Confronting Politics: The Role of Conflict Orientation in Shaping Political Debate

Previous research (Testa et al. 2014; Mutz 2015) finds that conflict orientation—individuals’ psychological predisposition towards conflict—conditions attitudes of people in the United States in the face of political disagreement. However, little research has been done into how conflict orientation influences the ways in which people engage in conversation that has the potential to become uncomfortable or contentious. While we argue that conflict orientation has a significant impact on the way college students discuss politics, results from series of interviews with undergraduate students about their thoughts and ideas regarding political incivility and campus free speech suggest that this is not the case. Instead, we find that deliberation and small-group conversation can bring both the conflict-avoidant and conflict-approaching into the political conversation.

Many people in the United States (henceforth, 'Americans') grew up hearing the adage that they 'should not talk politics at the dinner table.' Increasingly, Americans are also uncomfortable talking about politics anywhere else. According to a Pew Research Center poll, only 17% of those sampled in the U.S. reported they would be very comfortable talking about politics with someone they don't know well; half of survey participants said that talking politics with people they disagree with is stressful and frustrating (Pew Research Center 2019). However, talking about politics is seen by political communication scholars as integral to the democratic process: if citizens can't express their political ideas and opinions, leaders cannot accurately represent them. What's more, political conversation-especially with those with whom you disagree-is heralded by deliberative democrats for its ability to increase tolerance, improve the quality of opinions, and facilitate political engagement (Gastil 2008;Gastil & Black 2008Longo & Shaffer 2019. Others question whether conversation is truly 'the soul of democracy' (Mutz 2006;Schudson 1997) and criticize deliberative democracy for exacerbating existing inequalities in democratic participation (Sanders 1997;Young 2000). As political theorist Iris Marion Young argues, 'speech that is assertive and confrontational is here [in political deliberation] more valued than speech that is tentative, exploratory, or conciliatory' (Young 1996).
While what is considered effective deliberative reasoning has shifted over the past decade, the fact remains that political conversation is fundamentally about argument, which means that people who are uncomfortable with conflict and confrontation are going to find it more difficult than those who thrive in argumentative environments. This psychological predisposition, known as conflict orientation, shapes individuals' willingness to share political opinions online (Sydnor 2019a), their interpersonal discussions (Testa, Hibbing & Ritchie 2014) and trust in government (Mutz 2015).
In this paper, we argue that an individual's conflict orientation shapes their behavior in political conversations. Drawing on interviews with 47 American college students about online incivility, white nationalist speakers, and the prospect of contentious events on their own campus, we see some differences between the conflict-avoidant and conflict-approaching in the types of evidence mobilized in argumentation, but few differences in the frequency of their engagement in political conversation or in the quality of their conversation. These findings provide unique insight into the process of deliberation by drawing on behavior, tone, and language used in the discussion, rather than aggregate or inferred measures of political conversation. Focusing on small differences in the way participants engage in discussion, we find support for the practice of deliberative conversation around contentious issues, suggesting that even those who are uncomfortable with political discussions can find a way to talk about politics in a moderated, small-group setting. In other words, while conflict orientation may the deliberative poll-also argue that inequalities can be minimized in real deliberative settings as long as they are carefully constructed to be representative and offer a good context for considering the issues (Fishkin et al. 2017;Siu 2017). While the deliberative poll can be carefully crafted to minimize inequality and disrespect, contention in informal political talk is much harder to avoid. As Wells and colleagues demonstrate using the recall of Wisconsin Governor Scott Walker in 2012, those who were most likely to be at the center of unfriendly talk, like government workers, residents of economically distressed areas, and political outsiders, were more likely to stop talking about politics when the conversation centered around Walker's recall (Wells et al. 2017). Without the intentional sampling and moderation of a deliberative poll, casual conversation about politics suffered, and they suffered in different ways based on individuals' demographic characteristics.
