1. Introduction
Fear and hatred across partisan lines have grown dramatically in the United States in recent years (Iyengar et al., Reference Iyengar, Lelkes, Levendusky, Malhotra and Westwood2019). “Affective polarization” of this sort activates, and is sustained by, a range of psychological processes that erode trust, foster misperceptions, and foment mutual hostility across the partisan divide, all of which may damage democracy (Ahler and Sood, Reference Ahler and Sood2018, Finkel et al., Reference Finkel, Bail, Cikara, Ditto, Iyengar, Klar, Mason, McGrath, Nyhan, Rand, Skitka, Tucker, Bavel, Wang and Druckman2020, Moore-Berg et al., Reference Moore-Berg, Ankori-Karlinsky, Hameiri and Bruneau2020, Mernyk et al., Reference Mernyk, Pink, Druckman and Willer2022). Affective polarization may also cause citizens to prioritize partisan interests over democratic principles (Mason Reference Mason2018, McCoy et al. Reference McCoy, Rahman and Somer2018, Iyengar et al. Reference Iyengar, Lelkes, Levendusky, Malhotra and Westwood2019, Graham and Svolik Reference Graham and Svolik2020, Svolik Reference Svolik2020; c.f. Broockman et al. Reference Broockman, Kalla and Westwood2023).
In this paper, we experimentally evaluate a promising but understudied mechanism for reducing affective polarization: vicarious intergroup contact, whereby individuals who observe or learn about direct intergroup contact enjoy its benefits without actually participating in it (Wright et al., Reference Wright, Aron, McLaughlin-Volpe and Ropp1997). Vicariously reading about or watching members of one’s ingroup engage in civil, mutually respectful dialogue with outgroup members may generate many of the same benefits of direct contact while avoiding some of the potential limitations and adverse unintended consequences, as we discuss below. Indeed, a recent study comparing 25 “light touch” depolarization interventions found that a 4-minute advertisement depicting harmonious cross-partisan interactions in the UK produced the largest reductions in partisan animosity (Voelkel et al., Reference Voelkel, Stagnaro, Chu, Pink, Mernyk, Redekopp, Ghezae, Cashman, Adjodah and Allen2024). Yet to our knowledge, only four studies have examined the effects of vicarious cross-partisan contact in any rigorous, systematic way (Wojcieszak and Warner, Reference Wojcieszak and Warner2020, Huddy and Yair, Reference Huddy and Yair2021, Voelkel et al., Reference Voelkel, Chu, Stagnaro, Mernyk, Redekopp, Pink, Druckman, Rand and Willer2023, Reference Voelkel, Stagnaro, Chu, Pink, Mernyk, Redekopp, Ghezae, Cashman, Adjodah and Allen2024). (We describe these studies in further detail in Supplementary Information (SI) A.)
We advance this nascent literature by reporting results from a pre-registered, placebo-controlled randomized trial designed to test the effects of a 50-minute documentary depicting a direct contact workshop administered by Braver Angels, a nationwide nonprofit that seeks to reduce affective polarization among Americans.Footnote 1 During the workshop, which took place shortly after the 2016 US presidential election, Democrats and Republicans convened to interact with and learn about outgroup members in a structured, facilitated environment. The documentary features footage of the workshop and interviews with the organizers and participants—rank-and-file American voters who expressed policy preferences that are fairly typical of their respective parties (e.g., opposition to abortion among Republican participants, and support for universal health care among Democrats). The creators of the film partnered with us because they have preexisting ties to media production companies and sought an independent impact evaluation to inform their case for distributing the documentary more widely.
To evaluate the documentary, we recruited a nationally representative sample of 2,000 Americans from YouGov’s online panel in the summer of 2022. We contacted participants in three waves. In wave 1, we collected information on demographics and potential sources of treatment effect heterogeneity, and assigned participants to one of five treatment conditions: (1) the 50-minute Braver Angels documentary; (2) a 50-minute placebo nature documentary about wildebeest migration; (3) a 5-minute version of the Braver Angels documentary; (4) a second 5-minute version that emphasized partisan misperception correction; and (5) a pure control group.Footnote 2 We measured outcomes in waves 2 and 3, which were administered on average 5 and 50 days after treatment, respectively. Here, we focus primarily on our wave 2 results and the comparison between the Braver Angels and placebo documentaries.
Consistent with our pre-registered hypotheses,Footnote 3 we find that vicarious contact via the documentary reduced affective polarization, our primary attitudinal outcome of interest. Effects on our primary behavioral outcomes are more mixed: vicarious contact increased interest but not investment in future depolarization programming. The documentary also curbed stereotyping of out-partisans—one of the mechanisms that we hypothesized might explain the reduction in affective polarization. Consistent with some previous studies (Santoro and Broockman, Reference Santoro and Broockman2022, Voelkel et al., Reference Voelkel, Chu, Stagnaro, Mernyk, Redekopp, Pink, Druckman, Rand and Willer2023, Baron et al., Reference Baron, Blair, Choi, Gamboa, Gottlieb, Robinson, Rosenzweig, Turnbull and West2025), we find no evidence that the documentary mitigated anti-democratic attitudes. We do find, however, that it increased optimism about the survival of democratic institutions and bolstered faith in the efficacy of dialogue as a tool for political change. The documentary had larger effects on self-identified Democrats, although the difference with self-identified Republicans is not statistically significant. We find some suggestive evidence that these effects decayed over time.
