Misperceptions about out-partisans’ democratic values may erode democracy

Two studies (one preregistered) of Americans (N = 2200) drawn from a nationally representative panel show that both Democrats and Republicans personally value core democratic characteristics, such as free and fair elections, but severely underestimate opposing party members’ support for those same characteristics. Democrats estimate that the average Democrat values democratic characteristics 56% (in Study 1) and 77% (in Study 2) more than the average Republican. In a mirror image, Republicans estimate that the average Republican values democratic characteristics 82% (in Study 1) and 88% (in Study 2) more than the average Democrat. In turn, the tendency to believe that political ingroup members value democratic characteristics more than political outgroup members is associated with support for anti-democratic practices, especially among Republicans. Results suggest biased and inaccurate intergroup perceptions may contribute to democratic erosion in the United States.

www.nature.com/scientificreports/ understood as social norms, it becomes clear that democratic norms are not the aggregate individually held beliefs about the importance of democratic characteristics. Rather, they reflect mutual expectations about the degree to which others who participate in the same democratic arena value and abide by a shared set of rules that govern democratic practice 11 . Democratic norm strength requires partisans to not only adhere to the same set of rules, but also to expect conformity from those belonging to the opposite party. If voters come to believe that pro-democratic views and behaviors are not normative among opposing party members, and that opposing party members fail to stop democratic abuse by their elected representatives, they may justify their own party engaging in undemocratic behaviors. Why? Because such actions may appear to be the only way to preemptively protect their interests against imminent violation. Indeed, one of the essential functions of democratic norms is to ensure the ability for those who are currently out of power to fairly contest to regain it. To the degree that people come to believe that their opponents have no regard for such norms, the prospect of political loss becomes untenable and antidemocratic behaviors may come to be perceived as justified, even necessary.
Despite strong evidence that both Democrats and Republicans do value democratic principles, psychological biases-likely fueled by elite behavior-may lead members of both parties to believe that pro-democratic views are non-normative among political opponents. This theorizing is grounded in research on political sectarianism 12,13 , which documents how American political parties have morphed into mega identities that represent more than mere ideological coalitions 14 . As a result, political parties, like other forms of social group membership, foster an "us" vs. "them" way of thinking that gives rise to numerous forms of intergroup bias 15 . In this work, we focus on one form of bias-the tendency to misperceive the motives and beliefs of out-partisans.
A growing body of research highlights how this "us" vs. "them" psychology can lead partisans to attribute negative motives and beliefs to opposing party members and to exaggerate the size of ideological differences. For example, Democrats and Republicans in the United States wrongly assume that their ingroup members are motivated by love whereas outgroup members are motivated by hate. In turn, this motive asymmetry drives intergroup conflict 16 . Similarly, Democrats and Republicans severely overestimate the extent to which members of their opposing party dislike and even dehumanize members of their own party 17,18 . In turn, this misestimation drives reciprocal dislike and dehumaniztion, exacerbates conflict, and even leads people to endorse anti-democratic behaviors 17,18 . These biases extend to issue-based polarization. Studies repeatedly show that Democrats and Republicans severely overestimate the size of their ideological differences on important political issues 18,19 , which can fuel political obstructionism 19 . Notably, these biases generlaize to other partisan conflicts 20 and can fuel partisan violence 21 . This growing body of work suggests that a generalizable pattern whereby exagerated negative perceptions about what outgroup members believe and how they behave are associated with the tendency to justify one's own (or one's group's) negative attitudes and behaviors. Such a tendency may also help to explain democratic norm errosion.
Just as psychological biases lead partisans to attribute negative motives and beliefs to their opponents, we reasoned that similar biases may lead Democrats and Republicans to (incorrectly) believe that members of their own party value democratic principles more than do members of their opposing party. Such biases might be quite pronounced given recent high-profile accusations by political elites in the United States about their opponents' anti-democratic behavior. For example, widespread misinformation about voter fraud and the 2020 election being stolen may lead Republicans to believe Democrats have little regard for democratic principles. And to the contrary, Republican attempts to overturn legitimate election results and suppress voter turnout may lead Democrats to doubt Republicans' commitment to democracy. Because we hypothesized such biases stem at least in part from social identity processes-through which individuals are motivated to attribute more positive traits to ingroup members-we also reasoned that these biases would be more pronounced among Democrats and Republicans who more strongly identified with their party. Although the focus of our research is on Democratic and Republican citizens' perceptions of each other, we also posited that these same psychological biases apply to perceptions of political elites, whose anti-democratic behaviors have an outsized influence on democratic norm erosion and shape public opinion more broadly 22 .
We suggest that the tendency to believe that own-party members value democratic principles more than opposing party members may threaten democratic norm strength and practice. This misperception may upend the delicate balance of mutual deterrence, upholding democratic norms while generating incentives for preemptive anti-democratic actions. This could lead partisans to believe that the only way to compete for power is to engage in anti-democratic behaviors that they (wrongly) think outgroup members support. Put another way, even if Democrats and Republicans value democratic principles themselves, and believe that the violation of democratic principles is harmful for democracy, partisans who hold biased perceptions that such views are normative only among ingroup members may be more likely to support anti-democratic behaviors regardless. In so doing, similar to how pluralistic ignorance leads people to conform to views they disagree with 23 , individuals who hold negatively biased perceptions may behave in a manner that only reinforces out-partisans' (inaccurate) perception that they do not value democratic principles. As a result, negatively biased perceptions might contribute to (and be fueled by) a downward spiral of democratic practice.
In the present research, we test our ideas in two studies-one preregistered-with self-identified Democrats and Republicans randomly sampled from a nationally representative panel maintained by YouGov. Here, we outline our core research questions, which themselves articulate our broader theoretical model. We first sought to replicate prior research documenting robust support for democratic principles among both Democrats and Republicans 5 . Next, we sought to test our novel prediction that, despite Democrats' and Republicans' support for democratic principles, members of both parties would wrongly believe that average members of their own party value democratic principles more than average members of their opposing party. In Study 1, we also explore whether these same biased perceptions apply to Democrats' and Republicans' perceptions of how much elites (i.e., members of Congress) from each party value democratic principles. Informed by social identity theory 15  www.nature.com/scientificreports/ we also test whether biased perceptions are exaggerated among partisans who identify more strongly with their political party. Finally, we test whether individuals with more biased perceptions (i.e., those who more strongly believe that ingroup members value democratic principles more than outgroup members) are also more willing to endorse anti-democratic behaviors that advantage their own party. Across our research questions, we also explore differences by political party. Preregistration for Study 2, as well as well all data and code for both studies, are available at https:// osf. io/ h9dt8/? view_ only= 81b8d 2ff93 1c4e6 0b8ea 027e9 f848d af.

