MainOutgroup bias, ‘among the most well-documented and widely observed phenomenon in the social sciences’1, manifests in negative perceptions of and behaviour towards outgroups, who are viewed with suspicion and discriminated against in allocations of goods2. When outgroups are minorities, outgroup bias carries pernicious consequences across economic (job discrimination), health (for example, odds of long-standing illnesses and maternal mortality) and judicial/policing (bail decisions, traffic stops) domains, and overall lower pro-social behaviours towards outgroup members3,4,5,6,7,8,9. In the US context, racial and ethnic outgroup biases held by the dominant group have deep historical and institutional roots10, leading to polarized attitudes as well as discriminatory behaviour on many fronts. Bias against Black and Latino/a citizens is reflected in debates about virtually every important public policy issue, from immigration to policing and education11, and is similarly pervasive in online spheres12. The growth of Diversity, Equity and Inclusion (DEI) programmes designed to combat outgroup bias in both public, private (for example, higher education13) and government spheres (see the Biden Administration’s series of Executive Orders around DEI14,15,16) stands as further testament to the issue’s importance.Empathy, the act of taking the perspective and understanding the experiences of others17,18, is one tool that holds promise for attenuating the worst effects of outgroup bias, even towards heavily stigmatized ‘others’19. Previous work has emphasized either affective aspects of empathy, for example, when the focus is on the decision to engage in empathy20, or cognition, such as in the act of thinking through another’s perspective (sometimes referred to as ‘perspective taking’ or ‘mentalizing’)21,22.In keeping with the growing tendency to see empathy as multifaceted23, and neither purely cognitive nor purely affective, our conception of empathy focuses on imagining the experience of others (sometimes referred to as the ‘imagine him’ or ‘imagine other’ perspective to to distinguish it from imagining one’s self in a different situation24,25). This type of empathy, consistent with a host of other recent work26,27,28, involves more than simply inferring another’s mental state, but instead is ‘a process of feeling into, in which Person A opens him or herself in a deeply responsive way to Person B’s feelings and experiencing but without losing awareness that B is a distinct other self’29. Taking the perspective of others in this way is a powerful tool for shaping policy preferences and improving attitudes towards others more broadly17,18,30. Critically, the effects of empathy do not stop at attitudinal warmth, carrying over to affect partisan polarization30,31,32 and behaviours towards outgroups as well19,33.Given the scale of the problem and the ameliorative properties of empathy, it is no surprise that much attention has focused on the effects of empathy on outgroup bias34,35,36. Research in this domain has included empathy-based exercises embedded within intergroup contact scenarios, including work by refs. 37,38, which successfully utilize resource-heavy face-to-face conversations and perspective taking to reduce exclusionary attitudes towards outgroup members (see refs. 39,40 for recent reviews). Other works utilize interactive exercises41, online role-playing games42 (sometimes requiring specialized virtual reality hardware43,44), long videos45 and extended writing tasks46, while another thread of research suggests that outgroup bias can be overcome through accumulated life experiences that engender outgroup empathy47.However, even these promising interventions encounter three related issues. First, the ‘modal treatments’ in these studies are perspective-taking treatments that include second-hand or imagined contact with outgroups48. These designs beg the question: outside of scenarios in which experimenters assign people to engage in empathy-related tasks, how can decisions to engage in empathy be encouraged in the first place? That is, how can we overcome what one recent review referred to as the ‘challenge of motivation’49? Second, the interventions fielded thus far are costly and time consuming, often requiring careful training of enumerators and almost always additional costs in equipment, time and footwork. Finally, empathy arises more easily for ingroup than for outgroup members47,50 and some perspective-taking interventions can actually amplify egocentric biases51. This can lead to a vicious cycle that one scholar has called ‘the power of otherness to block empathy’52. This means that successfully encouraging empathy towards outgroups, specifically, is an even taller order than just encouraging it more broadly. Moreover, promoting only ‘parochial’ empathy (towards one’s own group) can be ‘counterproductive’20 and is unlikely to lead to the beneficial pro-social outcomes associated with general empathy. In addition to these larger substantive concerns, current work often suffers from a number of methodological issues, including small sample size, lack of pre-registration and outcomes that are nearly exclusively short term and attitudinal (only 7% of outcomes in a recent survey of the literature were behavioural48).The detrimental effects of outgroup prejudice and the salutary properties of empathy prompt our research question: How can we encourage greater empathy (and foster inclusive attitudes/behaviours) from members of the dominant (White) group towards racial and ethnic outgroups? Our argument is that, given its documented effects on pro-social behaviour, praise from peers is a promising candidate for promoting empathy as well as changing behaviours and attitudes towards racial outgroups. Moreover, leveraging praise draws upon peoples’ natural resources—the desire for positive esteem and admiration from peers—answering the call of a recent review to ‘align interventions with key psychological motivations’49. Our argument focuses additionally on the role of positive feelings and social norms as causal mechanisms linking peer praise to its effects on empathy and behavioural inclusion.Praise is a promising candidate for an intervention to promote empathy given its documented effects on pro-social behaviour (particularly in the large literature on child development53). Definitions of praise abound but generally agree that the concept centres on ‘positive evaluations…of another’s products, performances or attributes’54 (see also ref. 55). Praise can be about behaviour, effort or personal qualities, and can occur either ex-ante or ex-post whatever is being encouraged; in fact, the most common use of praise is in trying to increase effort levels, motivation or encouragement of specific behaviours in the future. It can be distinguished from highly similar concepts such as ‘encouragement’ (often associated with tasks with which a person is currently struggling or in which they performed negatively) and simple acknowledgment/feedback, which is inherently neutral and non-judgmental56,57. In line with a consensus that views behaviour and effort-specific praise as more effective than ‘personal praise’ (focused on peoples’ attributes or qualities58), we focus on praise for engaging in empathy randomly assigned to our respondents in advance of their choice to engage in empathy or not (ex-ante to cleanly identify the effects of the treatment on behaviour). While our focus on the