The Effect of Forced Exposure to Crosscutting Information: What Is the Effect of Broadcast News Shows That Deliver Opposing Opinions?

The underlying assumption of the echo chamber theory is that exposure to only one-sided opinions may lead to biased attitudes. What would happen, however, if the audience were forcibly exposed to cross-cutting information simultaneously?

Objectivity has been a holy grail of journalism, especially in the countries with the North Atlantic or liberal model of media and politics (Hallin & Mancini, 2004). One of simplest way to achieve at least surface objectivity is to devote equal time or space to competing views, in other words, to pursue mechanical objectivity. In the United States, the equal time rule enforces balanced coverage of elections by forcing media to allocate equal airtime to each candidate. Balanced reporting is even more necessary in a highly polarized political environment where the media is constantly under attack for unfair reporting from both liberal and conservative parties. In South Korea, where political polarization has become increasingly problematic, current affairs broadcast shows or news shows often invite commentators who represent viewpoints of two dominant rival political parties. Commentators usually have formal ties to the political parties, as former/current members of the Korean Assembly or party officials, or have informal but clear connections to the two parties, or are political experts with well-known partisan attitudes. In most cases, the audience can easily guess who favors a particular party even before listening to their comments. Guests typically express their opinions in line with their party’s position on the matter under discussion. Unlike an echo chamber, audiences are exposed to opinions from both liberal and conservative sources on broadcast shows. In other words, the audience is subject to a forced exposure to opposing opinions.

However, the effect of forced exposure to cross-cutting information is unclear and remains a subject worth testing empirically. At best, forced exposure to the opposing opinions may lead the audience to deliberate the arguments of both sides before forming their own opinion. Thus, the audience may have a better chance of making informed decisions. In this case, broadcast news may provide a countermeasure to the echo chamber phenomenon. However, if the majority of the audience forms their opinion mainly based on partisanship rather than the message itself even when they are exposed to both sides of arguments, the balanced format of broadcast news may not help make a better informed electorate. The audience will still remain as polarized and biased as when they are exposed only to one-side of an argument. In such a case, the only role of broadcast shows with commentators from rival parties is to allow the media to escape from criticism of unfair coverage. It would not be the echo chamber, a situation where one is surrounded only by like-minded others, but the partisan processing of the message that drives political polarization.

The purpose of this paper is to empirically test the effect of forced exposure to cross-cutting information through a broadcast show like those discussed above. Because it is not possible to distinguish the unique effect of each commentator’s argument in a real setting, that is through a survey, the current study tests the effect of such broadcast shows through an experiment.

Literature Review

The Internet and social network services (SNS) have radically changed the way citizens consume news. Instead of receiving news with limited ideological differences from a handful of TV or radio channels, users can choose from a far greater variety of media with ideologically biased messages. Combined with the fragmentation and polarization of news media, greater freedom of choice likely results in a situation where most people only consume media that support their preferred ideological slant (Min & Yun, 2018). Echo chamber theory raises concerns that people may insulate themselves in ideologically homogenous bubbles that could lead to political polarization. A large body of research on political behavior has found that people prefer to mingle with and receive information from people with similar perspectives (Huckfeldt & Sprague, 1995; McPherson et al., 2001). A proliferation of the echo chamber phenomenon may prohibit the formation of informed decisions and lead to a more polarized and fragmented public space, which results in a society that is less tolerant toward different opinions and toward those who hold such opinions (Mutz, 2002; Prior, 2007; Sunstein, 2001).

Cognitive dissonance theory provides one of the most compelling explanations of the echo chamber phenomenon. According to Festinger (1962), a person may limit exposure to opinions that contradict their own opinion on a matter of interest to avoid cognitive dissonance. Several studies have found that people experience psychological discomfort when exposed to dissonant information (Elliot & Devine, 1994; Harmon-Jones et al., 1996; Zanna & Cooper, 1974). People try to reduce such dissonance by avoiding messages that do not agree with their own stance (Gibbons et al., 1997; Kiesler & Pallak, 1976; Losch & Cacioppo, 1990). One frequently adopted practice for avoiding dissonant messages is to seek ideologically consonant spaces and interactions, hence, the emergence of ideologically homogeneous echo chambers. Studies have found that offline political dialogue mostly occurs among friends, family members, and like-minded “others” (Conover et al., 2002; Huckfeldt & Sprague, 1995; Kim et al., 1999). Theiss-Morse and Hibbing (2005) found that even voluntary associations become ideologically homogenous over time.

