measures of prejudice through increased trait essentialism. Results were similar when including social essentialism as a covariate.
Study 2: Indirect Effects of Experimental Condition on Measures of Partisan Prejudice Through Trait Essentialism.
Exploratory Question: Moderation by Political Party
Finally, we examined if the effects were moderated by political party. First, using PROCESS Model 1, we explored if the effect of experimental condition on trait and social essentialism differed by political party. There was no significant interaction between condition and party on either trait essentialism or social essentialism (ps > .184). Lastly, given the results from Study 1, we once again explored whether the relationships between essentialism and the measures of partisan prejudice differed across the two parties using PROCESS Model 1, controlling for experimental condition. Once again party did not moderate the relationship between social essentialism and either measure of prejudice (ps>.928). However, contrary to Study 1, party did not moderate the association between trait essentialism and the prejudice composite (B = 0.10, SE = 0.09; p = .260; 95% CI = −0.07, 0.27) or number of dislikes (B = 0.27, SE = 0.20; p = .179; 95% CI = −0.13, 0.67).
Study 2 Discussion
In terms of aim 1, once again, trait and social essentialism were significantly and positively correlated with each other, but had differential associations with partisan prejudice. Specifically, stronger endorsement of trait essentialism predicted less partisan prejudice whereas greater endorsement of social essentialist beliefs predicted greater partisan prejudice. In testing aim 3, we first performed a manipulation check making sure the primary message was interpreted as intended. Participants in the genetic message condition reported stronger genetic attributions and reported greater endorsement of trait essentialism of political ideology relative to those in the control condition. There was no significant effect of experimental condition on social essentialism. Additionally, reading an article about the genetic underpinning of political ideology indirectly predicted lower levels of partisan prejudice toward members of the opposing party through increased trait essentialism. Finally, analyses showed that political party did not moderate the effects of the experimental condition on essentialism. Contrary to Study 1 findings, the link between greater endorsement of trait essentialism and decreased levels of partisan prejudice on the composite measure did not differ for Democrats versus Republicans.
In Study 2, we explored correlations and started to investigate our experimental manipulation aim, examining how genetic messaging impacted beliefs. In Study 3, we merged all three aims, looking at relations between essentialism and prejudice, as well as exploring the mechanisms and the impact of media messaging. In order to further test the mechanisms, we added a two-item measure potentially linking trait essentialism to prejudice, in addition to including the political strife blame measure from Study 1. We also examined social essentialism messaging, in addition to the trait essentialism messaging used in this study.
Study 3
Given the practical constraints of relying predominantly on genetic-focused media messaging to decrease partisan prejudice, and in an effort to test the malleability of social essentialism beliefs, in this preregistered study (https://osf.io/y276g/) we tested the effects of a media message designed to minimize social essentialist beliefs while preserving similar levels of trait essentialism to those induced by the genetic narrative (the full message is available on the Open Science Framework). For example, the article explained that a person’s identification as Democrat or Republican is only a singular aspect of their multifaceted identity, and knowing someone’s political identity provides limited insight into their overall character and does not capture the complexity of who they are. It emphasized that knowing someone’s political affiliation tells us very little about the person’s core characteristics or values. Additionally, we incorporated language emphasizing trait essentialism. We compared responses to this message relative to the message from Study 2 solely focused on emphasizing trait essentialism by focusing on the genetic foundation of ideology. We predicted that the social essentialism reduction message, relative to the trait essentialism enhanced message about the genetic underpinnings of political ideology, would predict lower levels of social essentialism and thus would be negatively linked to partisan prejudice. Thus, we expected a negative indirect effect of condition on prejudice via a reduction in social essentialism. We also explored effects of these messages on trait essentialism. It seems plausible that even though the social essentialism reduction message included language emphasizing trait essentialism, it may still weaken trait essentialism compared with the genetic message. If this is the case, this message may fail to directly impact prejudice due to the opposing effects of these two belief systems. Finally, focusing on the mechanisms, we explored whether deeming people less responsible for their political ideology mediates the link between trait essentialism and partisan prejudice and whether, as shown in Study 1, blaming opposite partisans for political strife mediates the association between social essentialism and partisan prejudice.
Methods
Participants and Procedure
We sought a minimum sample needed to have adequate power to detect small to medium effects of analyses with two groups (effect = .35; power = .80; α = .05; n = 260; Faul et al., 2007) and to detect small to medium regression coefficients (.26) in bias-corrected bootstrap tests (power = .80; n = 148; Fritz & MacKinnon, 2007). We recruited 351 participants from the United States, with roughly equal representation from both major political parties, using CloudResearch’s Approved Group participants (Hauser et al., 2022). Participants were compensated $1.25 for their participation. Using the same approach to data cleaning as in the first two studies, in this study 315 people remained after data screening. Of these, 13 indicated they were Independents and were removed leaving a final sample of 302 (51.7% Democrat, 48.3% Republican; mean age = 41.74 years, age range = 20–75; 151 women, 150 men, and one non-binary participant; 25 Black, 14 East Asian, 235 White, 23 Latino/a/x, one Middle Eastern or North African, nine Native American or American Indian, one Hawaiian or other Pacific Islander, five South Asian, two bi/multiracial, one other race or ethnicity).
The methodology was similar to Study 2. Before completing the survey, participants were randomly assigned to read one of two articles: the article on the genetics of political ideology used and described in Study 2 (n = 148) or a new article created by the research team with an added social essentialism reduction message (n = 154).
Social Essentialism Reduction Message
We asked half of the participants to read an article that gave a message designed to decrease social essentialism. The title of the article was “We are more similar than different: Democrats and Republicans share similar values.” Specifically, the social essentialism reduction (SE-Reduction) message aimed to promote a nuanced understanding of the essence of others by explaining that the composition of individual identity is complex and cannot be reduced to a single aspect such as being a Democrat or Republican. For example, the article stated, “It’s important to understand that people are multifaceted and political affiliation is only a small part of someone’s identity.” Additionally, to try and offset the potential that such a message also reduces trait essentialist beliefs, we described people’s politics as often being shaped by things that are not controllable, including genes. The overall goal of this message was to leverage the prejudice-minimizing forces of reduced social essentialist beliefs.
