Friday, April 29, 2022

Tweets from American politicians have become more incivil over the last decade; uncivil tweets tended to receive more approval and attention, publicly indexed by large quantities of “likes” and “retweets”

Incivility Is Rising Among American Politicians on Twitter. Jeremy A. Frimer et al. Social Psychological and Personality Science, April 28, 2022.

Abstract: We provide the first systematic investigation of trends in the incivility of American politicians on Twitter, a dominant platform for political communication in the United States. Applying a validated artificial intelligence classifier to all 1.3 million tweets made by members of Congress since 2009, we observe a 23% increase in incivility over a decade on Twitter. Further analyses suggest that the rise was partly driven by reinforcement learning in which politicians engaged in greater incivility following positive feedback. Uncivil tweets tended to receive more approval and attention, publicly indexed by large quantities of “likes” and “retweets” on the platform. Mediational and longitudinal analyses show that the greater this feedback for uncivil tweets, the more uncivil tweets were thereafter. We conclude by discussing how the structure of social media platforms might facilitate this incivility-reinforcing dynamic between politicians and their followers.

Keywords: incivility, political polarization, Twitter, social media, affective polarization

Some said that "conservatives think liberals are stupid, and liberals think conservatives are evil," but not so: Both sides see opponents as more stupid than evil

People See Political Opponents as More Stupid Than Evil. Rachel Hartman, Neil Hester, Kurt Gray. Personality and Social Psychology Bulletin, April 28, 2022.

Abstract: Affective polarization is a rising threat to political discourse and democracy. Public figures have expressed that “conservatives think liberals are stupid, and liberals think conservatives are evil.” However, four studies (N = 1,660)—including a representative sample—reveal evidence that both sides view political opponents as more unintelligent than immoral. Perceiving the other side as “more stupid than evil” occurs both in general judgments (Studies 1, 3, and 4) and regarding specific issues (Study 2). Study 4 also examines “meta-perceptions” of how Democrats and Republicans disparage one another, revealing that people correctly perceive that both Democrats and Republicans see each other as more unintelligent than immoral, although they exaggerate the extent of this negativity. These studies clarify the way everyday partisans view each other, an important step in designing effective interventions to reduce political animosity.

Keywords: political polarization, affective polarization, meta-perceptions, social perceptions

Our aim was twofold: to assess whether cross-party perceptions of unintelligence and immorality are distinct, and to compare liberals’ and conservatives’ perceptions of each other. Although some work suggests that dimensional complexity should be higher for ingroup members, who are often perceived as more heterogeneous (Mullen & Hu, 1989), we find greater dimensional complexity for outgroup members. Across four studies, we found that for outgroup ratings, unintelligence and immorality fell into two separate categories. On the contrary, for ingroup ratings, unintelligence and immorality were best conceived of as one factor. This might be explained by the fact that people’s experiences of negative attitudes and emotions are often more complex than their experiences of positive attitudes and emotions (Koenig-Lewis & Palmer, 2014).

Contrary to Krauthammer’s popular quote, we found that both liberals and conservatives view each other as more unintelligent than immoral. Furthermore, we found that participants accurately thought both Democrats and Republicans view each other as more unintelligent than immoral. However, participants exaggerated the magnitude of disparagement, thinking political groups have more negative views of their opponents than they actually do.

These findings replicated across four studies, when the questions were asked abstractly (Studies 1, 3, and 4) or regarding specific voting behaviors (Study 2). While preparing this manuscript, we took advantage of the Coronavirus pandemic to test whether political perceptions changed during a period of political tension. The pandemic, which began as an apolitical health threat (Holzwarth, 2020), transformed into a highly partisan issue in the United States (Newport, 2020). People on the right were eager to reopen the economy, whereas people on the left worried about the health risk (Roubein, 2020). Republicans were accused of being callous about human lives, whereas Democrats were accused of not understanding the gravity of the virus’s economic effects (Hulse, 2020). On May 14, 2020, we asked 329 MTurk workers via CloudResearch (Litman et al., 2017) to think of the way Democrats and Republicans have been reacting to the pandemic and indicate the extent to which the six unintelligence and six immoral items apply to each group. Replicating the previous four studies, both Democrats and Republicans viewed each other as more unintelligent than immoral. This timely replication provides further support for our main finding (full analyses are reported in the Supplemental Materials).


