Testing the social pressure hypothesis: Does in-party social pressure reduce out-party empathy?

Abstract

Empathy is considered one of the most critical components for bridging political divides and reducing animosity between political groups. Yet, empathy between political opponents is rare. There is a growing concern that partisans do not empathize with out-partisans because they feel social pressure from fellow in-partisans not to do so. This article examines this social pressure hypothesis and draws two conclusions. First, on the surface, the hypothesis seems plausible: citizens perceive fellow in-partisans as comparatively disapproving of and reluctant to engage in out-party empathy, and naïve cross-sectional analyses suggest that this perception translates into lower empathy towards out-partisans. Second, however, experimental data suggest that this relationship is not causal. Expecting disapproval from fellow in-party members for empathizing with out-partisans does not lead to a significant reduction in intentions to empathize with out-partisans. Rather, exploratory analyses suggest that social pressure by the in-party increases empathy toward out-partisans and triggers disappointment toward in-partisans. This implies that partisans can resist social pressure from the in-party and might even compensate for in-partisans’ lack of out-party empathy. The results are supported by original cross-sectional and experimental survey data (⁠$N=\mathrm{2,535}$⁠) collected in the United States, an arguably most likely case for in-party social pressure to shape partisans’ intentions. The results have important implications for understanding the causes of and viable strategies for building empathy across political divides.

out-party empathy, social pressure, in-party meta-perceptions, polarization

Significance Statement

Researchers and public initiatives aim to boost empathy between Republicans and Democrats to reduce political hostility. This study explores whether feeling social pressure by fellow in-partisans not to empathize with the other side inhibits out-party empathy. Surprisingly, expecting in-party disapproval for empathizing with out-partisans does not reduce partisans’ empathy for political opponents. Rather, perceiving such polarizing dynamics within their own party leaves partisans disappointed and increases empathy toward the opposing party, possibly as compensation for in-partisans’ shortcomings. This highlights the potential of interventions to increase empathy toward out-partisans despite in-party resistance. Future research should further investigate the effectiveness of empathy interventions in the face of in-party resistance and how feelings of disappointment might be useful in bridging political divides.

Introduction

In 2016, conservative political commentator Glenn Beck called on his fellow conservatives to understand the perspective of the Black Lives Matter movement, encouraging them “to start listening to each other and getting out of our own little labeled bubbles” (1). This appeal to empathy sparked a significant backlash from conservative news outlets and on Twitter (now X), with Beck facing widespread criticism and accusations of betrayal from within his own ranks (1–3). In 2017, New York Times journalist Richard Fausset faced substantial pushback for interviewing a White Nationalist to understand his perspective, with these efforts being harshly criticized as severe taboo (4), see also (5). These examples illustrate that those who empathize with political opponents can face significant backlash from within their own ranks. Could this dynamic explain the reluctance of many to empathize with those outside their political circles (6, 7)?

The political landscape of many Western democracies, and especially in the United States, is increasingly shaped by political hostility and animosity across people with different political identities and positions (8, 9). Out-party empathy—defined here as the motivation to understand the reasons behind out-partisans’ thoughts, feelings, experiences, and perspectives—is often regarded as the solution to bridge such hostile divides (10–14). As a consequence, research and public initiatives are working on interventions to increase out-party empathy as a primary strategy for overcoming political animosity. However, such efforts may be useless if individuals are inhibited from engaging in out-party empathy, not because they lack the ability to empathize, but because they worry about what fellow in-party members might think of them if they did. Such in-party meta-perceptions, i.e. concerns about how one is perceived by fellow party members, including the fear of negative judgment, being labeled a traitor, facing punishment, or even social ostracism might make partisans opt not to even consider empathizing with the other side (15–18). To understand and address hostile political divides, we must consider the interaction between individuals and collectives, accounting for the social structures and dynamics involved (19).

Based on this premise, this article tests the in-party social pressure hypothesis: partisans exhibit less out-party empathy when they believe that fellow in-partisans think negatively about them for doing so. I collected data in the United States—arguably a most likely case for party identification and social pressure to reduce out-party empathy (9, 20, 21)—using an online survey (study 1, $N=\mathrm{1,046}$⁠) and an online survey experiment (study 2, $N=\mathrm{1,489}$⁠). In study 1, both Republicans and Democrats perceive fellow in-partisans to be less willing to engage in and more disapproving of out-party empathy than they personally are, which is associated with lower self-reported out-party empathy. However, study 2 fails to establish a clear causal link. Actively considering negative reactions from in-partisans for empathizing with out-partisans does not significantly reduce self-reported out-party empathy; instead, it is found to increase out-party empathy. Hence, thoughts about being viewed negatively by fellow partisans for empathizing with political opponents do not seem to be the main factor reducing general intentions to empathize with out-partisans. Alternative explanations for reduced out-party empathy are discussed, pointing towards the out-party as a more important factor in reducing out-party empathy compared to the in-party, suggesting that the regulation of empathy might primarily be determined by out-party hate and the characteristics of the individual being empathized with rather than by in-party love or the social pressure from fellow partisans.

Defining out-party empathy

What does it mean to empathize with political opponents? Empathy is a complex concept comprising both affective and cognitive elements (14, 22, 23). Affective empathy involves sharing and experiencing the same emotions as the individual being empathized with, a phenomenon also referred to as experience sharing (23). Cognitive empathy involves gaining an understanding of another person without necessarily sharing their emotions. This includes perspective taking, which entails “either (a) imagining [another] person’s mental states (imagine-other) or (b) imagining their own mental states as if they were the [other] person or were experiencing the [other] person’ situation (imagine-self)” (14, p. 375), as well as related cognitive processes such as mentalizing, i.e. “the observers’ capacity to draw explicit inferences about targets’ intentions, beliefs, and emotions” or thought perception, i.e. “observers’ detection of targets’ internal states” (23, pp. 1608–1609).

This article focuses primarily on cognitive empathy rather than affective empathy, highlighting processes like imagine-other perspective taking, mentalizing, and thought perception. Out-party empathy, as defined here, involves Person A gaining an understanding of the thoughts, feelings, experiences, and perspective of Person B, who aligns with a political party different from Person A’s. For instance, a Democrat would demonstrate out-party empathy by seeking to understand the thoughts, feelings, experiences, and perspectives that lead a Republican to, for example, oppose immigration. However, this does not mean that the Democrat will share the same emotions as the Republican, such as resentment towards immigrants (affective empathy/experience sharing) or imagining what they would think if they were a Republican who opposes immigration (imagine-self perspective taking). Instead, the focus is primarily on understanding the thoughts, feelings, and experiences that have influenced the Republican’s position. Such understanding is considered an important forerunner of deeper empathic processes (23).