Measures
Participants completed the same essentialism and prejudice measures as in the previous studies. They also responded to the political strife blame measure from Study 1 in addition to a measure of political ideology responsibility.
Political Ideology Responsibility
We assessed the extent to which people believe others are responsible for their political ideology with the following three items: “People are responsible for their political ideology,” “Someone’s political ideology is something that is under their control,” and “People should be blamed for their political orientation.” The third item was not reliable with the other two; thus, we computed the scale with the first two items. Higher numbers represent greater responsibility attributions.
Results
See Table 1 for scale means, standard deviations, reliabilities, and correlations between scales.
Aim 1: Essentialism and Prejudice
In support of our first set of hypotheses, trait and social essentialism were significantly, positively correlated with each other, and stronger endorsement of trait essentialism predicted lower levels of partisan prejudice on both variables, whereas endorsement of social essentialism predicted greater levels of prejudice (see Table 1).
Aims 2–3: Mediating Mechanisms and Media Messages
Next, we tested the prediction that the social essentialism reduction (SE-Reduction) message, compared with the genetic underpinnings/trait essentialism increase (TE-Increase) message, would predict lower levels of social essentialism, and we explored the effect of condition on trait essentialism as well as on mediating the mechanisms. We first conducted a MANOVA with experimental condition as the between-subjects factor and social essentialism and trait essentialism as the dependent variables. Tests of between-subjects effects revealed that participants in the SE-Reduction message condition reported weaker social essentialist beliefs (M = 4.32; SD = 1.07) relative to those in the TE-Increase condition (M = 5.05; SD = 1.03; F(1, 300) = 36.96, p < .001; partial η2 = .11; 95% CI = 0.50, 0.97). Participants in the SE-Reduction message condition also reported weaker trait essentialism (M = 2.47; SD = 1.12) relative to those in the TE-Increase condition (M = 2.98; SD = 1.20; F (1, 300) = 14.80, p < .001; partial η2 = .05; 95% CI = 0.25, 0.78). This is likely, in part, due to the positive correlation between the two beliefs. Thus, the social essentialism reduction messaging reduced both social and trait essentialism.
To test the prediction that the SE-Reduction message would predict lower levels of prejudice relative to the TE-Increase message, we conducted a MANOVA on the partisan prejudice composite and number of dislikes measures. There was a marginally significant between-subjects effect for the composite measure, with lower levels of partisan prejudice from those participants in the SE-Reduction message condition (M = −0.08; SD = 0.90) relative to those in the TE-Increase condition (M = 0.12; SD = 0.88; F (1, 297) = 3.89, p = .050; partial η2 = .01; 95% CI = 0.00, 0.40). 6
We then tested if the SE-Reduction message, relative to the TE-Increase message, indirectly predicted prejudice through weakened social essentialism. We used PROCESS Model 4 to compute bootstrap-based confidence intervals (95%) for the estimate of the indirect effects of condition on partisan prejudice. As can be seen in Table 5, relative to reading a message about the genetic underpinnings of political ideology, reading the social essentialism reduction message indirectly predicted lower levels of partisan prejudice on both measures through decreased social essentialism. However, the SE-Reduction message also indirectly predicted greater levels of partisan prejudice through weakened trait essentialism of political ideology.
Study 3: Indirect Effects of Experimental Condition on Measures of Partisan Prejudice Through Trait and Social Essentialism.
Next, we explored the role of deeming people responsible for their ideology and blaming partisans for political strife in the observed effects. We preregistered to analyze effects only looking at the hypothesized mediator, but here we present results from indirect effect analyses, using PROCESS Model 4, that include both mediators. 7 We tested the predictions that responsibility for political ideology might mediate the link between trait essentialism and prejudice while blame for political strife might mediate the association between social essentialism and prejudice. First, with trait essentialism as the predictor, responsibility for political ideology did not mediate the association as predicted. Unexpectedly, blame for strife acted as a suppressor variable, showing a positive indirect effect between trait essentialism and prejudice (see supplemental materials at OSF). Similar analyses with social essentialism as the predictor revealed a robust indirect effect of social essentialism on prejudice through blame for political strife on both measures of partisan prejudice, and there was also a small positive indirect effect of responsibility on number of dislikes (see Table 6).
Study 3: Indirect Effects of Social Essentialism on Measures of Partisan Prejudice Through Responsibility and Blame.
Exploratory Question: Moderation by Political Party
Finally, we explored if the effects were moderated by participants’ political party. First, moderation analyses revealed that political party did not moderate the effect of experimental condition on either trait essentialism or social essentialism (ps > .170). Next, once again party did not moderate the relationship between social essentialism and either measure of prejudice (ps>.303). Similar to Study 2 and contrary to Study 1, party did not moderate the association between trait essentialism and the prejudice composite (B = −0.00, SE = 0.09; p = .975; 95% CI = −0.17, 0.17) or number of dislikes (B = −0.24, SE = 0.20; p = .221; 95% CI = −0.64, 0.15).
Study 3 Discussion
In this study, we successfully replicated the associations between both types of essentialism and partisan prejudice. Trait essentialism and social essentialism were significantly, positively correlated with each other, but a stronger endorsement of trait essentialism predicted less partisan prejudice and a greater endorsement of social essentialist beliefs predicted more prejudice. Additionally, the social essentialism reduction message, relative to a message solely focused on the genetic underpinnings of ideology, did reduce social essentialist beliefs as intended. The social essentialism reducing message directly led to lower levels of partisan prejudice on one measure of prejudice, and the message indirectly predicted lower levels of prejudice across both measures through decreased levels of social essentialism. However, this message also led to a weaker endorsement of trait essentialism and thus there was simultaneously an opposing positive indirect effect, predicting greater levels of prejudice, through weaker trait essentialism. These effects were relative to a pure genetic message and thus implications with other types of control groups are less clear. Thus, in our next study, we added a no-treatment control.
