The day after the election, the hate and darkness got to be too much. I unfriended nearly everyone. Now I have three friends: my husband and children…. I do hold (Facebook) responsible for helping to spread disinformation that contributed to both the election results and the level of vitriol and hate in this country. I am building my own wall now.
—SUSAN BRUBAKER KNAPP, QUOTED IN SELYUKH (2016)
IN NOVEMBER 2016, National Public Radio asked listeners how their attitudes toward social media were changing in the wake of the presidential election. As the epigraph for this chapter suggests, a majority of the responses described difficulty focusing, questioning of friendships, and overall feelings of frustration and exhaustion. Some individuals explicitly noted that they were turned off by the incivility and vitriol expressed by friends and family on Facebook. As reporter Alina Selyukh (2016) wrote, “People are turning off TVs (one even canceled her cable—mass media are not off the hook, either), deleting social apps from their phones, rationing time spent on Facebook and Twitter, and shrinking their digital friend lists.” However, Selyukh also noted that another set of respondents expressed regret at not being more outspoken on social media and vowing to speak their minds more during Trump’s presidency. What led to these two opposing outcomes? Why did some people stop looking for information on Facebook while others turned to it even more frequently to air their views and learn about politics?
These divergent paths are the result of the different experiences the conflict-avoidant and conflict-approaching have in the face of incivility. Conflict orientation shapes media choice; we consciously and unconsciously associate certain types of media or specific media outlets with varying levels of conflict and incivility. Because politics is an inherently conflictual domain, conflict orientation affects the frequency with which people choose political news over entertainment. Once Americans decide to tune in, they are constantly faced with the decision of which outlet or which program to choose. Based on headlines, Twitter comments, or blog posts, do they click the link to a news story or keep scrolling? Do they watch The Bachelorette or the national news? Once they decide to watch a news program, what do they do when exposed to political incivility? In this chapter, I evaluate evidence for the information-search hypothesis that these choices are in part a function of the interaction between conflict orientation and the incivility we are exposed to in political programming.
WHAT TO WATCH?
In her NPR story, Selyukh noted that many people were canceling cable, turning off their televisions, and deactivating Facebook—they were exiting the political media environment entirely. With so many channels, websites, streaming services, and other media available, an increasing proportion of Americans choose not to engage with political media at all (Arceneaux and Johnson 2013; Arceneaux, Johnson, and Cryderman 2013; Prior 2007). Recent changes in the media environment suggest that viewers engage in a two-step process when making consumption decisions. First, they decide whether to tune in to political news. Then, if they choose news over entertainment, they select a program or channel on the basis of a variety of characteristics, including partisanship and the amount of conflict they will likely be exposed to once they tune in.
The decision to select political news over entertainment is in part a function of individual characteristics, especially political interest (Prior 2007). Looking at data from a 2010 Pew survey about media habits, Arceneaux and Johnson (2013) found substantial information differences among four types of audiences: the inattentive, broadcast-news viewers, Fox viewers, and MSNBC viewers. Those who choose to watch political news enjoy it more than their inattentive counterparts; they are more likely to report following major political issues, are more likely to vote, and answer more political knowledge questions (e.g., identifying the majority political party in the U.S. House of Representatives) correctly. Cable-news viewers are even more engaged with politics and political information. About 80 percent of talk-show viewers reported that they enjoy keeping up with the news “a lot,” while only 62 percent of broadcast-news watchers said the same (Arceneaux and Johnson 2013). In other words, people who tune in to political news enjoy and are more interested in politics than their inattentive peers. In chapter 2, I offered evidence that conflict orientation is correlated with political interest, such that people who are conflict-approaching are also more likely to be interested in politics. Given this relationship, we might also expect that the conflict-approaching are more likely to pay attention to political news across all formats and platforms. Indeed, Arceneaux and Johnson (2007) found that, when given a choice, the conflict-approaching will spend more time watching the uncivil punditry of Fox’s Hannity & Colmes than their conflict-avoidant peers. Based on this research, the tune-in hypothesis proposes that the conflict-approaching will be more likely to watch, listen to, or read political news than their conflict-avoidant friends.
Tune-in hypothesis: Conflict-approaching individuals will report consuming political media generally at a greater frequency than their conflict-avoidant counterparts.
Once an individual decides to tune in to political news, she is immediately faced with a new conundrum: where does she want to turn for that news? The increasingly diverse media environment spurred by the growth of cable and Internet sources only increases the likelihood that individuals will be aware of, and use, information about the tone of a show or format to make decisions about their media consumption. Communication scholars have long investigated the “gratifications” that attracted audiences to certain media and the ways in which individuals use media to fulfill social and psychological needs (Cantril 1942; Katz, Gurevitch, and Haas 1973; Papacharissi 2007; Ruggiero 2000). Rubin (1994) differentiates between ritualized, habitual use of media and instrumental use, in which people watch a show because it offers useful information or allows them to engage in social discussion of that show with their friends or peers. Those who have greater exposure to or affinity for media are more likely to engage in habitual media use, while more specific motives drive instrumental use of media. Applying uses-and-gratifications theory to the dynamic between conflict orientation and mediated incivility suggests that the conflict-avoidant and the conflict-approaching use media differently and receive differential benefits from that use. At minimum, given the emotional responses seen in chapter 3, the conflict-avoidant are less likely to feel psychologically fulfilled after being exposed to incivility than the conflict-approaching. Therefore, when choosing media to consume, they will actively avoid outlets or platforms that they know might increase their likelihood of exposure.
Building on uses-and-gratifications theory (or at least similar foundational assumptions), political scientists have also started to theorize about how and why individuals make their political news choices. Prior notes that “more choice leads to better sorting of the television by taste” (2007, 95). As a result, Americans are able to choose whose opinions they want to hear, exacerbating knowledge gaps and echo chambers. In an increasingly choice-driven media environment, academics and public figures have become concerned about partisan selective exposure—the individual’s decision to watch, read, or listen to like-minded media outlets. While evidence suggests that media are the venue most likely to expose citizens to diverse viewpoints, other research has shown that Americans select news reports based on the perceived affinity between their political preferences and those of the particular media outlet (Iyengar and Hahn 2009; Mutz and Martin 2001; Stroud 2011).
Partisan selective exposure is a concern because of its potential to increase polarization among the electorate (Stroud 2011; Sunstein 2009) and reduce cross-cutting exposure, or opportunities to “hear the other side” (Jamieson and Cappella 2008; Mutz 2006). However, others have argued that these findings are in part an artifact of experimental design: randomized exposure to ideologically biased programming leads to further entrenchment of attitudes in situations where individuals are forced to watch counter-attitudinal programming. When participants are given the ability to choose, they are found to be more open to arguments from across the political spectrum (Arceneaux, Johnson, and Cryderman 2013).