When empathy fails: the role of in-party social pressure

Why would partisans refrain from empathizing with out-partisans? Research argues that it is not because they lack the ability to empathize, but rather because they lack the motivation (23). “[F]or every reason to choose empathy, there is another reason to avoid it” (24, p. 39). Avoiding empathy is particularly pronounced in competitive or conflictual situations such as politics where empathizing may come at specific costs (25, 26). The decision to empathize then hinges on individuals weighing the associated costs and benefits. If empathy is perceived as beneficial, people are more likely to engage in it (10). Conversely, when empathy entails costs, individuals are less inclined to exhibit it (27, 28).

In their taxonomy of empathic failures, Jamil Zaki and Mina Cikara shed light on various factors that make it costly for people to empathize with others (25). One aspect is the role of social norms, encompassing perceptions about whether others typically engage in a specific behavior, whether others deem this behavior appropriate, and assumptions about the consequences of adhering to or breaking these norms (29, 30). Social norms significantly shape people’s political attitudes and intentions (31–33) as well as their actual political behavior (30). While there is often a focus on generalized social norms, here, I focus specifically on normative behavior defined by the in-group and emphasize the importance of understanding how social norms shape attitudes and behavior in the context of group dynamics (34–36). This understanding is particularly important in polarized political environments where aligning with a specific political party can significantly influence animosity towards political opponents (37).

There may exist an in-party norm against showing empathy towards political opponents, meaning that partisans typically refrain from empathizing with out-partisans, disapprove of such behavior, and socially sanction those who do not adhere to this norm. Research shows that empathizing with others indicates tolerance (18), openness to compromise, or even opinion change (17). While these traits are typically seen as positive, in polarized and contentious intergroup contexts, they can become undesirable because they may pose risks to the interests of the in-party. By engaging in out-party empathy, partisans risk giving out-partisans’ “perspectives an undeserved legitimacy or traction” (16, p. 107) and disadvantaging their own side (38). Based on this, “those who empathize across social divides might be repudiated by their own peers for doing so” (5, p. 1023). To avoid backlash from within their own ranks and not jeopardize their status in the in-party, “individuals are best served by limiting their cooperative efforts to ingroup members and withholding cooperation from […] noncoalition members” (23, p. 1612). After all, belonging to a group is a fundamental psychological need that is essential to maintain (39). Research shows that in-group norms can reduce the positive impact of inter-group contact and increase hostility towards out-groups (40, 41) and that in particular anti-social behavior, which might include withholding empathy from political opponents, is highly contagious when exhibited by those who are socially close (42). Based on this, I expect anticipating in-party lack and disapproval of out-party empathy to diminish the overall willingness to empathize with out-partisans.

But what do people actually believe about others who engage in empathy? This area of research is relatively new, and recent studies have produced mixed results. Some studies show that those who empathize with others are perceived as more tolerant, cooperative, and rational than those who avoid it (17), characteristics that are generally viewed positively in the political sphere (43, 44). People also consider it morally right to show the same empathy towards out-groups as towards their in-group. Conversely, they consider it morally wrong to show empathy exclusively towards either the in-group or the out-group, although the greatest moral disapproval targets those who show more empathy towards members of the out-group (vs. in-group) (45). This squares with other research suggesting that individuals have a negative view of those who empathize with political opponents. They view them as morally questionable (15) and dislike when they compromise with their political adversaries (46), potentially making partisans angry about people who might be leaving their political group (47). Likability especially decreases when co-partisans are uncommitted to their views and empathize with those holding particularly illegitimate beliefs (5, 17, 18). Yet, (17) found that those who empathized with extreme opponents were still more liked than those who refrained from doing so.

However, the crucial aspect might not be individuals’ actual evaluations of those who empathize with the other side but rather how people believe their peers would assess them. “[I]f people mistakenly believe that others discourage political-perspective seeking, they may abstain from it out of fear of social punishment” (17, p. 1798). Such (inaccurate) meta-perceptions—i.e. how people think others perceive them—have been found to influence political attitudes and intentions towards out-partisans^a (48–50). Hence, the inclination to empathize with political opponents may be shaped by such in-party meta-perceptions.

Study 1

Study 1 offers initial descriptive insights and correlational analyses on how partisans perceive themselves and others when it comes to out-party empathy. Based on an IRB-approved (Aarhus University Institutional Review Board: BSS-2023-068) and pre-registered^b original online survey of 1,199 US-Americans conducted on the online crowdsourcing platform Prolific, I test the pre-registered hypotheses that partisans are less empathic to out-partisans the less they believe in-partisans are empathic to out-partisans (H1) and the more they believe in-partisans disapprove of out-party empathy (H2).^c The sample is selected based on quotas of the US population in terms of age, gender, and ethnicity (simplified US Census). Analyses are restricted to self-identified Democrats and Republicans including leaners (⁠$N=\mathrm{1,046}$⁠).

Self-reported and perceived in-party engagement in as well as disapproval of out-party empathy are measured using additive indices that consist of eight items each. Depending on self-report or in-party perception, engagement in out-party empathy is measured by asking participants to what extent they agree or disagree with statements like “[I/Most [in-partisans]] want to understand the reasons for why [out-partisans] hold their opinions,” on a scale from $1=$ strongly disagree to $7=$ strongly agree. Similarly, self-reported/perception of in-party disapproval of out-party empathy is measured by asking, e.g. “To what extent do [you/most [in-partisans]] approve or disapprove of [in-partisans] wanting to understand the reasons for why [out-partisans] hold their opinions,” on a scale from $1=$ [I/Most [in-partisans]] strongly disapprove to $7=$ [I/Most [in-partisans]] strongly approve. Indices are recoded to range from 0–1 for better comparison. More details on the measures and survey procedure are reported in the Materials and methods section. The questionnaire containing all measured items is available on the Open Science Framework (OSF) repository: https://osf.io/y4jx8/.

Results study 1

What are partisans’ perceptions of how in-partisans view out-party empathy?