What happens when a person is exposed to messages from others with a similar ideological orientation? Interpersonal discussions with like-minded people reinforce existing attitude extremities (Bennett & Iyengar, 2008; Iyengar & Hahn, 2009; Stroud, 2010), by increasing confidence in one’s beliefs (Frey, 1986; Sunstein, 2001). Wojcieszak (2010) reported that active users of ideologically extreme online groups tend to believe in the superiority and prevalence of their own opinions over opposing opinions. Stronger confidence often leads to greater intolerance toward people with different opinions (Garrett et al., 2016). Thus, the echo chamber effect poses a real threat to democracy, which is most effective when citizens have substantiated beliefs (Downs, 1957).

However, research on selective exposure to media provides a mixed picture depending on the level of analysis. There is abundant empirical evidence of selective exposure at the micro level (Garrett, 2009; Iyengar & Hahn, 2009; Knobloch-Westerwick & Meng, 2009). However, macro-level results do not correspond with the concept of selective exposure. Surveys have found that people, especially those with greater political involvement, are often exposed to diverse media and dissonant messages (Dubois & Blank, 2018; Flaxman et al., 2016; Fletcher & Nielsen, 2018). Webster (2014) argued that mass media usage data do not support selective exposure; instead, people tend to consume a large variety of media with diverse ideological orientations. Twitter users engage with users with different ideological propensities, especially those across the left–right divide (Bright, 2018). Gentzknow and Shapiro (2011) reported that ideological segregation of online news consumption is low in absolute terms. Thus, contrary to echo chamber theory, people often allow themselves to be intentionally or unintentionally exposed to dissonant messages from different-minded others.

As discussed above, selective exposure is not possible in the TV and radio news shows that invite both liberal and conservative commentators at the same time. In a sense, the audience are forcibly exposed to both sides of an argument. Does exposure to opposing opinions necessarily lead to the formation of more informed and balanced opinions? Previous research has seldom investigated such kinds of situations. The effect of such broadcast news shows remains a subject where further empirical testing is needed. Therefore, this paper poses the following question.

RQ: What is the effect of forced exposure to cross-cutting information?

Exposure to cross-cutting information would certainly place the audience in a better position to form informed and balanced opinions than if they were in an echo chamber. However, there are substantial reasons to doubt this would actually be the case.

Social identity perspective proposes that citizens remain under-informed because, instead of actively collecting and carefully reviewing relevant information on a policy issue to make informed decisions, they rely on convenient shortcuts, specifically, partisan cues, to decide their stance on an issue. In other words, citizens choose to rely on partisan cues to support their preferred parties. Contrary to the theory of elaboration likelihood, the influence of partisan cues is not limited to peripheral and low-information processing (Cohen, 2003). Even in cases of high elaboration, people ignore or defy information that provides evidence that is not aligned with their political reference group. In addition, the way the news media deliver policy information may facilitate confrontational attitudes. Members of, or those with a clear connection to, rival parties are often invited to debate on current affairs or news shows. Such a format may lead the audience to view policy issues as partisan conflicts.

Exposure to messages with opposing political views may even backfire and aggravate political polarization. Previous research on the effects of exposure to conflicting messages found that people often counter messages with opposing views using motivated reasoning (Bail, 2015; Lord & Lepper, 1979; Nyhan & Reifler, 2010). Studies have reported the presence of backfire effects when respondents in experiments are exposed to information that corrects factual inaccuracies. To summarize, the effect of exposure to messages from both ideological orientations is a subject for empirical investigation. Thus, this paper proposes that the audience will evaluate the argument of a commentator differently depending on whether the perceived ideological orientation of the commentator matches their own political ideological propensity or not.

H1: The audience will evaluate the argument of a commentator more favorably when the perceived ideological orientation of the commentator is similar to their own ideological inclination compared to when the commentator is seen as neutral.

H2: The audience will evaluate the argument of a commentator more negatively when the perceived ideological orientation of the commentator is the opposite of their own ideological inclination compared to when the commentator is seen as neutral.

To test the hypotheses, an experiment was carried out.

Experiment

Stimuli

For the experiment, we created audio clips that simulated a radio talk show with two guests and one host. There were two variants. The first type had two guests with affiliations (to fabricated NGOs) that implied their ideological orientation—one guest belonged to “Democratic Liberal Solidarity” and the other was affiliated with the “National Conservative Federation.” We called this variant the partisan scenario. The second variant had seemingly neutral (fabricated) NGOs without any partisan cues—one guest was a member of the “Korea Policy Foundation” and the second guest was from the “Social Policy Research Institute.” We called the second variant the non-partisan scenario.