connection between praise and pro-social behaviour is not new, previous work has often centred on child–parent relationships (with an emphasis on adolescent populations59).While praise is a promising candidate for an empathy-encouraging intervention, extant literature suggests lessons for how to operationalize praise by leveraging the importance of peers in social networks60. A multidisciplinary literature on peer effect processes portrays this group as increasingly important upon broaching adulthood, with ‘respected’ peers and ones who share values particularly influential59. Peer effects occur across contexts as diverse as uptake of education, future planning, emotional happiness and economic and welfare outcomes61,62,63,64,65. This work dovetails with a parallel literature on social learning, which has demonstrated the importance of ‘learning the value of stimuli, actions or knowledge from others’ for outcomes specifically related to empathy66,67.Peer effects have two additional features that make them particularly relevant for our purposes. First, peer influence is particularly effective in encouraging pro-social behaviours and attitudes, including warmth towards outgroups and inclusive behaviour68,69,70. Paluck and colleagues, for example, tap into peer networks to encourage anti-conflict norms and behaviour in a middle-school setting71. One explanation for the efficacy of peer influence is that ‘external social motivations’ can reduce ‘parochial empathy’, the tendency to empathize more with one’s own group and (relatively) less with outgroups20. In fact, among a plethora of lukewarm (and sometimes negative) results from diversity programmes, one set of promising results comes from drawing on people’s desire to look good in front of their peers72. Second, and critically for our purpose, peer effects seem to persist over time62,73. This is particularly helpful in the context of interventions to promote empathy, where durable effects are rarely found48.Our argument highlights two mechanisms: positive emotions and norms, through which praise from peers might affect empathy-related outcomes, each related to a different component of our intervention. We focus on emotions given the natural connection between empathy and affect more generally—empathy is fundamentally ‘an affective response’74—as well as its more particular role in the first part of the empathetic process, when individuals choose whether to engage in or avoid empathy75. Indeed, the link between praise and positive emotions is a baseline expectation in much of the literature76. This is summed up by ref. 77, which writes that the ‘obvious and immediate outcome’ of praise is ‘simple, positive affect’. Moreover, while there are strong links between praise and positive emotions, there are also connections between positive emotions and many other beneficial outcomes78,79,80,81,82, with some recent work suggesting a link between positive mood and pro-social behaviour83 that might operate as a feedback loop or ‘virtuous cycle’84. Other recent studies have conclusively demonstrated that triggering positive emotions (towards outgroups) leads to behaviours such as higher outgroup donations and likelihood of approaching rather than avoiding85. Our theory emphasizes the role of happiness—the most common positive emotion86, but we note the possibility of other positive emotions at play in the causal effect chain.Our second possible mechanism originates in the ‘peer’ component of our intervention and focuses on norms: beliefs about what others do or how they expect us to behave87. Since praise is inherently non-neutral, one pathway through which it might operate is by affecting beliefs about what is valued by the sender of the praise (here, one’s peers)56. As noted above, work on social learning suggests that learning from others in this way might be particularly potent, even for ‘complex social phenomena’ such as empathy88 and there is a ‘robust’ connection between social norms related to diversity and intergroup attitudes and behaviours’60.One reason to suspect this causal mechanism is the relative importance of social norms, the power and reach of which ‘can hardly be overestimated’89. In fact, one recent work finds exactly this in a set of laboratory and field experiments: a norms treatment shifted the attitudes of non-marginalized groups towards outgroups and their overall feelings towards pro-diversity initiatives70 (see also refs. 60,90). In another set of studies, ref. 91 finds that ‘people imitate others’ pro-social behaviours’; instituting norms of generosity and pro-social behaviour affected helping behaviour, donation amounts and even respondents’ own feelings (suggesting that there may even be a link between our two posited mechanisms). More broadly, the strength and content of norms has been found to affect behavioural patterns across societies as well92. The ‘normative pathway’ we focus on is close to, but subtly distinct from a ‘priming’ mechanism in which peer praise increases the ‘mental accessibility’ of the concept of empathy or pushes people to evaluate the target group or individual ‘through the lens’ of empathy (see discussion in ref. 93).In this registered report, we develop and tested a non-invasive, low-cost intervention harnessing natural peer influence to encourage empathy and political/attitudinal inclusion towards racial and ethnic outgroups. Our intervention is praise from peers about engaging in empathy and comes in the form of a word cloud of real praise elicited in pilot studies, as well as text describing average thermometer ratings towards people who are empathetic. Since making group identity salient risks triggering parochial empathy only towards ingroup members, our treatment promotes only general empathy towards others (with no mention of outgroups, specifically)20.As highlighted in our ‘Scale-up proposal for peer praise’ figure in Supplementary Information Section C, our intervention is relatively easy and low cost to adapt to new online or in-person contexts. For example, to test the intervention in a field setting, one would field a small survey to generate real peer praise for empathy—among whatever group one is interested in treating—and then conduct a baseline survey of attitudes and behaviours before using the collected peer praise intervention to encourage empathy (either online or in person). The overall implementation costs will be particularly low for the many contexts in which data on attitudes towards others are already collected (for example, college orientations, DEI training at private companies and public institutions). The cost of the initial survey to generate praise is also small, and especially so when compared to in-person interventions that require between 10 and 20 minutes per person of contact with trained enumerators, multiplied by thousands of people38,94. Our supplementary materials include a downloadable packet of instructions for researchers to adapt our intervention to other contexts.The real peer praise that fuels our treatment comes from one of two sources in our studies: either a neutral group of ‘online peers’ on MTurk (in our main Peer Praise Study) or a more identity-relevant group of online co-partisans (in our parallel Co-Partisan Peer Praise Study). In both cases, we are leveraging the ‘external social motivations’ that recent reviews have suggested as promising to reduce the empathy gap towards outgroups20. In keeping with the concern that