Near the end of 2017 Pew Research Center released a number of “striking findings”, led off by the claim that “partisan divides dwarf demographic differences on key political values” (see the figure below from the Pew Research Center; see the full report here).  Specifically, splits have intensified over:

  • How much the government should do to help the needy
  • Racial discrimination and its impact
  • Immigration
  • Corporate profits
  • Environmental laws and regulations.

Furthermore, a declining share of Americans hold a mix of liberal and conservative values:

New research by Suhay, Bello-Pardo, and Maurer (2018) suggests that incivility in online political discourse may be contributing to these increases in polarization.

Full Reference: Suhay, E., Bello-Pardo, E., & Maurer, B. (2018). The polarizing effects of online partisan criticism: Evidence from two experiments. The International Journal of Press/Politics, 23, 95-115.

Abstract (Emphasis added):

Affective and social-political polarization—a dislike of political opponents and a desire to avoid their company—are increasingly salient and pervasive features of politics in many Western democracies, particularly the United States. One contributor to these related phenomena may be increasing exposure to online political disagreements in which ordinary citizens criticize, and sometimes explicitly demean, opponents. This article presents two experimental studies that assessed whether U.S. partisans’ attitudes became more prejudiced in favor of the in-party after exposure to online partisan criticism. In the first study, we draw on an online convenience sample to establish that partisan criticism that derogates political opponents increases affective polarization. In the second, we replicate these findings with a quasi-representative sample and extend the pattern of findings to social polarization. We conclude that online partisan criticism likely has contributed to rising affective and social polarization in recent years between Democrats and Republicans in the United States, and perhaps between partisan and ideological group members in other developed democracies as well. We close by discussing the troubling implications of these findings in light of continuing attempts by autocratic regimes and other actors to influence democratic elections via false identities on social media.


Suhay, Bello-Pardo, and Maurer begin with a review of recent findings on increasing partisan and ideological polarization in the United States and note that this rise coincides with a sharp increase in internet political discourse.  Several features of social media and online discussion forums “have the potential to contribute to partisan polarization, including high levels of opinionation, criticism of political adversaries, and uncivil discourse (Anderson et al., 2014; Levendusky, 2013)” (Suhay et al., 2018, p. 96).  The reported research consists of two experiments that investigated whether internet partisan criticism increases polarization.

Suhay et al. distinguish between affective polarization (partisans’ intense dislike of the opposition) and social polarization (a motivation and tendency to avoid associating with partisan opponents; see also, Iyengar et al. 2012; Iyengar and Westwood 2015; Mason 2016).  The authors suggest that ideological differences between Democrats and Republicans only partially explain the increase in both forms of polarization (see Iyengar et al. 2012; Iyengar and Westwood 2015, Pew Research Center 2016). They further suggest that social identity theory (Tajfel, Billig, Bundy, & Flament, 1971; Tajfel & Turner, 1979) may also help explain the increase in affective and social partisanship in the United States (see also, Huddy et al., 2015; Iyengar et al. 2012; Mason 2015).

Briefly, social identity theory contends that the existence of different social groups inevitably leads to affective and social polarization between members of those groups.  This is because individuals see the groups they belong to as extensions of themselves, and thus hold their own group (or groups) in higher esteem than other groups that they do not belong to/identify with.  Group successes and failures can feel like personal successes and failures (Huddy et al., 2015) and direct criticism of the ingroup may be perceived as a direct personal attack (Mackie et al., 2009).  In such circumstances, individuals may compensate by exhibiting greater ingroup favoritism, as well as increased hostility towards and derogation of rival outgroups.  Iyengar et al. (2012) and Mason (2016) have previously found evidence that this dynamic has contributed to the increased affective and social polarization in the United States.

Finally, Suhay et al. briefly review the rapid changes in the media landscape that have occurred in the United States with the emergence and growth of the internet.  They note that while greater participation in political discourse may have benefits, in general and when conversing with diverse others, there are at least two particular features common to online discourse that are concerning (p. 99):

First, much online discourse is low quality, with posts displaying group think, emphasizing opinion over fact, and including illogical argument and inaccurate factual claims (Groshek and Bronda 2016; Ruiz et al. 2011; Singer 2011). Second, online political discussions are overwhelmingly negative, dominated by criticism of political opponents (Berry and Sobieraj 2014; Park 2015; Suhay et al. 2015). With the prior point in mind, much of this criticism is not substantive but, rather, ad hominem (Berry and Sobieraj 2014). Some of this online negativity is overtly disrespectful and demeaning (Coe et al. Rains 2014; Gervais 2014; Mutz 2005). Such uncivil political discourse has been shown to delegitimize the political opposition (Mutz 2007), decrease open-mindedness (Borah 2012), and encourage copycat incivility (Gervais, 2014).