Our study makes at least five contributions to the growing literature on depolarization. First, as noted above, ours is one of only a handful of studies of vicarious contact across partisan (rather than, say, ethnic) lines, which, despite its promise, remains an under-explored mechanism for reducing partisan animosity. It is not obvious that findings from studies of other forms of vicarious intergroup contact will generalize to vicarious cross-partisan contact, as research on partisan animosity explicitly draws on a “distinct theoretical literature” (Paluck et al., Reference Paluck, Porat, Clark and Green2021, 537). Second, whereas most studies of cross-partisan contact (vicarious or otherwise) involve interventions that were designed and implemented by researchers, we evaluate the effects of a documentary produced by an established nonprofit with a nationwide profile that is committed to depolarizing the American public and, importantly, to disseminating the documentary as widely as possible. This increases the probability that the intervention we evaluate will be scaled up outside the context of the study itself. We also extend beyond the “light touch” interventions that have become popular in depolarization research (Voelkel et al., Reference Voelkel, Stagnaro, Chu, Pink, Mernyk, Redekopp, Ghezae, Cashman, Adjodah and Allen2024) and the prejudice reduction literature more broadly (Paluck et al., Reference Paluck, Porat, Clark and Green2021). Indeed, in their landmark review of the literature, Paluck et al. (Reference Paluck, Porat, Clark and Green2021, 536) lament the “scarcity of prejudice reduction research on the kinds of programs that are frequently called for in the real world, specifically entertainment and mass media interventions.” Our study helps to fill this gap.
Third, whereas most depolarization studies focus on measuring attitudes alone (Hartman et al., Reference Hartman, Blakey, Womick, Bail, Finkel, Han, Sarrouf, Schroeder, Sheeran, Van Bavel, Willer and Gray2022), our evaluation combines both attitudinal and behavioral outcomes. We also measure a wider range of attitudinal outcomes than most depolarization studies,Footnote 4 and a more realistic set of behavioral outcomesFootnote 5 designed to capture the extent to which participants might support and seek out future depolarization programming—an important outcome in itself, since durable depolarization likely requires reinforcement over time. Fourth, while most cross-partisan contact studies (vicarious or otherwise) measure outcomes immediately or only very shortly after treatment is administered (Hartman et al., Reference Hartman, Blakey, Womick, Bail, Finkel, Han, Sarrouf, Schroeder, Sheeran, Van Bavel, Willer and Gray2022), we measure outcomes in both the short (5 days) and medium (50 days) terms,Footnote 6 thus generating important insights into treatment effect durability while also mitigating concerns about social desirability bias and experimenter demand effects, which are likely to diminish over time. Our use of a nationally representative sample should also increase our ability to make credible inferences about the likely effects of the documentary on the U.S. public.
Finally, the vicarious contact intervention we studied (the Braver Angels documentary) was a byproduct of a direct contact intervention (a Braver Angels workshop), illustrating how vicarious contact may help extend the reach and magnify the impact of direct contact. The contact occurred in the wake of a highly contentious presidential election, thus showcasing the possibility of civil cross-partisan relations even at a moment of heightened partisan animosity. Equally important, the workshop on which the documentary was based was the subject of a previous experimental evaluation that used very similar attitudinal outcome measures (Baron et al., Reference Baron, Blair, Choi, Gamboa, Gottlieb, Robinson, Rosenzweig, Turnbull and West2025), thus providing a plausible benchmark for our treatment effect estimates. This can help inform the durability-scalability trade-off that is inherent to any comparison between direct and vicarious contact interventions, helping practitioners decide where to direct their efforts and funds (Hartman et al., Reference Hartman, Blakey, Womick, Bail, Finkel, Han, Sarrouf, Schroeder, Sheeran, Van Bavel, Willer and Gray2022, Littman et al., Reference Littman, Scacco, Weiss, Halperin, Hameiri and Littman2023). As discussed below, we estimate that the reach of the documentary is roughly equivalent to 10 direct contact workshops implemented by the same nonprofit, achieved in 12% of the time and at a fraction of the cost. Taken together, these findings indicate that scalable, inexpensive vicarious contact interventions can reduce affective polarization, at least in the short term.
2. Theoretical framework
One possible avenue for reducing intergroup animosity—including affective polarization—is through direct intergroup contact (e.g., Pettigrew and Tropp, Reference Pettigrew and Tropp2006, Scacco and Warren, Reference Scacco and Warren2018, Mousa, Reference Mousa2020, Paluck et al., Reference Paluck, Porat, Clark and Green2021, Weiss, Reference Weiss2021). While most studies of the “contact hypothesis” (Allport, Reference Allport1954) focus on prejudice between racial, ethnic, or sectarian groups, the effects of contact may extend to animosity across the partisan divide. For instance, recent research has shown that in-person cross-partisan workshops can reduce affective polarization and increase donations to depolarization initiatives, with effects lasting at least six months (Levendusky and Stecula, Reference Levendusky and Stecula2021, Baron et al., Reference Baron, Blair, Choi, Gamboa, Gottlieb, Robinson, Rosenzweig, Turnbull and West2025). Other studies have shown that online cross-partisan conversations can reduce affective polarization as well (Santoro and Broockman, Reference Santoro and Broockman2022, Combs et al., Reference Combs, Tierney, Guay, Merhout, Bail, Hillygus and Volfovsky2023, Tausch et al., Reference Tausch, Birtel, Górska, Bode and Rocha2024).
Overall, however, direct contact has yielded mixed results, and recent studies suggest the effectiveness of these interventions may be sensitive to small changes in program design and scope (Paluck et al., Reference Paluck, Green and Green2019, Santoro and Broockman, Reference Santoro and Broockman2022, Zhou and Lyall, Reference Zhou and Lyall2025). Moreover, direct contact interventions tend to be time- and resource-intensive; typically rely on competent, experienced facilitators; and require recruiting and matching participants across partisan lines—a challenge given that many Americans feel threatened by, and thus avoid, exposure to out-partisans (Enos, Reference Enos2014, MacInnis and Page-Gould, Reference MacInnis and Page-Gould2015, Dorison et al., Reference Dorison, Minson and Rogers2019). This may cause a dynamic whereby only those who are already interested in engaging with out-partisans agree to participate in direct contact interventions. Indeed, previous studies suggest that individuals are more likely to seek out direct contact when they believe that doing so will help them achieve a specific goal, such as making new friends or learning new skills (Kauff et al., Reference Kauff, Beneda, Paolini, Bilewicz, Kotzur, O’Donnell, Stevenson, Wagner and Christ2021). Conversely, others may find themselves forced into direct contact (e.g., in school or at work)—a loss of volition that may heighten the risk of backfire effects (Paolini et al., Reference Paolini, Harwood, Rubin, Huck, Dunn and Dixon2024). These limitations may diminish the prospects for direct contact to reduce partisan animosity at scale.