Examining the mechanisms in this study, we once again found that social essentialist beliefs predicted partisan prejudice through blaming opposite partisans for the political strife in society—a replication of Study 1. In addition, we found a small indirect effect through our new measure of responsibility for political ideology. However, responsibility for political ideology failed to mediate the negative effect between trait essentialism and partisan prejudice as expected. Though trait essentialism was associated with a reduced perception of responsibility, this perception did not in turn predict prejudice. This suggests that perceived responsibility may not be the mechanism through which trait essentialism predicts prejudice, or it may indicate a limitation of our two-item scale’s ability to accurately capture the construct. Finally, political party did not moderate the effects of the experimental condition on essentialism, and, similar to Study 2, the link between greater endorsement of trait essentialism and decreased levels of partisan prejudice did not differ for Democrats versus Republicans.
In Study 4, we focused on aims 1 and 3, continuing to test and hone our examination of the impact of media messaging. We did not examine the mechanisms in Study 4. In Study 5, we focused on identifying the mechanisms linking trait essentialism beliefs to lower levels of partisan prejudice. In Study 6, we again pursued all three aims by reporting correlations among constructs, examining the mechanisms, and manipulating beliefs via media messaging.
Study 4
In this preregistered study (https://osf.io/c5w3s/), we more comprehensively explored the role of media messaging on partisan prejudice by comparing both media messages to a no message control condition. In this study, participants either read one of the two messages from Study 3, the mainstream news message on the genetic underpinnings of political ideology or the social essentialism reduction message, or they read no message before completing measures. In addition to aiming to replicate the basic pattern of correlational relationships among trait essentialism, social essentialism, and prejudice, we predicted that the social essentialism reduction message, relative to a message about the genetic underpinnings of political ideology and the no message control, would predict lower levels of partisan prejudice and do so indirectly via a reduction in social essentialism. Finally, we also reported trait essentialism across experimental conditions, expecting similar effects found in previous studies.
Methods
Participants and Procedure
We sought to recruit a similar sample size as solicited in the first two experimental studies with 175 participants per condition. We recruited a panel of participants from the United States using CloudResearch’s Connect platform (Hartman et al., 2023), requesting equal representation from both major political parties. Five hundred and twenty-three participants completed the study and were compensated $1.25 for their participation. Using the same approach to data cleaning as in previous studies, in this study 505 people remained after data screening. Of these, 14 indicated they were Independents and were removed leaving a final sample of 491 (51.3% Democrat, 48.7% Republican; mean age = 41.22 years, age range = 18–80; 237 women, 252 men, and one non-binary participant; 50 Black, 40 East Asian, 372 White, 41 Latino/a/x, zero Middle Eastern or North African, five Native American or American Indian, zero Hawaiian or other Pacific Islander, eight South Asian, three bi/multiracial, two other race or ethnicity).
We randomly assigned participants to read either the article on the genetics of political ideology (n = 161) or the article with a social essentialism reduction message (n = 164), or they were not asked to read an article (n=166). Next, participants completed the measures of trait essentialism, social essentialism, and partisan prejudice.
Results
See Table 1 for scale means, standard deviations, reliabilities, and correlations between scales.
Aim 1: Essentialism and Prejudice
In support of our first set of hypotheses, trait essentialism and social essentialism were significantly and positively correlated with each other. Also, similar to earlier studies, stronger endorsement of trait essentialism predicted less partisan prejudice, whereas stronger endorsement of social essentialism predicted greater levels of partisan prejudice 8 (see Table 1).
Aim 3: Media Messaging
Next, we tested if messages impact essentialism (see Figure 3). We conducted a MANOVA with experimental condition (2 = Social Essentialism Reduction [SE-Reduction] Msg, 1 = Trait Essentialism Increase [TE-Increase] Msg, 0 = Control) as the between-subjects factor and trait and social essentialism as the dependent variables. There were significant effects of condition on both essentialist beliefs (see Table 7). Tukey’s HSD post-hoc tests showed that participants in the SE-Reduction condition reported significantly lower levels of social essentialism relative to those in both the TE-Increase condition and Control condition; the means in the TE-Increase and Control conditions were the same. Additionally, post hoc analyses showed that participants in the TE-Increase condition reported higher levels of trait essentialism relative to those in the Control condition, similar to Study 2 findings. The mean of trait essentialism in the SE-Reduction condition was in between the other two conditions and not significantly different from either.
Study 4: Means for Essentialism and Prejudice Across Experimental Conditions.
Study 4: Univariate Effects and Post-Hoc Tests for Effects of Experimental Condition on Trait and Social Essentialism.
Note: * p < .01 ** p < .001; TE-Inc = Trait Essentialism Increase Message; SE-Red = Social Essentialism Reduction Message; Cont = No Message Control; HSD = Honestly significant difference.
Next, to test the prediction that the SE-Reduction message would predict lower levels of prejudice relative to both the TE-Increase message and the Control condition, we conducted a MANOVA with experimental condition predicting partisan prejudice (see Figure 3). There were significant effects of condition on the prejudice composite and the number of dislikes (see Table 8). Tukey’s HSD post-hoc tests showed that participants in the SE-Reduction condition reported significantly lower levels of prejudice relative to those in the Control and TE-Increase conditions. Additionally, post hoc analyses revealed that participants in both the SE-Reduction and the TE-Increase condition reported fewer number of dislikes relative to those in the Control condition.
Study 4: Univariate Effects and Post-Hoc Tests for Effects of Experimental Condition on Prejudice Measures.
Note: * p < .05; TE-Inc = Trait Essentialism Increase Message; SE-Red = Social Essentialism Reduction Message; Cont = No Message Control; HSD = Honestly significant difference.
We tested if the SE-Reduction message, relative to the TE-Increase and Control messages, indirectly predicted prejudice through weakened social essentialism. We conducted preregistered indirect effect analyses using PROCESS Model 4 with condition (1 = SE-Reduction, 0 = TE-Increase and Control) as the predictor variable, trait and social essentialism as the mediators, and partisan prejudice as the outcomes. As shown in Table 9, compared with either reading no message or the TE-Increase message, reading the SE-Reduction message indirectly predicted lower levels of the prejudice composite and fewer dislikes about opposing partisans through decreased social essentialism. There were no significant indirect effects of trait essentialism.
Study 4: Indirect Effects of Condition on Measures of Partisan Prejudice Through Trait and Social Essentialism.
Note: Experimental Condition (1 = SE-Reduction Msg, 0 = TE-Increase Msg and Control).