Results
Democrats and republicans highly valued democratic characteristics. Consistent  ], η p 2 = 0.375). In relative terms, Republicans thought that own party members valued democratic characterizes 82% (in Study 1) and 88% (in Study 2) more than out-party members. These results are displayed in Fig. 2 In addition to our core questions, which pertain to perceptions about everyday citizens' values, we also asked Study 1 participants (in an exploratory follow-up survey conducted two months later) to predict the degree to which the average Democratic and Republican congressperson valued the same principles. Results  Americans (especially republicans) with more biased intergroup perceptions were more approving of anti-democratic practices. We theorized that Americans with more biased intergroup www.nature.com/scientificreports/ perceptions would also be more approving of anti-democratic practices. As expected, in both studies, individuals who more strongly believed that the average ingroup member valued democratic characteristics more than the average outgroup member were also more willing to subvert democratic principles, in practice, to help their party (  Fig. 4. Notably, a mini meta-analysis shows that the cross-study effect for Democrats was also non-significant, z = 1.46). As can also be seen in Fig. 4,

Discussion
In two studies conducted with large samples of American partisans drawn (N = 2200) from a nationally representative panel, we found that both Democrats and Republicans highly valued democratic characteristics. While this finding, which is consistent with prior research 5 , offers room for optimism about the health of American democracy, it also raises an important question: given high levels of individual support for democratic principles, why do so many partisans engage in anti-democratic behavior and tolerate it by their elected representatives 24,25 ? Our results provide one potential explanation for this paradox. By demonstrating that American partisans severely underestimate out-partisan support for bedrock democratic principles, we suggest that these intergroup misperceptions may independently jeopardize Americans' , and especially Republicans' , commitment to democracy. A core contribution of this research is our theorizing about the nature of democratic norms, with implications for thinking about democratic norm erosion. Rather than conceptualize democratic norm strength as an aggregate of individually held beliefs about the importance of democratic principles, we join others who have recently begun conceptualizing democratic norms as the aggregate of individuals' perceptions of how much others, and particularly out-partisans, value democratic principles 11 . If norm strength was measured using individually held opinions, our data would present a rather optimistic take. However, by focusing on perceptions-themselves a constitutive quality of social norms-our data demonstrate that democratic norm strength-and potentially practice-may be severely undermined by a partisan psychology of "us" vs. "them" thinking.