Vicarious intergroup contact offers a potential solution to these problems. Like direct contact, vicarious contact is rooted in the intuition that cooperative, mutually respectful intergroup interactions can reduce intergroup animosity, especially when participants are united by a common goal, have equal standing, and enjoy social and institutional support (Allport, Reference Allport1954). Vicarious contact likely activates both cognitive and emotional mechanisms among participants, which work in tandem. Cognitively, it may transmit information that corrects (excessively) negative stereotypes about the outgroup while also fostering perceptions of intergroup similarity (Wojcieszak and Warner, Reference Wojcieszak and Warner2020)—for instance, by convincing participants that their policy preferences are closer to those of the outgroup than they previously realized. Vicarious contact may also increase open-mindedness to outgroup views.
Emotionally, vicarious contact may promote intergroup empathy and reduce the anxiety that may otherwise arise when individuals think about, or interact with, outgroup members (Mazziotta et al., Reference Mazziotta, Mummendey and Wright2011, Vezzali et al., Reference Vezzali, Hewstone, Capozza, Giovannini and Wölfer2014). Vicarious contact may also mitigate the perception of threat (to values, to the country, to one’s ingroup, etc.) posed by outgroup members—likely a complex process involving both cognition and emotion. Individuals may also extrapolate beyond their own experiences of vicarious contact to make inferences about intergroup relations more generally. For example, they may become more optimistic about the efficacy of dialogue as a mechanism for promoting intergroup harmony; about the prospects for avoiding intergroup violence and promoting intergroup civility; and—to the extent that endemic intergroup conflict is perceived as a threat to democracy—about the survival of democratic institutions in the future.
The theory underlying vicarious contact posits that individuals can experience these benefits even if they merely observe intergroup contact, rather than participating in it themselves. Unlike direct contact, vicarious contact can be experienced in private, without raising concerns about social sanctions or unpleasant intergroup interactions. Individuals are also likely to enjoy greater discretion over their exposure to vicarious contact, since they can simply look away from contact they find unappealing. In this sense, the psychological barriers to entry may be lower for vicarious contact than for direct contact because individuals may find it less threatening to simply observe intergroup contact rather than participate in it themselves. Observing positive direct contact may also help establish civil, mutually respectful intergroup interactions as a social norm, making participants believe that intergroup interaction is both common and desirable. Finally, vicarious contact may magnify the effects of direct contact interventions. Direct contact may help depolarize those who actually engage in it, while vicarious exposure to direct contact may help depolarize those who merely witness it.
While scholarship on vicarious contact is still nascent (Paluck et al., Reference Paluck, Porat, Clark and Green2021), previous studies suggest that it can improve intergroup relations (Mazziotta et al., Reference Mazziotta, Mummendey and Wright2011, Vezzali et al., Reference Vezzali, Hewstone, Capozza, Giovannini and Wölfer2014). But there may also be limits to what contact—direct, vicarious, or otherwise—can achieve. For example, one observational study from Denmark finds that while (self-reported) direct contact across partisan lines is negatively correlated with affective polarization, the association is null among those who strongly identify with their political party (Thomsen and Thomsen, Reference Thomsen and Thomsen2023). The effects of contact in the US may similarly depend on the strength of participants’ party identification—for example, if partisan motivated reasoning induces individuals to resist the cognitive mechanisms that contact activates (Mason, Reference Mason2018). The effects may also be concentrated among particular partisan or ideological groups. For instance, previous studies suggest that conservatives are more likely than liberals to believe that “to compromise with one’s political opponents is dangerous because it usually leads to the betrayal of our own side” (Jost et al., Reference Jost, Sterling, Stern and Fishbach2017, Jost, Reference Jost2017). This may dampen the impact of cross-partisan contact among more conservative participants. We examine these and other potential moderators below.
3. Materials and methods
3.1. Recruitment and randomization
We conducted a pre-registered, placebo-controlled trial using a nationally representative sample of Americans (n = 2000) recruited from YouGov’s online panel. We fielded wave 1 of the study between June 27 and July 15, 2022. In total, 2,573 participants were randomized into one of the five conditions described in the introduction. From those 2,573 participants, YouGov created a nationally representative sample of 2,000 by matching participants’ demographic characteristics to the 2019 American Community Survey and voter files, as described in SI C. Participants who were assigned to one of the three treatment conditions (the full Braver Angels documentary or one of the two shorter films) or the placebo condition were then invited to watch the video to which they were assigned; participants who were assigned to the pure control condition were not shown a video. After participants finished watching the videos, we presented them with comprehension checks and three open-ended qualitative questions about the film.
In wave 2, we re-contacted all 2,573 participants who were randomized into one of the five conditions to complete a 10-15 minute survey, which we use to measure the study’s main outcomes. Participants were recontacted at least 24 hours after treatment; they completed the survey 5 days after treatment on average. 2,105 participants completed wave 2. Out of this wave 2 sample, YouGov created a nationally representative sample of 1,600 participants. We fielded wave 3 between August 16 and 22, 2022. Again, all 2,573 participants were re-contacted in wave 3; 1,612 completed the survey. From this latter sample, YouGov created a nationally representative sample of 1,360 participants. Descriptive statistics and balance tests are available in SI D and SI F, respectively. We discuss attrition below and in SI G.
3.2. Outcomes and measurement
We test the effect of the documentary on two main outcomes: (1) affective polarization, measured attitudinally, and (2) interest and investment in depolarization, measured behaviorally. Vicarious contact is likely to have the strongest and most lasting effects if it not only reduces affective polarization, but also inspires Americans to learn more about, participate in, and support depolarization initiatives and the organizations that implement them. In other words, vicarious contact is likely to be most effective if it not only changes attitudes, but alters behaviors as well.