More generally, a variety of personal and environmental characteristics influence the quality and quantity of individuals' engagement in political conversation, especially when that conversation leads to disagreement. While it may vary from country to country, everything from an individual's neighborhood context to the extent to which they are an extravert can shape their reaction to disagreement in political conversation (Gerber et al. 2012;McClurg 2006). Testa, Hibbing and Ritchie (2014), for example, find that individuals who hold a positive orientation towards conflict-who find conflicts exciting and comfortable to navigate-are more likely to be tolerant of diverse opinions when exposed to political disagreement. Those who hold stronger negative orientations towards conflict do not experience the same benefits in the face of disagreement. Focusing on the same psychological trait, Sydnor (2019) finds that the conflictapproaching are more likely to offer opinions about politics online, even in the face of incivility, while Wolak (2022) finds that a conflict-seeking orientation is the best predictor of an individual's enjoyment of talking politics. Conflict orientation, along with other psychological traits (see, for example, Hibbing, Ritchie & Anderson 2011;Kenski, Coe & Rains 2020), can influence individuals' reactions to political conversation and especially the crosscutting political communication inherent to deliberative dialogue.
While previous research has established the effects of political conversation vary with conflict orientation, less emphasis has been placed on how the conversation itself is shaped by these traits. As many scholars have demonstrated, one of the biggest hurdles to deliberative conversation is the social nature of conversation. Because individuals have the potential to build or damage relationships in the process of navigating political conversation (Eliasoph 1998Gerber et al. 2012Huckfeldt et al. 1995), the process may elicit increased anxiety and avoidance-especially among those who dislike confrontation and conflict in the first place. Therefore, we expect that conflict orientation will shape individuals' engagement in conversation-their likelihood of speaking first, their need to be prompted by a moderator/researcher, and the frequency with which they speak over the course of the conversation (H1). From a deliberative perspective, discussion partners should recognize themselves as, and be treated as, equal and legitimate contributors to the conversation with equal opportunities to participate in a debate. We suspect that the conflict-approaching may be more comfortable in these situations than the conflictavoidant, which may undermine their ability to openly participate in a discussion.
Not only do deliberative political conversations need participants to be equally engaged and empowered to speak, but they also require an atmosphere of mutual respect and civility (Ferree et al. 2002). This is true even in the context of hateful, racist, or misogynistic voices; we must consider what white-nationalists and others have to say before rejecting their input to the democratic conversation (Scudder 2019). However, some people will likely be better at what Scudder describes as 'uptake' than others; they will be more willing to listen even if the arguments they are listening to are hurtful to them or to other groups. The conflict-approaching are less concerned with confrontation and may therefore be more willing to listen to hateful speech, but they may also be more willing to use uncivil language themselves. Therefore, we expect that political conversation is more likely to contain incivility when the interlocuter is conflict-approaching than when he or she is conflict-avoidant (H2). The presence of incivility has complicated implications for the effectiveness of deliberation; as Guttmann and Thompson argue, deliberative democracy is predicated on civility and mutual respect, but can be ' consistent with impassioned and immoderate speech…even extreme nondeliberative methods may be justified as necessary steps to deliberation' (Gutmann & Thompson 1996: 90).
As discussed above, traditional conceptions of deliberative conversation require attention to specific forms of evidence, priviledging logical, fact-based arguments over narratives and personal storytelling. Other critics have argued that the focus on fact over personal narratives can marginalize groups, like nonwhite or female participants, that are already less likely to participate in political conversation. We expect that the same may be true of the conflict-avoidant. A personal narrative could be perceived as a less confrontational approach to making an argument-you are starting from your own perspective rather than trying to draw a general conclusion or recall specific facts (and risk being called out for being wrong). Therefore, we expect that personal narratives will be used more frequently by the conflict-avoidant, while the conflict-approaching will be more likely to use fact-based arguments (H3).
Finally, while some deliberation scholars (for example Saam 2018) argue that negative emotions can lead participants to exit deliberation, other research suggests that the relationship between emotions and deliberation are complex and shaped by procedures that help participants place their emotional experience in a larger democratic framework (Johnson, Morrell, & Black 2019). Previous research has found that the conflict-avoidant and conflict-approaching have different emotional responses to uncivil and contentious political conversations (Sydnor 2019a). While the conflict-avoidant report greater anxiety, disgust and anger in the face of incivility, the conflictapproaching are more likely to experience positive emotions like amusement or entertainment. We expect these differences to manifest in participants' discussion here as well, in ways that ultimately impact the quality of their conversation. To the extent that participants mention their emotional responses to political conversation, we expect that the conflict-avoidant will be more upset than the conflict-approaching (H4).

