CrossRef citations to date
Research Article

Testing the Effect of Cross-cutting Exposure to Cable TV News on Affective Polarization: Evidence from the 2020 U.S. Presidential Election



This study tests whether the consequences of cross-cutting exposure (CCE) to political news on affective polarization depend on the type of media individuals consume. Findings from the 2020 American National Election Studies panel data suggest that CCE to cable TV news, a medium frequently described as uncivil, may increase negative affect toward the out-party candidate, and that such effect may be mediated by feelings of anger and certainty in one’s political attitudes. These insights draw attention to elite incivility, in addition to citizens’ partisan predispositions, as a potential contributor to the growing affective divide in American politics.

Communication scholars are divided on the effect of cross-cutting exposure (CCE), or encountering counter-attitudinal political information, on affective polarization. Some argue that exposure to attitude-challenging perspectives would have a counter-effect on affective polarization for it would attenuate the strength of one’s previously held beliefs (Mutz, Citation2002; Price et al., Citation2002), while others found that exposure to disagreements may bolster individuals’ prior opinions, leading people to form a more extreme position toward the outgroup (Eveland, Citation2004; Wojcieszak, Citation2010). Partisan strength has often been described as among the most important factors in determining the consequences of CCE on interparty hostility – that is, those who have strong prior attitudes are more likely than weaker partisans to engage in directional processing of information that counters previous attitudes, which could further reinforce extreme positions (Bolsen et al., Citation2014). Still, relatively less attention has been paid as to how the quality of political messages distributed by media elites may explain the competing arguments regarding the effect of CCE on affective polarization.

This study asks whether “political incivility,” or disrespectful attacks that insult the opposing ideological group, serves as an account for the link between CCE and individuals’ tendency to dislike the out-party. Earlier research on CCE to uncivil messages and affective polarization has largely focused on political discussion that occurs in offline/online social networks (e.g., Suhay et al., Citation2018), highlighting individual citizen’s responsibility in furthering deliberative democracy. Of particular interest in this research is incivility in the news media, particularly that in cable TV news in the U.S., which has frequently been described as a medium that uses sensationalism, partial truths and ad hominem attacks on the opponents (Sobieraj & Berry, Citation2011; York, Citation2013). Drawing on a two-wave national panel data set, this article aims to: (1) examine the effect of cross-cutting cable TV news exposure on affective polarization in the context of the 2020 U.S. presidential election, and (2) suggest anger and attitude certainty as potential mediators through which encountering dissonant views contributes to a more polarized political landscape. In doing so, the present study attempts to call attention to elite incivility as a possible contributor to citizen animosity toward ideological opponents.

Cross-cutting Exposure to Cable TV News and Affective Polarization

Scholars define affective polarization as the tendency of those identifying with a political party to view the outgroup negatively and ingroup positively (Iyengar et al., Citation2019). Though related to disagreement on policy preferences (i.e., issue polarization), affective polarization in particular underscores the animosity between the parties, or “feeling” toward the in- versus the outgroup (p. 130). Social identity theory suggests that merely identifying with a group is sufficient to provoke negative feelings toward the outgroup (Mason, Citation2016), which makes partisans, or individuals who are strongly attached to a party, more likely than nonpartisans to be affectively polarized against one another. Despite the ongoing debate over whether the mass public has gradually become more divided on political issues (e.g., Fiorina et al., Citation2008), a near consensus holds that interparty hostility has increased substantially over the last few decades among average citizens in America (Iyengar et al., Citation2019).

Partisan news sources in the high-choice media environment have often been cited as a salient contributor to partisan animosity, as selective exposure to pro-attitudinal information may reinforce preexisting ingroup-outgroup feelings. The concept of cross-cutting news exposure has attracted scholarly attention for this matter, being regarded as the antithesis of likeminded exposure that could potentially bridge the affective divide. Deliberative theorists argue that, because encountering dissimilar opinions would make people aware of multiple perspectives, CCE would increase attitudinal ambivalence, leading individuals to hold more moderate opinions and balanced political judgments (Mutz, Citation2002; Price et al., Citation2002). However, this somewhat optimistic view may not be applicable to partisans in a sharply divided two-party system since they tend to process dissonant information in a biased manner to rationalize or even bolster their prior views (Eveland, Citation2004).