Following Baron et al. Reference Baron, Blair, Choi, Gamboa, Gottlieb, Robinson, Rosenzweig, Turnbull and West2025, we measured affective polarization using an index of five itemsFootnote 7 which were standardized and averaged (Cronbach’s α = 0.744):
(1) Feeling thermometer: We asked respondents how “warmly” they feel toward the inparty and outparty on a scale of 0 to 100. We then calculated the difference between these two numbers.Footnote 8
(2) Social distance: We asked respondents how comfortable they would feel having out-partisans as (1) close personal friends and (2) neighbors, and (3) how comfortable they would feel if their best friend married an out-partisan.
(3) Trust: We asked respondents how often they believe they can trust the inparty and outparty to “do what is right for the country” on a scale of 1 to 5, where 1 indicates almost never and 5 indicates almost always. We then calculated the difference between these two numbers.
(4) Outparty threat (binary): We asked respondents whether they believe out-partisans constitute a “serious threat” to the United States.
(5) Negative partisanship (binary): We asked respondents whether they identify with their party primarily out of support for the inparty or primarily out of opposition to the outparty.
Our pre-registered affective polarization index thus benchmarks trust and warmth toward the outparty against trust and warmth toward the inparty. We also construct an alternative version of the index that does not subtract outparty warmth from inparty warmth or outparty trust from inparty trust (Cronbach’s α = 0.766).Footnote 9 This alternative index uses outparty warmth and outparty trust alone. While it was not pre-registered, this approach to operationalizing out-partisan animus is more standard in the literature (Hartman et al., Reference Hartman, Blakey, Womick, Bail, Finkel, Han, Sarrouf, Schroeder, Sheeran, Van Bavel, Willer and Gray2022).
We measured interest and investment in depolarization in three ways. First, in wave 2, we asked participants whether they were interested in signing up for the Braver Angels newsletter, which features news about upcoming depolarization workshops and events. At the end of the survey, we provided participants who expressed interest with the link to sign up for the newsletter and tracked whether they clicked the link. Unfortunately, we were not able to track whether they actually signed up for the newsletter. We instead assume that if respondents expressed interest in signing up and also clicked a link to do so later in the survey, then they were genuinely interested in learning more about depolarization.Footnote 10
Second, in wave 2, we gave participants the option of donating some of their study compensation to three different depolarization organizations: Living Room Conversations, AllSides for Schools, and Braver Angels.Footnote 11 Finally, in wave 3, we gave participants the opportunity to take part in one-on-one, online cross-partisan conversations run by the organization Unify America,Footnote 12 which is independent of Braver Angels. Participants who expressed interest in taking part in these conversations were redirected to the Unify America website using a special link. We tracked whether they clicked the link. Unify America also informed us whether participants signed up after clicking the link, and whether they actually participated in a conversation.
3.2.1. Secondary outcomes and mechanisms
We also tested the effects of the Braver Angels documentary on two secondary outcomes. First, we expected that by mitigating affective polarization, vicarious contact might have the salutary side effect of increasing participants’ optimism about the prospects for reducing hostility and fostering civility across partisan lines, and for ensuring the longevity of American democratic institutions. We asked respondents to rate how optimistic they are about the survival of democratic institutions, the restoration of civility between Democrats and Republicans, and the ability of American democracy to overcome polarization (on a scale of 1-5). We also asked them to rate how effective they believe dialogue is as a tool for change (on a scale of 0-100), and how long they believe it will take to rebuild trust across the partisan divide (ranging from 0-5 years to never).
Second and relatedly, we expected that vicarious contact might also strengthen participants’ commitment to democratic norms and procedures. While scholars continue to debate the causal relationship between affective polarization and commitment to democracy (Broockman et al., Reference Broockman, Kalla and Westwood2023), recent research suggests that highly polarized citizens are often willing to trade off democratic principles for partisan gain (Graham and Svolik, Reference Graham and Svolik2020). We hypothesized that by reducing affective polarization, vicarious contact might weaken this tendency. We thus asked participants to rate how likely they would be to support anti-democratic practices that benefit their own party in six hypothetical scenarios.Footnote 13 We then standardized and averaged these items into an index.
Finally, we posited several mechanisms through which the documentary might mitigate affective polarization. First, vicarious contact might encourage participants to question negative stereotypes about out-partisans and potentially adopt positive stereotypes instead. We asked participants to rate how well different characteristics describe members of both their own party and the outparty. Some characteristics were positive (patriotic, honest, generous, intelligent, and open-minded) while others were negative (mean, selfish, and hypocritical). We construct two standardized indices of positive and negative stereotypes. Second, vicarious contact might encourage open-mindedness to outparty views and greater willingness to consider the perspectives of outpartisans. Third, vicarious contact might increase empathy toward outgroup members and their concerns.Footnote 14 Fourth, vicarious contact might weaken participants’ belief that Democrats and Republicans hold irreconcilable views on key policy issues. We asked participants to rate on a scale of 0-100 how strongly they believe Democrats and Republicans disagree on three issues in particular: abortion, paid family leave, and same-sex marriage. Vicarious contact may also mitigate beliefs that partisans are excessively divided.Footnote 15
3.3. Analysis strategy
For most of the analyses reported in this paper, we estimate the intention-to-treat (ITT) effect of the documentary using a simple bivariate regression comparing mean outcomes among participants assigned to the full-length Braver Angels documentary to mean outcomes among participants assigned to the placebo nature documentary. We privilege the matched over the full sample because this is what we pre-registered, but also because we are most interested in how the documentary would affect a representative sample of Americans. Additionally, the matched sample exhibits no differential attrition between the Braver Angels and placebo documentary conditions, allowing us to make more reliable inferences. We weight all observations by the product of the inverse probability of being included in the matched nationally representative sample and the inverse probability of completing the survey. We describe our procedure for constructing these weights below and in SI G.2. We also report results for the full sample in SI I.1.Footnote 16
To test for treatment effect heterogeneity by party ID and other potential moderators, we simply interact each moderator with our indicator for assignment to the Braver Angels documentary.