Political polarization is at an all-time high. To bridge political divides, researchers and organizations need a solid understanding of how partisans perceive each other. The present studies show that, although liberals and conservatives often seem to disagree about moral values, the two groups still disparage each other’s intelligence more than each other’s morality. This finding replicates the results of the Axios poll (Hart, 2018), but runs counter to the Pew Research Foundation poll (Pew Research Center, 2019a). This may be a result of the question framing (i.e., “compared to the average American”). Future research should investigate the discrepant findings.

Two (competing) theories in the social psychology literature highlight the importance of morality in politics: According to Moral Foundations Theory (Graham et al., 2009Haidt, 2012), liberals and conservatives disagree on many issues because they differ in their moral foundations. According to the Theory of Dyadic Morality (Schein & Gray, 2018), liberals and conservatives share the same moral mind, and therefore, if they come to understand this fact, they should find common ground. The findings from our studies suggest that morality is just part of the story, and perhaps not the most important part. If unintelligence, rather than immorality, drives perceptions of political groups, future research and interventions should aim to facilitate recognition of the other group’s knowledge and intelligence, rather than focus primarily on their morality.

Our fourth study found that political groups tend to overestimate the degree to which they view each other as unintelligent and immoral. This finding replicates similar findings in the literature: Partisans overestimate the extremity of positions held by each group (Ahler, 2014Chambers et al., 2006Lees & Cikara, 2020Levendusky & Malhotra, 2016Van Boven et al., 2018Yudkin et al., 2019) and think each side dehumanizes the other more than they actually do (Moore-Berg et al., 2020). Ahler (2014) has demonstrated that alleviating misperceptions is often beneficial not only in correcting the meta-perceptions but also in mitigating the attitudes themselves. Future research should explore this method for reducing affective polarization.


We acknowledge several limitations to the present research. First, we restricted our data collection to American participants; thus, our findings may not necessarily generalize to partisan groups in other cultures. Second, the data we collected, to the extent they can be generalized to the American population, only reflect the participants’ perceptions of their outgroups at the time the data were collected. Notably, we collected our data prior to the storming of the Capitol on January 6, 2021. Significant events such as this one may have a large impact on political perceptions. However, the public discourse surrounding the event appeared to reflect our finding: On social and mass media, observers framed the right-wing protesters as “misled,” “brainwashed,” and “manipulated” (Hale, 2021Kristof, 2021Lewis, 2021).

In addition, as we only assessed perceptions of unintelligence and immorality, we do not exhaustively describe liberals’ and conservatives’ perceptions of each other. Partisans likely use many other negative adjectives to describe their opponents. Our study has high face validity, in that the focus on “stupid” and “evil” reflects cultural discourse about political groups, but it is far from a comprehensive overview of partisan disparagement. Furthermore, we acknowledge that we did not provide partisans with the opportunity to report positive views of their opponents.

Finally, there is the possibility that the general pattern of political outgroup unintelligence ratings being greater than immorality ratings is not a finding that is specific to political outgroups, but rather is a characteristic of any intergroup perceptions. Although it is certainly possible that one could observe the same pattern in other intergroup contexts, there are reasons to believe that these same patterns do not generalize across all ingroup–outgroup perceptions. For example, both men and women who endorse benevolent sexist beliefs (Glick & Fiske, 1996) are more likely to rate the gap between unintelligence and immorality to be larger for women than for men because here women are stereotyped as being both pure and in need of protection (i.e., incapable). Even in the case of other antagonistic groups, it is not necessarily the case that these same patterns occur. For example, atheists are uniquely seen by Christians as being highly immoral (but not necessarily unintelligent; Gervais, 2013Gervais et al., 2011), whereas Christians are seen by atheists as being less competent in science, in part because of the perceived conflict between science and religion (Rios et al., 2015Simpson & Rios, 2019).

Finally, one might argue that unintelligence perceptions were higher than immorality perceptions because of the items’ wordings. Perhaps we worded the unintelligence items more negatively, causing partisans to endorse them more. However, if this were the case, we would not have expected the same pattern of results (unintelligence > immorality) for the ingroup ratings. Another objection one might raise is that people are just averse to seeing others as evil. This may be true, but first, participants did endorse the immorality items to some degree, and second, we can conclude that despite the pervasiveness of political antipathy, partisans are still somewhat reluctant to view each other as immoral.