Before presenting the main pre-registered analysis, I provide an exploratory overview of partisans’ perceptions of how others view out-party empathy, compared to themselves. Figure 1, Panels A1 and A2 show that both Democrats and Republicans perceive fellow in-partisans to be, on average, less empathic to out-partisans than they report themselves to be (Republicans: $\mathrm{\Delta }M=-0.15,t(631)=-7.9,P<0.001$⁠; Democrats: $\mathrm{\Delta }M=-0.13,t(1436)=-10.9,P<0.001$⁠). Out-partisans are perceived to be least empathic to out-partisans on average (compared to self-report: Republicans: $\mathrm{\Delta }M=-0.29,t(635)=-15.2,P<0.001$⁠; Democrats: $\mathrm{\Delta }M=-0.33,t(1440)=-28.4,P<0.001$⁠; compared to in-party perceptions: Republicans: $\mathrm{\Delta }M=-0.14,t(639)=-7.8,P<0.001$⁠; Democrats: $\mathrm{\Delta }M=-0.20,t(1447)=-18.5,P<0.001$⁠). Figure 1B1 and B2 show similar patterns for the disapproval of in-partisans engaging in out-party empathy. Both Republicans and Democrats see fellow in-partisans as, on average, more disapproving of when fellow in-partisans engage in out-party empathy than they report themselves to be (Republicans: $\mathrm{\Delta }M=0.14,t(640)=8.5,P<0.001$⁠; Democrats: $\mathrm{\Delta }M=0.12,t(\mathrm{1,447})=11.9,P<0.001$⁠).

Fig. 1.

A1 and A2) Smoothed kernel density estimates of self-reported out-party empathy as well as perceptions of in-partisans and out-partisans is shown. Measures range from 0–1 with higher values reflecting more out-party empathy. B1 and B2) Distributions for the extent to which Democrats and Republicans report themselves and perceive other in-partisans to disapprove of out-party empathy are shown. Measures range from 0–1 with higher values reflecting stronger disapproval of in-partisans engaging in out-party empathy. Dotted lines represent mean values.

Are perceptions of how in-partisans view out-party empathy related to partisans’ self-reported out-party empathy?

Yes. Figure 2 displays estimated OLS coefficients for the relationships between perceived in-party reluctance (H1) and disapproval (H2) of out-party empathy and partisans’ self-reported degree of out-party empathy. The figure presents coefficients from various pre-registered models incorporating different types of control variables (see Materials and methods section for control rationale). Perceiving in-partisans as more hesitant to empathize with out-partisans correlates with lower average self-reported out-party empathy (Bivariate model: $b=-0.366,\mathrm{se}=0.037,P<0.001$⁠). This association weakens when additional controls beyond sociodemographics are included. Similarly, partisans demonstrate lower empathy towards out-partisans on average when they perceive in-partisans as more disapproving of such empathy (Bivariate model: $b=-0.375,\mathrm{se}=0.042,P<0.001$⁠). Again, these associations become less robust when controls beyond sociodemographics are added, and they become statistically insignificant when indicators of out-party animosity are incorporated in the model (⁠$b=-0.054,\mathrm{se}=0.044,P>0.05$⁠). While most of the results support both hypotheses, they are less consistent when controls beyond sociodemographic characteristics are added.^d This raises the question of whether these correlations are spurious or indeed signify a causal relationship.

Fig. 2.

Associations between perceptions about in-party approaches to out-party empathy and self-reported out-party empathy. The figure displays unstandardized Ordinary Least Squares (OLS) regression coefficients with 90% (thick lines) and 95% (thin lines) confidence intervals calculated based on robust standard errors. The dependent variable is self-reported out-party empathy, ranging from 0–1 with higher values indicating stronger out-party empathy. Reported effects are based on different pre-registered model specifications. The bivariate specification includes only the relevant independent variable (either perceived in-party reluctance towards or disapproval of out-party empathy), ranging from 0–1 where higher values indicate a stronger reluctance towards or disapproval of out-party empathy, respectively. The other three specifications gradually introduce different levels of control (C1–C3) with each level incorporating the control variables from the previous level. Control Level 1 (C1) augments the bivariate model with sociodemographic factors (age, gender, education, race, and party identification). Control Level 2 (C2) supplements C1 by adding self- and other-perceptions (perceiving out-party empathy as legitimization, in-party reluctance towards or disapproval of out-party empathy (depending on which one has not been included in the model as the main independent variable), degree of self-uncertainty). Control Level 3 (C3) extends C2 by including indicators of political hostility (affective polarization, contact with political opponents, out-party harassment, adherence to democratic norms). All details regarding the included control variables are reported in the Materials and methods section.

Study 2

Study 2 takes an experimental approach to investigating whether the patterns observed in study 1 are causal. Its main aim is to determine whether perceived social pressure from one’s own party discourages individuals from empathizing with the opposing party. I conducted an online survey experiment with a pre-registered^e and IRB-approved (Aarhus University Institutional Review Board: BSS-2023-130-S2) between-subjects design. The study involves a convenience sample of 1,489 US-Americans, evenly distributed across partisanship (50.16% Republicans, 49.83% Democrats) and gender (49.9% female, 49.4% male, 0.7% other), using the crowdsourcing platform Prolific. Participants are randomly assigned to one of the three conditions: social cost (⁠$N=498$⁠), social benefit (⁠$N=484$⁠), or baseline (⁠$N=507$⁠).^f Based on data from study 1 and other research (15, 17), participants are provided with information indicating either disapproval (social cost condition) or approval (social benefit condition) of engaging in out-party empathy by fellow in-partisans. To reinforce the treatment and increase perceived social pressure, participants are also asked to write down in detail the negative (social cost condition) or positive (social benefit condition) reactions they expect from in-partisans for empathizing with out-partisans.

Participants in the social cost condition read:

We recently conducted a survey to understand the beliefs and values of ordinary citizens who identify as [in-party]. Here’s what we discovered: A large share of [in-partisans] we surveyed say that most [in-partisans] disapprove of fellow [in-partisans] who try to understand the perspective of [out-partisans]. These findings align with another recent study showing that people dislike politically like-minded individuals who listen to or engage with those from the other side.

Imagine what fellow [in-partisans] might think of you if you showed understanding for the perspective of [out-partisans]. What negative reactions do you anticipate from fellow [in-partisans]?

Participants in the social benefits condition read:

We recently conducted a survey to understand the beliefs and values of ordinary citizens who identify as [in-party]. Here’s what we discovered: The majority of [in-partisans] we surveyed approves of fellow [in-partisans] who try to understand the perspective of [out-partisans]. These findings align with another recent study showing that people like politically like-minded individuals who listen to or engage with those from the other side.

Imagine what fellow [in-partisans] might think of you if you showed understanding for the perspective of [out-partisans]. What positive reactions do you anticipate from fellow [in-partisans]?

Subsequently, participants report their own intentions to empathize with out-partisans (measured the same way as self-reported out-party empathy in study 1), along with other measured outcomes. In the baseline condition, participants first report their degree of out-party empathy along with other measures and are only at the end of the survey asked to describe potential reactions from in-partisans for empathizing with out-partisans without receiving any information about in-party approval or disapproval. More details on measures and analytical models employed to analyze the experiment are reported in the Materials and methods section. The full questionnaire is available on the OSF repository: https://osf.io/y4jx8/.