3.4. Attrition
Given that we asked participants to consent to take part in the study before knowing their treatment assignment, we were concerned about participants dropping out due to the length of the Braver Angels documentary. We were especially concerned about the possibility of differential attrition across treatment conditions.Footnote 17 Beyond including a placebo group, we took several additional steps to mitigate this risk. First, as part of the informed consent process, we told participants that they might be asked to watch a 50-minute documentary, and asked them to consent only if they were certain they would be able to watch. Second, we informed them that they would receive extra compensation (in the form of a $20 Amazon gift card) if they were assigned to watch a documentary. Third, we used a second informed consent filter to attempt to screen out likely attriters: after participants gave informed consent, we reminded them again that they might be asked to watch a 50-minute documentary, and gave them another opportunity to opt out of the study prior to randomization. Patterns of attrition reveal both that our concerns were justified—we observe significantly higher attrition in the two 50-minute documentary conditions—and that our attempts to mitigate differential attrition between the full-length Braver Angels film and the placebo documentary of the same length were effective. See Table SI.9 for further details.
We thus focus on comparing participants assigned to the full-length Braver Angels documentary with those assigned to the placebo nature documentary in the nationally representative sample, where we observe no differential attrition between treatment conditions. As we show in SI G, observable differences between attriters and non-attriters in this sample are substantively small, with standardized mean differences (SMDs) close to 0 for most covariates. Levels of compensation and the amount of time spent watching a film as part of the study are identical for these two groups of participants, ensuring greater comparability.Footnote 18 Compliance rates for both the Braver Angels documentary and the placebo were high. 94% of respondents assigned to the placebo answered an attention check question correctlyFootnote 19, while 95% of respondents assigned to the Braver Angels documentary correctly answered one of two manipulation check questions.Footnote 20 There was no statistically significant difference in compliance between the two groups (p = 0.42).Footnote 21
As an additional precaution, we use inverse probability weights (IPWs)Footnote 22 to correct for the small and statistically insignificant levels of differential attrition that we observe between the Braver Angels and placebo nature documentary groups. See SI G.2 for further details on the construction of these weights.
4. Results
4.1. Primary outcomes: Affective polarization and interest and investment in depolarization activities
Consistent with our hypotheses, relative to the placebo nature documentary, we find that the Braver Angels documentary reduced scores on our pre-specified affective polarization index by 0.14 standard deviations (
$\beta \!\!=\!\! -0.140$, p = 0.039, N = 584), and reduced scores on a more conventional index focused on outparty animus by 0.163 standard deviations (
$\beta \!=\! -0.163$, p = 0.017, N = 584).Footnote 23 Figure 1 visualizes these results. The figure reports the ITT of the Braver Angels documentary at wave 2, which participants completed 5 days after treatment on average.Footnote 24 We operationalize affective polarization using the five-item index described above, which we standardize for ease of interpretation. Larger values indicate greater affective polarization.
Vicarious intergroup contact reduced affective polarization among a nationally representative sample of americans.
We show in SI I.1 that our results are robust to alternative specifications that do not include weights. We also show in SI I.1 that the ITT on our pre-specified index is consistently (albeit marginally) smaller than the ITT on our alternative index focused on outparty animus, implying that the documentary reduced affective polarization primarily by softening partisan animosity, rather than by closing the gap between perceptions of the inparty and outparty.
Results for our behavioral outcomes are more mixed, as we show in Figure 2. Consistent with our hypotheses, relative to placebo, we find that participants who were assigned to watch the Braver Angels documentary were 7.7 percentage points more likely to click a link to sign up for the newsletter (
$\beta = 0.077$, p = 0.035, N = 509). However, contrary to our hypotheses, we find no evidence that the documentary increased investment in depolarization organizations. If anything, we find suggestive evidence for the opposite: relative to placebo, participants who were assigned to watch the Braver Angels documentary were 3.7 percentage points less likely to make any donations (
$\beta \!=\! -0.037$, p = 0.369, N = 583), though this difference is substantively small and not statistically significant at conventional levels.Footnote 25 One possible explanation for these mixed effects on our behavioral outcomes is that YouGov respondents expected their compensation to be commensurate with the amount of time they spent completing the study; while the Braver Angels film appears to have stimulated their interest in future Braver Angels programming, it apparently did not convince them to forfeit any of their expected compensation for watching a relatively long documentary. It is possible that we would have observed more positive effects on donations if we had provided respondents with a windfall separate from their compensation, but we cannot be sure.
Vicarious contact increases interest but not investment in depolarization activities.
4.2. Secondary outcomes: Democratic optimism and anti-democratic attitudes
Relative to placebo, we find that the Braver Angels documentary increased participants’ optimism about the survival of democratic institutions (
$\beta = 0.319$, p = 0.001, N = 583) and the restoration of civility between Democrats and Republicans (
$\beta = 0.431$, p = 0.000, N = 583), and also strengthened participants’ belief in the efficacy of dialogue as a tool for change (
$\beta = 0.457$, p = 0.000, N = 528). While the documentary moderately increased participants’ belief that non-violent change is possible, this effect is substantively smaller and not statistically significant at conventional levels (
$\beta = 0.176$, p = 0.135, N = 583). Figure 3 displays these results.Footnote 26
Vicarious contact increased optimism and strengthened belief in the efficacy of dialogue.
Contrary to our hypotheses, however, we find no evidence that the documentary mitigated (or exacerbated) anti-democratic attitudes (
$\beta \!\!=\!\! -0.080$, p = 0.263, N = 583). Figure 4 displays these results. This is consistent with other studies showing that interventions that reduce affective polarization do not necessarily affect attitudes toward democracy (Voelkel et al., Reference Voelkel, Chu, Stagnaro, Mernyk, Redekopp, Pink, Druckman, Rand and Willer2023, Baron et al., Reference Baron, Blair, Choi, Gamboa, Gottlieb, Robinson, Rosenzweig, Turnbull and West2025). Taken together, these results suggest that while interventions aimed at mitigating polarization may not have a direct effect on participants’ willingness to serve as checks on anti-democratic behavior among elites, they may engender optimism about democracy and bolster commitment to dialogue.