In both experiments, participants were randomly assigned to one of three experimental groups (control, anti-Democrat, anti-Republican) and read news articles discussing a partisan debate between Democrats and Republicans.  After reading the article they were then exposed to reader comments.  All three experimental groups were exposed to two negative but nonpartisan comments.  The anti-Democrat and anti-Republican groups were then exposed to three comments with party-specific incivility (those in the anti-Democrat group saw anti-Democrat comments, while those in the anti-Republican group saw anti-Republican comments). After reading the article and the comments, affective evaluations of Democrats and Republicans were assessed.  In Study 1 affective evaluations of then-President Obama were also assessed, and in Study 2 desire for social distance from political opponents was also assessed in two ways.  The first, asked participants how happy they would be if a member of their family married 1) a Republican and 2) a Democrat.  A difference score (Republican happiness – Democrat happiness) was then computed.  Second, Suhay et al. also asked people how important it was to them to live in a community “where most people held political views similar to your own,” with higher scores indicating “very important.”

Study 1 consisted of 424 participants recruited via Mechanical Turk, and was run in the Spring of 2013.  The content of the news article focused on the automatic spending cuts that were triggered on March 1, 2013 when Congress failed to reach an agreement to reduce the deficit.  About 46% of participants were Democrats and 35% were Independents.  Just 13% were Republicans.  Demographic information (including party affiliation) was obtained prior to randomly assigning participants to conditions.

Study 2 consisted of 542 participants recruited through Qualtrics Panels.  Thus, the dataset in Study 2 was representative for age, gender, and census region.  An equal number of Democrats and Republicans were also obtained.  As with Study 1, demographic information (including party affiliation) was obtained prior to randomly assigning participants to conditions.


In Study 1 the effects of the anti-Republican treatment were stronger than the effects of the anti-Democrat treatment.  Exposure to anti-Republican comments drove Republicans and Independents to evaluate the Republicans more favorably than Democrats.  This effect was also found when assessing evaluations of Barack Obama.  Exposure to anti-Republican comments drove Republicans and Independents to evaluate Obama less favorably than Democrats. The Democrats sampled, on average, did not react to the stimuli.  In other words, the effects in Study 1 were driven by the reactions of Independents and Republicans.

The findings of Study 2 mirrored those of Study 1 for feeling thermometer ratings and for feeling about marrying a political opponent.  Exposure to anti-Republican comments drove Republicans and Independents to evaluate the Republicans more favorably than Democrats, and to report greater happiness at a family member marrying a Republican.

For the likeminded community variable, Republicans were more likely to report that living in a likeminded community was very important to them after exposure to anti-Republican comments.  Likewise, Democrats were more likely to report that living in a likeminded community was very important to them after exposure to anti-Democrat comments.


In this study, “participants who read two or three reader comments critical of their own or the opposing party were, on average, more likely to (1) rate the in-party higher than the out-party on a feeling thermometer, (2) express greater happiness at the prospect of a family member marrying a co-partisan (as opposed to someone in the out-party), and (3) state they prefer living near politically likeminded people” (Suhay et al., 2018, p. 107).  Additionally, “in Study 1 Independents and Republicans were also (4) more likely to disapprove of President Obama after exposure to partisan criticism” (Suhay et al., p. 107).

Interestingly, the effects appear to be strongest among those exposed to the anti-Republican stimuli: “Effects are strongest among those who received the anti-Republican stimulus and with respect to the Democratic feeling thermometer. In short, there is a pattern of heightened reaction to Democrats complaining about the opposition party that registered mainly as party polarization with respect to the Democratic feeling thermometer” (Suhay et al., 2018, p. 108).

There are at least two limitations to this set of studies.  First, participants are limited to the United States and the hypotheses were not tested outside this context.  Second, online communication, particularly about news, is rapidly changing.  Many news organizations are eliminating their online comment sections and political conversations are increasingly occurring on mediums such as Twitter.  Although Suhay et al. (2018) suggest their findings would likely extend beyond online comment sections, this has yet to be demonstrated.

Suhay et al. (2018) close by arguing that eliminating comment sections (or other forms of online political discourse) “is a ‘solution’ that currently raises many more problems than it solves” (p. 109).  They recommend shifting to a system that decreases the anonymity of each individual user as a way to improve civility in online political discourse.

Why These Findings are Important

They provide preliminary evidence that antagonistic partisan criticism online may be contributing to growing affective and social polarization within the United States.

Opinions expressed are those of the author(s). Publication does not imply endorsement by Heterodox Academy or any of its members. We welcome your comments below. Feel free to challenge and disagree, but please try to model the sort of respectful and constructive criticism that makes viewpoint diversity most valuable. Comments that include obscenity or aggression are likely to be deleted.