Results study 2

Does the manipulation effectively lead partisans to think that the in-party disapproves of them empathizing with out-partisans?

Yes. Figure 3A shows differences between experimental groups in perceived in-party disapproval of out-party empathy. As predicted, participants in the social cost condition are on average statistically significantly more likely to think that in-party members disapprove of them empathizing with the other side compared to participants in the social benefit condition (⁠$b=0.158,\mathrm{se}=0.012,P<0.001$⁠, Cohen’s $d=0.84$⁠). Exploratory analyses comparing both treatment groups to the baseline show that the baseline lies in between.^g

Fig. 3.

A) Mean perceived in-party disapproval of out-party empathy across experimental conditions, used as a manipulation check is shown. Values range from 0–1, higher values indicate stronger perceived in-party disapproval. B) Mean self-reported out-party empathy across experimental conditions used to test H1 is displayed. Values range from 0–1, higher values indicate stronger out-party empathy. C) The proportion of participants who chose to read about the perspective of an out-partisan (vs. in-partisan) across experimental conditions is displayed. Values are dummy-coded, where $0=$ participant chose to read about in-partisan perspective, $1=$ participant chose to read about out-partisan perspective. D) Mean likelihood of being in contact with out-partisans across experimental conditions is presented. Values range from 0–1, higher values indicate a higher likelihood to be in contact with out-partisans. E) Mean feelings towards out-partisans across experimental conditions are shown. Values range from 0–1, higher values indicate warmer feelings. For all panels, error bars represent 95% confidence intervals. Thick brackets display pre-registered comparisons. Dashed brackets display exploratory comparisons. ***$P<0.001{,}^{**}P<0.01{,}^{*}P<0.05$⁠, $\mathrm{ns}=P>0.05$⁠.

Does anticipated in-party disapproval (vs. approval) of out-party empathy reduce self-reported out-party empathy?

Not significantly. As shown in Fig. 3B, the observed differences in self-reported out-party empathy between the social cost and social benefit conditions are surprisingly small and do not reach statistical significance at the conventional 95% level (⁠$b=-0.026,\mathrm{se}=0.015,P>0.05$⁠, Cohen’s $d=0.11$⁠). According to the pre-registration, hypothesis 1 is formally rejected.^h The data neither support the pre-registered interaction hypothesis that those with a stronger need to belong to their in-party are less inclined to empathize with out-partisans when they perceive disapproval from their in-party (see Supplementary Material 3.7.1).ⁱ

Surprisingly, exploratory comparisons to the baseline conditions show that those who expect in-party disapproval increase out-party empathy (⁠$b=0.066,\mathrm{se}=0.016,P<0.001$⁠, Cohen’s $d=0.26$⁠), simliar to those who expect in-party approval (⁠$b=0.092,\mathrm{se}=0.016,P<0.001$⁠, Cohen’s $d=0.37$⁠).

What could explain these findings? Why do individuals only weakly and not significantly decrease out-party empathy when considering in-partisans’ disapproval vs. approval? Is this finding outcome specific? Furthermore, what might account for the surprising increase in out-party empathy among those anticipating in-party disapproval compared to baseline? The next paragraphs aim to illuminate these questions based on exploratory analyses.^j

Exploratory Analysis I: does in-party disapproval affect alternative measures of out-party empathy?

In addition to the pre-registered primary outcome measure of out-party empathy, I collected several alternative measures that allow me to explore whether the statistically insignificant difference in out-party empathy between the social cost and social benefit condition is outcome-specific. These alternative measures include (i) a behavioral indicator of out-party empathy in which participants had to choose between reading an article offering insights to either the perspective of an out-party member or the perspective of an in-party member, (ii) the likelihood of being in contact with out-partisans, as well as (iii) feelings towards out-partisans on a feelings thermometer.

Figure 3C shows that participants in the social cost condition are, on average, not statistically significantly more or less likely to choose to read about the perspective of an out-partisan (vs. in-partisan) than those in the social benefit condition (⁠$b=-0.029,\mathrm{se}=0.032,P=0.37$⁠, Cohen’s $d=0.058$⁠). Nor are there differences when comparing social cost and benefit conditions to baseline (social cost: $b=0.001,\mathrm{se}=0.032,P>0.05$⁠, Cohen’s $d=0.00$⁠; social benefit: $b=0.030$⁠, $\mathrm{se}=0.032,P>0.05$⁠, Cohen’s $d=0.06$⁠). Neither are participants in the social cost condition significantly more or less likely to be in contact with out-partisans compared to participants in the social benefit condition (Fig. 3D; $b=-0.010,\mathrm{se}=0.015,P>0.05$⁠, Cohen’s $d=0.04$⁠). However, just as for out-party empathy, partisans in both the social cost and the social benefit conditions significantly increase their willingness to be in contact with out-partisans compared to baseline (social cost:  $b=0.044,\mathrm{se}=0.015,P<0.01$⁠, Cohen’s $d=0.18$⁠; social benefit:  $b=0.054,\mathrm{se}=0.015,P<0.001$⁠, Cohen’s $d=0.23$⁠). The same applies to average feelings towards out-partisans. There is no statistically significant difference between social cost and benefit conditions (Fig. 3E; $b=-0.028,\mathrm{se}=0.016,p>0.05$⁠, Cohen’s $d=0.11$⁠), however, partisans increase their liking of the out-party in the social cost and social benefit conditions compared to baseline (social cost:  $b=0.032,\mathrm{se}=0.015,P<0.05$⁠, Cohen’s $d=0.13$⁠; social benefit:  $b=0.060,\mathrm{se}=0.015,p<0.001$⁠, Cohen’s $d=0.25$⁠). The patterns found in the main analysis thus extend to alternative measures closely related to out-party empathy.

Exploratory analysis II: why does in-party disapproval increase out-party empathy compared to baseline?