Vicarious contact does not affect support for anti-democratic actions.
4.3. Mechanisms: Stereotypes, mass perceptions of policy differences, outgroup empathy, and open-mindedness
In our PAP, we specified several possible mechanisms that might explain the documentary’s effects on affective polarization and support for depolarization as a goal. Consistent with our hypotheses, we find that the Braver Angels documentary strengthened positive stereotypes about the outparty (
$\beta \!\!=\!\! 0.201$, p = 0.062, N = 581) while weakening negative ones (
$\beta \!\!=\!\! -0.202$, p = 0.043, N = 581), though the former effect is only weakly statistically significant at conventional levels.Footnote 27 We believe that stereotyping is one plausible mechanism through which vicarious contact reduces partisan animosity. Similarly, the film improved measures of open-mindedness (
$\beta \!=\! 0.196$, p = 0.093, N = 488) and empathy for the outgroup (
$\beta \!\!=\!\! 0.227$, p = 0.089, N = 489), although both of these effects are only weakly statistically significant at conventional levels. Figure SI.8 in SI I.1.3 presents the results of these analyses.
Contrary to our hypotheses, however, we find no evidence that the documentary mitigated participants’ belief that Democrats and Republicans are irreconcilably divided on key policy issues (
$\beta\!\! =\!\! -0.036 $, p = 0.550, N = 536). This is true even when breaking this measure down by issue: abortion (
$\beta\!\! =\!\! 0.080$, p = 0.484, N = 518), paid family leave (
$\beta\!\! =\!\! -0.139$, p = 0.180, N = 488), or same-sex marriage (
$\beta\!\! =\!\! -0.101$, p = 0.365, N = 507). While not decisive, these results cast doubt on perceived partisan agreement—at least with respect to policy issues—as a mechanism through which vicarious contact reduces affective polarization. We also find no evidence that the film affected respondents’ perceptions of division or unity between Republicans and Democrats (
$\beta \!\!\!=\!\!\! 0.168$, p = 0.169, N = 545). These results suggest that stereotyping, open-mindedness, and empathy for the outgroup may serve as mechanisms through which vicarious contact reduces affective polarization, while perceived similarity of policy positions or perceptions of division and unity more generally do not. Vicarious contact may persuade participants of the value of dialogue across partisan lines without erasing perceived, or real, disagreements between partisan groups.
This combination of findings should help assuage concerns that the documentary may have created a false sense of cross-partisan unity or fostered overly optimistic perceptions of cross-partisan agreement on divisive issues. Viewers appear to have become more convinced of the value of cross-partisan dialogue, more optimistic about the prospects for civility and democratic stability in the future, less prone to stereotyping out-partisans, and more inclined toward cross-partisan empathy and open-mindedness, without also becoming pollyannish about the cleavages that continue to divide the two parties. These mixed results should also mitigate concerns about social desirability bias, since we find beneficial effects on some outcomes that might be susceptible to social desirability concerns (e.g., optimism about the survival of democratic institutions) but null effects on others (e.g., belief in the possibility of non-violent change). If social desirability bias explained our results, we would expect to observe consistently beneficial effects across these outcomes. But we do not.
4.4. Extensions: Treatment effect heterogeneity and persistence over time
Does the effect of vicarious contact vary by participants’ partisan and ideological leaning? Our results suggest that it might, though our heterogeneous treatment effect analyses have lower statistical power than our ITT analyses, so we interpret them more cautiously. For compactness, we focus on affective polarization, our primary attitudinal outcome, in most of our analyses of treatment effect heterogeneity. In Figure 5 we show that the film’s beneficial effect on affective polarization is driven disproportionately by self-identified Democrats (
$\beta\!\! =\!\! -0.191$, p = 0.047, N = 584). Among Republicans, the effect is substantively small and statistically indistinguishable from zero (
$\beta\! =\! -0.070$, p = 0.461 N = 584), though, importantly, the difference between these two conditional average treatment effects (CATEs), captured by the coefficient on the interaction term, is not statistically significant (
$\beta\! =\! 0.121$, p = 0.369, N = 584). We show in SI I.3.5 that our results are substantively similar when we include covariates to partial out observable differences between Democrats and Republicans. (These latter analyses with covariate adjustment were not prespecified.)
Additional analyses examining treatment effect heterogeneity by ideology (i.e., liberal or conservative) and relative confidence in liberal vs. conservative media outlets are consistent with the findings above. The negative ITT on affective polarization appears to be concentrated among more liberal participants, and among participants who express more trust in liberal media outlets. However, the coefficients on the interaction terms are not consistently statistically different from zero, and we are careful not to over-interpret the CATEs to imply that the documentary is only effective among Democrats and liberals. (We present these figures in SI I). Finally, SI I.3.2 presents CATEs on optimism and anti-democratic attitudes by party ID. While we find no evidence of statistically significant differences between sub-groups, the results do offer directions for future research. For instance, it appears that the film mitigated anti-democratic attitudes among self-identified Republicans (though we interpret these results cautiously due to low statistical power). Examining whether different depolarization strategies are needed for Democrats and Republicans strikes us as a promising avenue for future research.
Vicarious contact reduced affective polarization primarily among Democrats.
We also test whether the effect of vicarious contact on affective polarization decays over time using wave 3 of the survey, administered approximately 50 days after treatment. As we discuss in SI G, we observe statistically significant differential attrition between the Braver Angels and placebo nature documentary groups in wave 3, raising the risk of bias in our treatment effect estimates. Fortunately, we have access to a rich array of pre-treatment covariates that we can use to diagnose and attempt to correct for differential attrition using IPWs. Figure 6 suggests that the effect observed in the short term decays to a statistical null over time. This is true for both our pre-registered affective polarization index (
$\beta = -0.115$, p = 0.110, N = 506) and our alternative index focused on outparty animus only (
$\beta = -0.119$, p = 0.143, N = 506). These wave 3 ITT estimates for the two indices are, respectively, 82% and 73% as large as their corresponding wave 2 ITT estimates. While this evidence of decay is consistent with our priors, we interpret it somewhat cautiously due to the aforementioned differential attrition.