The Conversations
To investigate conversational patterns and their variation across conflict orientation, we conducted a series of interviews with 47 college students at a small liberal arts school in the southwest United States. Participants were recruited through the introductory American Politics course and through flyers around campus; those who did not receive course credit were entered to win a gift card to a local business. The sample included 47 participants (33 women, 13 men, and 1 who declined to identify) and the majority of the participants identify as white (82%), while 7% identify as black or African-American, 2% as East Asian, 4.5% as South Asian, and 2% as Native Hawaiian or Pacific Islander. Because the participants were selected from a student population, the average age of the sample skews younger than the general U.S. population (M = 20.1).
It is important to note that the student sample used here is not representative-of Americans as a whole or even of the student body at the university-nor were participants randomly selected. Both factors limit the generalizability of the results to a broader population. Moreover, the opt-in nature of the recruitment process opens up the possibility that those who choose to participate are different in meaningful ways from those who are not participating. For example, the conflict-avoidant may be less likely to participate in a study that they know is about political discussion. That being said, we see the data derived from this sample of individuals in this specific context as useful and emblematic of the types of discussions that might occur among groups of students around a contentious political event. The dynamics and constraints observed by the researchers are likely to be similar to those seen in other political conversations between a small number of students who are acquaintances or friends (Weiss 1994).
Before beginning the conversations, participants were taken to a private room, asked to complete an initial questionnaire, and then to watch a brief CNN clip about white nationalist Richard Spencer's visit to Auburn University in March 2017. Spencer's visit to Auburn was one of many speaking engagements on college campuses by members of the far right in the United States that provoked protest and unrest, as well as cries to protect free speech and concerns about the 'fragility' of college students in the U.S. (see Bollinger 2019). The clip used in this study documents the tension surrounding Spencer's speech at the university, reporting on the arrest of protestors and his inflammatory rhetoric. When a student asks Spencer for ' a direct reason as to what is wrong with diversity,' he responds: 'What is wrong with diversity? It makes the world ugly. It makes the world lose all meaning. It's a way of bringing to an end a nation and a culture that was defined by white people' (CNN 'White Nationalist' 2017). While many participants were not familiar with the specific incident depicted in the clip, as college students, they were familiar with the debate over free speech on college campuses that accompanied talks like Spencer's.
After his statement, the video clip ended and participants were asked to read a set of comments in response to the video. 1 After they read the comments, the researcher and another student entered the participant's room and the two students were asked about their reactions to the video and the comments-about the tone of the commentary, about its appropriateness for online discussion, and about how they would respond. The second student appeared to be another participant but was actually a confederate following a general script in response to the researcher's questions. The script was based on the confederate's true opinions to minimize the possibility that the participant realized the confederate was playing a role in the experiment. Regardless of the participant's responses, she noted that the video was provocative and inflammatory, and that conversation around issues like diversity had to balance civility with passionate activism. Her presence in the experiment was designed to make the participant feel more like he or she was in a conversation than a one-onone interview. The presence of the confederate allows us to understand the participants' comments as deliberation, rather than the simple sharing of an individual opinion. Together, the confederate and participant evaluated the role of civility and incivility in political conversation and weighed the pros and cons of having a controversial speaker come to campus.
After this discussion of the tone of Spencer's remarks and the subsequent online commentary, participants were told by the researcher that Richard Spencer was being invited to their campus and were given the opportunity to discuss their reactions and to participate in political activities, like signing a petition or taking a campaign button supporting or opposing Spencer's arrival on campus. Several elements of this procedure involved deception: the comments in response to the video were created by the researchers and there were no plans to bring Spencer to the campus in question. 2 We felt this deception was necessary in order to generate authentic reactions to the possibility of Spencer speaking on campus; research has found differences in participants' behavior when they believe a situation is real and when they are considering hypotheticals (Armor & Sackett 2006;Kang et al. 2011). However, each interview generated effective conversation about participants' understanding of the free speech debate on college campuses and offered insight into their approach to political talk. Table 1 provides two examples of the flow of conversation between the researcher, confederate and participant (the full interview protocol is available in Appendix B).
When looking at how participants spoke about the tenor of the conversation surrounding Spencer, and the hypothetical reactions of the campus community, we focused on four different components of their engagement. First, we examined how much and when they spoke. Were they quick to jump in to the conversation, or did they wait for prompting from the researcher? How many times did they speak over the course of the two conversations?