Because participants in the TE-Increase condition reported higher levels of trait essentialism relative to those in the Control condition, we explored if trait essentialism mediated differences in prejudice between these conditions. We used a similar indirect effect analytic approach as above, but the predictor variable was the contrast between those in the TE-Increase condition relative to the Control condition. Trait essentialism did not mediate the effect of the condition on the partisan prejudice composite, 9 but did mediate the number of dislikes (indirect effect = −0.14, SE = 0.06, 95% CI = −0.28, −0.04).
Exploratory Question: Moderation by Political Party
Finally, we explored if the effects were moderated by participants’ political party. First, moderation analyses revealed that political party did not moderate the effect of experimental condition (SE-Reduction vs TE-Increase and Control) on social essentialism or condition (TE-Increase vs. Control) on trait essentialism (ps > .367). Next, the association between social essentialism and both measures of prejudice were not moderated by party (ps > .352). However, like in Study 1, political party moderated the association between trait essentialism and the prejudice composite such that the association was only significant for Democrats, not Republicans (see Table 10 and Figure 4).
Study 4: Conditional Effects of Trait Essentialism on Partisan Prejudice for Democrats and Republicans.
Study 4: The Association Between Trait Essentialism and Partisan Prejudice as a Function of Political Party.
Study 4 Discussion
Once again, we replicated the basic associations between trait essentialism, social essentialism, and partisan prejudice. Both trait and social essentialism were significantly, positively, correlated with each other. Yet, a stronger endorsement of trait essentialism predicted less partisan prejudice and a greater endorsement of social essentialist beliefs predicted more prejudice.
Additionally, we built on Study 3 by testing the genetic underpinnings message relative to a social essentialism reduction message and this time we included a no message control. We predicted and found that the social essentialism reduction message, relative to both a message solely focused on the genetic underpinnings of ideology or a no message control condition, did reduce social essentialist beliefs as intended. Additionally, the social essentialism reduction message led to lower levels of the partisan prejudice composite relative to the other two conditions and fewer number of dislikes relative to the control condition. Furthermore, the message indirectly predicted lower levels of prejudice through decreased levels of social essentialism relative to the other two conditions. We also explored the effects of the conditions on trait essentialism. Inconsistent with the finding from Study 3, there was no difference in trait essentialism across the two message conditions. However, we did replicate the findings in Study 2; those in the TE-Increase condition reported greater endorsement of trait essentialism relative to those in the control condition. Moreover, there was only an indirect effect of the TE-Increase message relative to the control condition through trait essentialism on the number of dislikes. Finally, we once again found that the link between a stronger endorsement of trait essentialism and decreased scores on the partisan prejudice composite was found only for Democrats and not Republicans.
Study 5
In Study 5, we focused on our second aim—understanding the mediating mechanisms. We conducted a cross-sectional study focused on better measuring the constructs linking trait essentialist beliefs with lower levels of prejudice. We tested the predictions that perceiving others as less responsible for their political ideology and not attributing malicious intent to them would mediate the negative associations between trait essentialism and prejudice, and that blaming opposite partisans for political strife would mediate the positive associations between social essentialism and partisan prejudice. Additionally, we once again explored whether the negative link between trait essentialism and partisan prejudice was only found for Democrats and not Republicans.
Methods
Participants and Procedure
We sought a minimum sample of 250 and recruited 303 participants from the United States using CloudResearch’s Connect platform (Hartman et al., 2023), compensating them $1.25 for participation. Using custom prescreening, we again solicited half Republicans and half Democrats. Using the same data quality and screening techniques as in the previous studies, in this study, 293 people remained after data screening. Of these, nine indicated they were Independents and were removed, leaving a final sample of 284 (49.6% Democrat, 50.4% Republican; mean age = 38.40 years, age range = 18–80; 118 women, 163 men, three non-binary; 32 Black, 14 East Asian, 215 White, 19 Latino/a/x, five Middle Eastern or North African, four Native American, zero Hawaiian or other Pacific Islander, two South Asian, six bi/multiracial, two other race or ethnicity).
Procedures were similar to Study 1 with minor changes to the measures and the addition of two measures: a new political ideology responsibility measure and an intent to harm measure.
Measures
Participants responded to the same measures as in Study 2 with two minor changes: the wording in the social essentialism measure was changed to contain wording pertaining only to parties (not ideologies), and the political ideology scale was changed from a seven-point scale to a six-point scale. 10 The two new scales used a seven-point scale from strongly disagree to strongly agree.
Political Ideology Responsibility
We worded all responsibility questions in the direction of others not being responsible. We used the two items from Study 3, reworded: “People are NOT responsible for their political ideology,” and “Someone’s political ideology is something that is NOT under their control.” We also included three new face-valid items: “[OppIdeoPpl] shouldn’t be blamed for the turmoil in politics; it’s just their inherent ideology,” “It’s unfair to hold [OppIdeoPpl] responsible for today’s political problems, given the deeply rooted nature of their beliefs,” and “The political divide today isn’t just the fault of the [OppIdeoPpl]; their beliefs are a product of factors beyond their control.” Items were recoded such that higher numbers represent greater perceived responsibility.
Political Ideology Intent to Harm
We assessed the extent to which participants believe that people with the opposite ideology intend to do harm with five items: “[OppIdeoPpl] express their views not to harm others, but because they genuinely believe in them,” “When [OppIdeoPpl] advocate for their positions, it is NOT a reflection of an intention to harm the opposing side,” “Most [OppIdeoPpl] don’t push their agenda to intentionally provoke the other side,” “The views held by [OppIdeoPpl] are more about their own convictions than about undermining the opposition,” and “[OppIdeoPpl] might hold views that are different, but that doesn’t mean they have malicious intentions.” Items were recoded such that higher numbers represent greater perceived intent.
Results
See Table 1 for means, standard deviations, and correlations between scales.
Aim 1: Essentialism and Prejudice
Once again, trait essentialism and social essentialism were significantly, positively, and moderately correlated with each other (see Table 1). Next, stronger endorsement of trait essentialism negatively predicted the prejudice composite and the number of dislikes regarding the opposite party and social essentialism showed the opposite pattern, positively predicting both measures of partisan prejudice.