Possible causal mechanisms. A natural question is why Democrats and Republicans underestimate each
other's support for Democratic principles. One explanation, grounded in social identity theory 15 , is that the same psychological processes that lead partisans to overestimate affective and ideological polarization influence intergroup perceptions more broadly [18][19][20] .
A second explanation may be that partisans infer beliefs and values from actions, even if these actions are not representative of all party members. For example, Democrats who witnessed some Republicans storm the Capitol in the aftermath of the 2020 election have good reason to suspect that anti-democratic values are normative among some Republican voters, who continue to vote for elected representatives who scholars warn consistently undermine democratic principles 2,3 . Such perceptions are likely only exacerbated by a hyperpolarized media landscape 26 , in which members of both parties are constantly exposed to narratives of anti-democratic behavior committed by out-partisans. In such a hyperpolarized landscape, even asymmetric anti-democratic attitudes could fuel symmetric misperceptions and kickstart a downward spiral.
A third related explanation is that anti-democratic behavior committed by partisan elites may trickle down to influence perceptions of partisan citizens 22 , whose votes ultimately give power to elites themselves. Our finding that partisans attribute more anti-democratic values to elites from their opposing (and even own) party provides some support for this explanation. However, given the asymmetric nature of some of our findingswhich demonstrate a stronger relation between biased perceptions and support for anti-democratic behavior by Democrats and Republicans-this account does not, on its own, explain the symmetrical nature of observed biased misperceptions.
A fourth explanation is that political campaigns and political media benefit from exaggerating the threat posed by out-partisans. Political campaigns (and aligned media organizations) must mobilize their supporters to donate funds, get out the vote, attend rallies, and call legislators-all costly activities. Increasingly, they must do so on a national level. More even-handed political media is incentivized to broadcast conflictual, threat-laden content to attract viewers. This creates a kind of escalation. Since voters may become desensitized to threats, the threshold for activation may climb higher and higher, requiring a greater "dose" of outsider threat for mobilization and attention. Future research should aim to test these four possible causal mechanisms.
The present research also contributes to an emerging interdisciplinary literature on the role of inaccurate intergroup perceptions in democratic decline. Notably, prior research demonstrates that correcting inaccurate intergroup perceptions about affective polarization can reduce toxic polarization and even partisan violence 21 . However, such corrections-which to date have focused on perceptions about affective and ideological polarization-have yet to counteract anti-democratic attitudes 27 . It is plausible that these interventions may fall short because they have not directly targeted perceptions about democratic values. Our results suggest that democratic erosion may be caused, at least in part, by a misguided fear that out-partisans do not share the same democratic values and, as a result, will not abide by the rules of the game. Thus, one potentially promising intervention strategy would be to directly assure Americans of their opposing party's robust support for shared democratic principles. We are aware of two not yet published attempts to do so, both of which followed a preprint of an earlier version of this work. One experiment yielded significant but only small effects 28 , suggesting that, given a hyperpartisan media, anti-democratic behavior by partisan elites, and recent examples of citizens storming the capitol at the behest of their party leader, it may be difficult for an informational intervention to effectively counteract the many forces fueling partisan misperceptions. However, a separate set of studies reported in a recent preprint not only conceptually replicates our general finding that Democrats and Republicans who most fear that out-partisans will subvert democracy are themselves more willing to subvert democratic principles, but also extends this finding by experimentally showing that an intervention correcting inaccurate misperceptions about out-partisans' commitment to democracy promotes adherence to democratic norms 29 . Future research is needed to identify what makes some informational interventions more effective than others.
Alternative explanations for our findings. While our research, and emerging complementary work 29 , suggests that biased intergroup perceptions may help to explain the seeming paradox between Americans' support for democratic principles and the simultaneous erosion of democratic practice, it is important to consider www.nature.com/scientificreports/ alternative explanations and the potential for bidirectional processes. We review four non-exhaustive possibilities, each of which deserves independent inquiry. One alternative explanation could be that negative partisanship 30 and affective polarization 12 are so strong that they override Americans' otherwise pro-democratic attitudes. Thus, it is possible that the cross-sectional relation observed in the present research-between misperceptions and anti-democratic attitudes-emerges from a third variable. While this may explain part of our observations, the asymmetrical relation between misperceptions and anti-democratic views, with stronger relations among Republicans than Democratic, suggests that partisan animosity on its own does not offer a full mechanistic explanation. Relatedly, one might ask what causes the asymmetry in the first place? One potential explanation could be that our studies were conducted shortly after the Democrats took control of the presidency and congress, making the stakes much larger for Republicans, who were ousted from political power. This may help to explain why the relation between biased perceptions and support for anti-democratic behaviors was stronger for Republicans in both studies. Nonetheless, given the cross-sectional nature of our data, we are not able to fully rule out third-variable effects or to confidently identify the source of observed asymmetry. Future experimental and/or longitudinal research is needed to address these questions.
A second alternative explanation for this incongruence may be that partisans have different views about what actions constitute a threat to democratic principles. For example, voter suppression laws are often framed as protecting election integrity. This suggests that proponents of anti-democratic policies may recognize democratic principles' widespread popularity and market anti-democratic proposals in pro-democratic language. This may help to explain why many Republican voters and elites defended efforts to overturn the 2020 presidential election as a way to protect democratic processes. Relatedly, racial animus underlying support for Donald Trump 31 , as well as racialized conceptions of who is American 32 , may foster alternative views about how to defend democracy. As a result, anti-democratic behaviors that disproportionately disenfranchise racial minorities, such as voter suppression 33 , can be perversely seen as either protecting American democracy or hurting it, depending on the accuracy of one's preconceptions. Future research should explore these possibilities.
A third alternative explanation is that pro-democracy views may not be as deeply held as many Americans self-report. For example, it is possible that Democrats or Republicans overestimate the extent to which they and their own party value democratic norms (rather than underestimate how much out-partisans value democratic norms). This could result from social desirability bias 34 or stem from a gap between the values individuals hold in abstract and how they put these values into practice. For example, partisans may view democratic principles that they support as fungible and worth sacrificing for competing values, such as power. And for many segments of the population (e.g., White Christians), for whom increasing diversity threatens political power, there may be a newfound tension between pro-democratic views and pro-group views [35][36][37][38] . This broader context may also help to explain why a willingness to subvert ostensibly held democratic values was higher among Republicans. This question remains ripe for future research.
Fourth, it is also possible that the relation between democratic norm adherence and (mis)perceptions about out-partisans commitment to democracy is bidirectional. That is, just as partisans may be more willing to subvert democratic principles that they hold important if they believe that out-partisans won't play by the rules, partisans may also seek to downplay out-partisans' commitment to democracy to justify their own (or their group's) antidemocratic behavior. Thus, misperceptions may help individuals maintain a positive view of their group and/or serve as a moral license to continue undemocratic behavior.
Democratic strength requires that citizens adhere to-and remove from office elites who violate-bedrock principles of democracy. The maintenance of strong democratic norms ensures that parties have a fair chance to compete for power and that minority rights are protected. While it is important that citizens value democratic principles, doing so may not be enough to ensure that they also uphold such principles in practice. Democratic norm strength requires that partisans from all parties not only personally value democratic principles, but also believe that members of their own and opposing parties do the same. This mutual perception, which itself can be conceptualized as a metric of democratic norm strength, is easily jeopardized in a sectarian political climate marked by a hyper-partisan psychology. Democratic health requires attention to this psychology.