While much has been studied and predicted about the role of individual predispositions in shaping the way one responds to ideologically incongruent opinions, some suggest that the degree of civility or politeness in dissimilar opinions might determine the effectiveness of CCE in mitigating negative attitudes toward the out-party members. Literature on offline political talk among citizens suggests that cross-cutting discussion that involve disagreement can facilitate political learning and promote tolerance toward competing ideological positions (e.g., Mutz, Citation2002), but such arguments have generally not been supported by studies on online political discussion, especially those on social media sites. Scholars have expressed skepticism that CCE on social media should induce political deliberation, largely since user-generated messages in online platforms tend to be much more uncivil compared to those in face-to-face interactions, as overtly demeaning, illogical criticism of opponents dominates online political discussions (Suhay et al., Citation2018). Experimental evidence suggests that a user may become more entrenched in her/his existing views when exposed to opposing views on social media (Bail et al., Citation2018), and that exposure to uncivil postings that attack one’s ingroup may generate feelings of aversion, reducing users’ open-mindedness (Borah, Citation2013) thus the potential for effective deliberation.

The relative lack of research concerning CCE to uncivil professional news media is perhaps due to the perceptions that (1) exposure to counter-attitudinal partisan media is very unlikely in a high-choice media environment, and that (2) political discourse that is highly demeaning and disrespectful of the opponent is an occurrence usually reserved for the often-anonymous ordinary citizens on the Internet. While most individuals indeed have the tendency to seek out information that supports their preexisting political beliefs, research has shown that citizens in a complex news media environment may constantly encounter/consume opposing political views (Dubois & Blank, Citation2018) – a recent national survey on political media use found that roughly a quarter of Americans who identified with a political party got news from counter-attitudinal sources, although many of them said they distrust those sources (Jurkowitz et al., Citation2020).

More importantly, cable TV news outlets in recent years have become a subject of criticism due to their partisan style of reporting aimed at appealing to a profitable niche audience. Cable TV news may not be the sole medium with an explicitly partisan slant, but news outlets such as Fox News, CNN and MSNBC have garnered considerable controversy largely due to their uncivil mode of delivering political news contents. Scholars report that cable news programs are significantly more likely than other news sources (e.g., print newspaper and websites) to contain insulting language, name calling, mockery and misrepresentative exaggeration in their political comments (Gervais, Citation2019; Sobieraj & Berry, Citation2011). York (Citation2013) argues that cable TV news channels, being pressured to fill the ever-widening 24/7 news hole, often choose to air negative, combative stories to fill the hole and present them in an overly emotional, hostile fashion to increase ratings, which tend to result in messages that emphasize entertainment and political division over hard facts (p. 110). Earlier findings suggest that exposure to attitude-consonant cable TV news contents aimed at provoking visceral responses from the audience may make partisan viewers even more extreme (Levendusky, Citation2013) and increase affective polarization regardless of audiences’ previous partisan leanings (Lu & Lee, Citation2019).

The core question in the present study lies in whether exposure to counter-attitudinal cable TV news channels mitigates or exacerbates affective polarization. Several studies have hinted at the positive effect of cross-cutting media use on affective de-polarization, conforming to the previous notion that CCE in general should result in attitudinal ambivalence and tolerance toward opposing viewpoints. It should be noted, however, that much of the evidence has been obtained from either an analysis of mass media sources other than cable TV news (e.g., news websites and blogs; Garrett et al., Citation2014), combined measures of cable and network/mainstream TV programs (e.g., Lu & Lee, Citation2019) or research conducted overseas (e.g., Conroy-Krutz & Moehler, Citation2015).