Example 1: Conversation about the Civility/Incivility of the Video and Comments Section
Researcher: Ok, so in you all's opinions, do you think it's better for people to be passionate in engaging in political discourse regardless of how civil or uncivil that conversation may be? Or do you think that civility and political discussion is more important than how passionately people may engage within that conversation? Participant: That's a hard question.
Confederate: I would say maybe just a little bit of both, because I think you do need to be passionate about what you are saying because you believe in it hopefully, so that you're able to be passionate about it. But also, I think you need to be civil. Because if you're uncivil, nobody's really gonna be listening to you, or value what you say because you're being so negative [and] hostile about it. So I think a little bit of both is important.
Participant: Yeah. 'Cause I think civility kicks in within having a good discussion or good lecture or anything, like he was having questions and answers. You have to have some sort of, you can't just be, like, yelling over each other, not letting anyone speak, ' cause that's not going to convince anyone or change anyone's mind that that's a hopeful outcome, situation. Confederate: I'm fine with having conservative people here, of course, but I don't want any alt-right-thinking people to think that this is their pass to be hateful and mean to others.
Participant: Right, but I just don't think that College Republicans have paid for Richard Spencer to come so I think that would be a lot of it, is where is this funding coming from. 'Cause if it's from university funds or my tuition that's somehow going to this, absolutely the fuck not. I don't know. There's just a lot of parameters that I need to know.
We coded each participants' conversations for these characteristics in order to assess their participation in the discussion. Second, we were interested in their arguments. Did they push back against statements made by the confederate? What kinds of evidence did they provide for their opinions? One of the authors and a research assistant coded for six different kinds of evidence that students brought into their discussions: the use of facts, cause-and-effect logic, emotional reactions, personal experience, community (university) values, and societal values (see Appendix A for codebook and interrater agreement). Of these evidence types, facts were used the least-by only 17% of the sample. Each of the other six was used at least once by over 50% of the sample (emotional reactions: 51%, personal stories: 53%, cause-and-effect: 57%, community values: 68%, and societal values: 40%).
Finally, we were interested in the tenor of the conversation. Did they get emotional as they talked about their reactions to Spencer's rhetoric? Did they use uncivil language-defined in this context as name-calling, obscenities, insults, shouting or aspersions-in describing him or the supposed online commentators? As we discuss below, while many students got emotional, only a few used any sort of uncivil rhetoric. This data was merged with participants' conflict orientation as measured in the questionnaire completed prior to watching the video clip.
Conflict orientation was measured using the 75-item Conflict Communication Scale (Goldstein 1999). Participants' scores on each item were combined into an additive index and standardized on a zero to 1 scale, where 1 indicates the most conflict-approaching score possible on the scale, and zero indicates the most conflict avoidant. We were able to calculate the conflict orientation of 40 of our participants. 3 On average, the sample was just slightly conflict-approaching (M = 0.55, SD = 0.11), a somewhat unexpected leaning, as women tend to be more conflictaverse (Coffe & Bolzendahl 2017;Wolak 2020) and the sample is almost 70% women. At the same time, as Figure 1 shows, the distribution of orientations across the sample approximates a normal distribution (skewness = 0.53) with a mean near the scale's midpoint, much like the distribution of other, more representative samples (see, for example, Sydnor 2019a).

Results
To test our hypotheses, we divided the sample into two groups based on their conflict orientation-those whose scores were greater than 0.5 (the ' ambivalent' midpoint) and those whose scores were lower. According to this distinction, sixteen of the participants qualify as conflict avoidant while 24 are conflict approaching. We then used one-tailed two-sample t-tests with unequal variances or tests of proportions to assess the difference ways in which the conflict-avoidant and conflict-approaching engaged in political conversation. In short, these empirical analyses produce few statistically significant differences between conflict avoidant and conflict approaching groups. Our first hypothesis was that the conflict-avoidant would be less engaged in the conversation than the conflictavoidant. We had several ways of assessing ' engagement': we looked at whether the participant spoke first (instead of the confederate), how many times they spoke, and if they required prompting from the researcher (see Appendix C for analysis). Across all three measures, there were no statistically significant differences between the conflict-approaching and the conflict-avoidant.