Aim 2: Mediating Mechanisms
Next, we conducted indirect effects analyses. An initial factor analysis on all of the items from the political strife blame scale and the ideology responsibility and intent to harm scales confirmed the three separate factors. And, we included all three mediators in both analyses. First, using PROCESS Model 4, we estimated the indirect effect of trait essentialism on partisan prejudice. There was a significant negative indirect effect of trait essentialism on the prejudice composite through both decreased perceptions of responsibility and intent and there was a significant negative indirect effect on number of dislikes through responsibility (see Table 11). There were no effects through blame. Second, we estimated the indirect effect between social essentialism and prejudice. As shown in Table 12, the positive associations between social essentialism and both the prejudice composite and number of dislikes were mediated by blame, and there was also a positive indirect effect on both through responsibility. There was a positive indirect effect of intent on the prejudice composite only.
Study 5: Indirect Effects of Trait Essentialism on Measures of Partisan Prejudice Through Responsibility, Intent, and Blame.
Study 5: Indirect Effects of Social Essentialism on Measures of Partisan Prejudice Through Responsibility, Intent, and Blame.
Exploratory Question: Moderation by Political Party
Finally, we once again explored whether the relationships between both measures of essentialism and the measures of partisan prejudice differed across the two parties. We found a significant interaction between trait essentialism and political party on the prejudice composite (see Table 13), but not number of dislikes (p = .877). This interaction revealed that the negative association between trait essentialism and partisan prejudice was only significant for Democrats, not Republicans (see Figure 5). There were no significant interactions between social essentialism and party on either outcome measure (ps > .496). Finally, because we identified two mechanisms linking trait essentialism to prejudice in this study, we conducted a moderated mediation analysis, Process Model 8, to examine if the mediation between trait essentialism and prejudice through responsibility and intent was moderated by party. The indices of moderated mediation were significant for both mediators showing that the indirect effects of trait essentialism through responsibility (Index = −0.06, SE = 0.03; CI = −0.12, −0.01) and intent (Index = −0.15, SE = 0.06; CI = −0.26, −0.04) were only significant for Democrats and not Republicans.
Study 5: Conditional Effects of Trait Essentialism on Partisan Prejudice for Democrats and Republicans.
Study 5: The Association Between Trait Essentialism and Partisan Prejudice as a Function of Political Party.
Study 5 Discussion
In this study, we once again showed the contradictory linkages between trait and social essentialism with partisan prejudice. We also focused on the mechanisms underlying these opposing effects. We identified two mechanisms linking trait essentialism with lower levels of partisan prejudice: participants did not perceive individuals as responsible for their political ideology and did not attribute malicious intent to them (this was significant for the composite measure only). Although in Study 3 we did not find support for perceived responsibility as a mediator, we used a two-item measure that was focused on high levels of responsibility and controllability. In this study, we used a five-item measure with wording focused on lower levels of perceived responsibility for political ideology, which may have more effectively captured the essence of the mechanism. We also once again replicated findings from Studies 1 and 3 showing that blaming opposite partisans for political strife mediates the positive associations between social essentialism and partisan prejudice. However, we also showed that perceiving others as responsible for their political ideology and attributing malicious intent to opposite partisans also helped explain the association between social essentialism and partisan prejudice. Additionally, exploratory analyses once again showed that the link between greater endorsement of trait essentialism and decreased levels of partisan prejudice was found only for Democrats and not Republicans on the composite partisan prejudice measure.
Study 6
In this preregistered study (https://osf.io/5sg64/), we sought to more stringently test our model by addressing the limitations of our previous experimental approach, which focused on reducing prejudice by either increasing trait essentialism or decreasing social essentialism. In this study, we included three conditions—one designed to decrease trait essentialism, one to decrease social essentialism, one condition was a no-message control. In addition to aiming to replicate the basic pattern of correlational relationships among trait essentialism, social essentialism, and prejudice, we examined trait and social reduction messages, relative to a control and each other. Specifically, we investigated both direct and indirect links to prejudice via essentialism and blame.
Methods
Participants and Procedure
We recruited a similar sample size as solicited in the first experimental studies with 175 participants per condition. We recruited a panel of participants from the United States, using CloudResearch’s Connect platform (Hartman et al., 2023), requesting equal representation from both major political parties. A total of 526 participants completed the study and were compensated $1.25 for their participation. Using the same approach to data cleaning as in the previous studies, in this study 518 people remained after data screening. Of these, 11 indicated they were Independents and were removed, leaving a final sample of 507 (49.9% Democrat, 50.1% Republican; mean age = 37.34 years, age range = 18–74; 273 women, 231 men, and three non-binary participants; 90 Black, 41 East Asian, 345 White, 41 Latino/a/x, two Middle Eastern or North African, seven Native American or American Indian, three Hawaiian or other Pacific Islander, three South Asian, 13 bi/multiracial, three other race or ethnicity).
We randomly assigned participants to either read the article with the trait essentialism reduction message (n = 161), the article with a social essentialism reduction message (n = 173), or they were not asked to read an article (n = 173). Participants completed the same measures as in Study 5, except we omitted the list of likes and dislikes measure to save time. 11 And once again, a factor analysis on all of the items from the three attribution scales—perceived responsibility for ideology, perceived intent to do harm, and blame for political strife—confirmed three separate factors.
Trait Essentialism Reduction Message
One-third of participants read the message titled, “Political ideologies: Individuals’ experiences and choices shape beliefs.” The goal of this message was to promote the belief that individuals’ political ideology is not fixed, but changeable and controllable. It highlighted that political beliefs can evolve through personal experiences and conscious choices. Participants were required to view the article for at least 15 seconds and to describe the main message of the article to a friend before proceeding to the measures.
Social Essentialism Reduction Message
Another third of participants read the message titled “We are more similar than different: Democrats and Republicans share similar values.” This article, modeled after the SE-Reduction message used in Studies 3 and 4 and designed to be parallel to the TE-Reduction message, aimed to reduce the belief that political party affiliation defines an individual’s character and beliefs. For example, the article emphasized that members of opposing parties do not have fundamental natures that divide them, but they often share core values and characteristics. Participants viewed the article for fifteen seconds and answered the reading comprehension question before proceeding.