In the study perhaps most similar to the current research, Gervais’s experiment (Gervais, Citation2019) found that counter-attitudinal exposure to uncivil tweets from cable TV outlets/personalities (e.g., CNN and Bill O’Reilly) that attacked the outgroup with negative adjectives and hyperbole, stimulated affective reactions from the viewers, which could eventually drive the audience to engage in a “combative form of partisanship.” The fact that counter-attitudinal incivility generated significantly more unpleasant feelings from the subjects than did civil messages from the same source (pp. 643–644) suggests incivility, rather than CCE per se, as a factor that could potentially contribute to affective polarization. While little to no work has yet been done regarding the effect of CCE to cable TV news programs on affective polarization in the context of the 2020 presidential election, this study expects cross-cutting cable TV news exposure to have a positive relationship with affective polarization, based on the notion that uncivil attacks from an ideologically incongruent news source could induce animosity toward the outgroup. Conceptualizing cable TV news as “uncivil political media,” I hypothesize:

H1: Cross-cutting exposure to cable TV news (W1) is positively related to affective polarization (W2).

In the following section, I provide a review on the extant studies concerning audience’s emotional response to uncivil dissonant messages to delve deeper into the psychological mechanism underlying the above proposed hypothesis.

Anger and Attitude Certainty as Potential Mediators

Among the extensive list of negative emotions that could stem from exposure to partisan media contents, anger has frequently been a subject of interest for studies on CCE. Classified under the broad category of negative affective states, anger in a political context may refer to emotion that individuals experience when their political goals have been interfered or threatened (Carver & Harmon-Jones, Citation2009; Nabi, Citation1999). If the goals of a potential voter lie in confirming her/his valued issue positions (Lu & Myrick, Citation2016) and achieving electoral victories (Gervais, Citation2019), counter-attitudinal arguments that challenge one’s existing beliefs would inhibit/threaten such goals, thus making CCE more likely than likeminded exposure to trigger anger among audiences.

Although CCE may also induce feelings of anxiety/fear, according to the appraisal-tendency framework, which posits that emotions have discrete antecedents and effects (Suhay & Erisen, Citation2018), anger rather than anxiety would play a mediating role in the relationship between CCE and affective polarization. Anger is elicited when individuals are “certain of who is responsible for the offense” and feel that they are “capable of countering the threat” (Gervais, Citation2019, p. 638; Suhay & Erisen, Citation2018, p. 797). This distinguishes anger from anxiety and fear, which are known to arise due to a lack of certainty and perceived personal control (Lu & Myrick, Citation2016). In this sense, CCE to cable TV news would generate anger among party identifiers who are well aware of the source of threat (i.e., their political outgroup), especially prior to a highly contested election when they feel efficacious about their ability to offset the threat through voting.

One of the most cited consequences of anger is negative affect toward the outgroup. Anger would lead individuals to protect their existing political beliefs, and to heavily rely on ideological cues and biased processing to do so when consuming new information (Lazarus, Citation1991). Despite people’s tendency to become less reliant on previously held convictions and more accepting of attitude-challenging information when feeling anxious (MacKuen et al., Citation2010), anger motivates people to attack and argue against the outgroup for the purpose of defending themselves (Nabi, Citation1999). Angry individuals may attribute blame for unpleasant events to the outgroup members, perceiving them as the source of anger (Keltner et al., Citation1993).

CCE, from a deliberative theorist’s point of view, may not automatically elicit anger, but the presence of incivility in political commentary would make anger more likely among those exposed to attitude-challenging messages. Anger is triggered when one perceives injustice, unfairness or a violation of what s/he believes should be accepted standards, and is thus typically directed toward others and the ideological outgroup (Gervais, Citation2019). Political incivility that involves invectives and ridicule to attack the audience’s ingroup, coupled with flattering rhetoric to praise the viewers’ opponents of moral and intellectual superiority (Sobieraj & Berry, Citation2011), should elicit perceptions that such information violates a traditional standard of justice and objectivity. That being said, political incivility manifested on cross-cutting TV programs may elicit greater anger compared to uncivil contents on print/text-based media, as the audiovisual nature of the medium allows the use of cynical tones, shouting and sarcastic facial expressions to discredit the opponents (Gervais, Citation2019). Given anger’s potential to contribute to a climate of interparty hostility, I propose the following hypothesis:

H2: Cross-cutting exposure to cable TV news (W1) is positively related to affective polarization (W2) through anger (W1).