All the commentators’ arguments were identical. The only difference between the two scenarios was the commentators’ affiliation. We did not use party membership as a partisan cue because party structure in South Korea is not stable enough to foster enduring and strong party loyalty. Even though South Korean political structure has had two dominant rival parties for over 30 years, there have been many fluctuations. Many small- and medium-sized parties with various ideological orientations have come and gone over the last 30 years. Instead, we decided to use words that show organizations’ ideological dispositions.

The author acted as the host. The commentators were both women, one in her 20s and another in her 40s. None of the voice actors had formal voice-acting training. Similar to popular radio talk show formats, the two guests sequentially discussed four topics: the abolition of local assemblies, stock transaction tax reform, amendment of the Minsik-I Act, and Broadcasting Act reform regarding the approval of general programming and news television channels. These issues are discussed in more detail below. The two guests expressed different opinions on all four topics. We attempted to make the arguments of the two commentators as rational and persuasive as possible. The two guests took turns commenting on the four issues; the “liberal” commentator spoke first in the first and the third topics while the “conservative” commentator answered the remaining two topics first. In the audio clip, the guests discussed the four issues for approximately five minutes.

The authors selected topics that were not salient issues at the time of the experiment and had not received much media attention to minimize the possibility of strong preexisting opinions on the topics. In addition, we chose topics on which the two rival parties did not have a fixed position, so that even a participant with great political knowledge would not be able to guess the partisanship of the speaker. The first topic, local assembly abolition, is a recurring issue in municipal elections that does not receive extensive media coverage. The two dominant parties’ positions remain ambiguous. The second topic, securities transaction tax reform, received substantial media attention, especially during the most recent presidential election; however, it is an issue that only interests those who participate in the stock market. We deemed that our participants, college students, were unlikely to pay attention to the issue. In addition, party positions on these issues are not very different. The third topic—the revision of the Minsik-I Act—refers to the amendment of two related laws: the Road Traffic Act and the Act on the Aggravated Punishment of Specific Crimes (henceforth, the Aggravated Punishment Act). In a tragic car accident, a seven-year-old child was killed while crossing a street at a pedestrian crossing. In response to mounting public pressure, the National Assembly passed the amendment of the two laws, which greatly increased the penalty for drivers in car accidents in school zones.^[1] Owing to the severity of the punishment, the reform was criticized for being poorly planned. It was a relatively popular issue in 2021 when the reform was carried out, but only among those who drove frequently. The last topic referred to reforms of the Broadcasting Act and the approval of general programming and television news channels. The Broadcasting Act stipulates that general broadcasting channels and television news channels must receive approval for a fixed duration (3–5 years) from the regulator, the Korea Communications Commission. This provision has recently come under attack because the explosive growth in the size of news media is making such an intervention increasingly less necessary. Despite its importance, it is an issue that only interests media policy experts. We concluded that party positions would not be salient or fixed on all of four issues.

The script for the non-partisan scenario (not the audio clip) was evaluated by 16 college students before the experiment. Testers found the commentators’ arguments not skewed ideologically.

Experimental Procedure

We recruited college students who were enrolled in two elective classes. A total of 200 hundred students opted to participate in the experiment. The participants received an additional credit for participation. Participants were randomly assigned to the two scenario groups. Each group had 100 participants. The experiments were conducted on May 17 and 20, 2022. Participants listened to the audio clips in a classroom through a loudspeaker. The participants were told that they were evaluating an audio clip of a pilot episode for a new radio talk show. After listening to the clip, the participants completed questionnaires that asked about their ideological orientation (using a five-point Likert-type scale: conservative, somewhat conservative, neutral, somewhat liberal, liberal), political involvement, credibility (perceived), and ideological orientations of the commentators (five-point Likert-type scale: conservative, somewhat conservative, neutral, somewhat liberal, liberal), and how much they agreed with the positions of the two commentators on each issue (five-point Likert-type scale: strongly agree, somewhat agree, neutral, somewhat disagree, strongly disagree). Additionally, the questionnaire contained two questions for each topic (eight in total) to measure how much the participants had learned from listening to the discussion. In addition, the learning scores were calculated separately for each commentator. The sum of the correct answers was used as the learning score. Additional information on gender and media usage of the participants was also collected. The respondents were debriefed after the experiment was completed.