Arceneaux and Johnson’s (2013) active-audience theory specifically incorporates choice in the media environment into models of the relationship between media coverage and individual attitudes. Their theory rests on two propositions. First, an active audience is scrutinizing and evaluating the messages it receives, engaging in motivated reasoning to accept viewpoints that fit in their worldview and to dismiss those that challenge their beliefs. Second, their active audience makes “purposive viewing decisions” (10). Entertainment seekers select out of the news environment, diluting the effects of highly partisan cable talk shows by removing the “inadvertent audience.” The effects of partisan news coverage on those who remain—those who chose to watch broadcast or cable news—will be different from the effects on the inadvertent audience. Many of the same concerns can be raised about exposure to incivility and conflict in political media. If you never watch talking heads shout at one another on cable television, being exposed to that sort of exchange in an experimental context could be more startling or remarkable than if you regularly watch cable shows. Incivility’s effects on willingness to hear the other side, consumption of political information, and beliefs about government legitimacy could be exacerbated by an individual’s ability to choose to be exposed to it, or that ability to choose might actually mitigate incivility’s problematic impact on political behavior. Arceneaux and Johnson’s (2007) findings provide some support for both sides of the question. On one hand, the negative effects of televised incivility on trust go away once people can avoid being exposed to it. On the other, even those who chose to watch incivility felt as though they had less say in government functions—less internal efficacy—than those who were not exposed.
Citizens select the media they use to gather political information on the basis of an outlet’s partisan congruence, but also on a particular format’s likelihood of introducing information in an uncivil, confrontational manner. For some citizens—the conflict-avoidant—the fact that a media personality is known for her vitriolic approach to the news will dissuade them from seeking her perspective. Recall my students watching the Hannity clip. Some remarked in class discussion that they never watched cable news because they knew that it would lead to the kind of finger-pointing and name-calling they had just witnessed. For others, this knowledge was enough to draw them in and make them regular viewers. While some of us are capable of making this distinction on a show-by-show basis, others form more general, format-based assessments of the likelihood of exposure to incivility. In other words, Americans’ preferences for newspapers or cable television are shaped by their perceptions that these types of media are more or less uncivil.
In order to assess the extent to which individuals choose particular media environments on the basis of their likelihood of being exposed to incivility, we must first categorize different types of media outlets as either high- or low-conflict. Because the swath of possible media outlets is so vast, I limit my categorizations to the platform level, seeking to distinguish cable television from network television, and Internet sources like blogs from social media like Twitter, but without delving into the differences between, say, Hannity and the Nightly News with Lester Holt. By virtue of their breadth, these categories are naturally a bit unwieldy, but they are grounded in research into the presence and perception of conflict and incivility on each platform.
Just as citizens intuitively (albeit at times incorrectly) sort media outlets into liberal or conservative camps, they also believe that certain types of media are more likely to use uncivil tone and language than others. These perceptions are partially a function of the attributes of each platform; individuals perceive visual and audio media as more uncivil than textual media even when the message is the same (Sydnor 2018). The sources used, topic of an article, and outlet—blog or news site—can influence perceptions of incivility in the same message (Anderson et al. 2014; Coe, Kenski, and Rains 2014; Thorson, Vraga, and Ekdale 2010). However, some media are also more likely to present uncivil messages than others. Sobeiraj and Berry (2011; Berry and Sobeiraj 2014) developed thirteen categories of language and expressions that comprise outrageous behavior to reach their conclusion that cable news and talk radio are much more likely to introduce uncivil discourse than blogs (Berry and Sobeiraj 2014, 45). Papacharissi (2004) also investigates incivility in online forums and, even when separating uncivil and impolite talk into two categories, finds that around 50 percent of messages contain either impolite or uncivil commentary. Looking at comments posted in response to newspapers’ online articles, Coe et al. (2014) found that more than one in five comments on online articles (22 percent) written for the Arizona Daily Star were uncivil. Research suggests that cable television, talk radio, and Internet sources are the most likely venues for uncivil political communication.
The Internet has been identified as particularly hospitable to incivility. As Chen (2017) writes, “the de-individuation of online discourse along with the lack of conversational cues plus the speed with which a comment can go public and viral online foments a perfect storm of sorts for incivility to flourish and cause harm” (64). Deindividuation, or the subsuming of one’s individual identity to a group identity, makes people feel more connected to their weaker social ties than they really are, leading them to share personal information that they might not typically share face to face. However, this same tendency can make individuals online feel freer to act in socially undesirable ways. Just like Eagles fans who smashed windows and overturned a car after their team’s 2018 Super Bowl win, commenters online feel more uninhibited because there is a perception that their words are more difficult to trace back to them. Many comment streams or Internet forums allow participants to post anonymously, which has been found to increase the amount of incivility in the comments section (Santana 2014), but even when people have to sign their own name or link to their Facebook account, they feel hidden in the vast expanse of the web (Chen 2017).
This sense of deindividuation is more complicated on social media. Some 67 percent of American adults report getting at least some of their news from social-media sites (Shearer and Gottfried 2017). Although these sites primarily link to other outlets such as blogs and traditional news organizations’ websites, networked individuals frequently share these links while adding their own opinions, commentary, or reactions. Incivility is alive and well on platforms like Twitter and Facebook, where journalists, politicians, and thousands of other people regularly experience gendered harassment, flaming, and trolling (Chen 2017; Oz, Zheng, and Chen 2017). However, affordances of different social media lead to variation in the presence of incivility. For example, Oz, Zheng, and Chen (2017) found that posts responding to White House tweets were more likely to be uncivil than posts responding to the White House on Facebook. In part, this is because users control who they are “friends with” on Facebook or “following” on Twitter, and these connections are frequently driven by real-life acquaintance or shared interests. While these sites might slightly increase the diversity of one’s social network (Kim, Hsu, and de Zúñiga 2013), other research suggests that Twitter users, in particular, tend to seek out clusters of other users who are politically homogeneous (Himelboim, McCreery, and Smith 2013). Though homogeneity does not preclude the use of uncivil language, in many cases it reduces the presence of incivility. This tendency toward like-minded networks allows me to categorize social media as a low-conflict source of political information, even while the Internet more generally is labeled high-conflict.
While “new” media like cable, talk radio, and blogs are frequently uncivil, traditional forms of political media do not demonstrate the same level of rude and impolite language. Berry and Sobeiraj (2014) use newspapers as a control in their evaluation of outrage on blogs and cable television, finding that newspaper columnists are likely to have about six instances of outrage in a column (in comparison, cable television and radio segments contain, on average, twenty-three to twenty-four incidents). Cable programming is frequently tied to political incivility (York 2013; Gervais 2014), whereas network news has been found to minimize individuals’ assessments of politics as uncivil (Forgette and Morris 2006).