Why might partisans who anticipate disapproval by in-partisans for empathizing with out-partisans increase out-party empathy compared to baseline? To better understand this, I explore participants’ feelings towards the in-party across experimental conditions. Figure 4A illustrates the average values of different discrete feelings reported by participants in response to in-party reactions to out-party empathy. In the social cost condition where participants anticipated negative reactions, the predominant feeling reported is disappointment (mean $=0.36,\mathrm{sd}=0.39$⁠). In the social benefit condition, participants react positively to the anticipated positive reactions by in-partisans, primarily expressing feelings of hope (mean $=0.47,\mathrm{sd}=0.41$⁠), closely followed by feelings of happiness (mean $=0.38,\mathrm{sd}=0.39$⁠), pride (mean $=0.35,\mathrm{sd}=0.4$⁠), and enthusiasm (mean $=0.25,\mathrm{sd}=0.36$⁠). The feelings reported in the baseline condition fall between those reported in the social cost and social benefit conditions. In addition to this, Figure 4B shows general feelings toward in-partisans. Participants in the social cost condition slightly reduce their positive feelings towards the in-party compared to participants in the social benefit condition (⁠$b=-0.027,\mathrm{se}=0.013,P<0.05$⁠, Cohen’s $d=0.14$⁠) with no statistically significant differences comparing social cost and benefit conditions to baseline (social cost: $b=-0.012,\mathrm{se}=0.012,P>0.05,$ Cohen’s $d=0.06$⁠; social benefit: $b=0.015,\mathrm{se}=0.013,P>0.05,$ Cohen’s $d=0.08$⁠).

Fig. 4.

A) Means of different discrete feelings towards anticipated reactions of in-partisans across experimental groups are shown. Values range from 0–1, with higher values indicating stronger feelings. B) Mean feelings towards the in-party across experimental conditions are shown. For all panels, error bars represent 95% confidence intervals. In B), dashed brackets display exploratory comparisons. ***$P<0.001{,}^{**}P<0.01{,}^{*}P<0.05,\mathrm{ns}=P>0.05$⁠.

Discussion

An increasing body of research argues that partisans avoid empathizing with those from the other side due to social pressure, i.e. partisans fear disapproval from fellow in-partisans for being empathic towards out-partisans (5, 15–17). The present paper investigates the validity of this argument. While initial cross-sectional analyses suggest that this relationship is plausible—partisans perceive in-partisans as comparatively less willing to engage in out-party empathy and more disapproving of others doing so, which is associated with a decrease in self-reported out-party empathy—experimental data shows that this relationship is not causal. Despite actively considering negative reactions from fellow in-partisans for empathizing with the out-party (vs. potential in-party praise for doing so), participants do not significantly reduce out-party empathy, and any substantial decrease observed is minimal at best. Further explorations show that anticipating in-party disapproval for out-party empathy neither affects closely related measures to out-party empathy such as choosing to read about the perspective of an out-partisan, willingness to be in contact with an out-partisan, nor feelings towards the out-party.

Surprisingly, out-party empathy increases when participants anticipate disapproval from fellow in-partisans compared to baseline. Warm feelings toward the out-party and the willingness to be in contact with them also increase. Further exploratory analyses show that partisans react with disappointment to in-party disapproval of out-party empathy but do not like the in-party less. These results indicate that, contrary to expectations, partisans do not diminish their general intentions to empathize with out-partisans when they anticipate disapproval from within their own ranks. Rather, like those who believe their party would approve of them engaging in out-party empathy, those who anticipate disapproval maintain a comparable level of out-party empathy and even increase empathy towards the out-party compared to baseline. At the same time, the prospect of negative reactions from in-partisans for engaging in out-party empathy makes partisans disappointed in the in-party but does not reduce the general liking of the in-party.

How can we understand these findings? One interpretation is that partisans prioritize their personal convictions over perceived expectations from the in-party. When partisans sense disapproval from fellow partisans of empathizing with the other side, it creates a clash between their own values and those of the in-party (51, 52). Personal values play a more important role in determining behavior than social norms in environments where few others exhibit the behavior in question or where there is high behavioral variance, but less so in polarized environments (53). However, even in polarized environments like those between Republicans and Democrats, personal values might still significantly impact intentions despite anticipated social pressure from the in-party. For example, study 1 shows that partisans report being more approving of fellow in-partisans empathizing with out-partisans than they perceive the in-party to be, and research shows that empathy is widely seen as a moral virtue (45), even when directed towards political opponents with extreme views (17). In addition, recent polls from the Pew Research Center suggest that both Democrats and Republicans are frustrated with the current state of politics (54) and see cooperation between parties among the top national issues to address (55). When partisans realize that their in-party discourages empathy across party lines, it conflicts with their personal values of being empathic, bridging divides, and reducing partisan conflicts. This mismatch then leads to disappointment with the in-party, which fails to live up to the individual’s expectations. Despite this disappointment, however, partisans maintain their loyalty to the in-party and do not like it less. Partisans then seem to compensate for their party’s disapproval of empathy towards the out-party by boosting their own empathy levels. This is analogous to a parent–child relationship: When a child misbehaves, the parent is disappointed and seeks to rectify the child’s misbehavior by overcompensating it, while their love for the child remains unchanged.

Another alternative explanation for the results can be attributed to social desirability bias. Rather than responding according to their genuine beliefs, participants may have based their answers on what they deem to be socially acceptable. Although participants were intended by design to be influenced by social considerations, social desirability bias may still have influenced their responses. The goal was to manipulate the social desirability of the in-party; however, a higher-level social desirability bias could still override the social pressure from the in-party. Such social desirability could have influenced the results in at least two ways:

(i) The treatment making partisans anticipate disapproval from their in-party for showing empathy towards the out-party might not have generated enough social pressure to override a broader sense of social desirability associated with such empathic behavior. In an anonymous online survey setting, participants may not have been concerned about immediate consequences from in-partisans for expressing empathy towards out-partisans but may have felt a stronger inclination to conform to overarching social norms. If participants were prompted to display empathy in the presence of their openly disapproving in-party, the results might be different (42). However, previous studies have observed effects even in contexts where explicit social sanctions are absent (31, 36). While the present studies focus on partisans’ intentions to empathize with out-partisans, a critical question concerns the suppression of empathetic actions toward out-partisans. Future research should delve into the behavioral outcomes of such intentions. It is crucial to explore whether partisans act in line with their genuine intentions, are swayed by what the in-party deems socially desirable, or adhere to a broader societal understanding of what is socially desirable beyond the in-party.

Another way in which social desirability could have influenced the results is that (ii) perceiving disapproval from the in-party for empathizing with out-partisans ironically might have increased the perceived desirability of such behavior. This parallels the phenomenon where trying to suppress a thought actually makes it more prominent (56). Consequently, partisans exhibit comparable levels of out-party empathy towards those who anticipated approval from their in-party for this behavior.