Anger is positively associated with the level of confidence. Earlier studies have shown a stronger association between certainty of thoughts about an event or a group of people and anger compared to that and other negative emotions such as fear and sadness (Blankenship et al., Citation2013). Scholars believe that anger would activate self-assurance, assertiveness and confidence in attitudes to prepare individuals to defend convictions (Carver & Harmon-Jones, Citation2009), suggesting the direction of the possible causal relationship between anger and attitude certainty. Based on the existing literature on political attitudes, this study uses the terms confidence and certainty interchangeably since it is common to do so when the gist of the discussion revolves around “how confident, certain, or sure people are in the validity of their thoughts and attitudes” (Briñol et al., Citation2018, p. 693).

Here, it is worth noting that attitude certainty, which refers to “the subjective sense of conviction, or confidence, with which one holds one’s attitude” (Cheatham & Tormala, Citation2015, p. 1537), is conceptualized as a subdimension of attitude strength. Matthes et al. (Citation2010) argue that attitude certainty may capture attitude strength more accurately than other indicators such as attitude extremity since one could hold a non-extreme attitude toward an issue, but feel confident that their attitude about the issue is correct (p. 777). Researchers found that attitudes held with certainty foster willingness to engage in different types of political actions and to openly express one’s opinions (for a review, see Cheatham & Tormala, Citation2015), and considering that anger is an “approach emotion” associated with action tendencies (Nabi, Citation1999), one may expect anger to be a possible antecedent of certainty in attitudes.

From a media psychology perspective, anger would increase attitude certainty for it activates heuristic processing of information. Earlier experimental findings show that anger, compared to neutral and other negative emotions, led to more stereotypic judgments of information, greater attention to the superficial cues of the message and less attention to the arguments presented in the message (e.g., Tiedens & Linton, Citation2001). Negative emotions may boost willingness to acquire knowledge about politics (Marcus et al., Citation2000), but anger in particular would lead to shallow processing of evidence and hostile judgments of cross-cutting information based on ideological cues such as perceived source credibility (Tiedens & Linton, Citation2001). Weeks (Citation2015) suggests that anger, rather than general negative affect, facilitates partisan motivated reasoning of political information, which in turn may reinforce feeling of confidence/conviction in one’s attitude. Based on the notion that a lack of tolerance toward counter-attitudinal views contributes to animosity toward political opponents, the following hypothesis is put forward:

H3: Anger (W1) is positively related to affective polarization (W2) through attitude certainty (W2).

Though rarely stated explicitly in the existing literature on CCE, the above-cited studies hint at a potential serial relationship between exposure to counter-attitudinal cable TV news and affective polarization through anger, where attitude certainty mediates the link between anger and negative affect toward the opposing party. This study additionally asks whether exposure to cross-cutting incivility directly relates to attitude certainty. If CCE that promotes tolerance for diverging viewpoints were to result in “attitudinal ambivalence” or opinion uncertainty, as deliberative theorists suggest, I expect CCE to uncivil media to have a direct association with attitude certainty, a concept that has typically been regarded as the flip side of attitudinal ambivalence (Luttrell et al., Citation2020). While studies have shown that attitude-threatening information could strengthen one’s existing attitude toward an issue (e.g., Schmuck et al., Citation2020), due to the relative lack of literature linking CCE to certainty/confidence in attitudes, I pose the following as a research question:

RQ: Is cross-cutting exposure to cable TV news (W1) positively related to affective polarization (W2) through attitude certainty (W2)?

My expectations and inquiry regarding the relationship between the key variables are illustrated in as a causal path diagram.

Figure 1. Conceptual mediation model of cross-cutting cable TV exposure and affective polarization.

Figure 1. Conceptual mediation model of cross-cutting cable TV exposure and affective polarization.



Data for this study came from the 2020 American National Election Studies (ANES) time series survey. The dataset is based on a two-wave panel survey of nationally representative sample of U.S. citizens of voting age. Pre-election survey was conducted between August 18, 2020 and November 3, 2020 (response rate: 36.7%) and post-election data were collected between November 8, 2020 and January 4, 2021 (reinterview rate: 90.0%). To test the hypotheses regarding CCE, respondents who self-identified as Independents or affiliate with other parties were excluded from further analyses. This brought the sample size to 4,044 respondents, all of whom participated in both pre- and post-election surveys.