We also only have limited support for hypothesis two, that the conflict-approaching will use more incivility than their conflict-avoidant peers. There were only seven instances of incivility throughout the entirety of the study; 22% of the conflict-approaching used some sort of incivility while 6% of the conflict-avoidant (1 individual) did the same. For example, one conflict-approaching participant referred to Spencer's talk as a 'pathetic appeal,' while others called Spencer a 'pissed off little white man,' ' a bigot,' and 'unintelligent'. This difference just barely passes the threshold for statistical significance at p = 0.099. Regardless of the model specification or measurement strategy used (all of which are described in detail in Appendix C), we see little difference between the conflict approaching and conflict avoidant in their willingness to engage in conversation or speak uncivilly.
While our first two hypotheses did not play out as expected, we nonetheless have a rich qualitative picture of the ways in which incivility was deployed as part of the deliberation. Participants treated each other with mutual respect; although they did not always extend the same courtesy to Spencer's white nationalist ideas, they did listen to and consider them before expressing their dismay through the use of incivility. Incivility was deployed as an expression of heightened emotion that was frequently integrated into a fuller engagement with the topic and presentation of arguments and opinions. For example, when asked what they thought of Spencer and his ideas, Mia, a conflict-approaching (CCS score = 0.59) black student, immediately stated, 'He's like a human version of wet garbage. He's just a horrible human being.' But she continued her thought, moving from name-calling to an empathetic reflection on the impact of Spencer's words on other members of the community: 'We're just so diverse with race, and background and perspective. He's someone who hates diversity. Why would you invite someone who hates diversity to such a diverse place? I don't understand.' Listening to Spencer's argument that diversity makes the world ugly prompted an emotional reaction among many participants, some of whom used incivility alongside other forms of reason-giving to articulate their own opinions and arguments for why he should not be allowed on their campus. Few students were quick to turn to incivility and that when they did, it was usually as an emotional expression of moral sentiment that still advanced deliberation (Krause 2008).
In our third hypothesis, we expected the conflictavoidant and conflict-approaching would use different types of evidence when discussing the tenor of the online comments and the prospect of Richard Spencer visiting their campus. Specifically, we expected the conflictavoidant-the group that potentially would feel more uneasy in the context of a political discussion-to draw more on their personal lives and narratives in making the case for or against Spencer's presence on campus than the conflict-approaching. Here, we find support for our hypothesis. Seventy-three percent of the conflict-avoidant used a personal story in their discussions, while 45% of the conflict-approaching did the same (p = 0.036; see Figure 2 for the breakdown for all forms of evidence).
These personal stories took a variety of forms, from reflections on the 2016 election, to their family and friends' experiences, to their own challenges in navigating contentious politics. Among both the conflict-avoidant and conflict-approaching, the use of personal stories reflected only negative attitudes towards Spencer coming to campus and the use of hate speech on any platform. When evaluating the impact Spencer might have on the campus environment, Aria (CCS = 0.56) recalled her experience on campus after the 2016 election: 'right after the election, right after it, there was a bunch of talking around campus that was like, no safe spaces, MAGA, all that stuff, all the hateful stuff.' A more conflict-avoidant student, Hannah (CCS = 0.44), also drew on past experience to express concern about how the campus would feel if Spencer spoke: 'I took this class…[where] we sat split ideologically, and you could tell who was where and it was fine, it was chill until the day after the election. Everyone lost their ever-loving mind and we were talking to each other in ways where I was like, "I have never heard you use that language before."' Students, especially the conflict-avoidant ones, were quick to draw on their personal past experience with incivility and disagreement on campus to make the case for avoiding future disagreement.
While some believed Spencer would lead to the same sort of harassment they and their friends had experienced after the 2016 election, others were more hopeful that their friends from across the aisle would find rhetoric like Spencer's equally distasteful. As Harper (CCS = 0.70) explained, 'I even have a couple friends that are conservative, but they wouldn't think like that. One of my really good friends, he's fairly conservative; he's definitely Republican, but his roommate is Muslim and pretty liberal, and he is from the Middle East, but they're best friends. I'm going to have trouble seeing if anybody's going to actually enjoy that and support it.' Reagan, a conflict-avoidant student (CCS = 0.42) reflected on how attending college had expanded her worldview in ways that made Spencer's rhetoric less appealing: 'Growing up in a racist, very conservative middle-class family, I didn't really know, I didn't really know anything and then I came to [this university] and I'm so glad that I was educated on white privilege, race, [and] class.' While both the conflict avoidant and conflict approaching employed personal stories and narratives in their thoughts about contentious political speech, the conflict-avoidant were more likely to use these stories to demonstrate the negative repercussions of bringing extreme or radical speakers to campus.