Finally, SI I.3.4 presents wave 3 results for the optimism outcomes. While the treatment effect appears to decay for most of these outcomes, the ITT on belief in the efficacy of dialogue remains substantively large and statistically significant, even 50 days after the intervention (
$\beta \!\!\!\!=\!\!\!\! 0.364$, p = 0.001, N = 465). This estimate is approximately 80% as large as the wave 2 ITT estimate, indicating a rate of decay similar to the rate for our affective polarization outcomes.
The effect of vicarious contact on affective polarization decayed over time.
5. Discussion
We conducted a pre-registered, placebo-controlled trial to test whether a mass media vicarious intergroup contact intervention—delivered in the form of a 50-minute documentary film—can depolarize a nationally representative sample of Americans. We find that vicarious contact reduced affective polarization and increased interest but not investment in depolarization activities approximately 5 days after exposure. While vicarious contact did not mitigate anti-democratic attitudes, it did increase participants’ optimism about the survival of democratic institutions and the ability of Americans to overcome polarization, and strengthened participants’ belief in the efficacy of dialogue as a tool for change. Our results suggest that the documentary may have reduced affective polarization by correcting stereotypes of out-partisans; we find more suggestive evidence that it may have also increased open-mindedness and built empathy toward out-partisans. We find no evidence that the documentary reduced affective polarization by instilling the perception that Americans are less divided along partisan lines than participants previously believed, or that they are less divided on specific policy issues. We also find that the effects of the documentary appear to be concentrated among participants who self-identified as Democrats, as ideologically liberal, and as more trusting of liberal media outlets.
This study advances the intergroup contact literature by showing that vicarious contact through a 50-minute media intervention can reduce partisan polarization. Unlike traditional forms of contact that require direct interactions, our research demonstrates that merely observing interactions through mass media can similarly promote positive outcomes, such as increased optimism and empathy. These findings extend the traditional scope of intergroup contact theory by offering long-sought-after evidence of vicarious contact’s effects on outgroup prejudice and stereotyping (Paluck et al., Reference Paluck, Green and Green2019, Littman et al., Reference Littman, Scacco, Weiss, Halperin, Hameiri and Littman2023). To date, only four studies (summarized in SI A) have rigorously examined this possibility, and of these four studies, only two study contact between citizens, while evaluates a (very short) media intervention that does not depict American partisans. Our study thus advances existing research on intergroup contact that largely elides the question of whether vicarious rather than direct contact can reduce prejudice and improve intergroup relations.
Demonstrating that mass media can serve as an accessible, scalable tool for fostering positive intergroup relations has important practical implications as well, suggesting that practitioners should continue scaling up and innovating on this format in the future. Moreover, the fact that the effects of the documentary were most pronounced among Democrats suggests that the impact of mass media interventions may be shaped by ideology, emphasizing the importance of understanding the target audience when designing vicarious contact interventions. While we prioritized aggregate effects across the U.S. population by using a nationally representative sample, future work should more directly interrogate treatment effect heterogeneity by oversampling particular groups.
How cost-effective is vicarious contact relative to more traditional direct contact interventions? One way to estimate cost-effectiveness is to benchmark our ITT estimates against a previous study evaluating the same in-person direct contact workshop featured in the documentary, run by the same organization (Baron et al., Reference Baron, Blair, Choi, Gamboa, Gottlieb, Robinson, Rosenzweig, Turnbull and West2025). Recently, Littman et al. Reference Littman, Scacco, Weiss, Halperin, Hameiri and Littman2023 have called for precisely this type of benchmarking in intergroup contact research. The comparison is illustrative but should be interpreted with caution: while our study and Baron et al. Reference Baron, Blair, Choi, Gamboa, Gottlieb, Robinson, Rosenzweig, Turnbull and West2025 use very similar measures to operationalize attitudinal affective polarization, the samples, experimental comparisons, and timing of outcome measurement differ. Baron et al. Reference Baron, Blair, Choi, Gamboa, Gottlieb, Robinson, Rosenzweig, Turnbull and West2025 sampled college students, compared participants who were assigned to the workshop to an empty control, and measured outcomes 14 days after treatment; our study uses a nationally representative sample of Americans, compares participants who were assigned to the documentary to a placebo, and measures outcomes an average of 5 days following treatment.Footnote 28 Moreover, we view the workshops and documentary as complements rather than substitutes, as we discuss below. With this caveat in mind, using our pre-specified version of the affective polarization index, the workshop and documentary reduced affective polarization by 0.184 and 0.140 standard deviations, respectively. This implies that the effect of the 50-minute documentary is roughly three-quarters (approximately 76%) of the magnitude of the effect of an all-day in-person workshop.Footnote 29
Perhaps unsurprisingly, the effects of the documentary thus appear to be smaller and potentially less durable than the effects of the in-person workshop. In this sense, the documentary is perhaps best used as a booster to prior in-person contact or as a gateway to further engagement. That said, the potential for scale offered by the documentary remains compelling, especially relative to more standard interventions involving in-person or online contact, which have higher costs in terms of staff and participant time and reach a fraction of the people. Each in-person Braver Angels workshop of the type featured in the documentary brings together a maximum of 32 peopleFootnote 30 at a time for a full day, while 351 Americans watched the 50-minute documentary as part of our study alone. This is roughly equivalent to the reach of 10 workshops achieved in 12% of the time and at a fraction of the cost. By way of illustration, 1,000 workshops would reach 32,000 people and would cost roughly $1,130,000 (or $1,130 per workshop, according to Braver Angels). Even if one were to pay viewers $10 (well above the federal minimum wage of $7.25/hr) to watch the film, one would still reach 113,000 people with the same amount of money.