Cross-cutting Exposure to Cable TV News

In the first wave of the survey, a dichotomous item asked whether respondents “have heard anything about the presidential campaign” from TV programs, which allowed this study to further exclude respondents who did not use any TV sources for subsequent analyses. The survey included 16 cable TV news shows, all of which were either from Fox News, CNN or MSNBC (see, Dilliplane et al., Citation2013 for validity and reliability of self-reported media use measures).Footnote1 Based on earlier works on partisan slant of cable TV networks (Gervais, Citation2019; York, Citation2013), Fox News was coded as a conservative cable news source, while CNN and MSNBC were regarded as liberal sources.

Respondents were asked to select any of the 16 shows that they watch “at least once a month” (Mentioned = 1; Not mentioned = 0) – seven of them aired on Fox News (e.g., Hannity, Tucker Carlson Tonight), while the rest nine programs were either from CNN or MSNBC (e.g., The Lead with Jake Tapper, The Rachel Maddow Show). Following the previous study on partisan media use and affective polarization (Lu & Lee, Citation2019), the final CCE variable (M = .49, SD = .34) was calculated by dividing the total number of consumed counter-attitudinal shows by the total number of consumed cable TV news programs. For instance, if a Democrat who watched 10 cable news programs indicated that three of them were from Fox News, the CCE score for that respondent was 0.3. This calculation method was also used to create the variable on likeminded exposure to cable TV news to be included as a control variable in the subsequent analyses (see, for average number of programs viewed).ō

Table 1. Average number of media sources viewed.


Since feelings of anger, according to the cognitive appraisal framework of emotion, tend to be experienced and reported as outrage, anger or irritation (O’Mara et al., Citation2011), this study relied on three measures of anger in the pre-election survey. Respondents were asked to indicate how outraged/angry/irritated they feel about how things are going in the country on a scale from 1 (not at all) to 5 (extremely). The three items were averaged to create the final variable on anger (M = 3.72, SD = 1.01, Cronbach’s α = .90).

Attitude Certainty

Given that angry voters during an election may form a strong opinion on multiple issues facing the nation (Schill & Kirk, Citation2017), this study used a series of dichotomous choice items in the post-election survey that asked respondents whether they believe that: (a) Russia tried to interfere in the 2016 presidential election; (b) Trump’s administration deported more unauthorized immigrants during the first three years than Obama’s administration; (c) most scientific evidence shows childhood vaccines cause autism; (d) world temperatures have risen on average over the last 100 years; (e) COVID-19 was developed intentionally in a lab; and (f) hydroxychloroquine is a safe and effective treatment for COVID-19. Respondents were then asked to indicate how confident they feel about their answers to the above questions on a 5-point scale (1 = not at all, 5 = extremely). The latter indicators, regardless of the correctness of respondents’ beliefs, were averaged to assess attitude certainty (M = 3.55, SD = .78, Cronbach’s α = .78).

Affective Polarization

This study used ANES’s “feeling thermometer” index of Donald Trump and Joe Biden to measure affective polarization, considering that animosity between the parties during a presidential election year would perhaps be best manifested through feelings toward in- vs. out-party candidate for the presidency. In both pre- and post-election waves, respondents indicated how cold/warm their feelings were toward Trump and Biden on a 101-point scale from 0 (extremely cold) to 100 (extremely warm). Following the widely used method to measure affective polarization (Iyengar et al., Citation2019), the dependent variable was obtained by taking the absolute value (i.e., magnitude) of the difference in the thermometer ratings of the two candidates (W1 M = 72.24, SD = 26.43; W2 M = 73.81, SD = 27.18). reports the zero-order two-tailed correlations between the key variables.

Table 2. Pearson’s correlations between key variables.