What are the implications of this? Irrespective of whether these results can be explained by participants’ genuine convictions, by a broader social desirability bias, or other alternative explanations, they generally suggest that the in-party does not play a significant role in partisans’ general intentions to empathize with those from the other side. Drawing attention to polarizing dynamics coming from the in-party does not strengthen these polarizing dynamics. Instead, individuals are able to resist the dynamics and overcompensate for the in-party. This holds true even for one of the most deeply affectively polarized landscapes, the United States (9, 21). If there is no decrease in empathy towards out-party members due to anticipated in-party social pressure in the United States, it is unlikely to manifest in countries where party identification is not as central and polarization is less pronounced.

An additional notable implication of an exploratory analysis is the potential significance of the feeling of disappointment. It warrants further investigation into when and why people become disappointed with others in politics and how it influences their attitudes, intentions, and behaviors. Disappointment appears to be an overlooked yet potentially influential emotion in shaping political preferences and actions (57). This calls for more research on the causes and consequences of disappointment in the political realm.

If perceived social pressure by the in-party does not decrease intentions to empathize with out-partisans, what other factors might be at play? It is very likely that hostile attitudes towards the out-party outweigh perceived social pressure from the in-party (45). Existing research strongly suggests that individuals are hesitant to empathize with individuals from opposing parties simply because of their partisan affiliation (25, 58, 59). Hence, the regulation of empathy might first and foremost hinge on the person empathized with rather than on concerns about how one might be perceived by one’s peers for empathizing with this person. This challenges the idea that in-group favoritism determines out-group derogation (37) and points to that intentions not to empathize with out-partisans could be driven mostly by out-group hate. The data presented here support this notion. Study 1 shows that the correlation between perceived in-party disapproval and self-reported out-party empathy disappears after controlling for indicators of out-party hostility (Fig. 2). Additional exploratory analyses show that reduced out-party empathy is highly correlated with out-party hate, however, not with in-party love (see Supplementary Material 2.2.3). Future studies should delve deeper into the relative significance of in-party and out-party dynamics, as well as their interplay, in regulating out-party empathy.

However, the downregulation of empathy may involve more than mere dislike for the opposing party. Partisans might strategically avoid empathizing with the out-party because they are worried about ending up in a weaker and disadvantaged position (38). They could also be concerned about out-partisans acting in bad faith rather than good faith and being taken advantage of if they were too empathic to them. This notion is echoed in the idea that “empathy engenders an asymmetry that empowers an (imaginary) other; the empathetic person is at the same time emptied out and weakened.” (60, p. 60). Study 1 reveals that out-party members are perceived as the least empathic towards their political opponents (Fig. 1), reflecting the widespread belief that “those on the other side don’t get [us]” (61). These perceptions are associated with reduced levels of empathy towards the out-party (see Supplementary Material 2.1), consistent with prior research showing how meta-perceptions about the out-party influence attitudes towards them (48–50, 62), but see (63, 64). Hence, when empathizing implies putting oneself in a comparatively weaker and more disadvantaged position toward someone who potentially acts in bad faith rather than good faith, refraining from it appears rational and in line with reciprocal norms.

Alternatively, the reluctance to empathize could simply be due to cognitive reasons rather than resentment or strategic considerations. Empathy requires mental effort (27, 28), and partisans may opt out of the additional cognitive load of empathizing with those who have very different views from their own, simply out of convenience.

These alternative explanations suggest that empathy is more influenced by the characteristics of the person being empathized with than by social pressure from fellow partisans. However, an alternative explanation of reduced empathy towards out-partisans that does not discount the influence of the in-party is that partisans decrease empathy not because they fear in-party disapproval for empathizing with out-partisans, but rather because they believe their in-party approves of not being empathic. In other words, the individual could abstain from empathy in the hope of being praised for it (65). Study 2 shows that partisans appear to adhere more closely to their in-party when these express approval rather than disapproval (see also (66)). Future studies should thus explore not only how the costs of empathy may diminish it but also how the benefits of not being empathic contribute to its reduction.

While these alternative explanations might all play a role, future studies should delve deeper into the relative importance of different factors shaping empathy in different contexts. Rather than focusing solely on one explanation, such as either in-party or out-party dynamics, researchers should contrast multiple explanations to discern their relative and collective influence on out-party empathy (19). Moreover, these dynamics warrant particular attention in real-world settings beyond online surveys, where behavioral data can offer richer insights. This approach is crucial for identifying and effectively addressing the potential and challenges of empathy to bridge hostile political divides.

Conclusion

There has been growing concern that social pressure within political parties could discourage individuals from empathizing with members of opposing parties (5, 15–17)—a critical tool to bridge political divides. The evidence presented in this article alleviates these concerns. Although cross-sectional analyses suggest that perceived disapproval from the in-party for empathizing with out-partisans is associated with a decrease in out-party empathy, experimental data reveal no statistically significant causal effect. Experimentally manipulating in-party disapproval (vs. approval) did not result in statistically significant decreases in out-party empathy. Neither did social pressure by in-partisans not to empathize with out-partisans affect other forms of engagement with the out-party, such as reading about an out-party opinion, being in contact with out-partisans, or liking the out-party. Surprisingly, partisans exhibit increased out-party empathy (along with increased warmth toward out-partisans and greater willingness to engage in contact with them) when they anticipated negative reactions from in-partisans compared to baseline. This unexpected rise in out-party empathy in anticipation of in-party disapproval may be linked to feelings of disappointment toward in-partisans. These findings suggest that partisans can resist and defy pressure by the in-party if they want. Considering cases such as Glenn Beck and Richard Fausset, it seems likely that both were aware of a potential backlash from within their own ranks. Yet, this awareness did not stop them from seeking to understand perspectives that differ significantly from their own.

Materials and methods

Study 1

Data

Study 1 is based on an IRB-approved (Aarhus University Institutional Review Board: BSS-2023-068) and pre-registered^k online survey of 1,200 Americans conducted between 2023 July 28–30 via the crowdsourcing platform Prolific. All 1,200 participants provided informed consent to the study, successfully completed one of two attention tests, and answered all questions in the survey.^l One duplicate ID was identified, and all corresponding responses were removed, reducing the sample to 1,199 participants. Each participant was compensated USD 1.83 (1.45 BP) for completing the survey. As pre-registered, final analyses included only self-identified Democrats and Republicans, including leaners, resulting in a final sample size of $N=\mathrm{1,046}$⁠.

Measures

This section describes the measurement of the key variables reported in study 1. In addition to these key variables, several other variables were measured in the survey. A comprehensive overview of all measured variables and descriptives can be found in Supplementary Material 1. Find questionnaires with exact measurements on the OSF repository: https://osf.io/y4jx8/.