researchers asking hypothetical questions will be rewarded with hypothetical answers, our design combines naturalistic real praise with a range of behavioural outcome measures, including donation behaviour, letter writing to the White House and a validated behavioural task in which respondents make real decisions about whether to engage in empathy or avoid it95,96, answering calls to focus our studies on behaviour as well as attitudes49.In addition to measuring (empathy and political) behaviour alongside attitudes, our research design incorporates three additional and novel features. First, our design takes up the challenge set out by two recent reviews48,49 to investigate whether effects from empathy-related interventions persist over time (analogous to recent work on long-term effects of correcting misperceptions97). To that end, we field a longitudinal survey experiment in which respondents are re-interviewed 1 week after the initial treatment in our main study. This provides a more stringent test of the intervention while also helping to shed greater light on mechanisms. Second, our main study features a placebo in Wave 1, addressing potential concerns about how the peer praise treatment works and ruling out alternative explanations. Finally, our parallel Co-Partisan Peer Praise Study uses a treatment in which praise comes from co-partisans, allowing us to investigate whether the efficacy of praise depends on how ‘identity relevant’ the source of praise is98.Our light-touch intervention and behavioural task have been extensively tested across seven pilot studies and a total of 2,466 participants (14,704 observations accounting for multiple trials; timeline of pilot studies presented in Fig. 1). Our approach of ‘rigorous incrementalism’, embodied by our seven pilot studies, established several crucial steps in the causal process, honed our peer praise treatment and generated clear baselines against which we will be able to judge the efficacy of our treatment in encouraging empathy towards racial and ethnic outgroups. Specifically, our pilots found:1.a general aversion to empathy that required a 10% premium in wages to overcome.2.respondents perceived engaging in empathetic description as more demanding, costly, difficult and anxiety producing than engaging in pure description.3.evidence of the effectiveness of peer praise in motivating empathy towards generalized ‘others’ (that is, without respondents knowing the identity of the target).4.evidence of the role of positive emotions and norms as causal mechanisms, both of which were supported to varying degrees in our pilots.Fig. 1: Pilot study timeline.Number of observations are presented for each pilot as N, with number of respondents as n, for studies with multiple trials of the main task per respondent.Full size imageIn this registered report, we focused on the effect of peer praise on empathy, attitudes and measures of behavioural and political inclusion. In all cases, our interest was in attitudes and behaviour among White respondents in the United States towards Black Americans and non-White Latino/a respondents, both being contexts where outgroup bias is particularly durable and pernicious. Our first set of six hypotheses in our ‘Peer Praise Study’ investigated the direct, immediate effect of praise from a neutral group of online peers on empathy and attitudinal and behavioural inclusion (see Peer praise design Table 1). The ‘peers’ in this part of our study were other MTurk workers, based in part on evidence that the MTurk community functions similarly to more traditional workplaces99. Online communities such as MTurk are increasing in scope and depth100 and function as a community through which workers can sustain and build identities101,102, build reputations103 and even garner better wages104.Table 1 Peer praise design tableFull size tableH1 focuses on the outcome of empathetic behaviour, preempting concerns about social desirability bias or impression management on the part of our respondents, while H2 uses a more common measure of self-reported empathy105,106. In line with work referenced earlier that links praise with positive emotions and pro-social behaviours as well as attitudinal warmth, we also posit that respondents who receive peer praise for empathy are more willing to engage in behaviour that is politically inclusive towards racial outgroups as well as view those same groups more warmly. H3 focuses on the effects of the peer praise treatment on political inclusion, for which we elicit two semi-behavioural outcomes: willingness to (1) donate to civil rights and racial justice advocacy groups (UnidosUS, Black Lives Matter) and (2) advocate on behalf of racial/ethnic equity to the White House. Those two measures are combined into an index of ‘semi-behavioural’ political inclusion41,107,108. H4 assesses the extent to which peer praise can promote attitudinal inclusion, measured through an index of (1) respondent-reported social distance and (2) thermometer measures towards Black and Latino/a outgroup members107,108,109.Our main Peer Praise Study also assessed the durability of our peer praise intervention via a longitudinal design in which respondents are recontacted 1 week following treatment (consort diagram of study presented in Fig. 2). As a recent review noted, one of the biggest challenges of interventions to improve attitudes and behaviours towards outgroups is whether proposed methods have any lasting effect48. We measure two outcomes for the longitudinal component of the study: empathetic behaviour (H5) and an attitudinal index (H6).Fig. 2: Peer Praise Study diagram.The design features three arms (A, B, C) where the first two arms are control and peer praise, and the last is placebo. All respondents are asked pre-treatment covariate questions and complete a practice trial of the behavioural empathy task, after which they are randomized into one of the three A, B, C arms with probabilities (0.4, 0.4, 0.2). Respondents then see each of the outcome questions (‘DVs’) in random order, then finish by answering questions about task difficulty, preference and post-treatment covariates. This completes the Wave 1 survey. Wave 2, fielded a week later, follows up with respondents in Arms A and B to survey DVs and post-treatment covariates again.Full size imageOur Peer Praise Study also includes two additional sets of analyses that are exploratory in nature, designed to probe mechanisms and ensure the validity of our treatment. To assess our proposed mechanisms, we both measured respondents’ level of happiness and their perceptions of norms around empathy, as well as use the longitudinal study to calibrate the plausibility of different causal paths. To the latter point, any average treatment effect (ATE) found in the longitudinal follow-up is less likely to be driven by (fleeting) positive emotions induced by our intervention. To ensure that the average treatment effect of peer praise was not driven by extraneous elements of the praise intervention (for example, the colours of the text), our main study included a placebo arm in which respondents also receive peer praise, but for ‘description’ rather than empathy. If the elements of the treatment that matter for promoting empathy are only presentational, then ‘praise for description’ should also cause a higher likelihood of choosing to engage in empathy in our choice task (choice task example images presented in Fig. 3). Below we summarize the first six