Of course, other options for distribution are available and are arguably more realistic. According to Google Ads, a highly conservative digital advertising rule-of-thumb for a general U.S. population suggests that the average cost-per-view of a YouTube video is $1. This implies that, for every dollar spent on advertising, we can reasonably expect a yield of one view of the documentary. Even if we assume that half the people who begin watching the documentary will not finish it, with a $1,000,000 budget, a depolarization documentary could still reach 500,000 Americans.Footnote 31
An even more cost-effective approach would be mass dissemination over popular television networks or streaming platforms; indeed, this is what our partners had in mind when they asked us to evaluate the documentary’s impact. For example, films that stream on Netflix are usually purchased by the company from the creators but are not normally advertised, thus reducing costs.
While it is impossible to know how the Braver Angels documentary would be received, by way of illustration, Netflix provides a publicly available breakdown of the number of views of films on its platform.Footnote 32 In the 6-month period between July and December 2023, Netflix hosted 9,395 films. The lowest number of views for any film in that 6-month period, as well as the modal number, was 100,000. Approximately 40% of films had 100,000 views. A few documentaries analogous in length and quality of production (though not subject matter) to the one we study here had between 200,000 and 300,000 views.Footnote 33
Mass distribution of films highlighting vicarious contact across partisan lines also has a very recent precedent. For example, in 2021, a documentary featuring vicarious contact, called Reunited States, was nationally distributed on multiple major platforms.Footnote 34 Other content, such as a long-running CNN series called The United Shades of America, has also regularly featured footage of positive interactions across the partisan divide. Distribution of Reunited States was conducted with support from our partner, Braver Angels, and community screenings were funded by private foundations. A similar approach—leveraging horizontal networks such as Braver Angels’ local chapters (which exist in all 50 states)—could be used to disseminate the documentary we study here, which has already been viewed 27,000 times on the Braver Angels YouTube channel without any mass distribution or advertising at all.
At the same time, any type of scaling is not without barriers. While some political and media elites may express interest in depolarization or bipartisanship, many elites are motivated to polarize the electorate for political or financial gain. They, therefore, may be uninterested in (or even opposed to) the distribution of depolarizing content, and in the goal of depolarization more broadly. Depolarizing content may also provoke some amount of backlash as it reaches larger swaths of the highly polarized American public. Of course, these headwinds are likely to hinder any depolarization effort conducted at scale.
Another barrier to scale is simply the length of the documentary and its ability to hold the attention of a viewer with minimal interest in depolarization. It was for this reason that we developed 5-minute versions of the documentary. While we cannot make robust comparisons given differential attrition between the films of different lengths, we nevertheless find no evidence that the long documentary was more effective than the short ones in reducing affective polarization or generating commitment to depolarization (as measured by donations). The results are presented in Figure SI.14 in SI I.3.7. We do find, however, that the long documentary was more effective at stimulating interest in depolarization (as measured by newsletter signups); at strengthening belief in the efficacy of dialogue; and (more weakly) at increasing optimism about the prospects for restoring civility across partisan lines. But again, we interpret these results cautiously given the extent of differential attrition between these experimental conditions.Footnote 35 Comparing shorter and longer mass media interventions strikes us as a promising avenue for future research.
6. Conclusion
In sum, our findings suggest that mass media interventions featuring vicarious intergroup contact can serve as a relatively cheap and scalable tool to promote depolarization in the United States, at least in the short term. These interventions can reduce affective polarization, foster optimism, and highlight the potential for dialogue to bridge partisan divides. Our study broadens the scope of traditional intergroup contact theory by demonstrating that vicarious contact, particularly through mass media, can reduce polarization without the need for direct, face-to-face interactions. More broadly, our findings suggest that mass media interventions may be a powerful mechanism for addressing large-scale social issues, including the urgent problem of partisan polarization.
Our findings also raise important questions for future work that can advance our understanding of contact theory in light of our rapidly evolving media ecosystem. First, as media and information infrastructures shift toward more video-based modalities and platforms (i.e., series vs. videos, TikTok vs. YouTube, etc.), a better understanding of how different types of video content can be harnessed to reduce intergroup prejudice remains an important task and should be explored further. Second, as a greater proportion of social interactions migrate to the virtual world, it becomes ever more important to understand the potentially disparate effects of different types of contact. For instance, as in-person interactions become rare, it is possible that they could also become more or less effective relative to virtual contact that might take place more frequently. Exploring possible trade-offs between the frequency and impact of different forms of interaction is a fruitful and important future direction.
Third, our results point to the possibility that vicarious contact may serve as a gateway to other forms of exposure to, and dialogue with, members of the outparty. While partisan segregation in-person and online creates barriers to exposure, a more fragmented media landscape also offers multiple plausible gateways for indirect contact. Individuals could engage in multiple types of contact over time, and we do not know enough about how different types of contact might enhance or mitigate each other. We hope future researchers will continue to explore the promise and limits of vicarious contact to inspire more meaningful engagement and dialogue across party lines.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/psrm.2025.10039. To obtain replication material for this article, https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/HVSWTH.
Acknowledgement
We are grateful to Peter Yarrow for creating the Braver Angels documentary, as well as for his insights and partnership in carrying out this research. Peter passed away on January 7th, 2025, before the publication of this article. Peter was a visionary individual who pushed us to make academic rigor useful to the public good and humbled us with his wisdom. This article would not have been possible without him and we dedicate it to his memory.
We thank Braver Angels for allowing their work to be filmed and rigorously evaluated. Funding for this research was provided by Operation Respect. We thank Unify America for their collaboration, Roman Gallardo for research assistance, and Niyoshi Parekh for video editing.
Funding declaration
Funding for this research was provided by Operation Respect and was exclusively used to cover the cost of data collection.
Competing interests
The authors declare no competing interests.
Ethics
The study was approved by the Brown University IRB (protocol #2022003227). All participants provided informed consent and were compensated for their participation in the study.
Open access