Results
See Table 1 for scale means, standard deviations, reliabilities, and correlations between scales.
Aim 1: Essentialism and Prejudice
In support of our first set of hypotheses, trait essentialism and social essentialism were significantly and positively correlated. Importantly, stronger endorsement of trait essentialism predicted lower levels of partisan prejudice, while stronger endorsement of social essentialism predicted higher levels of prejudice (see Table 1).
Aims 2–3: Mediating Mechanisms and Media Messaging
Next, we tested the predictions that the trait essentialism reduction message would lead to lower levels of trait essentialism and that the social essentialism reduction message would lead to lower levels of social essentialism compared with the control and each other. We conducted a MANOVA with experimental condition (2 = Social Essentialism Reduction [SE-Reduction] Msg, 1 = Trait Essentialism Reduction [TE-Reduction] Msg, 0 = Control) as the between-subjects factor, and trait essentialism and social essentialism as the dependent variables. There were significant effects of condition on both trait and social essentialist beliefs (see Figure 6 and Table 14). Tukey’s HSD post-hoc tests showed that participants in the TE-Reduction condition reported significantly lower levels of trait essentialism compared with those in the Control condition, and marginally lower levels than those in the SE-Reduction condition. The means for trait essentialism in the SE-Reduction and Control conditions were not significantly different. Additionally, participants in the SE-Reduction condition reported lower levels of social essentialism relative to those in the Control condition and marginally lower levels than those in the TE-Reduction condition. Those in the TE-reduction condition also reported lower levels of social essentialism compared with the control condition.
Study 6: Means for Essentialism Across Experimental Conditions.
Study 6: Univariate Effects and Post-Hoc Tests for Effects of Experimental Condition on Trait and Social Essentialism.
Note: * p < .01 ** p < .001; TE-Red = Trait Essentialism Reduction Message; SE-Red = Social Essentialism Reduction Message; Cont = No Message Control; HSD = Honestly significant difference.
Next, to test the predictions that the TE-Reduction message would lead to greater levels of prejudice relative to the other conditions and that the SE-Reduction message would lead to lower levels of prejudice relative to the other conditions, we conducted an ANOVA with experimental condition predicting partisan prejudice. There was no significant effect of condition on prejudice (p = .497).
We then tested if the TE-Reduction message indirectly predicted prejudice through weakened trait essentialism. Using PROCESS Model 4, we conducted indirect effects analyses with condition (1 = TE-Reduction, 0 = SE-Reduction and Control) as the predictor variable, trait essentialism as the mediator, and partisan prejudice as the outcome. As shown in Table 15, the TE-Reduction message indirectly predicted greater levels of prejudice through decreased trait essentialism, compared with either reading no message or the SE-Reduction message. Similarly, we tested if the SE-Reduction message indirectly predicted prejudice through weakened social essentialism. Using the same analytic approach (1 = SE-Reduction, 0 = TE-Reduction and Control), the SE-Reduction message indirectly predicted lower levels of prejudice through decreased social essentialism, compared with reading no message or the TE-Reduction message (see Table 15).
Study 6: Indirect Effects of Condition on Measures of Partisan Prejudice Through Trait and Social Essentialism.
Note: Trait Essentialism Reduction Condition (1 = TE-Reduction Msg, 0 = SE-Reduction Msg and Control); Social Essentialism Reduction Condition (1 = SE-Reduction Msg, 0 = TE-Reduction Msg and Control).
Next, we conducted indirect effects analyses to examine the three attribution mediating mechanisms—perceived responsibility for ideology, perceived intent to do harm, and blame for political strife—first by using self-report measures of essentialism and then using experimental condition as the predictor. First, using PROCESS Model 4, we estimated the indirect effect of self-reported trait essentialism, with higher values indicating stronger essentialism beliefs, on partisan prejudice through all three attribution mediators. There was a significant negative indirect effect of trait essentialism on prejudice through perceived responsibility for ideology but not the other two attribution measures (see Table 16). Next, we examined the indirect effect of the TE-Reduction condition, compared with the other two conditions, on partisan prejudice. Analyses revealed there were no significant indirect effects of TE-Reduction condition on prejudice through any of the attribution mediators. However, pre-registered exploratory analyses revealed a significant positive serial indirect effect of the TE-Reduction condition on prejudice through reduced trait essentialism (M1) and in turn increased perceived responsibility (M2; details available in supplemental materials at OSF).
Study 6: Indirect Effects of Trait Essentialism and Trait Essentialism Reduction Condition on Partisan Prejudice.
For social essentialism, we first estimated the indirect effect between self-reported social essentialist beliefs and prejudice. As shown in Table 17, the positive association between social essentialism and prejudice was mediated by blame for political strife and perceived intent to harm but not political responsibility. Lastly, we examined the indirect effect of the SE-Reduction condition, compared with the other two conditions, on partisan prejudice. Analyses revealed no significant indirect effects of SE-Reduction condition on prejudice through any of the mediators. Pre-registered exploratory analyses showed significant serial negative indirect effects of the SE-Reduction condition on prejudice through reduced social essentialism (M1), which led to lower levels of blame and perceived intent to harm (M2; details available in supplemental materials at OSF).
Study 6: Indirect Effects of Social Essentialism and Social Essentialism Reduction Condition on Partisan Prejudice.
Exploratory Questions: Moderation by Political Party
Finally, we explored whether effects were moderated by participants’ political party. Moderation analyses revealed that party marginally moderated the effect of TE-Reduction condition (vs. the other two conditions) on trait essentialism, with the condition significantly reducing trait essentialism for Republicans (B = −0.59, SE = 0.15; p < .001; 95% CI = −0.89, −0.29) but not for Democrats (B = −0.17, SE = 0.16; p = .286, 95% CI = −0.48, 0.14). Political party did not moderate the effect of SE-reduction condition (versus the other two conditions) on social essentialism (p = .260).
Next, political party once again moderated the association between trait essentialism and the prejudice composite, with the association being significant only for Democrats and not for Republicans (see Table 18 and Figure 7). Finally, the association between social essentialism and prejudice was not moderated by party (p =.401).