To find further evidence that cable and network news differed in their level of incivility, I conducted content analysis of 666 television news segments from MSNBC, Fox, CNN, NBC, and ABC.1 I assessed the presence of civility and incivility by looking for language similar to that coded by Sobieraj and Berry and recorded whether each source, throughout the course of the segment, used any of three civil and four uncivil communication strategies. Specifically, I measured the presence or absence of the following civil approaches: indication that an opponent’s policies would positively change American values or institutions, acknowledgment of common ground, and use of complementary language or praise of an opponent. Incivility was coded as present if a source placed blame on his or her opponents, used hyperbolic language to characterize his or her opponent (“outrageous”), used pejorative language (“racist,” “liar”), and/or described the opposition with derogatory adjective (“reckless,” “weird”). From these seven items, I created additive measures of incivility and civility for each segment, counting the total number of uncivil and civil incidents present, regardless of type.
In this analysis, certain types of uncivil language are used more frequently across all television news outlets, while other types are much more prevalent on cable news. Table 4.1 shows the percentage of segments that contain at least one use of each type of language. While not approaching Sobeiraj and Berry’s finding of outrage incidents in 100 percent of the cable-television news sample, I did find that 70 percent of my sample contained at least one of the five measures of uncivil language. Sources were most likely to use blame in dealing with their opponents, and least likely to accuse others of lying. There was substantially less evidence of civil language, with only 18 percent of the total sample demonstrating any of the three types of civility coded: praise for the other side, acknowledgment of common ground, or support for American values. Of those three, sources were most likely to acknowledge common ground with their detractors.
TABLE 4.1 PERCENTAGE OF SEGMENTS WITH CIVIL OR UNCIVIL LANGUAGE, BY MEDIA OUTLET
| |
FOX |
MSNBC |
CNN |
ABC |
NBC |
TOTAL |
| Any Uncivil Language |
78 |
80 |
68 |
55 |
50 |
70 |
| Blame |
63 |
54 |
48 |
31 |
39 |
51 |
| Hyperbole |
37 |
54 |
30 |
24 |
24 |
34 |
| Accusations of lying |
17 |
24 |
9 |
6 |
2 |
12 |
| Name-calling |
39 |
47 |
21 |
22 |
19 |
28 |
| Threatens American values |
26 |
35 |
20 |
14 |
11 |
22 |
| Any Civil Language |
17 |
23 |
20 |
20 |
7 |
18 |
| Praise |
7 |
5 |
7 |
2 |
4 |
6 |
| Common ground |
11 |
18 |
14 |
12 |
4 |
13 |
| Bolsters American values |
4 |
4 |
4 |
6 |
2 |
4 |
Table 4.1 records what percentage of segments contained particular types of civility and incivility. It does not capture how frequently multiple forms of incivility were present within the same segment. It is one thing to blame one’s opponents for political problems, but another to do so while also calling them lying idiots who are destroying America. Segments were not likely to contain multiple identifiers of civil language. On average, across all five news outlets, a segment contained less than one incident of civil language (
). However, when incivility was used, it was frequently used multiple times. Cable and network news averaged two types of uncivil language per segment (
). However, the distribution of uncivil incidents per segment is strongly right-skewed; while the averages for both types are relatively low, 25 percent of segments contain more than three uncivil incidents.
There is also variation in the prevalence of highly uncivil segments across media outlets, with Fox and MSNBC containing averages of 3.2 and 2.5 incidents per segment, respectively. CNN, NBC, and ABC contained, on average, 1.7 or fewer incidents of uncivil language.2 Cable networks had 10 percent more overall uncivil language than network television. These results suggest that network news programming is more civil than the shows on cable news. This distinction reinforces previous findings that television news is highly uncivil and provides clear evidence that it is cable, not network television, that drives this association.
To summarize, the research outlined here establishes that political media can be divided into high- and low-conflict sources. Specifically, cable television, Internet news and blogs, and talk radio are classified as high-conflict, while newspapers, network television, and social media are low-conflict. These categorizations can be summarized as follows:
CATEGORIZATION OF MEDIA BY LEVEL OF CONFLICT
| High-Conflict Media |
Low-Conflict Media |
| Cable television |
Newspapers (including their online versions) |
| Internet news/blogs |
Network television |
| Talk radio |
Social media (Twitter, Facebook) |
Survey data suggest that citizens perceive these differences in incivility across media platforms and blame cable, Internet, and talk radio for the increase in political incivility (Weber Shandwick, KRC Research, and Powell Tate 2013).
Individuals will seek to match their attitudes toward conflict to media that fit their predispositions. Those who are entertained by incivility and argumentation can select the high-conflict media programming found on cable television, online blogs, and talk radio, while their conflict-avoidant counterparts can seek information from low-conflict environments such as newspapers, network television, and social media. Given the widespread public perception that politics lacks civility and the argument that the media are to blame for this problem, uses-and-gratifications theory, active-audience theory, and research on selective exposure suggest that certain political coverage appeals to people who have a taste for argument and incivility. Those who do not like this conflict will turn away from media coverage of politics (per the tune-in hypothesis) or turn to a few select platforms or outlets. Given the distinctions between types of media laid out here, I expect conflict-avoidant individuals to prefer low-conflict media formats, while the conflict-approaching turn to high-conflict programming.
Media-choice hypothesis: Conflict-approaching individuals will report preferences for high-conflict sources, such as Internet-only sources and cable television. Conflict-avoidant individuals will report preferences for media outlets that could be perceived as more civil—specifically, network television and newspapers.
Given this understanding of selective exposure to incivility, the conflict-avoidant have two exit points on their path to learning about politics: they can choose not to tune in to politics in the first place, or they can tune in but select specific outlets or platforms on which they are less likely to experience vitriolic language. On the one hand, this sounds great—only those people who can tolerate incivility will be frequently exposed to it—but disparities in media consumption have been shown to exacerbate inequalities in political knowledge and electoral turnout (Prior 2007). When people decide not to listen to political radio or catch the evening news, they learn less about what is going on in politics and are far from the ideal democratically engaged citizens. On the other hand, those who persist in tuning in to politics are not exposed solely to the types of information best suited to learning about politics. In some cases, they are exposed to the very thing they were trying to avoid, and this exposure changes the way they look for and interpret future information. Once individual feel anxious, they tend to seek out and remember threatening information about that policy at the expense of other issues, thereby applying an unbalanced set of considerations to their evaluation of a policy or political situation (Albertson and Gadarian 2015). Reading an article that makes one anxious about immigration, for example, will lead to additional search for information about the threats posed by immigration. It stands to reason that in the face of incivility, those who are content to experience conflict will take a different approach to processing information than those who want to avoid it.