registered hypotheses:H1: Peer praise increases likelihood of choosing to empathize (compared to no peer praise).H2: Peer praise increases self-reported empathy (compared to no peer praise).H3: Peer praise increases political inclusion (compared to no peer praise).H4: Peer praise increases attitudinal inclusion (compared to no peer praise).H5: Peer praise increases empathetic behaviour (measured 1 week following treatment) (compared to no peer praise).H6: Peer praise increases empathetic attitudes (measured 1 week following treatment) (compared to no peer praise).Fig. 3: Behavioural empathy task illustration.Top: example of an image (Race = White, Valence = Fearful) randomly drawn from a deck of cards, presented to respondents before choice task. Bottom: respondents choose between DESCRIBE and FEEL buttons to select their task.Full size imageOur second set of pre-registered hypotheses concerned the source of praise and are tested via a parallel study entitled Co-Partisan Peer Praise (depicted in the Co-partisan peer praise design Table 2). While H1–H4 assessed the efficacy of a peer praise treatment in which the praise came from a neutral group of online peers, it is possible that online peers might be too diffuse a group to matter to respondents, particularly if the efficacy of the treatment in promoting empathy depends on how much respondents value the opinion of the source of praise. While some extant work on peer effects in online networks99,102 suggests optimism that more generic peer praise might promote empathy, it is still possible that praise from a highly salient social identity grouping might be more effective.Table 2 Co-partisan peer praise design tableFull size tableTo explore whether co-partisan peer praise can affect empathy, political inclusivity and attitudinal warmth towards racial/ethnic outgroups, we pre-registered and fielded our Co-Partisan Peer Praise Study, identical to Wave 1 of the Peer Praise Study except for the source of the praise, which comes from co-partisans rather than MTurk workers (control group receives nothing). Co-partisans are a particularly important group in the United States, where partisan identity has been described as one of Americans’ ‘most salient social identities’110 (see also ref. 111). The outcomes for H7–H10 (empathy behaviour, self-reported empathy, political inclusion and attitudinal warmth) exactly parallel H1–H4 and are summarized below:H7: Co-partisan peer praise increases likelihood of choosing to empathize (compared to no co-partisan peer praise).H8: Co-partisan peer praise increases self-reported empathy (compared to no co-partisan peer praise).H9: Co-partisan peer praise increases political inclusion (compared to no co-partisan peer praise).H10: Co-partisan peer praise increases attitudinal inclusion (compared to no co-partisan peer praise).Additional exploratory analyses investigate the relative efficacy of co-partisan peer praise compared to the general MTurk peer praise. Because the tendency to empathize correlates with partisan identity112,113, we anticipate finding heterogeneous treatment effects by party in our exploratory analysis. Given that Democrats have been described as the party of inclusivity114, one might expect Democratic respondents to respond more to praise for empathy than Republican respondents, although conversely, if the correlation between party identity and empathy towards racial outgroups is high enough, it is possible that ‘ceiling effects’ among Democrats might suggest the opposite prediction. The consort diagram for the Co-Partisan Peer Praise Study can be found in Fig. 4.Fig. 4: Co-partisan peer praise diagram.The design features two main treatment arms, co-partisan peer praise and control. All respondents are asked pre-treatment covariate questions and complete a practice trial of the behavioural empathy task, after which they are randomized into one of the two arms with equal probability. Respondents then see each of the outcome questions (‘DVs’) in random order, then finish by answering questions about task difficulty and preference, and post-treatment covariates.Full size imageResultsWe fielded our survey experiment Study 1 Wave 1 on N = 5,303 adult respondents registered in the United States on MTurk in July and August 2025. Of Study 1 Wave 1 respondents, 90.31% were successfully recontacted for Wave 2 (an attrition rate between waves of 9.69%). Given the attrition of roughly a tenth of the respondents across waves in Study 1, we conducted registered Manski bound calculations on our hypotheses tested with longitudinal data (H5 and H6). Study 2 was fielded in August 2025 on N = 4,404 respondents.In our results below, we distinguish between analyses that were (1) registered and powered upon; that is, our main hypotheses, H1–10, (2) registered but not powered upon (that is, listed in registration) and (3) purely exploratory analyses. As per our registration, analyses only include respondents who passed both attention checks and completed the survey. Main results for the peer praise for empathy and co-partisan peer praise studies are presented in Figs. 5 and 6. Summary regression tables for H1–H6 and H7–H10 can be found in Supplementary Information Sections D and E respectively.Fig. 5: Study 1 estimated average treatment effects (ATE) for H1–H6.X axis is estimated ATE of peer praise compared to control, Y axis are hypotheses tested. For each hypothesis except H5 and H6, the ATE is plotted with 95% CI; for H5 and H6, we plot the Manski bounds for ATE which take into account attrition experienced in the follow-up. Wave 1 hypotheses are in purple, while long-run Wave 2 hypotheses are in yellow. Respective N sizes for H1–H6 tests are: N = 4,516, N = 5,081, N = 5,081, N = 5,081, N = 3,525, N = 3,947.Full size imageFig. 6: Study 2 estimated ATE for H7–H10.X axis is estimated ATE of co-partisan peer praise compared to control, Y axis are hypotheses tested. For each hypothesis, the ATE is plotted with 95% CI. Respective N sizes for H7–H10 tests are: N = 2,628, N = 2,914, N = 2,914, N = 2,914.Full size imagePeer praise for empathy increases political inclusion towards outgroups but does not change attitudes or prompt empathy behaviour (pre-registered)Our first set of hypotheses concern the effect of the ‘peer praise for empathy’ treatment on four outcomes: the choice to empathize in a behavioural task (H1), self-reported empathy (H2), political inclusion (H3) and attitudinal inclusion (H4). Our outcomes thus represent a broad range of attitudinal (H2, H4) and behavioural (H1, H3) outcomes.For attitudinal outcomes, we do not detect statistically distinguishable effects of the peer praise intervention. Specifically, estimates for both self-reported empathy (H2: β = −0.011; 95% CI [−0.038, 0.017]; p = 0.442) and our index of attitudinal inclusion, which combines social distance and warmth towards racial and ethnic outgroups (H4: β = −0.0031; 95% CI [−0.013,0.007]; p = 0.563) are centred near zero with confidence intervals including positive and negative values. Similarly, for the behavioural outcome of choosing to engage in empathy, respondents randomized into the peer praise condition were not statistically more likely than those in the control arm to choose to engage in empathy with a racial or ethnic outgroup (H1: β = −0.012; 95% CI [−0.044,0.019]; p = 0.437).Importantly, the absence of statistically significant effects for H1, H2 and H4 should not be interpreted as evidence of the absence of an