Method
Participants. Respondents for both studies were sampled from YouGov, which maintains a large opt-in panel of Americans, from which we drew nationally representative samples using YouGov's Active Sampling methodology. Respondents were compensated directly by YouGov for participating in this research. See A second question asked whether respondents identified as strong or weak members of their respective political party. This question also allowed respondents to indicate whether they leaned toward one party or the other, although, after removing independents, all partisans identified as either strong or not very strong partisans.
Actual and perceived importance of democratic characteristics. Democratic characteristics were selected from Bright Line Watch 5 to represent four key principles. (1) "Elections are conducted, ballots counted, and winners determined without pervasive fraud or manipulation. " (2) "All adult citizens enjoy the same legal and political rights. " (3) "Government agencies are not used to monitor, attack, or punish political opponents. " And (4) "Law enforcement investigations of public officials or their associates are free from political influence or interference. " Participants rated how important they found these characteristics on a sliding scale (0 = not at all important to 100 = extremely important Analytic strategy. Results were analyzed using weighted regressions. Weights supplied by YouGov were applied to provide population-level estimates. Respondents were weighted to their respective parties using propensity scores including age, gender, race/ethnicity, years of education, and region. Weights were post-stratified based on these criteria as well as 2016 and 2020 presidential vote choice. For main analyses, we contrast coded political party (− 0.5 = Republican, 0.5 = Democrat). Marginal means and simple slopes were calculated using dummy codes. For analyses involving party strength moderation, strong partisans were coded 0.5 and weak partisans were coded − 0.5. Simple slopes were again analyzed using dummy codes.
To investigate associations between biased meta-perceptions and support for anti-democratic practices, we subtracted outgroup norm predictions from ingroup norm predictions. To ease parameter interpretation, we divided this score by 100. Removing observations >|3| studentized deleted residuals did not meaningfully influence results.
Given the asymmetrical relation between biased perceptions and one's own willingness to violate democratic norms, which was stronger for Republicans than Democrats, we calculated a cross-study effect by conducting a mini meta-analysis. Following recommended practices, we converted observed t-values to correlations, which were weighted and pooled to calculate a cross-study z-score 39 .

Data availability
The datasets and code generated and analyzed for both studies are available in the Open Science Foundation repository at https:// osf. io/ h9dt8/? view_ only= 81b8d 2ff93 1c4e6 0b8ea 027e9 f848d af. www.nature.com/scientificreports/ and members of the Beyond Conflict team for helpful comments. We thank Roman Gallardo for his assistance with study programming.