To assess these divergent approaches, scholars have focused on information processing in the absence of choice. In other words, once individuals are exposed to incivility, how does this change what they learn or what information they seek out next? Research suggests that incivility plays a role in one’s ability to find and recall political messages. Minich, Mendoza, and Brown (2018) have found that when an argument is framed civilly, participants are more likely to click on subsequent links to news stories. While civility appears to keep people engaged with news longer, studies of its effects on recall are mixed. In comparison with those who watched a civil campaign ad, Brooks and Geer (2007) found a small, not statistically significant decrease in factual recall among those who were exposed to an uncivil ad. However, Mutz (2015) found that incivility increased attention to content, substantially increasing individuals’ recall of issue arguments. Incivility frames political material in a way that is—for some people—more entertaining (Jasper 1998; Lepper 1994; Mutz 2015). Clearly, then, framing political information in an uncivil manner changes the way in which individuals seek out or avoid additional political news.
When given the choice, the conflict-avoidant are expected to turn away from incivility and the conflict-approaching to move toward it. But once both groups have been exposed to uncivil political news, research in political psychology and communication suggests that their behavior will run counter to our expectations and intuitions. Rather than turning away from incivility, the conflict-avoidant will lean in, seeking out more of the very thing that makes them uncomfortable. This biased information processing could stem in part from the anxiety the conflict-avoidant feel as a result of incivility. Just as the individual who is worried about immigration cannot stop reading news about the threat posed by immigrants, the conflict-avoidant consumer cannot stop watching people shout at one another and use derogatory language.
Alternatively, this tendency could be a result of the linkage between conflict-avoidance and the Big Five trait agreeableness. As discussed in chapter 2, conflict-avoidant individuals are more likely to score highly on agreeableness, and therefore to be naturally good-natured and cooperative. Agreeable individuals are susceptible to demand effects: they continue to do things not because they enjoy them, but because the tasks are expected or because they will improve social harmony. They may see exposure to political incivility as the expected result of turning on the news, and therefore continue to seek it out because that is what is expected of them as political news consumers. The studies presented here cannot fully explore which of these mechanisms most likely shapes the relationship between conflict-avoidance, incivility, and information search. Instead, I focus on the primary and somewhat counterintuitive hypothesis that the conflict-avoidant will search for more incivility after being exposed to it in political media.
A different body of research helps us understand why the conflict-approaching, faced with incivility, might engage in a type of satisficing behavior, essentially saying “OK, that’s enough, I can turn to other things now.” Mood-management theory suggests that people turn to media to regulate their broader emotional state. Individuals who are in a good mood are less concerned about whether the media they are about to consume have a positive or negative tone, whereas those who are in a bad mood look for media that will improve it (Knobloch-Westerwick 2006). For example, Knobloch (2003) finds that people experiencing positive affective states spend more time on positive web content than those with mediocre moods, but less time with the same content than those in negative moods. In the previous chapter, I demonstrated that the conflict-approaching are more likely to experience positive emotions in the face of incivility than their conflict-avoidant peers. To these individuals, uncivil coverage of politics is perceived as positive content; it is something they enjoy spending time viewing. This is not to say they dislike civil political news, but because they are already in a good mood in the wake of uncivil political media, they will put less effort into seeking out media options that will maintain that mood. Sufficiently entertained by the uncivil news they have been exposed to, the conflict-approaching look at less incivility and less political news because they have already met their media needs.
Information-search hypothesis: The conflict-avoidant will be more likely to seek out uncivil media clips after being exposed to incivility. The conflict-approaching will turn to civil news after being exposed to incivility.
TESTING THE EFFECTS OF CHOICE
To test the first two hypotheses—that citizens seek to expose themselves to uncivil media only to the extent that their conflict orientation allows—I use data from Project Implicit and the first Mechanical Turk (MTurk) study. Each survey asks a set of questions from the Conflict Communication Scale and invites participants to report their typical media consumption and political participation in the past year (see appendix A for specific survey questions).
The two sets of studies take different approaches to measuring patterns of media use. Project Implicit focuses on daily consumption patterns. I asked participants, “During a typical week, how frequently do you watch/read/listen to…network television/cable television/radio/ Internet/newspaper?” Participants were asked to select from a range of zero to seven days. I use this measure to estimate the amount of time spent tuning in to political news, constructing an average of the number of days a week that participants consumed any of the five types of news. A lower average amount of weekly consumption of any news suggests that participants were choosing other activities or entertainment over political information. This measure offers a general understanding of the extent to which individuals are selecting whether or not to consume political information, in addition to providing comparative consumption statistics for different types of media.
Frequency measures like those used in the Project Implicit study are sometimes challenged as an effective measurement technique because they are difficult for individuals to report correctly (Price 1993; Schwarz and Oyserman 2001). The belief is that these self-reporting measures lead to inflated reports of media usage. Dilliplane, Goldman and Mutz (2013) argue that rather than focusing on the frequency of study participants’ media consumption, as in the Project Implicit study, researchers should ask participants to select their regularly viewed programs from a list. The program-list approach reduces the cognitive burden on participants and is found to provide a more reliable estimate of actual media consumption (Dilliplane et al. 2013). Therefore, in MTurk Study 1, I asked participants to respond to the question “Which of the following is your main source of political news and information?” Participants chose among eight options: Internet-only sources, newspapers, network television, cable television, radio, social media such as Facebook and Twitter, talking with others, or saying they did not really follow political news.3 They were then asked to report their second major source of political news and information. Although these questions are not an entirely accurate reflection of Dilliplane et al.’s program-list strategy, or of the choice environment in which participants select where to get their political news, it does reduce concerns about overreporting and encourages them to choose their preferred news source. Responses to this question shed light on whether conflict-avoidant and conflict-approaching individuals prefer different types of media as their primary source of political information.
Table 4.2 displays descriptive statistics for both methods of measuring media consumption. Across both sets of studies, the Internet is turned to most frequently as the preferred media source—an average of 4.5 days a week, and the first choice of 35.2 percent of the MTurk sample.4 Cable and network television are used at relatively similar frequencies (both are watched 2.9 days a week, on average) and are the most-preferred source of a similar percentage of participants (16.8 percent report a preference for cable television and 12.3 percent for network television). Radio ranks next, with Project Implicit participants tuning in around 2.7 days a week and only 5.7 percent of the participants in MTurk Study 1 reporting that radio is their preferred source for political information. The additional media choices offered to Mechanical Turk respondents—social media, talking to others, and not following political news—were collectively the preferred outlets of just over 10 percent of the sample.
TABLE 4.2 MEDIA CONSUMPTION HABITS OF THE SAMPLES
| |
AVERAGE NUMBER OF DAYS PER WEEK PARTICIPANTS REPORTED USE |
PERCENT REPORTING MEDIUM AS FIRST-CHOICE OUTLET |
| Internet sources |
4.5 |
35.2 |
| Newspaper |
1.8 |
19.6 |
| Network television |
2.9 |
12.3 |
| Cable television |
2.9 |
16.8 |
| Radio |
2.7 |
5.7 |
| Social media such as Facebook and Twitter |
— |
4.7 |
| Talking with others |
— |
3.1 |
| Don’t really follow political news |
— |
2.7 |
The major discrepancy across the two measurement strategies is in the prevalence of newspaper use and the number of individuals reporting that newspapers are their preferred source of news. Project Implicit participants read newspapers only an average of 1.8 days a week, making them the least used of the five sources. However, almost 20 percent of MTurk users reported that newspapers were their preferred news source, making it the second most popular media source after the Internet. This disparity likely stems from differences in the question wording. In Project Implicit, participants were asked about how frequently they read print newspapers. Mechanical Turk participants, on the other hand, were asked specifically to report their use of newspaper websites as part of the newspaper category, rather than as an Internet source. Therefore, people who visit newspaper websites were captured as using the Internet in the Project Implicit sample but as reading newspapers in MTurk Study 1.