effect. To assess whether these null results are informative, we conducted pre-registered Bayesian analyses comparing the null to alternatives of substantively meaningful magnitude. These analyses yield Bayes factors below one for H1, H2 and H4, indicating evidence favouring the null relative to effects of the size the study was designed to detect. Results are robust to Benjamini–Hochberg adjustments for multiple testing.While the peer praise intervention evinced no effects on any attitudinal measure and no effect on the choice to engage in empathy with a member of a racial or ethnic outgroup, it did cause an increase in our index of ‘political inclusion’ (H3: β = 0.031; 95% CI [0.012, 0.049]; p = 0.00156), which combines two outcomes: a donation task in which respondents pay a portion of their bonus to the Black Lives Matter (BLM) movement or UnidosUS and a letter-writing task in which respondents write an anonymous letter to the White House in support of prioritizing racial and ethnic equity policies.Study 2 was designed in a parallel fashion to Study 1, but the ‘praise for engaging in empathy’ originated from co-partisans rather than the more neutral group of online peers. Across outcomes, we again do not detect statistically distinguishable effects: self-reported empathy (H8: β = − 0.004; 95% CI [−0.035, 0.028]; p = 0.809), attitudinal inclusion (H10: β = − 0.004; 95% CI [−0.015, 0.007]; p = 0.455), choice to engage in empathy with outgroup members (H7: β = 0.006; 95% CI [−0.032, 0.044]; p = 0.744) and political inclusion (H9: β = 0.014; 95% CI [−0.011, 0.039]; p = 0.273). Bayesian analyses for H7–H10 similarly yield Bayes factors favouring the null relative to substantively meaningful alternatives (Bayes factors all Article Google Scholar Pager, D., Bonikowski, B. & Western, B. Discrimination in a low-wage labor market: a field experiment. Am. Sociol. Rev. 74, 777–799 (2009).Article PubMed PubMed Central Google Scholar Karlsen, S. & Nazroo, J. Y. Relation between racial discrimination, social class, and health among ethnic minority groups. Am. J. Public Health 92, 624–631 (2002).Article PubMed PubMed Central Google Scholar Arnold, D., Dobbie, W. & Yang, C. S. Racial bias in bail decisions. Q. J. Econ. 133, 1885–1932 (2018).Article Google Scholar Goncalves, F. & Mello, S. A few bad apples? Racial bias in policing. Am. Econ. Rev. 111, 1406–1441 (2021).Article Google Scholar Saluja, B. & Bryant, Z. How implicit bias contributes to racial disparities in maternal morbidity and mortality in the United States. J. Womens Health 30, 270–273 (2021).Article Google Scholar Stepanikova, I., Triplett, J. & Simpson, B. Implicit racial bias and prosocial behavior. Soc. Sci. Res. 40, 1186–1195 (2011).Article Google Scholar Pierson, E. et al. A large-scale analysis of racial disparities in police stops across the United States. Nat. Hum. Behav. 4, 736–745 (2020).Article PubMed Google Scholar Acharya, A., Blackwell, M. & Sen, M. The political legacy of American slavery. J. Polit. 78, 621–641 (2016).Article Google Scholar Hartman, T. K., Newman, B. J. & Scott Bell, C. Decoding prejudice toward hispanics: group cues and public reactions to threatening immigrant behavior. Polit. Behav. 36, 143–163 (2014).Article Google Scholar Freelon, D., Marwick, A. & Kreiss, D. False equivalencies: online activism from left to right. Science 369, 1197–1201 (2020).Article PubMed CAS Google Scholar Adida, C. L., Lo, A., Rhee, I. & Williams, S. Do equity, diversity and inclusion (EDI) requirements change student political attitudes? Polit. Groups Identities 12, 958–967 (2024).Diversity, Equity, Inclusion, and Accessibility in the Federal Workforce Executive Order 14035 (Federal Register, 2021).Advancing Racial Equity and Support for Underserved Communities Through the Federal Government Executive Order 13985 (Federal Register, 2021).Advancing Equity, Justice, and Opportunity for Asian Americans, Native Hawaiians, and Pacific Islanders Executive Order 14031 (Federal Register, 2021).Decety, J. & Cowell, J. M. Friends or foes: is empathy necessary for moral behavior? Perspect. Psychol. Sci. 9, 525–537 (2014).Article PubMed PubMed Central Google Scholar Waal, F. B. M. d The antiquity of empathy. Science 336, 874–876 (2012).Article PubMed Google Scholar Batson, C. D., Chang, J., Orr, R. & Rowland, J. Empathy, attitudes, and action: can feeling for a member of a stigmatized group motivate one to help the group?. Pers. Soc. Psychol. Bull. 28, 1656–1666 (2002).Article Google Scholar Behler, A. M. C. & Berry, D. R. Closing the empathy gap: a narrative review of the measurement and reduction of parochial empathy. Soc. Personal. Psychol. Compass 16, e12701 (2022).Article Google Scholar Zaki, J. Empathy: a motivated account. Psychol. Bull. 140, 1608–1647 (2014).Article PubMed Google Scholar Davis, M. H. in Handbook of the Sociology of Emotions (eds Stets, J. E. & Turner, J. H.) 443–466 (Springer, 2006).Muradova, L. & Arceneaux, K. Reflective political reasoning: political disagreement and empathy. Eur. J. Polit. Res. 61, 740–761 (2022).Article Google Scholar Stotland, E. Exploratory investigations of empathy. Adv. Exp. Soc. Psychol. 4, 271–314 (1969).Batson, C. D.The Altruism Question: Toward a Social-Psychological Answer (Psychology Press, 2014).Batson, C. D., Early, S. & Salvarani, G. Perspective taking: imagining how another feels versus imaging how you would feel. Pers. Soc. Psychol. Bull. 23, 751–758 (1997).Article Google Scholar Batson, C. D. & Ahmad, N. Y. Using empathy to improve intergroup attitudes and relations. Soc. Issues Policy Rev. 3, 141–177 (2009).Article Google Scholar Gubler, J. R., Karpowitz, C. F., Monson, J. Q., Romney, D. A. & South, M. Changing hearts and minds? Why media messages designed to foster empathy often fail. J. Polit. 84, 2156–2171 (2022).Article Google Scholar Barrett-Lennard, G. T. The empathy cycle: refinement of a nuclear concept. J. Couns. Psychol. 28, 91–100 (1981).Article Google Scholar Baron, H. et al. Can Americans depolarize? Assessing the effects of reciprocal group reflection on partisan polarization. Polit. Behav. 47, 1363–1388 (2025).Bayram, A. B. & Holmes, M. Feeling their pain: affective empathy and public preferences for foreign development aid. Eur. J. Int. Relat. 26, 820–850 (2020).Article Google Scholar DeSante, C. D. & Smith, C. W. Fear, institutionalized racism, and empathy: the underlying dimensions of Whites’ racial attitudes. PS Polit. Sci. Polit. 53, 639–645 (2020).Article Google Scholar Faulkner, N. ‘Put yourself in their shoes’: testing empathy’s ability to motivate cosmopolitan behavior. Polit. Psychol. 39, 217–228 (2018).Article Google Scholar Galinsky, A. D. & Moskowitz, G. B. Perspective-taking: decreasing stereotype expression, stereotype accessibility, and in-group favoritism. J. Pers. Soc. Psychol. 78, 708–724 (2000).Article PubMed Google Scholar Weyant, J. M. Perspective taking as a means of reducing negative stereotyping of individuals who speak english as a second language. J. Appl. Soc. Psychol. 37, 703–716 (2007).Article Google Scholar Shih, M., Wang, E., Trahan Bucher, A. & Stotzer, R. Perspective taking: reducing prejudice towards general outgroups and specific individuals. Group Process. Intergroup Relat. 