Control Variables

The following demographic variables measured in Wave 1 were held constant throughout the analyses to control for potential confounds: age (M = 53.64, SD = 16.85), gender (female: 56.4%), education (M = 4.69, SD = 2.00, Range = 1 [less than high school] to 8 [professional school or doctoral degree]), household income (M = 12.35, SD = 6.60, Range = 1 [under $9,999] to 22 [$250,000 or more]), race (White: 75.7%). Also included as statistical controls were party identification (1 = Democrats [53.4%], 2 = Republicans [46.6%]), partisan strength (weak partisans: 30.4%, strong partisans: 69.6%) – both of which were shown to significantly affect the level of affective polarization (Lu & Lee, Citation2019) – along with newspaper use (e.g., The New York Times; 45.5%), Internet sites use to follow presidential campaign (e.g., HuffingtonPost.com; 71.1%), mainstream TV news use (e.g., ABC World News Tonight; M = .13, SD = .13, Range = 0–1, Cronbach’s α = .64) and likeminded exposure to cable TV news (M = .51, SD = .34, Range = 0–1), all of which were assessed in the pre-election survey. Considering “anxiety” as a distinct negative emotional response that has the potential to affect the level of anger (Marcus et al., Citation2006), the three items that asked how afraid/worried/nervous respondents feel about how the things are going in the country (1 = not at all, 5 = extremely) in Wave 1 were averaged to create the anxiety variable (M = 3.65, SD = 1.10, Cronbach’s α = .92) to be controlled throughout the analyses.


To test H1, two types of ordinary least square regression analyses were conducted: (1) time-lagged, which assessed the association between CCE in Wave 1 and affective polarization measured in Wave 2, and (2) autoregressive, where affective polarization in Wave 1 was introduced as a control in the lagged model. Compared to a lagged model, an autoregressive model would provide a clearer evidence of causation as the results “reflect the influence of predictor variables in [Wave 1] on the outcome variable in [Wave 2] above and beyond the causal influence of prior levels of the outcome variable on itself” (Shah et al., Citation2005, p. 549). As presented in , exposure to counter-attitudinal cable TV news was positively associated with affective polarization across the lagged (β = .04, p < .05) and the autoregressive model (β = .03, p < .05). Interestingly, likeminded exposure to cable TV news, in this particular dataset, was not significantly related with affective polarization (β = .01, p = .37).Footnote2 Though not hypothesized, CCE did not remain significant at the .05 level when anger (W1) was additionally entered as the final block of the autoregressive model (β = .02, p = .08). Anger had a significant positive effect on affective polarization (β = .06, p < .001).

Table 3. Time-lagged and autoregressive regression models testing the effect of cross-cutting cable TV news exposure on affective polarization.

To test the hypotheses and research question regarding the two proposed mediators (i.e., anger and attitude certainty) in the association between CCE and affective polarization, PROCESS Model 6 (Hayes, Citation2013) with 95% bias-corrected confidence intervals and 5,000 bootstrap samples was used to test the theoretical mediation model presented in , controlling for affective polarization measured in Wave 1. summarizes the standardized regression coefficients of the hypothesized relationships.

Figure 2. Results of mediation analysis, standardized coefficients (N = 3,343).

Note. Values in parentheses denote standard error.
Figure 2. Results of mediation analysis, standardized coefficients (N = 3,343).

The results supported H2, as anger significantly mediated the path from CCE to the level of hostility toward the opposing candidate (b = .12, SE = .08, bootstrapping CI = [.01, .31]) (see, for the indirect effect of CCE on affective polarization through the hypothesized mediators). Anger and attitude certainty were positively correlated (β = .15, p < .001) and the path from attitude certainty to affective polarization was statistically significant (β = .09, p < .001), lending support for H3. The results confirmed the sole RQ for this study, which asked whether CCE relates to affective polarization through attitude certainty (b = .24, SE = .12, bootstrapping CI = [.01, .49]). CCE in the conceptual mediation model was not directly associated with animosity toward the outgroup candidate, as its relationship with the outcome variable was fully mediated by anger and attitude certainty.

Table 4. Indirect effect of cross-cutting TV exposure on affective polarization.