A total of 26 cases were not included in the final analysis due to incomplete answers - 16 cases from the partisan scenario group and 10 cases for non-partisan scenario group. In the end, the data of 174 cases – 84 for partisan scenario group, 90 for non-partisan scenario group, were used for statistical analysis.

Analysis

To test the effect of partisan cues and affiliations on the participants’ perceptions of each commentator’s ideological orientation, the study carried out a paired sample t-test on the commentators’ perceived ideology. The effect of the commentators’ perceived ideological orientation on attitudes toward the commentators’ arguments was tested using independent sample t-tests. We divided participants into three ideological groups. Those who answered “conservative” and “somewhat conservative” were assigned to a group labelled “Conservatives,” and those who answered “liberal” and “somewhat liberal” were assigned to the “Liberals” group. The remaining participants were labeled “Neutral.” We carried out two Multivariate Analysis of Variance (MANOVA) tests separately for the Liberals and Conservatives with 8 variables for attitudes towards each speaker as independent variables. In addition, we conducted two additional independent sample t-tests on the learning scores to determine whether there was any difference in the learning scores between the scenarios (the perceived ideological orientation of the commentator) and participants’ ideological orientation. SPSS version 28 was used for statistical analysis.

Results

In recent years, gender difference in ideological orientation among young generations, especially among those in their 20s, has been repeatedly reported by surveys. Such a difference was also observed in the last presidential election where 58.7% of male voters in their 20s voted for the conservative candidate, while 58% of female voters of the same age group chose the liberal candidate (Han, 2023). We compared gender difference in ideology in order to check whether the experiment participants show similar tendencies to the general population.

As is shown in Table 1, female participants showed significantly different ideological inclinations compared with male participants. Gender differences in ideological orientation corresponded to the general population, where young women are reportedly more liberal than young men (Han, 2023).

To check whether the stimuli functioned as expected, we conducted two paired sample t-tests of the perceived differences between the ideologies of the commentators for each scenario. Table 2 summarizes the test results.

As shown in Table 2, the first commentator, who had an affiliation to a supposedly conservative organization, was perceived as significantly more conservative in the partisan scenario than the second commentator (mean -1.381, p < .001). The perceived ideology score of the first commentator, who had an affiliation to a presumably conservative entity, was 3.61 on a 5-point Likert scale, while that of the second commentator, who had had an affiliation to supposedly liberal organization, was 2.23; there was a 1.38 points difference on average. In the non-partisan scenario, the perceived ideology score of the first commentator was 3.12, and that of the second commentator was 2.84; there was only a .278 points difference in the scores between the two commentators. The results clearly showed that participants in the experiment perceived commentators’ ideological inclinations differently in the partisan scenario, while there was no statistically significant difference in the non-partisan scenario. This result confirms that partisan cues influence perceptions of the ideology of commentators.

As can be seen in Table 3, Liberals showed a significant difference in attitudes towards speakers’ comments based on scenarios=, (F = 2.202, Wilks’ Λ = .747, partial η² = .253, p < .05). Table 4 shows that the difference is driven mainly by the different attitudes towards two commentators’ arguments on issue 4. On the fourth issue, in the case of the revision of the Broadcasting Act, the Liberals evaluated the first commentator more negatively when she was seen as politically conservative (MD = 1.69, SD = 0.74) than when she was perceived as neutral (MD = 2.38, SD = 0.94), F = 0.897, p < .05, while they evaluated the second commentator more favorably when she was perceived as liberal (MD = 2.53, SD=0.80) than she was seen as neutral (MD = 1.87, SD = .88), F = .703, p < .05. There were no significant differences in other issues.

Tables 5 provides a summary of a MANOVA test for the Conservatives. Conservatives showed no significant difference at the .05 significance level in attitudes towards speakers’ comments based on scenarios, F = 2.113, Wilks’ Λ = .686, partial η² = .314, p = .059. As Table 6 shows, in the case of the amendment of the Aggravated Punishment Act, the Conservatives evaluated the first commentator more positively when she was perceived as conservative (MD = 3.33, SD = 0.87) compared to when she was seen neutral (MD = 2.23, SD = 1.23), while they evaluated the second commentator more unfavorably when she was perceived as liberal (MD = 1.0, SD = 1.06) than when she appeared as neutral (MD = 1.86, SD = 1.17). There was no significant difference in other issues.

From the results, we concluded that there is not enough evidence that people tend to judge the opinions of a person with the same ideological disposition more favorably and those of a person with an opposite ideological propensity unfavorably. Thus, Hypotheses 1 and 2 were not supported.