CHOOSING POLITICS AND CHOOSING A SOURCE
According to the tune-in hypothesis, individuals first make a choice about whether to tune in to politics or do something else. This choice is influenced by a given individual’s conflict orientation. The conflict-avoidant should spend less time consuming political news than their conflict-approaching peers.
The Project Implicit study asks about how many days a week participants use different types of media, so we can use that question to examine whether the conflict-approaching spend more time watching, listening to, or reading political news. Looking first at the results of bivariate linear regression5 of media consumption on conflict orientation, it is clear that those individuals who are more comfortable with conflict are likely to consume all forms of media with greater frequency. A one-unit increase in conflict orientation (or a one-unit shift toward a conflict-approaching inclination) results in a very small increase in the number of days an individual reports consuming each type of media. As figure 4.1 shows, these small increases compound rapidly and lead the most conflict-approaching individuals to consume many forms of media one full day more frequently than their most conflict-avoidant counterparts. Each of these differences is statistically significant, with the exception of the two television categories.
FIGURE 4.1 Average frequency of media use, by conflict orientation.
Note: Linear predictions derived from bivariate regressions of media use on conflict orientation. Shaded areas represent 95% confidence intervals.
Source: Project Implicit.
These bivariate analyses offer support for my claim that people who are more comfortable with conflict choose to consume more political news; the regression coefficients are in the expected directions even when they fail to approach statistical significance. However, this analysis uses observational data and therefore requires the introduction of controls to increase our confidence that it is truly conflict orientation that is shaping these preferences and not some unobserved characteristic that is correlated with both conflict orientation and media preferences. In chapter 2, I established that conflict orientation is tied to a range of demographic and political characteristics that also are likely to affect the frequency with which one consumes political media; once I control for these demographic characteristics—interest, ideology, party identification and strength, age, gender, personality, and education level—conflict orientation’s effects on the frequency with which individuals are exposed to political media and their reported preference for particular sources disappear (see table 4.3). This could suggest that conflict orientation is not guiding political behavior, at least in the context of media consumption, or that the effects of conflict orientation are mediated by these demographic characteristics.
TABLE 4.3 REGRESSION OF CONFLICT ORIENTATION ON MEDIA CONSUMPTION, CONTROLLING FOR DEMOGRAPHIC INFLUENCES
| |
AVERAGE WEEKLY USE |
NEWSPAPER |
NETWORK TV |
CABLE TV |
RADIO |
INTERNET |
| Conflict orientation |
0.01 |
0.01 |
−0.0008 |
0.009 |
0.02 |
0.02 |
| |
(0.009) |
(0.014) |
(0.015) |
(0.013) |
(0.014) |
(0.013) |
| Political interest |
0.58** |
0.40** |
0.59** |
0.62** |
0.44** |
0.83** |
| |
(0.047) |
(0.074) |
(0.080) |
(0.072) |
(0.079) |
(0.070) |
| Party ID |
−037** |
−0.51* |
−0.08 |
−0.46* |
−0.23 |
−0.57** |
| |
(0.134) |
(0.212) |
(0.229) |
(0.206) |
(0.226) |
(0.200) |
| Party strength |
−0.11 |
−0.20 |
−0.33 |
−0.11 |
0.06 |
0.06 |
| |
(0.102) |
(0.161) |
(0.174) |
(0.157) |
(0.172) |
(0.152) |
| Age |
0.04** |
0.04** |
0.06** |
0.06** |
0.05** |
0.01* |
| |
(0.004) |
(0.006) |
(0.006) |
(0.006) |
(0.006) |
(0.005) |
| Education |
0.03 |
0.09** |
−0.08* |
−0.09** |
0.10** |
0.10** |
| |
(0.021) |
(0.033) |
(0.035) |
(0.032) |
(0.035) |
(0.031) |
| Hispanic |
0.22 |
−0.19 |
0.85** |
0.77** |
0.20 |
−0.50* |
| |
(0.163) |
(0.257) |
(0.277) |
(0.249) |
(0.273) |
(0.243) |
| Black |
0.55** |
0.05 |
1.48** |
0.88** |
0.48 |
−0.15 |
| |
(0.167) |
(0.263) |
(0.284) |
(0.255) |
(0.280) |
(0.249) |
| Extraversion |
0.02 |
0.01 |
0.04 |
0.02 |
0.03 |
−0.03 |
| |
(0.015) |
(0.024) |
(0.025) |
(0.023) |
(0.025) |
(0.022) |
| Agreeableness |
0.008 |
−0.02 |
0.04 |
0.02 |
0.003 |
0.02 |
| |
(0.022) |
(0.035) |
(0.038) |
(0.034) |
(0.037) |
(0.033) |
| Conscientiousness |
0.06** |
0.07** |
0.09** |
0.10** |
0.05 |
−0.03 |
| |
(0.016) |
(0.028) |
(0.030) |
(0.027) |
(0.030) |
(0.027) |
| Emotional stability |
0.02 |
0.001 |
0.07* |
0.004 |
0.0002 |
0.004 |
| |
(0.018) |
(0.028) |
(0.030) |
(0.027) |
(0.029) |
(0.026) |
| Openness |
−0.03 |
0.003 |
−0.11** |
−0.09* |
0.02 |
0.03 |
| |
(0.022) |
(0.035) |
(0.038) |
(0.034) |
(0.037) |
(0.033) |
| Constant |
−0.73** |
−1.37** |
−0.69** |
−1.23** |
−1.31** |
0.92** |
| |
(0.195) |
(0.309) |
(0.333) |
(0.300) |
(0.328) |
(0.292) |
| R2 |
0.18 |
0.17 |
0.24 |
0.27 |
0.21 |
0.27 |
| N |
993 |
993 |
992 |
992 |
993 |
993 |
For example, political interest is moderately correlated with conflict orientation and is likely to influence how regularly we look at political media.6 It is possible that conflict orientation will play a different role in the media habits of the extremely politically interested than it would for those not at all politically interested. Figure 4.2 displays the results of a regression of the frequency of media consumption on conflict orientation, political interest, and the interaction between the two.7 The figure demonstrates that the effects of conflict orientation on the frequency of media consumption change at different levels of political interest and across types of media. While there is no clear pattern across all types of media, conflict orientation influences consumption habits differently for those with different levels of political interest. In many cases, those who are very or extremely politically interested (the top two lines on each graph) look different from those who are not at all or somewhat interested in politics.