Out-party empathy

Out-party empathy was measured on three levels: (i) self-report, (ii) perception of in-partisans, and (iii) perception of out-partisans. Each level utilized the same set of eight items, albeit with modifications specific to the respective level. Example items are: “I want to understand the reasons for why [out-party] hold their opinions” for self-reported out-party empathy; “Most [in-party] want to understand the reasons for why [out-party] hold their opinions” for perceptions of in-partisans’ degree of out-party empathy; “Most [out-party] want to understand the reasons for why [in-party] hold their opinions” for perceptions of out-partisans’ degree of out-party empathy. All items were measured on a Likert scale from $1=$ strongly disagree to $7=$ strongly agree. Values were recoded to range from 0–1 and combined into three indices taking the respective row means (see Cronbach’s alpha values in Supplementary Material 1.4). Higher values indicate higher out-party empathy. However, for the main analysis, perceived out-party empathy by the in-party was reverse coded with higher scores reflecting perceptions of the in-party being reluctant to empathize with out-partisans.

Disapproval of out-party empathy

Disapproval of out-party empathy was measured at two different levels: (i) self-report and (ii) perception of in-partisans. Each level utilized the same set of eight items, albeit with modifications specific to the respective level. For self-reported disapproval, participants were for example asked “To what extent do you approve or disapprove of if a [in-party] wants to understand the reasons for why [out-party] hold their opinions,” on a scale from $1=$ I strongly disapprove to $7=$ I strongly approve. For perceptions of in-partisan disapproval, participants were for example asked “To what extent do most [in-party] approve or disapprove of if a [in-party] wants to understand the reasons for why [out-party] holds their opinions,” on a scale from $1=$ Most [in-party] strongly disapprove to $7=$ Most [in-party] strongly approve. Values were recoded to range from 0–1 and combined into two indices taking the row mean (see Cronbach’s alpha values in Supplementary Material 1.4). Higher values reflect stronger disapproval of out-party empathy.

Models

In line with the pre-registration linear OLS regression models are used to regress self-reported out-party empathy on perceived in-party reluctance towards (H1) or disapproval of out-party empathy (H2). Four different models with different levels of control were calculated for each hypothesis: First, bivariate associations containing only the independent variable of interest. Then, stepwise control variables were added to achieve different levels of control: (C1) added sociodemographic control variables such as age, gender, education, race, and party identification. (C2) built on C1 and added either perceived in-party disapproval (for H1) or perceived in-party reluctance (for H2), as well as perceived out-party reluctance towards out-party empathy, perceiving out-party empathy as legitimization, and self-uncertainty. (C3) built on C2 and added indicators of out-party hostility such as affective polarization, contact with political opponents, harassment by other parties, and adherence to democratic norms. Unless otherwise noted, results do not change substantially by model specification and when inattentive participants are excluded (see Supplementary Material 2.1).

Study 2

Data

This study was pre-registered^m and IRB-approved (Aarhus University Institutional Review Board: BSS-2023-130-S2). Data were collected on the crowdsourcing platform Prolific 2023 December 1–3. It aimed for a convenience sample of 1,500 participants, balanced on Partisanship (50% Republicans, 50% Democrats) and Gender (50% male, 50% female) using Prolific’s screening questions. All participants had to provide informed consent before entering the study. Participants were also screened based on inattention to two out of two attention checks (in line with Prolific’s policies).ⁿ Eleven participants did not enter the correct submission code and were therefore excluded. This leaves a total of 1,489 participants (49.36% female, 50.16% Republicans, median age $=$ 43, 72.33% White, 10% Black, median education: bachelor’s degree or equivalent). Each participant was compensated $1.89 (PB 1.50) for completing the survey.

Measures

This section describes the measurement of the key outcome variables reported in study 2. In addition to these key variables, several other variables were measured. A comprehensive overview of all measured variables and descriptives can be found in Supplementary Material 1.1. Find all exact measurements in the questionnaire on the OSF repository: https://osf.io/y4jx8/.

Perceived in-party disapproval

To assess successful manipulation, participants in all three conditions were asked at the end of the survey “To what extent do you think fellow [in-partisans] approve or disapprove of you if you…,” followed by a battery of six items inspired by disapproval measures from study 1, for instance, “tried to learn about how [in-partisans] see the world to better understand their point of view.” Participants rated each item on a Likert Scale from $1=$ in-partisans strongly disapprove to $7=$ in-partisans strongly approve. Items were recoded to range from 0–1 and combined into an index by taking their row mean (see Cronbach’s alpha values in Supplementary Material 1.4). Higher values indicate higher perceived in-party disapproval.

Out-party empathy

The primary dependent variable is assessed similarly to self-reported out-party empathy in study 1, using eight items combined to an index ranging from 0 to 1. Higher values indicate greater out-party empathy.^o

Reading about the perspective of out-partisans

For a more behavioral outcome of out-party empathy, participants are asked to choose between reading an opinion piece in which an ordinary Republican or Democrat shared their perspective about the upcoming presidential election and explained their thoughts, feelings, and experiences that led them to support their political party.^p Participants are assigned a value of 0 if they chose to read about the perspective of an in-partisan and 1 if they chose to read about the perspective of an out-partisan.

Contact to out-party

Participants rate how likely they would be to engage in a political discussion with a [out-partisan], spend occasional social time with a [out-partisan], be next-door neighbors with a [out-partisan], be close friends with a [out-partisan], marry a [out-partisan], on a scale from $1=$ extremely unlikely to $7=$ extremely likely. Items were combined into an additive index by taking their row mean and recoding from 0–1, with higher values indicating a higher likelihood of being in contact with out-partisans (see Cronbach’s alpha values in Supplementary Material 1.4).

Feelings towards in- and out-party, and affective polarization

Feelings towards in- and out-partisans are measured using simple feelings thermometers, ranging from $0=$ cold and negative to $100=$ warm and positive. Values were recoded to range from 0–1, with higher values representing warmer feelings.

Discrete feelings towards in-partisan reactions

Participants are asked what feelings the anticipated reactions of fellow in-partisans triggered in them. Following a procedure by (67), participants first “select any of the following emotions that you feel when thinking about the [social cost condition: negative/social benefit condition: positive/baseline condition: empty] reactions from [in-partisans] that you just described,” choosing any of the following feelings: hopeful, proud, enthusiastic, happy, sad, disappointed, distressed, upset, angry, ashamed, afraid, anxious, hostile, disgusted, embarrassed, guilty, other (please indicate below). They then indicate on a scale from 0 (not at all) to 100 (extremely) the extent to which they felt the feelings they had just selected when they thought about the reactions of fellow in-partisans. Feelings that were not selected in the first phase were given a value of 0. Values are recoded to range from 0–1, with higher values indicating stronger feelings.