Two additional paired-sample t-test were carried out to test differences in learning. The analyses did not reveal any statistically significant differences between the groups. Table 7 summarizes the results of the t-test on learning between the Liberals and the Conservatives in the partisan scenario.

Discussion

The results confirmed that the respondents perceived commentators differently depending on their affiliations, irrespective of the content of their comments. Participants in the partisan scenario judged the first commentator, who belonged to a presumably conservative organization, to be significantly more conservative than the second commentator, who presumably belonged to a liberal organization.

To answer the research question, contrary to our expectations, the findings suggest that forced exposure to balanced opinions leads to formation of balanced and informed opinions, reducing the possibility of echo chamber. The analyses failed to support either H1 or H2. While MANOVA test results were significant in case of Liberals, they were not significant for the Conservatives; the p-value (.059) was slightly above the .05 significant level. The additional analyses of individual issues revealed that in only one out of the four issues, both the “Liberals” and the “Conservatives” judged the comments of a commentator who had the same perceived ideology more favorably. Both of the groups judged comments from a commentator with the opposite ideology more unfavorably on the same issues.

However, it seems to be premature to dismiss the possibility of partisan decision making entirely while watching such broadcasting shows. The analyses did find that the Liberals and the Conservatives in our experiment behaved as the research hypotheses anticipated in two issues: the liberals in the last issue, and the conservatives in the third issue. It is unclear why the Liberals demonstrated expected behaviors only for the fourth topic or why the Conservatives acted as they did on the third topic.

The final issue relates to media regulation reform. As discussed above, the conservative commentator argued for the abolition of institutional entry barriers to the TV news market by removing approval from the regulator requirement. The deregulatory approach to media regulation is in line with right-wing parties’ traditional attitudes. On the contrary, left-wing parties tend to emphasize public obligations, especially of public broadcasters, and prefer a more regulated broadcasting market. In this sense, we can argue that the Liberals in our experiment showed a similar attitude toward media deregulation to that of liberal-minded population.

One possible explanation for this outcome is that people do not always make partisan judgements on all types of policy issues. In other words, the activation of partisan attitudes may depend on the issue itself. Currently, we do not have enough information to guess which issues have a higher chance of activating partisan judgement. Future studies with more diverse types of issues may provide further insights into this question.

Conclusion and Limitations

To summarize, this study was able to confirm that a simple partisan cue of a commentator’s affiliation was sufficient to perceive them as having a certain ideological inclination. The experiment, however, was not able to find sufficient evidence that people tend to evaluate the comments of a commentator whose perceived ideology coincides with their own ideological inclination more favorably and vice versa, regardless of the content of their arguments. Participants’ perception of a commentator’s ideological propensity influenced their opinion of the commentator’s arguments only in two out of eight cases. Thus, the hypotheses are not supported.

The findings of the current study suggest that broadcast news shows with commentators representing positions of rival political parties at least do not automatically trigger partisan attitudes in the audience even when commentators’ ideological propensities are clear. The experiment participants evaluated commentators’ arguments mainly based on the message, not the perceived ideology of the commentator. A popular broadcast show format with two opposing partisan commentators may help the audience form well-informed and balanced opinions, as was intended.

However, we cannot not rule out the possibility of partisan decision making entirely. In two cases out of eight, participants favored arguments of a commentator with a similar ideology inclination and disfavored those of a commentator with opposing ideology. The current study could not identify the type and the characteristics of issues that trigger this connection. Additional experiments with alternative designs are needed.

Future studies should consider a more diverse set of issues. Contrary to popular perceptions, South Koreans do not show clear differences in policy preferences based on their ideological orientation on most issues except those related to North Korea and national security (Park et al., 2012). Future studies could include issues related to the North–South Korea relationship and national security. In addition, this study avoided political issues out of fear that they may trigger partisan message processing more easily. However, it is on political issues that commentators usually show the greatest differences. Issues that are directly related to the two major political parties may trigger stronger reactions.

Future research could also differentiate issue salience. We deliberately avoided issues that received significant public attention. Issues on which people are likely to have strong existing attitudes may trigger different results compared to less well-known issues.

The participants in the experiment were recruited exclusively from a cohort of college students. A limitation of this practice is that people in their 20s tend to have relatively weaker party loyalty or ideological orientation. In addition, they tend to be less interested in politics. Future studies would benefit from recruiting participants from diverse age groups.

The study simulated a popular TV or radio show format as closely as possible because our interest was as theoretical as it was practical. Stimuli with alternative message delivery designs may produce different results.