12, 565–577 (2009).Article Google Scholar Broockman, D. & Kalla, J. Durably reducing transphobia: a field experiment on door-to-door canvassing. Science 352, 220–224 (2016).Article PubMed CAS Google Scholar Kalla, J. L. & Broockman, D. E. Reducing exclusionary attitudes through interpersonal conversation: evidence from three field experiments. Am. Polit. Sci. Rev. 114, 410–425 (2020).Article Google Scholar Tropp, L. R. & Barlow, F. K. Making advantaged racial groups care about inequality: intergroup contact as a route to psychological investment. Curr. Dir. Psychol. Sci. 27, 194–199 (2018).Article Google Scholar Vorauer, J. D. & Petsnik, C. in Psychological Intergroup Interventions: Evidence-based Approaches to Improve Intergroup Relations (eds Halperin, E. et al.) 31–44 (Routledge, 2024).Adida, C. L., Lo, A. & Platas, M. R. Perspective taking can promote short-term inclusionary behavior toward Dyrian refugees. Proc. Natl Acad. Sci. USA 115, 9521–9526 (2018).Article PubMed PubMed Central CAS Google Scholar Simonovits, G., Kezdi, G. & Kardos, P. Seeing the world through the other’s eye: an online intervention reducing ethnic prejudice. Am. Polit. Sci. Rev. 112, 186–193 (2018).Article Google Scholar van Loon, A., Bailenson, J., Zaki, J., Bostick, J. & Willer, R. Virtual reality perspective-taking increases cognitive empathy for specific others. PLoS ONE 13, e0202442 (2018).Article PubMed PubMed Central Google Scholar Herrera, F., Bailenson, J., Weisz, E., Ogle, E. & Zaki, J. Building long-term empathy: a large-scale comparison of traditional and virtual reality perspective-taking. PLoS ONE 13, e0204494 (2018).Article PubMed PubMed Central Google Scholar Pashak, T. J. et al. Empathy diminishes prejudice: active perspective-taking, regardless of target and mortality salience, decreases implicit racial bias. Psychology 9, 1340–1356 (2018).Article Google Scholar Bientzle, M., Eggeling, M., Kanzleiter, M., Thieme, K. & Kimmerle, J. The impact of narrative writing on empathy, perspective-taking, and attitude: two randomized controlled experiments on violations of covid-19 protection regulations. PLoS ONE 16, e0254501 (2021).Article PubMed PubMed Central CAS Google Scholar Cikara, M. & Fiske, S. T. Bounded empathy: neural responses to outgroup targets’ (mis)fortunes. J. Cogn. Neurosci. 23, 3791–3803 (2011).Article PubMed PubMed Central Google Scholar Paluck, E. L., Porat, R., Clark, C. S. & Green, D. P. Prejudice reduction: progress and challenges. Annu. Rev. Psychol. 72, 533–560 (2021).Article PubMed Google Scholar Čehajić-Clancy, S. & Halperin, E. Advancing research and practice of psychological intergroup interventions. Nat. Rev. Psychol. 3, 574–588 (2024).Article Google Scholar Sirin, C. V., Valentino, N. A. & Villalobos, J. D. Seeing Us in Them: Social Divisions and the Politics of Group Empathy (Cambridge Univ. Press, 2021).Eyal, T., Steffel, M. & Epley, N. Perspective mistaking: accurately understanding the mind of another requires getting perspective, not taking perspective. J. Pers. Soc. Psychol. 114, 547–571 (2018).Article PubMed Google Scholar Segal, E. A., Gerdes, K. E. & Lietz, C. A. Assessing Empathy (Columbia Univ. Press, 2017).Waugh, W., Brownell, C. & Pollock, B. Early socialization of prosocial behavior: patterns in parents’ encouragement of toddlers’ helping in an everyday household task. Infant Behav. Dev. 39, 1–10 (2015).Article PubMed PubMed Central Google Scholar Kanouse, D. E., Gumpert, P. & Canavan-Gumpert, D. in New Directions in Attribution Research Vol. 3 (eds Harvey, J. H. et al.) 97–115 (L. Erlbaum Associates, 1976).Kelly, F. D. & Daniels, J. G. The effects of praise versus encouragement on children’s perceptions of teachers. Individ. Psychol. 53, 331–341 (1997).Google Scholar Henderlong, J. & Lepper, M. R. The effects of praise on children’s intrinsic motivation: a review and synthesis. Psychol. Bull. 128, 774–795 (2002).Article PubMed Google Scholar Pety, J., Kelly, F. & Kafafy, A. E. A. The praise-encouragement preference scale for children. Individ. Psychol. 40, 92–101 (1984).Google Scholar Mueller, C. M. & Dweck, C. S. Praise for intelligence can undermine children’s motivation and performance. J. Pers. Soc. Psychol. 75, 33–52 (1998).Article PubMed CAS Google Scholar Brechwald, W. A. & Prinstein, M. J. Beyond homophily: a decade of advances in understanding peer influence processes. J. Res. Adolesc. 21, 166–179 (2011).Article PubMed PubMed Central Google Scholar Murrar, S. & Brauer, M. in Psychological Intergroup Interventions (eds Halperin, E. et al.) 45–57 (Routledge, 2024).Duflo, E. & Saez, E. Participation and investment decisions in a retirement plan: the influence of colleagues’ choices. J. Public Econ. 85, 121–148 (2002).Article Google Scholar Duflo, E., Dupas, P. & Kremer, M. Peer effects, teacher incentives, and the impact of tracking: evidence from a randomized evaluation in Kenya. Am. Econ. Rev. 101, 1739–1774 (2011).Article Google Scholar Fowler, J. H. & Christakis, N. A. Dynamic spread of happiness in a large social network: longitudinal analysis over 20 years in the Framingham Heart Study. BMJ 337, a2338 (2008).Article PubMed PubMed Central Google Scholar Bertrand, M., Luttmer, E. F. P. & Mullainathan, S. Network effects and welfare cultures. Q. J. Econ. 115, 1019–1055 (2000).Article Google Scholar Carrell, S., Fullerton, R. & West, J. Does your cohort matter? Measuring peer effects in college achievement. J. Labor Econ. 27, 439–464 (2009).Article Google Scholar Pärnamets, P., Espinosa, L. & Olsson, A. Physiological synchrony predicts observational threat learning in humans. Proc. R. Soc. B 287, 20192779 (2020).Article PubMed PubMed Central Google Scholar Pan, Y., Novembre, G. & Olsson, A. The interpersonal neuroscience of social learning. Perspect. Psychol. Sci. 17, 680–695 (2022).Article PubMed Google Scholar Barry, C. M. & Wentzel, K. R. Friend influence on prosocial behavior: the role of motivational factors and friendship characteristics. Dev. Psychol. 42, 153–163 (2006).Article PubMed Google Scholar Balconi, M. & Canavesio, Y. Emotional contagion and trait empathy in prosocial behavior in young people: the contribution of autonomic (facial feedback) and Balanced Emotional Empathy Scale (BEES) measures. J. Clin. Exp. Neuropsychol. 35, 41–48 (2013).Article PubMed Google Scholar Murrar, S., Campbell, M. R. & Brauer, M. Exposure to peers’ pro-diversity attitudes increases inclusion and reduces the achievement gap. Nat. Hum. Behav. 4, 889–897 (2020).Article PubMed Google Scholar Paluck, E. L., Shepherd, H. & Aronow, P. M. Changing climates of conflict: a social network experiment in 56 schools. Proc. Natl Acad. Sci. USA 113, 566–571 (2016).Article PubMed PubMed Central CAS Google Scholar Dobbin, F. & Kalev, A. Getting to Diversity: What Works and What Doesn’t (Harvard Univ. Press, 2022).Lépine, A. & Estevan, F. Do ability peer effects matter for academic and labor market outcomes? Labour Econ. 71, 102022 (2021).Article Google Scholar Hoffman, M. L. in Emotions, Cognition, and Behavior (eds Izzard, C. E. et al.) 103–131 (Cambridge Univ. Press, 1984).Sirin, C. V., Valentino, N. A. & Villalobos, J. D. The social causes and political consequences of group empathy. Polit. Psychol. 38, 427–448 (2017).Article Google Scholar Morton, J. S., Mikolajczak, M. & Luminet, O. New perspectives on the praise literature: towards a conceptual model of compliment. Curr. Psychol. 41, 6038–6050 (2022).Delin, C. R. & Baumeister, R. F. Praise: more than just social reinforcement. J. Theory Soc. Behav. 24, 219–241 (1994).Article Google Scholar Erez, A. & Isen, A. M. The influence of positive affect on the components of expectancy motivation. J. Appl. Psychol. 87, 1055–1067 (2002).Article PubMed Google Scholar Foo, M.