To see whether these findings extend across different media channels, post hoc analyses were carried out replicating the above procedure to investigate the association between affective polarization and exposure to cross-cutting political information on internet news sites, which was shown to have a significant correlation with the dependent variable when included as a control in the previous autoregressive model (β = .03, p < .05). The pre-election survey included items that asked respondents to indicate news websites they consume at least once a month. After excluding respondents who did not use any Internet sites (N = 2,817), ideological stance of each source was identified based on previous political communication studies (e.g., Peterson et al., Citation2021). Six of the 15 internet sites were categorized as having a liberal stance (e.g., HuffingtonPost.com), six as neutral (e.g., news.Yahoo.com) and three as conservative sources (e.g., Breitbart.com). Accordingly, variables on exposure to likeminded and cross-cutting internet sites were computed (Mlikeminded = .52, SD = .39; Mcross-cutting = .12, SD = .37). Results indicated that CCE to internet news sites was negatively related to affective polarization in the autoregressive model (β = −.03, p < .05), while likeminded exposure had a significant positive association with interparty hostility (β = .05, p < .001), controlling for the use of TV news and neutral websites. The post hoc mediation analyses revealed that CCE to news websites was not significantly associated with anger (β = .01, p = .60) and attitude certainty (β = .02, p = .21).


In view of the widespread discussion about the role of audience predispositions in determining the effect of CCE in reducing or exacerbating affective polarization (e.g., Levendusky, Citation2013), the present study sought to examine the possibility that the outcome of CCE may also be contingent on the civility of media discourse that individuals are exposed to. Conceptualizing cable TV news as politically uncivil media channels that are prone to the use of disrespectful, insulting or hyperbolic statements about the outgroup, I employed the ANES panel data collected before and after the 2020 presidential election to test the relationship between CCE to political incivility and the degree to which citizens view the opposing candidate more negatively than their own.

The first key finding suggests that exposure to counter-attitudinal cable TV news programs may significantly contribute to affective polarization, which is in line with the previous experimental results that indicated how exposure to dissonant views may not enhance tolerance for the ideological outgroups when the cross-cutting messages themselves are uncivil (Bail et al., Citation2018; Gervais, Citation2019). The current research extends earlier findings by tapping into the recent nationally representative panel sample of adults collected in a presidential election year and by directly linking CCE frequency to affective polarization. Also demonstrated in the results are the potential mechanisms through which CCE contributes to animosity toward the outgroup, as the hypothesized mediation model showed significant mediating effects of anger and attitude certainty on the relationship between exposure to uncivil dissonant media and negative sentiment toward the opposing candidate. I additionally report that attitude certainty partially mediated the path from anger to affective polarization, a finding that conforms to the notion that anger, which has frequently been cited as a predictor of motivated reasoning (e.g., Weeks, Citation2015), may fuel feelings of confidence in attitudes among party identifiers.

The post hoc findings in this study allude to the possibility that the aforementioned insights might be specific to cable TV news, or media channels that are widely recognized as uncivil/impolite political information sources. The present study demonstrated a significant mitigating effect of CCE to internet sites on affective polarization, as deliberative theorists and earlier communication studies on CCE have suggested. CCE to internet sites, a medium that has rarely been cited in the literature on political incivility, was neither associated with anger nor attitude certainty in the conceptual mediation model. These results extend prior research that has demonstrated the contrasting effect of online and cable news use on valuing the exposure to divergent viewpoints. While online news, which tends to share the same norms of fairness and balance as their offline content (Borah et al., Citation2013), was found to encourage the audience to believe that CCE is a valuable experience, cable news discouraged them to do so as it delegitimizes opposing viewpoints (p. 412). The insights from post hoc analyses add weight to the central thesis of this research that the kinds/quality of news produced and distributed by media elites, rather than the very exposure to cross-cutting viewpoints, may play a determining role in influencing citizen’s feelings toward political opponents, adding “media type” to the ongoing discussion over the potential factors that determine the effect of CCE on affective polarization.