Study 6: Conditional Effects of Trait Essentialism on Partisan Prejudice for Democrats and Republicans.
Study 6: The Association Between Trait Essentialism and Partisan Prejudice as a Function of Political Party.
Study 6 Discussion
Once again, we replicated the bivariate associations between trait essentialism, social essentialism, and partisan prejudice. Both trait and social essentialism were significantly and positively correlated. However, stronger endorsement of trait essentialism predicted less partisan prejudice, while greater endorsement of social essentialist beliefs predicted more prejudice. We used manipulations to examine the effects of messages designed to decrease trait essentialism, decrease social essentialism, and a no-message control. Our manipulations were generally successful, as both essentialism reduction conditions reported significantly lower levels of essentialism compared with the control condition. There were also marginally lower levels of essentialism in each reduction condition compared with the other.
Although the essentialism messages did not directly impact prejudice, they did so indirectly through changes in essentialism. Specifically, the trait essentialism reduction message predicted greater levels of partisan prejudice indirectly via a decrease in trait essentialism. This link to prejudice is similar to recent work that finds growth mindsets (which are equivalent to low trait essentialism) can increase stigma-related outcomes (Burnette et al., 2024). Conversely, the social essentialism reduction message predicted lower levels of partisan prejudice indirectly via a reduction in social essentialism. This replicates the robust link between social essentialism and prejudice (Agadullina & Lovakov, 2018).
We also examined the mediating mechanisms. Similar to Study 5, we found a significant negative indirect effect of trait essentialism on prejudice through the perception that others are not responsible for their political ideology. However, unlike Study 5, the indirect effect through malicious intent was not significant in this study. Consistent with findings from Studies 1, 3, and 5, blaming opposite partisans for political strife mediated the positive associations between social essentialism and partisan prejudice. Additionally, similar to Study 5, attributing malicious intent to opposite partisans also helped explain the association between social essentialism and prejudice. Examining the mediating mechanisms by looking at the experimental conditions as predictors revealed no significant indirect effects of the essentialism reduction conditions on prejudice through any of the attribution mediators. However, pre-registered exploratory analyses showed that the trait essentialism reduction condition had a significant positive indirect effect on prejudice through reduced trait essentialism, which in turn increased perceived responsibility. Additionally, the social essentialism reduction condition had a significant negative indirect effect on prejudice through reduced social essentialism, leading to lower levels of blame and perceived intent to harm. That is, we only saw indirect effects on prejudice through attributions when we considered the link to self-reports of essentialist beliefs first.
Finally, exploratory analyses once again indicated that the relationship between greater endorsement of trait essentialism and decreased levels of partisan prejudice was observed only among Democrats, not Republicans.
General Discussion
We had three principal aims. First, our primary aim was to illustrate that, despite the positive relationship between trait and social essentialism, they differentially predict prejudice. Across six studies, our findings consistently demonstrated a positive correlation between these two beliefs, and trait essentialism predicted lower levels of partisan prejudice, whereas social essentialism predicted greater levels of prejudice. We conducted a mini-internal meta-analysis to synthesize effects across the six studies (Goh et al., 2016). First, for the positive correlation between trait and social essentialism, the average effect was M r = .22, Z = 8.40, p < .001. Second, in terms of the average effect for the negative link between trait essentialism and the prejudice composite measure the link was M r = −.19, Z = −7.18, p < .001. Third, for the average effect for the positive link between social essentialism and the prejudice composite the link was M r = .33, Z = 12.82, p < .001. Overall, effects replicated across all six studies. The positive link between social essentialism and prejudice is especially meaningful and suggests a potential point on which to intervene. In terms of an intuitive benchmark, this effect is larger than the effect of taking an antihistamine on symptom reduction (r = .11) and larger than the alleviating effect of pain medication (r = .14). The effect is similar to the link between cooler temperatures and higher elevations (r = −.34; Funder & Ozer, 2019).
Our second aim was to identify the mechanisms through which trait and social essentialism differently predict prejudice. We found that trait essentialism predicted lower levels of prejudice through perceiving individuals as less responsible for their political ideology, with some support for also attributing less malicious intent to them. Conversely, social essentialism predicted greater prejudice through blaming opposite partisans for political strife and attributing malicious intent to them, with some evidence also for holding them responsible for their ideology. This is the first study to our knowledge to show these opposing mechanisms linking essentialism to prejudice in the context of political attitudes.
Our third aim was to examine if we could manipulate these beliefs via ecologically valid media messaging. Initially, we focused on reducing prejudice by increasing trait essentialism and decreasing social essentialism. However, we then shifted to a more nuanced theoretical approach, manipulating both types of essentialism independently trying to reduce them and comparing these conditions to a control to clearly examine their independent effects on prejudice. In line with expectations, both essentialism beliefs were amenable to change, though, perhaps not surprisingly, considering the positive link between the beliefs, manipulating them completely independently proved partially successful. Importantly, these media messages had meaningful implications for partisan prejudice both directly and through their influence on essentialism. In Study 2, the genetic article increased trait essentialism, which in turn lowered levels of partisan prejudice relative to a control. In the next two studies, the social essentialism reduction message, compared with the genetic article and a control, reduced social essentialism and reduced prejudice both directly and indirectly through social essentialism. However, in Study 3, the social essentialism reduction article was also associated with reduced trait essentialism relative to the genetic article, which indirectly predicted greater levels of prejudice through weaker trait essentialism. In our final experiment, we successfully reduced both types of essentialism relative to a control and marginally relative to each other. Although the essentialism messages did not directly impact prejudice, they did so indirectly through changes in essentialism. The trait essentialism reduction message predicted greater levels of partisan prejudice indirectly via a decrease in trait essentialism. Conversely, the social essentialism reduction message predicted lower levels of partisan prejudice indirectly via a reduction in social essentialism.
Finally, we explored how political party might moderate links. This was exploratory in nature without clear theoretical reasoning for what might occur. Party affiliation did not moderate the links between social essentialism and prejudice in any study, suggesting that these associations are strong for people of both parties. However, in four of the six studies, the associations between trait essentialism and partisan prejudice was not robust for Republicans. 12 The effects showed that Democrats who held weak trait essentialism beliefs reported stronger prejudice relative to Republicans. These findings were not predicted, and it is unclear what might be driving these party differences. They may have to do with Democrats’ increasing and greater tendency to view Republicans as closed-minded (Pew Research Center, 2022); this closed-mindedness might be related to greater attributions of responsibility, a belief that links trait essentialism to prejudice. Future research should work to understand the nature of these party differences.