Models

In line with the pre-registration linear OLS regression models are used to compare means between the social cost and the social benefit conditions, treating the social benefit condition as baseline category. For exploratory analyses, I also compare means in the social cost and social benefit conditions with the baseline condition. If not otherwise noted, results are robust to excluding participants who are inattentive to attention checks as well as potential straightliners, and to controlling for potential imbalances in the randomization. In cases of heteroskedasticity, robust standard errors are computed. These include all models comparing any outcome between the social cost and social benefit conditions, except for the out-party feelings thermometer. Additionally, robust standard errors are used when comparing both treatment groups to the baseline, but only for outcomes out-party contact, choosing to read about the out-party perspective, and in-party feelings thermometer.

Notes

a

Although these studies mainly focus on how meta-perceptions about the out-party influence attitudes toward the out-party, rather than how meta-perceptions about the in-party influence attitudes toward the out-party.

b

https://osf.io/4akjv.

c

These hypotheses are labeled H3 and H4 in the pre-registration. Besides changing the order of hypotheses, I rephrased the original hypotheses by replacing the words “people” and “accurate understanding” with “partisans” and “out-party empathy.” Furthermore, study 1 was designed to test additional pre-registered hypotheses proposing that partisans have less out-party empathy the less they perceive the out-party to be empathic toward their in-party and the more they perceive engaging in out-party empathy to legitimize out-partisans” views. These hypotheses are excluded in the main manuscript due its primary focus on in-party dynamics shaping out-party empathy, but are reported in Supplementary Materials 1.2 and 2.1.

d

I also hypothesized that effects are stronger the more partisans believe that out-party empathy legitimizes out-partisans’ views or the more partisans feel uncertain about themselves. I found support for the former but not the latter. Partisans who think out-party empathy legitimizes their views tend to reduce out-party empathy more the more they think in-partisans disapprove of it. This effect, however, becomes statistically insignificant when controlling for indicators of out-party hostility. There is no significant interaction with self-uncertainty. However, exploratory analyses suggest that partisans with a higher need to belong to the in-party show lower empathy toward out-partisans if they perceive in-partisans as unwilling to engage in or disapproving of such empathy. See Supplementary Material 1.2 for an overview of all pre-registered hypotheses, Supplementary Material 2.1 for all pre-registered analyses, and Supplementary Material 2.2 for all exploratory analyses of study 1.

e

https://osf.io/fxcu9.

f

Randomization checks reveal no statistically significant differences on pre-treatment measures across conditions, except for one: participants in the social cost condition are, on average, slightly younger than those in the social benefit condition (⁠$b=-1.94$ years, $P<0.05$⁠). This difference becomes statistically insignificant (⁠$P>0.05$⁠) when adjusting for multiple comparisons (Bonferroni adjustments). Including age as a control variable in the analyses does not substantially change the results (see Supplementary Material 3.4).

g

To assess the robustness of the manipulation, I explore participants’ qualitative descriptions of how they think fellow in-partisans would react to them empathizing with out-partisans. Results are robust: participants in the social cost condition name more negative reactions than participants in the social benefit condition, with the baseline condition falling in between. See Supplementary Material 3.6.

h

Controlling for age to account for imbalances across conditions or exclusion of participants who failed attention checks do not substantially change results. However, excluding potential straightliners (277 participants) yields a somewhat larger and statistically significant effect in the expected direction (⁠$b=-0.039,\mathrm{se}=0.016,P<0.05$⁠, Cohen’s $d=0.17$⁠). Further exploration into sub-items of the out-party empathy index reveals statistically significant differences (⁠$P<0.05$⁠) in the expected direction for two out of eight items (“I try to learn about how [out-party] see the world to better understand their point of view” and “I don’t bother to understand the reasoning behind the opinions or beliefs of [out-party]”). Find robustness checks in Supplementary Material 3.

i

Exploratory analyses also revealed no statistically significant differences between Republicans and Democrats or between strong and weak partisans, see Supplementary Material 3.7.3.

j

Given the findings, I present a checklist of seven steps in Supplementary Material 3.1 to examine alternative explanations for discovering a statistically nonsignificant effect, as recommended by (68). This checklist provides support for considering the effect as statistically insignificant.

k

https://osf.io/4akjv.

l

10 participants exceeded the maximum time allowed to spend on the survey, as automatically set by Prolific to 44 min). Thirty-four participants were screened out for inattention on two of two attention tests

m

https://osf.io/fxcu9.

n

One hundred and six participants exited the study due to reasons such as not providing consent, failing attention checks, exceeding time limits, or dropping out. See Supplementary Material 3.3 for details on attrition rates throughout the experiment and across experimental conditions. The only statistically significant distinction noted was that participants who dropped out in the cost and benefit conditions tended to be older compared to those in the baseline condition. Adjusting for multiple comparisons (Bonferroni adjustments), rendered all comparisons statistically insignificant (⁠$P>0.05$⁠). Controlling for age does not change the results (see Supplementary Material 3.4).

o

Three of the six items underwent minor grammatical adjustments to improve readability. For example, the item in study 1, “I don’t waste much time trying to understand [out-partisans’] point of view” was revised in study 2 to “I don’t waste much time trying to understand the point of view of [out-partisans].”

p

The opinion piece was created using ChatGPT, which participants were informed about in the debriefing.

Acknowledgments

I am grateful to Michael Bang Petersen, Lasse Lindekilde, Kristian Vrede Skaaning Frederiksen, Troels Bøggild, Molly Moore, Robb Willer, Jamil Zaki, Mina Cikara, Leonie Huddy, David Broockman, members of the American Political Behavior Lab at the University of California, Berkeley, members of the Political Behavior and Institutions Section at Aarhus University, participants of the Political Psychology pre-conference at the American Political Science Association Annual Meeting 2023, and participants of the Directions of Polarization, Social Norms, and Trust in Societies Workshop at MIT 2023 for their invaluable feedback at different stages of this article.

Supplementary Material

Supplementary material is available at PNAS Nexus online.

Funding

This research is funded by two grants from the Political Science Department at Aarhus University.

Author Contribution

L.P.: Conceptualized and conducted studies 1 and 2, analyzed the data, and wrote the article.

Previous Presentation

Results were previously presented at the Political Psychology pre-conference at the American Political Science Association Annual Meeting (2023 August 1–September 3) as well as at the Directions of Polarization, Social Norms, and Trust in Societies Workshop at MIT (2023 December 1–2).

Data Availability

All data, scripts, and materials are accessible on the respective OSF project in anonymous format: https://osf.io/y4jx8/. To ensure anonymity, responses to open-ended questions will not be made available.

Author notes

Competing Interest: The author declares no competing interests.