-D., Uy, M. A. & Baron, R. A. How do feelings influence effort? An empirical study of entrepreneurs’ affect and venture effort. J. App. Psychol. 94, 1086–1094 (2009).Article Google Scholar Oswald, A. J., Proto, E. & Sgroi, D. Happiness and productivity. J. Labor Econ. 33, 789–822 (2015).Article Google Scholar Storbeck, J., Dayboch, J. & Wylie, J. Fear and happiness, but not sadness, motivate attentional flexibility: a case for emotion influencing the ability to split foci of attention. Emotion 19, 655–664 (2019).Article PubMed Google Scholar Subramaniam, K. & Vinogradov, S. Improving the neural mechanisms of cognition through the pursuit of happiness. Front. Hum. Neurosci. https://doi.org/10.3389/fnhum.2013.00452 (2013).Aknin, L. B., Van de Vondervoort, J. W. & Hamlin, J. K. Positive feelings reward and promote prosocial behavior. Curr. Opin. Psychol. 20, 55–59 (2018).Article PubMed Google Scholar Layous, K., Nelson, S. K., Kurtz, J. L. & Lyubomirsky, S. What triggers prosocial effort? A positive feedback loop between positive activities, kindness, and well-being. J. Posit. Psychol. 12, 385–398 (2017).Article Google Scholar Čehajić-Clancy, S., Jamshed, N., Olsson, A. & Momčilović, A. From inspiration to restoration: moral elevation as a catalyst for improving intergroup relations in contexts of conflict. J. Pers. Soc. Psychol. 130, 77–92 (2026).Davidson, R. in Handbook of Emotion (eds Feldman Barrett, L. et al.) 143–154 (Guilford Press, 1993).House, B. R. How do social norms influence prosocial development? Curr. Opin. Psychol. 20, 87–91 (2018).Article PubMed Google Scholar Olsson, A., Knapska, E. & Lindström, B. The neural and computational systems of social learning. Nat. Rev. Neurosci. 21, 197–212 (2020).Article PubMed CAS Google Scholar McDonald, R. I. & Crandall, C. S. Social norms and social influence. Curr. Opin. Behav. Sci. 3, 147–151 (2015).Article Google Scholar Stangor, C., Sechrist, G. B. & Jost, J. T. Changing racial beliefs by providing consensus information. Pers. Soc. Psychol. Bull. 27, 486–496 (2001).Article Google Scholar Nook, E. C., Ong, D. C., Morelli, S. A., Mitchell, J. P. & Zaki, J. Prosocial conformity: prosocial norms generalize across behavior and empathy. Pers. Soc. Psychol. Bull. 42, 1045–1062 (2016).Article PubMed Google Scholar House, B. R. et al. Universal norm psychology leads to societal diversity in prosocial behaviour and development. Nat. Hum. Behav. 4, 36–44 (2020).Article PubMed Google Scholar Chong, D. & Druckman, J. N. Framing theory. Annu. Rev. Polit. Sci. 10, 103–126 (2007).Article Google Scholar Kalla, J. L. & Broockman, D. E. Which narrative strategies durably reduce prejudice? Evidence from field and survey experiments supporting the efficacy of perspective-getting. Am. J. Polit. Sci. 67, 185–204 (2023).Article Google Scholar Cameron, C. D. et al. Empathy is hard work: people choose to avoid empathy because of its cognitive costs. J. Exp. Psychol. Gen. 148, 962–976 (2019).Article PubMed Google Scholar Ferguson, A. M. & Inzlicht, M. Reliability of the empathy selection task, a novel behavioral measure of empathy avoidance. Behav. Res. Methods 55, 2638–2651 (2023).Carnahan, D., Bergan, D. E. & Lee, S. Do corrective effects last? Results from a longitudinal experiment on beliefs toward immigration in the US. Polit. Behav. 43, 1227–1246 (2021).Article Google Scholar Simas, E. N., Clifford, S. & Kirkland, J. H. How empathic concern fuels political polarization. Am. Polit. Sci. Rev. 114, 258–269 (2020).Article Google Scholar Martin, D., Hanrahan, B. V., O’Neill, J. & Gupta, N. Being a turker. In Proc. 17th ACM Conference on Computer Supported Cooperative Work and Social Computing 224–235 (Association for Computing Machinery, 2014).Scholz, T. Digital Labor: The Internet as Playground and Factory (Routledge, 2012).Irani, L. Difference and dependence among digital workers: the case of Amazon Mechanical Turk. South Atl. Q. 114, 225–234 (2015).Article Google Scholar Wang, X., Zhu, H., Li, Y., Cui, Y. & Konstan, J. A community rather than a union: understanding self-organization phenomenon on Mturk and how it impacts turkers and requesters. In Proc. 2017 CHI Conference Extended Abstracts on Human Factors in Computing Systems, 2210–2216 (Association for Computing Machinery, 2017).Brawley, A. M. & Pury, C. L. Work experiences on Mturk: job satisfaction, turnover, and information sharing. Comput. Hum. Behav. 54, 531–546 (2016).Article Google Scholar Kaplan, T., Saito, S., Hara, K. & Bigham, J. Striving to earn more: a survey of work strategies and tool use among crowd workers. Proc. AAAI Conf. Hum. Comput. Crowdsourc. 6, 70–78 (2018).Arceneaux, K. Anxiety reduces empathy toward outgroup members but not ingroup members. J. Exp. Polit. Sci. 4, 68–80 (2017).Article Google Scholar Williamson, S. et al. Family matters: how immigrant histories can promote inclusion. Am. Polit. Sci. Rev. 115, 686–693 (2021).Article Google Scholar Winter, N. & Berinsky, A. J. What’s your temperature? Thermometer ratings and political analysis. In Annual Meeting of the American Political Science Association (APSA, 1999).Enos, R. D. & Gidron, N. Exclusion and cooperation in diverse societies: experimental evidence from Israel. Am. Polit. Sci. Rev. 112, 742–757 (2018).Article Google Scholar Weisz, E., Ong, D. C., Carlson, R. W. & Zaki, J. Building empathy through motivation-based interventions. Emotion 21, 990–999 (2021).Article PubMed Google Scholar Iyengar, S. & Krupenkin, M. Partisanship as social identity; implications for the study of party polarization. Forum 16, 23–45 (2018).West, E. A. & Iyengar, S. Partisanship as a social identity: implications for polarization. Polit. Behav. 44, 807–838 (2022).Article Google Scholar Allamong, M. B. & Peterson, D. A. Screw those guys: polarization, empathy, and attitudes about out-partisans. Polit. Psychol. 42, 349–369 (2021).Article Google Scholar Clifford, S., Kirkland, J. H. & Simas, E. N. How dispositional empathy influences political ambition. J. Polit. 81, 1043–1056 (2019).Article Google Scholar Levitsky, S. & Ziblatt, D. How Democracies Die (Crown, 2018).Kass, R. E. & Raftery, A. E. Bayes factors. J. Am. Stat. Assoc. 90, 773–795 (1995).Article Google Scholar Merton, R. K. The Matthew effect in science: the reward and communication systems of science are considered. Science 159, 56–63 (1968).Article PubMed CAS Google Scholar Shamay-Tsoory, S. G. & Hertz, U. Adaptive empathy: a model for learning empathic responses in response to feedback. Perspect. Psychol. Sci. 17, 1008–1023 (2022).Article PubMed Google Scholar McCammon, S. How Empathy Came to be Seen as a Weakness in Conservative Circles (NPR, 2025).Chater, N. & Loewenstein, G. The i-frame and the s-frame: how focusing on individual-level solutions has led behavioral public policy astray. Behav. Brain Sci. 46, e147 (2023).Article Google Scholar Hara, K. et al. A data-driven analysis of workers’ earnings on Amazon Mechanical Turk. In Proc. 2018 CHI Conference on Human Factors in Computing Systems https://doi.org/10.1145/3173574.3174023 (Association for Computing Machinery, 2018).Bashri, M. F. & Kusumaningrum, R. Sentiment analysis using latent dirichlet allocation and topic polarity wordcloud visualization. In 2017 5th International Conference on Information and Communication Technology (ICoIC7) 1–5 (IEEE, 2017).Mustaqim, T., Umam, K. & Muslim, M. Twitter text mining for sentiment analysis on government’s response to forest fires with vader lexicon polarity detection and k-nearest neighbor algorithm. J. Phys. Conf. Ser. 1567, 032024 (2020).Grimmer, J., Roberts, M. E. & Stewart, B. M. Text as Data: A New Framework for Machine Learning and the Social Sciences (Princeton Univ. Press, 2022).Martinez, J. E. Facecraft: race reification in psychological research with faces. Perspect. Psychol. Sci. 20, 182–194 (2025).Ma, D. S., Correll, J. & Wittenbrink, B. The Chicago face database: a free stimulus set of faces and norming data. Behav. Res. Methods 47, 1122–1135 (2015).Article PubMed Google Scholar American National Election Studies. 