While the current research provides compelling evidence that the effect of CCE on affective polarization may vary across political news channels individuals are exposed to, it also comes with several limitations. Although this study examined citizens’ use of cable TV outlets that have been deemed uncivil in the previous media literature (e.g., York, Citation2013), it lacks information on the actual level (or the kinds) of incivility respondents were exposed to, which calls for content analytic works on cable TV programs aired prior to the 2020 election to provide a more accurate assessment into the effect of uncivil dissonant news on affective polarization. Another noticeable limitation in the present study, along with several other works on CCE, regards the use of the term “exposure” to broadly delineate citizens’ information consumption experience since it does not distinguish incidental encountering of counter-attitudinal views from purposeful selection of such information. Although social networking sites (SNSs) have the potential to incidentally expose users to cross-cutting partisan news and significantly contribute to positive/negative affect toward the other party (Iyengar et al., Citation2019), SNS use was not measured in the pre-election survey, which did not allow this study to consider SNS as an antecedent of CCE, nor its use as a statistical control in the analyses.

It should also be noted that the method employed in this work cannot be regarded as a definitive test of causality, as some of its findings are based on cross-sectional data – e.g., CCE and anger were only measured in pre-election survey. In encouraging future experimental research on CCE to use the current mediation model as a guidance to more rigorously test the temporal order of the variables in interest, I anticipate further improvements to the survey measures adopted in this research. Anger, for instance, was measured by asking respondents how angry they feel about “how things are going in the country.” While such feeling may entail certainty that responsibility can be ascribed to a particular external agent (Rico et al., Citation2017; see, also Keltner et al., Citation1993), the measures could be additionally refined to specify the target of audience anger. To delve deeper into the present CCE construct, which was obtained by adding up the total number of counter-attitudinal programs a respondent watch in a month, future studies may address the consumption frequency of each program – since regular viewers might be the most affected (Levendusky, Citation2013) – as well as the degree of its heterogeneity to a respondent’s preexisting political attitudes.

Despite not being initially hypothesized, the current work yielded some unforeseen results worth noting for future efforts in this area. In the chosen dataset, exposure to attitude-consonant cable TV programs was not shown to have a significant effect on the dependent variable, thus did not confirm the previously described role of likeminded partisan media exposure in increasing affective polarization (e.g., Levendusky, Citation2013). Adding to the ongoing debate over whether exposure to ideologically congruent news indeed causes affective polarization (Iyengar et al., Citation2019), this study raises the possibility that likeminded media elites were unable to generate additional animus toward out-party candidates (Westwood et al., Citation2019) in the context of the already emotionally polarized 2020 election (Zhu et al., Citation2021). Moreover, though respondents were significantly more likely to get news on consonant rather than dissonant internet sites (t[2816] = 35.65, p < .001), such difference was not observed in their cable TV news use (t[3342] = 1.45, p = .11), prompting the question as to what leads individuals to consume incongruent cable TV programs. Considering that consuming opinion-challenging information can better prepare one to engage in political debate, and that those who perceive themselves as fair-minded would be more likely to engage with dissimilar ideological perspectives (Garrett, Citation2009), the findings underscore the need to consider various underlying factors that explain why audiences might willfully consume cross-cutting news.

Citizens’ inherent political attitudes that shape how they respond to new information, along with their preference for news that reinforce their views, have long served as a prima facie explanation for voters’ negative feelings toward the ideological outgroup. While the current work should expand the existing scholarly discourse on the potential factors that shape the effect of CCE on affective polarization, perhaps its greater contribution lies in attempting to hold media elites accountable for contributing to the heightened antagonism between party supporters in American politics. As long as cable TV outlets celebrate the uncivil nature of their news products, the virtue of hearing from the other side in fostering deliberative democracy may remain debatable.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information


The author received no financial support for the research, authorship, and/or publication of this article

Notes on contributors

Hyungjin Gill

Hyungjin Gill (M.A. University of Missouri-Columbia) is a doctoral candidate at the University of Wisconsin-Madison School of Journalism and Mass Communication. His research centers on political communication, with a specific focus on examining the effect of social interactions on political information evaluations and public opinion perceptions.


1 For alternative approaches to measuring political news use, see, Prior (Citation2013).

2 The models were re-tested using the “total number of cable TV programs viewed” instead of the proportion measures. The significance of the initial findings remained unchanged.


Reprints and Corporate Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

To request a reprint or corporate permissions for this article, please click on the relevant link below:

Academic Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

Obtain permissions instantly via Rightslink by clicking on the button below:

If you are unable to obtain permissions via Rightslink, please complete and submit this Permissions form. For more information, please visit our Permissions help page.