FIGURE 4.2 Media use changes with conflict orientation and political interest.
Note: Figures represent linear predictions from a multivariate regression that includes an interaction between interest and conflict orientation.
Source: Project Implicit.
Looking first at the frequency with which individuals at various levels of political interest report weekly consumption of any form of media, those who are somewhat interested in politics stand out as having conflict orientation shape their consumption. Somewhat-interested individuals who are the most conflict-avoidant report consuming political media on any platform 2.5 days a week on average, while the most conflict-approaching participants who are somewhat interested in politics report exposure to political media close to three days a week. The same pattern is true for the frequency with which the somewhat politically interested tune in to radio and Internet media. Those at the most approaching end of the spectrum report using the radio or Internet one full day more than their conflict-avoidant peers.8
Ultimately, the evidence suggests that conflict orientation plays a role in the decision to tune in to political media, but that it plays a much smaller role than other relevant characteristics, such as one’s interest in politics. Those who are interested in politics are more likely to spend their time consuming the news. But once they make that choice, they are faced with an array of outlets across many media platforms. The media-choice hypothesis proposes that conflict orientation also shapes behavior at this second decision point. If conflict-avoidant individuals know that they are more likely to be exposed to disagreement, shouting, and nasty talk on cable news than they are when reading a newspaper, they will turn to the newspaper. Therefore, we now turn to the analysis of Mechanical Turk Study 1, which asked people which types of media they preferred, rather than how frequently they watched each.
When media consumption is measured as a preference for a particular type of media, rather than the frequency with which individuals consume each type, the results more accurately reflect my hypotheses. As figure 4.3 shows, conflict-approaching individuals are more likely to list cable television, Internet-only sources like the Huffington Post, or cable television as their top source for political news. Conversely, they are less likely than their conflict-avoidant counterparts to state that network television or social media are their preferred news source. Participants’ preferences for newspapers did not follow the hypothesized pattern. I expected that conflict-avoidant individuals would express a greater preference for newspapers than the conflict-approaching, but the relationship appears to run in the other direction, with the conflict-approaching indicating a greater probability of choosing hard-copy newspapers as their preferred news source.
FIGURE 4.3 Probability of selecting each type of media as one’s preferred source.
Note: Participants were asked to name their preferred source of political information. Each figure represents the predicted probability of selecting a given medium as one’s first choice, given conflict orientation. Predicted probabilities were calculated from a bivariate logistic regression.
Source: Mechanical Turk Study 1.
Although these findings support my overall expectations, it is important to take them with some caution. Only one relationship is statistically significant: that between conflict orientation and the preference for network television. Here, the most conflict-avoidant participants have about a 25 percent chance of selecting low-conflict network TV as their favorite source of political news, while the most conflict-approaching choose network television only about 5 percent of the time. In some cases, small sample sizes make statistical accuracy difficult. As shown in table 4.2, 35 percent of participants reported that the Internet was their preferred source, and the remaining 65 percent spread themselves across the other seven media options. Once these groups are broken down by conflict orientation, there are only a few individuals in each category, leading to high standard errors in the analysis.
In an attempt to overcome this sampling problem, I analyzed the data from MTurk Study 1 again, this time with individuals placed in one of three categories based on their CCS scores. Following Mutz (2015), I divide the CCS into equal thirds. Those in the lowest third are conflict-avoidant, the middle is considered conflict-ambivalent, and the highest third are conflict-approaching. I then compare the average probability of choosing each news source for those who are highly avoidant versus highly approaching. Looking at conflict orientation in this way, Bonferroni estimates from a series of one-way ANOVAs suggest that the conflict-avoidant and conflict-approaching have different preferences for network television and social media. About 3.7 percent more conflict-avoidant than conflict-approaching individuals prefer network television (F(918) = 4.81, p = 0.008), while the conflict-avoidant are about 5 percent more likely to prefer social media (F(918) = 2.63, p = 0.072). These results lend further support to the initial bivariate estimates, and offer weak evidence that conflict orientation drives individuals’ preferences for particular media platforms.
Ultimately, these findings offer only weak support for my expectations that conflict orientation guides individuals in two key decisions: whether to consume political media and, if so, what media platforms to turn to. Each set of measures has its own weaknesses and is a relatively blunt measure for assessing individuals’ specific media choices at the program level. Furthermore, neither directly assesses individuals’ choices in relation to their exposure to incivility, instead relying on categorization of platforms as more or less uncivil. Therefore, the next section focuses on the information-search hypothesis, examining how individuals’ media consumption and information-searching behavior change when they are exposed to political incivility.
SELECTIVE EXPOSURE AND THE SEARCH FOR INFORMATION
Evidence from the observational data is mixed about the extent to which conflict orientation guides individuals’ choice to consume political news or their choice of media platforms. However, the third hypothesis proposes that conflict orientation’s influence on information-search habits is activated upon exposure to political incivility. Conflict-approaching individuals will be satisfied by this initial experience with incivility and seek out less news and less uncivil news than they would if they watched a civil political clip first. The conflict-avoidant, on the other hand, will continue to look for information about politics and, ironically, will be drawn to incivility regardless of whether they are looking at politics or entertainment.
Mechanical Turk Study 2 was designed to evaluate this hypothesis. The experiment assessed media choice in the style of Arceneaux and Johnson (2013). After participants reported their preferences for news, entertainment, civil media, and uncivil media, they were asked to make a choice between four headlines that varied on each of the two dimensions—civility and news/entertainment. The control group made this choice without forced exposure to any additional media; the treatment groups were first asked to watch either a civil or an uncivil video clip. This setup allows us not only to see how conflict orientation shapes media choice in the absence of prior exposure, but also how prior preferences for certain types of news, combined with exposure to incivility or civility, might moderate the effects of conflict orientation on participants’ choices.
The lack of exposure to either a civil or an uncivil interaction in the control condition allows us to determine what effects stem from exposure to uncivil political news and what come simply from being exposed to political news at all. Participants assigned to a treatment condition were shown either the civil or uncivil CSPAN clip of coverage of recent congressional hearings about Planned Parenthood described in chapter 3. After exposure to the treatment condition, the subjects indicated how civil and informative each clip was on a scale from 0 to 4, where 0 indicated “not at all” and 4 signified “extremely.” Participants’ assessments of these two characteristics were used as a manipulation check;9 a two-sample t-test suggests that participants found the civil treatment to be significantly more civil (
, sd = 0.99).