This research has both theoretical and practical implications. We add to the growing body of literature focused on understanding the role of cognitive factors, such as misperceptions and stereotypes about opposite partisans (Ahler & Sood, 2018; Iyengar et al., 2019; Levendusky & Malhorta, 2016) and the salience of partisan identities (Iyengar et al., 2019; Levendusky, 2018), in partisan animosity. The cognitive perspective we bring to this research sheds light on relatively unexplored aspects of partisan prejudice by drawing from the rich literatures on mindsets and essentialism. In turn, our research makes important contributions to both of those bodies of literature, enriching our understanding of the complex dynamics involved in partisan beliefs and prejudice. Aligned with research findings across many domains, ranging from sexual prejudice (Hoyt, Morgenroth, et al., 2019) to mental illness stigma (Haslam & Kvaale, 2015), in the current work, we show that there are consistent countervailing influences of fixed beliefs about attributes (trait essentialism) and fixed beliefs about groups of people (social essentialism) on prejudicial attitudes in the domain of political ideology.
Our demonstration of the mechanisms underlying these differential links from essentialism to partisan prejudice makes novel and important contributions to the literature as the findings highlight the different facets of blame that can help explain each of the pathways. For example, consistent with previous research, trait essentialism beliefs predicted lower levels of prejudice by reducing attributions related to perceived personal responsibility (Haslam & Kvaale, 2015; Hegarty, 2010; Hoyt & Burnette, 2020). However, although our initial measure of responsibility and controllability did not effectively mediate the effect, we found a significant indirect effect when using a more robust measure with wording that emphasized lower levels of perceived responsibility for political ideology. This underscores the importance of precise wording in accurately assessing responsibility and blame.
Additionally, similar to previous research, social essentialism beliefs led to increased prejudice through negative attributions regarding the outgroup, specifically blaming them for political strife (Agadullina & Lovakov, 2018; Yzerbyt & Rogier, 2001). We also found that social essentialism beliefs predicted prejudice through the personal-level attributions, including attributing malicious intent to them and holding them responsible for their ideology. This suggests that the mechanism of blame in social essentialism operates at both personal and group levels. These findings highlight the complex ways in which essentialist thinking can shape prejudicial attitudes by influencing multiple dimensions of blame.
In addition, our theoretical approach and empirical findings can help clarify and extend findings from two studies exploring the relationship between essentialism and partisan prejudice. Wilson’s (2018) dissertation found that different facets of essentialism predicted partisan prejudice in opposite directions, consistent with our model and findings. Furthermore, Suhay et al.’s (2016) correlational work showed a positive link between lay beliefs in the biological basis of ideology and intolerance, despite much theoretical work predicting that such fixed beliefs should be associated with greater tolerance. Our theoretical approach suggests they may have tapped into social essentialist, as opposed to trait essentialist, beliefs.
On a practical level, our research can inform future research on interventions designed to promote greater partisan harmony. For example, in a recent review of interventions aimed at decreasing partisan animosity, the TRI framework identifies three levels of interventions: changing Thoughts, building Relationships, and transforming Institutions (Hartman et al., 2022). Our research has implications for interventions intervening at the level of thoughts as we identified pliable belief systems that are potent precursors to partisan prejudice. The effectiveness of media messaging to influence both belief systems suggests that such messaging offers potential pathways for practical interventions to mitigate partisan animosity. Additionally, although we were focused on media messaging, there are other approaches, such as interventions focused on highlighting commonalities (Hartmann et al., 2022), that can be leveraged to weaken social essentialism beliefs. Overall, more work is needed that finds the most powerful message for shifting beliefs in a way that maximizes the benefits while reducing the costs. One such approach might be a type of compensatory messaging that can weaken blame while also reducing social essentialist thinking (e.g., Burnette et al., 2017).
Before putting these findings into practice, a few limitations and areas of future inquiry must be discussed. As with all online research samples, data quality is at risk when responses are from inattentive humans and bots. To enhance the data quality of our online samples, participants were recruited via CloudResearch, and we implemented open-ended and multiple-choice attention checks to filter out unreliable responses. Nonetheless, the generalizability of our findings may only extend to online samples, and future research should aim to replicate effects within in-person settings and with more diverse populations. Also, by investigating our research questions within the United States’ current political climate and context of political messaging, the generalizability of our findings is yet to be explored across cultures and political systems. Next, although the articles used in the first experimental studies are high in ecological validity, this has the potential to compromise internal validity. The social essentialism reduction article used in Studies 3 and 4 was designed with two somewhat conflicting messages, one aimed to reduce social essentialism and one to try and maintain an elevated level of trait essentialism (e.g., low controllability of politics). We took this approach to see if we could craft messages with practical utility. In our final study, we sought to bolster our understanding of messaging and prejudice by focusing on trait and social essentialism reduction messages in independent conditions and included a control group. These messages were carefully designed to be parallel, manipulating only the essentialism beliefs. This final study suggests we can manipulate these beliefs, albeit not completely independently. The findings also highlight that links to prejudice via messaging are most likely driven by more proximal impacts on cognitions, including self-reports of essentialism and attributions.
Conclusion
Across six studies, we showed that even though trait essentialism and social essentialism were positively related, they nonetheless predicted prejudice in different ways: trait essentialism predicted lower levels of prejudice, and social essentialism predicted greater prejudice. Exploring the mechanisms, we found that trait essentialism predicted lower levels of prejudice through fostering the perception that others are less responsible for their political ideology, with some support for not attributing malicious intent to them. In contrast, social essentialism predicted greater prejudice through blaming opposite partisans for political strife and attributing malicious intent to them, with modest support for holding them responsible for their ideology. Finally, mainstream media messaging influenced both belief systems with implications, at least indirectly, for prejudice, suggesting potential pathways for practical interventions to mitigate partisan animosity.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material is available online with this article.