2016 time series study (datafile). ANES https://electionstudies.org/data-center/2016-time-series-study/ (2017).Andreychik, M. R. Feeling your joy helps me to bear feeling your pain: examining associations between empathy for others’ positive versus negative emotions and burnout. Pers. Individ. Dif. 137, 147–156 (2019).Article Google Scholar Ben-Ner, A., McCall, B. P., Stephane, M. & Wang, H. Identity and in-group/out-group differentiation in work and giving behaviors: experimental evidence. J. Econ. Behav. Organ. 72, 153–170 (2009).Article Google Scholar Linos, K., Jakli, L. & Carlson, M. Fundraising for stigmatized groups: a text message donation experiment. Am. Polit. Sci. Rev. 115, 14–30 (2021).Article Google Scholar Buchanan, L., Bui, Q. & Patel, J. K. Black lives matter may be the largest movement in US history. The New York Times (3 July 2020).Francis, M. M. & Wright-Rigueur, L. Black lives matter in historical perspective. Annu. Rev. Law Soc. Sci. 17, 441–458 (2021).Article Google Scholar Minta, M. D. Diversity and minority interest group advocacy in congress. Polit. Res. Q. 73, 208–220 (2020).Article Google Scholar UnidosUS - GuideStar Profile. https://www.guidestar.org/profile/86-0212873Hart, S. G. & Staveland, L. E. Development of NASA-TLX (task load index): results of empirical and theoretical research. Adv. Psychol. 52, 139–183 (1988).Article Google Scholar Davis, M. H. Measuring individual differences in empathy: evidence for a multidimensional approach. J. Pers. Soc. Psychol. 44, 113–126 (1983).Article Google Scholar Neuliep, J. W. & McCroskey, J. C. The development of a US and generalized ethnocentrism scale. Commun. Res. Rep. 14, 385–398 (1997).Article Google Scholar Buhrmester, M., Kwang, T. & Gosling, S. D. Amazon’s Mechanical Turk: a new source of inexpensive, yet high-quality data? Perspect. Psychol. Sci. 6, 3–5 (2011).Guess, A. M. & Munger, K. Digital literacy and online political behavior. Polit. Sci. Res. Methods 11, 110–128 (2023).Article Google Scholar Coppock, A. & McClellan, O. A. Validating the demographic, political, psychological, and experimental results obtained from a new source of online survey respondents. Res. Polit. https://doi.org/10.1177/2053168018822174 (2019).Lovett, M., Bajaba, S., Lovett, M. & Simmering, M. J. Data quality from crowdsourced surveys: a mixed method inquiry into perceptions of Amazon’s Mechanical Turk masters. Appl. Psychol. 67, 339–366 (2018).Article Google Scholar Berry, C., Kees, J. & Burton, S. Drivers of data quality in advertising research: differences across Mturk and professional panel samples. J. Advert. 51, 515–529 (2022).Article Google Scholar Enns, P. K. & Rothschild, J. Do you know where your survey data come from? Medium https://medium.com/3streams/surveys-3ec95995dde2 (2022).Lo, A., Renshon, J. & Bassan-Nygate, L. A practical guide to dealing with attrition in political science experiments. J. Exp. Polit. Sci. 111, 147–161 (2024).Lin, W. Agnostic notes on regression adjustments to experimental data: reexamining Freedman’s critique. Ann. Appl. Stat. 7, 295–318 (2013).Berinsky, A. J., Margolis, M. F., Sances, M. W. & Warshaw, C. Using screeners to measure respondent attention on self-administered surveys: which items and how many? Polit. Sci. Res. Methods 9, 430–437 (2021).Article Google Scholar Manski, C. Nonparametric bounds on treatment effects. Am. Econ. Rev. 80, 319–323 (1990).Google Scholar Göritz, A. S. in Online Panel Research: A Data Quality Perspective (eds Callegaro, M. et al.) 154–170 (Wiley Online Library, 2014).Lo, A. adelinelo/Praise-and-Empathy: Replication package for “The effect of praise from peers on empathy and political inclusion towards racial or ethnic outgroups” Zenodo https://doi.org/10.5281/zenodo.18749063 (2026).Download referencesAcknowledgementsFor comments and suggestions that aided the manuscript, we thank C. Adida, D. D. Choi, S. Clifford, E. Hartman, D. Kanter, R. McDermott, Y. Nygate, H. Pifarré i Arolas, D. Posner, M. Schwarze, C. V. Sirin and J. Weeks. J.R. was supported by the Institute for Humane Studies as well as a grant from the University of Wisconsin-Madison Office of the Vice Chancellor for Research and Graduate Education with funding from the Wisconsin Alumni Research Foundation and the University of Wisconsin-Madison Trice Faculty Award. A.L. was supported by funding provided by the Office of the Vice Chancellor for Research at the University of Wisconsin-Madison with funding from the Wisconsin Alumni Research Foundation. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.Author informationAuthors and AffiliationsDepartment of Political Science, University of Wisconsin-Madison, Madison, WI, USAAdeline Lo & Jonathan RenshonKennedy School of Government, Harvard University, Cambridge, MA, USALotem Bassan-NygateAuthorsAdeline LoView author publicationsSearch author on:PubMed Google ScholarJonathan RenshonView author publicationsSearch author on:PubMed Google ScholarLotem Bassan-NygateView author publicationsSearch author on:PubMed Google ScholarContributionsA.L. and J.R. conceived and designed the experiments and contributed materials/analysis tools. A.L. and L.B.-N. performed the experiments and analysed the data. A.L., J.R. and L.B.-N. wrote the paper.Corresponding authorCorrespondence to Adeline Lo.Ethics declarationsCompeting interestsThe authors declare no competing interests.Peer reviewPeer review informationNature Human Behaviour thanks Sabina Cehajic-Clancy, Oliver Christ and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.Additional informationPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Extended dataExtended Data Fig. 1 Real peer praise gathered for engaging in empathetic behavior.Real peer praise gathered for engaging in empathetic behavior in panel (a); Real peer praise by co-partisan group Democrats (b) and Republicans (c).Extended Data Fig. 2 Power analyses.Left panel: Study 1 (Peer Praise Study) Power analysis. X-axis is N size, Y-axis is power level. Data points represent average power from 10,000 simulations at a given N level for multiple-hypothesis adjusted H1 through H6. To power at 0.95 for all 12 hypotheses, we require N=5,300. Right panel: Study 2 (Co-Partisan Peer Praise Study) Power analysis. X-axis is N size, Y-axis is power level. Data points represent average power from 10,000 simulations at a given N level for multiple-hypothesis adjusted H7 through H10. To power at 0.95 for all 4 hypotheses, we require N=4,400.Supplementary informationSupplementary Information (download PDF )Supplementary Tables B1–E21, Figs. B1–D45, Information on Pilot Studies; Information on measurements/power calculations, and survey instrument; Information on Studies 1 and 2.Reporting Summary (download PDF )Peer Review file (download PDF )Rights and permissionsOpen Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.Reprints and permissionsAbout this article