After the subjects had answered these questions (or in the case of the control group, immediately after completing the pretest questionnaire), they were given the option to watch more clips related to politics and entertainment. If they chose to watch a clip, viewers were given the choice to continue watching additional clips or finish this section. There were a total of five clips to watch, each lasting approximately one minute: two political news clips and three reality television clips. Both types of clips had at least one civil and one uncivil option. Each clip was titled so as to indicate whether it depicted conflict, as well as the subject matter. As table 4.4 shows, phrases such as “activists feud” and “accuses teammates of lying, double-crossing” were used to indicate conflict and incivility, whereas phrases such as “Poll: Americans want a hearing for next justice” or “Contestants reflect on their success” were used to indicate civility and a lack of conflict. The uncivil political news clip featured two commentators arguing rudely and talking over one another. The uncivil reality television clip depicted two members of a team interrupting one another and accusing one another of lying and name-calling. The civil political news clip featured a commentator reporting on a poll. The civil reality television clip showed two members of a team complimenting one another and acting in a jovial manner. Because each clip was explicitly titled and subjects were allowed to select the clip of their choice (as well as when to finish this section), we were able to monitor their search habits.
TABLE 4.4 HEADLINES PARTICIPANTS COULD CHOOSE TO WATCH
| CATEGORY |
HEADLINE |
| Political |
Civil |
CNN: “Poll: Americans want a hearing for next justice” |
| |
Uncivil |
CNN: “GOP activists feud over Trump’s potential damage to the party” |
| Entertainment |
Civil |
The Amazing Race: “Contestants reflect on their success in plate-spinning” |
| |
Civil |
Master Chef: “Chef and judge disagree about pasta dish” |
| |
Uncivil |
Survivor: “Sierra accuses teammates of lying, double-crossing her” |
When evaluating the choices participants made to engage with the five headlines, I looked for three key measures: how many headlines participants clicked on, whether those headlines were civil or uncivil, and how long participants spent watching the videos associated with each headline. Across the three conditions, on average, participants clicked on about two headlines, with some selecting all five (3 percent) and others ignoring them entirely (24.5 percent) and moving on to the next section of the experiment (
, sd = 1.13). Those who did choose to watch clips overwhelmingly selected headlines that were civil (80 percent of clips selected were civil) and about entertainment topics (75 percent). They spent a little over a minute on the pages containing the videos associated with each headline (
, sd = 68.5 seconds). Even though people were more likely to select civil videos overall, the most popular single selection was the CNN piece about the feud between GOP activists, which was viewed by 51.7 percent of those who chose to view at least one clip, followed by the civil clip from Master Chef, which was seen by 39.7 percent of participants who chose to engage with the headlines. The least popular video was the clip from Survivor, which was viewed by 17.6 percent of participants.
Having described the contours of the experiment, I can now assess the extent to which conflict orientation, prior media preferences, and exposure to incivility shaped individuals’ search for additional information. Before examining the effects of conflict orientation and individual preferences for civil or uncivil media, it is worth noting that the two are related. A one-tailed, two-sample t-test demonstrates that participants who report preferences for uncivil political and entertainment television have, on average, a higher score on the conflict communication scale (mean = −1.63, sd = 0.27) than those who prefer civil news (mean = −2.36, sd = 0.45, p = 0.075). This relationship lends additional support to the two hypotheses presented earlier: conflict orientation shapes the kinds of media people prefer, and those who like conflict are also more likely to prefer shows that introduce incivility and disagreement.
That being said, preferences for civil or uncivil news play less of a role in an individual’s decision to search for information or select uncivil clips than does conflict orientation. We see this first when looking at how many videos participants chose to watch. Individuals’ preference for incivility and the match between their preferencs and the level of civility in the treatment both played a statistically insignificant role in the decision to click more articles. However, conflict orientation did play a significant role, both on its own and in conjunction with the treatments. As you can see in figure 4.4, conflict-avoidant participants, on average, clicked on about two stories after being exposed to either treatment. The conflict-approaching behaved a little differently. Those who saw the civil condition also clicked on two or more articles, on average, but those who watched an uncivil clip first were less likely to click, averaging only one additional article after the treatment.
FIGURE 4.4 Incivilty reduces information search among the conflict-approaching.
Note: Linear predictions derived from regression of the interaction of conflict orientation and treatment assignment, controlling for demographic and political variables.
Source: Mechanical Turk Study 2.
We see the same pattern if we look at how much time participants allocated to watching the videos. The graph on the left side of figure 4.5 displays the probability that an individual would spend any time watching a selection of videos from the options presented. As you can see, the conflict-avoidant are highly likely to spend time watching additional clips, regardless of whether they were first exposed to civil or uncivil political news. The conflict-approaching are also very likely to devote a nonzero amount of time to watching additional videos after watching a civil clip, but this tendency drops dramatically if they watch an uncivil clip first. A conflict-approaching individual who watched an uncivil clip had only about a 30 percent probability of allocating time to watching more clips. The same pattern plays out when we look at the right side of the graph, which estimates the average number of seconds spent watching across treatments and conflict orientations. The conflict-avoidant spend about 75–110 seconds watching clips, regardless of whether they saw civility or incivility first. The conflict-approaching spend less time watching once they have already been exposed to incivility; on average, they spend just under a minute less time reading in the wake of watching an uncivil clip.
FIGURE 4.5 Incivility reduces time spent watching videos by conflict-approaching.
Note: The figure on the left presents results from a logistic regression, while the right-hand figure displays the results of a OLS model. Both evaluate the effects of conflict orientation and treatment assignment, controlling for demographic and political variables.
Source: Mechanical Turk Study 2.
We might look at these data as evidence that in this case, exposure to incivility is having a negative impact on the conflict-approaching, but not on the conflict-avoidant. After all, more exposure to information and increased media consumption make people better democratic citizens (Dahl 1967; Downs 1957; Lupia and McCubbins 1998). However, as Albertson and Gadarian (2015) show, not all information-searching habits are normatively positive. When individuals are anxious, for example, they are more likely to look for information about the threat, place greater trust in government “experts,” and support protective policies. Therefore, we might expect the conflict-avoidant—who we know are more anxious once exposed to incivility—to spend more time than the conflict-approaching looking at additional clips because of their feelings of anxiety. The conflict-approaching are not only looking for less information once exposed to incivility, but they are also engaging in different search patterns when they do choose to look at video clips.
Given this relationship between anxiety and a focus on threatening material, I am interested in two patterns of information search. First, are participants seeking out the uncivil clips when they are given the option to watch additional videos? The conflict-avoidant may be watching more clips than the conflict-approaching and gravitating toward incivility more frequently. Results from a multinomial logit model suggest this is the case. Figure 4.6 displays the predicted probability of choosing either only civil videos or a mix of civil and uncivil clips, given the treatment and an individual’s conflict orientation. On the one hand, the conflict-approaching seem to move away from incivility after they have been exposed to it once. When participants are assigned to the civil treatment, the conflict-approaching have a 70 percent probability of looking at a mix of civil and uncivil videos, while the avoidant have a 60 percent chance of gravitating toward the civil clips. Conflict-approaching participants exposed to the uncivil treatment, however, are far more likely to seek out only civil videos when they get to choose what to read, with an 80 percent probability of choosing only civil clips, as opposed to only 15 percent likelihood of choosing a mix. In other words, once the conflict-approaching have been exposed to incivility once, they appear to move on. If they only see civility in the treatment clip, they are more likely to seek out incivility when given the choice. In contrast, conflict-avoidant individuals who are exposed to incivility continue to watch uncivil video clips, as we would expect from individuals responding to an anxiety-inducing stimulus. This decision to continue to seek out incivility, as we will see in the next chapter, has further effects on the conflict-avoidant individuals’ decisions about participating in and talking about politics.
FIGURE 4.6 Participants search for different types of articles if they see incivility first.
Note: Figures represent predicted probabilities of selecting civil or uncivil articles across conflict orientations after viewing a particular treatment, controlling for demographic and political variables.
Source: Mechanical Turk Study 2.
In addition to examining individuals’ search for incivility, I was interested in whether participants sought out political or entertainment clips. Here, the predicted probabilities (see figure 4.7) derived from the multinomial logit model suggest that civility makes the conflict-approaching more likely to choose to watch political videos. After watching a civil political clip, the conflict-avoidant are more likely to choose entertainment videos, while the conflict-approaching seek out a mix of entertainment and politics (only one participant in the entire sample chose only political videos). Once they have been exposed to political incivility, there is no statistical difference in conflict-avoidant individuals’ likelihood of choosing politics or entertainment—they have a 50–50 probability of selecting a political clip. However, the conflict-approaching are more likely to turn solely to entertainment, choosing a mix of political and entertainment about 65 percent of the time.
FIGURE 4.7 Conflict-approaching participants search for a mix of entertainment and politics after civil treatment.
Note: Figures represent predicted probabilities of selecting political or entertainment articles across conflict orientations after viewing a particular treatment, controlling for demographic and political variables.
Source: Mechanical Turk Study 2.
Together, these two sets of results present some challenges for our assumptions about how individuals seek out political information. Political science has long accepted that individuals are suboptimal information-seekers who use shortcuts and heuristics to make political decisions (Converse 1964; Lupia 1994). The findings here suggest that the tone of political communication plays a role in an individual’s decision-making process: if political information is conveyed in an uncivil manner, it can lead to a focus on that incivility among some people and a decision to turn to entertainment among others. For the conflict-approaching, the choice to select civil video clips in the wake of exposure to incivility is encouraging, while the decision to avoid political news runs contrary to what we might hope for in encouraging democratic citizenship.
THE MEDIATING ROLE OF EMOTIONS
In explaining the effects described, I hypothesized that the relationships between conflict orientation, exposure to incivility, and information search habits were at least partly a result of the differing emotions individuals felt upon watching the uncivil video clip. Mechanical Turk Study 2 lets me test the role of each of these emotions by using Baron and Kenny’s (1986) approach to mediation models. The first step in their procedure is to demonstrate that the initial variable—in this case, the interaction between conflict orientation and incivility—has an impact on the outcome: information search habits.10 The findings in the previous pages demonstrate that this is the case. The second step requires demonstrating that the initial variable correlates with the mediator: emotion. Chapter 3 demonstrates that conflict orientation and incivility interact such that conflict-avoidant individuals experience negative emotions in the face of incivility and conflict-approaching individuals experience positive emotions.
The final two steps require the establishment of a correlation between the mediator variable, the various emotional responses, and the outcome variable, holding the initial variable constant. At this stage, the mediation analysis breaks down. Across all five emotional reactions (anxiety, disgust, anger, enthusiasm, and amusement) and the three outcomes of interest, no emotional response has a statistically significant impact on the number of videos participants clicked after exposure to the treatment, nor on the amount of time they spent reading those articles or on the proportion of articles that were uncivil or political. In short, there is minimal evidence that any single emotional response can explain why the interaction between conflict orientation and incivility shapes individuals’ search for political information.
However, this is not to say that emotions play no mediating role. As I suggested earlier in this chapter, individuals’ overall mood, rather than specific emotions, may also play a role in shifting their approach to consuming information. Therefore, I ran the mediation models a second time, using a dummy variable for “positive mood” if a participant had reported any of the three positive emotions and another, “negative mood,” if a participant had indicated experiencing anxiety, disgust, or anger. The interaction between conflict orientation and incivility has no effect on the negative mood variable; there is only a direct effect of incivility on participants’ experience of negative emotions. Negative mood does not appear to moderate the relationship between incivility, conflict orientation, and information search.
Positive mood, in contrast, does have a mediating effect on the relationship between incivility, conflict orientation, and individuals’ decision to seek out information. Those individuals who were in a positive mood were slightly less likely to seek out information than their peers who did not report positive emotional reactions to the same treatments. Figure 4.8 displays the results of the final regression models, including positive mood as a mediator. The effect of positive mood is small—reducing information search by about half a video clip—but offers support for the expectations derived from mood-management theory. Positive mood leads to less information search, explaining a portion of the relationship between conflict orientation, incivility, and information search.
FIGURE 4.8 Positive mood mediates the join effect of incivility and conflict orientation.
Note: Linear predictions derived from regression of the interaction of conflict orientation and treatment assignment and reported positive mood.
Source: Mechanical Turk Study 2.
***
Previous research shows that individuals are active consumers of media content, particularly in today’s choice-filled environment. Americans can choose to watch the nightly news or scroll through the New York Times app, or step away from politics altogether and turn on The Bachelor. These choices are shaped by their interest in politics, their mood, and the emotions they experience as a result of exposure to certain information. My results are consistent with this work and build upon it, demonstrating that the choices we make about our media environment are also shaped by our orientation toward conflict. The conflict-approaching spend more time using media to learn about politics. The positive affect they feel when watching incivility leaves them with a sense of satisfaction that leads them to subsequently seek out fewer uncivil video clips than when they watch civil news. However, this same civil news encourages them to keep learning about politics; in the civil condition, the conflict-approaching are more likely to incorporate political videos into what they watch. To summarize: in the face of incivility, the conflict-approaching can step back and say “we’re good, we don’t need more nastiness,” while still engaging with media content. In the face of civility, they are more likely to seek out political information.
The conflict-avoidant, in contrast, face an uphill battle in their quest to learn about politics. They begin by being less interested in politics, and therefore spend fewer days of the week tuning in to political news. When exposed to political news, they make a decision to further engage, but they are just as likely—if not more likely—to seek out entertainment content than a mix of politics and entertainment. What is more, they are likely to engage with new clips that are also uncivil, consistent with research suggesting that anxiety leads individuals to pay more attention to perceived threats. However, anxiety does not serve a mediating role, according to this data set, raising the possibility of other explanations for the somewhat counterintuitive finding that when we are conflict-avoidant, we actually search for more rude and nasty conversations. In the context of this experiment, the conflict-approaching are not behaving as the ideal citizen when it comes to the search for political information. The conflict-avoidant are even further disadvantaged and discouraged by the political system.