Our final hypothesis was that the conflict-avoidant and the conflict-approaching would engage in different types of emotional expression, although this did not appear to be the case. While past research (Sydnor 2019a) suggests that the conflict-avoidant should express more negative emotions and the conflict-approaching should feel more positively about contentious conversation like the ones they witnessed and imagined were headed to their campus, we did not see this difference in this sample. Instead, the conflict-approaching were statistically more likely to use their emotional reactions-both good and bad-as evidence for their position towards Spencer's visit to campus. Fifty-eight percent of the conflict approaching drew on their emotional reactions to make an argument, while only 38% of conflict-avoidant students did the same thing (p = 0.09).
Regardless of whether the individual was conflictavoidant or conflict-approaching, the overwhelming emotional reaction was negative. Over half of the participants who expressed strong emotions in the discussion conveyed feelings of sadness, anxiety or anger at the video of Spencer speaking at Auburn, at the comments they read in response to the video, and about the prospect of Spencer visiting their university. These emotional displays ranged from, 'I thoroughly hate this man, I really do,' (Maya, CCS = 0.51) to, 'that stressed me out I was like oh my god,' (Ellie, CCS = 0.76), and with one student even claiming, 'that was very, very horrible and [made me] disappointed in humanity.' (Mia, CCS = 0.59) Others recognized how their own demographic characteristics allowed them to be less emotional. As Victoria (CCS = 0.39) noted, 'It doesn't affect me, because I'm white, so I don't feel repercussions from what they're saying, but if I was a person of color, I would definitely be more upset.' Even those who found it funny weren't amused so much as appalled or shocked; as Adaline (CCS = 0.56) reflected on her reaction to the video and comments, 'I was laughing. It was just so ridiculous.' For many students, these negative emotional reactions were a starting point from which they began their reflections and arguments about the impact of a visitor like Richard Spencer. While some research on deliberation suggests that emotions like anger and outrage interfere with the process of deliberation (after all, juries-heralded as deliberative bodies-are explicitly charged that they must not be swayed by sentiment or sympathy), the students in these discussions see their negative emotions as another form of evidence against Spencer's presence on campus.
Beyond our three hypotheses, we noticed a few other patterns in the use of evidence that help us understand how students deliberated on the potential for controversial political events on campus. First, those participants who were more likely to have a sense of or identify with the university community-older students, women (the campus is over 50% women), and white participants-were also more likely to talk about the ways in which Spencer's visit and the rhetoric present in the online comments were not representative of the university community. Sixty-eight percent of interviews used university values and norms as evidence for or against their opinion about the invitation of Spencer to campus. Some students connected their views about incivility to the university's approach to campus safety. As Adaline noted, someone who '[has] a lot of faith in [the] school and the people that go here,' would most likely expect the school to protect them and their beliefs regardless of what they may be, even though '[the university] is a very leftleaning, liberal school.' This belief in protection of ideas and values is seen in Alaina's hope that, '[the university] would take its students into consideration before actually allowing him on campus because they send those little sweet emails that are like, "We hope you're all safe and we want to make you feel safe," but, how actually are going to make people feel safe if you allow him on campus?' Those with the strongest university identity had the greatest expectations that the university has an obligation to protect its students by denying access to individuals that don't share their values or who could incite physical violence or disorder on campus. As they wrestled with which values to prioritize as part of the deliberative process, students' understanding of the community's culture and norms was one of the most common pieces justifications for their opinions.
We also coded for three other types of evidence that participants frequently used in their political conversations: the use of fact-based statements (17% of conversations), cause and effect statements (57%), and references to broader societal or American values (40%). While we found no statistical differences across conflict orientation or demographic characteristics in their use, they nonetheless indicate important approaches to student deliberation about the question of uncivil language online and the invitation of controversial speakers on campus. These forms of evidence showed that students were drawing connections between potential incivility on campus and other spaces in which they are exposed to nasty rhetoric and hate speech. As Aaron (CCS = 0.40) explained, 'Twitter for example, or Reddit or YouTube, they take down videos or posts that have hate speech on it. It's a company policy, and the company has the right to do that.' However, he was one of the few who ultimately advocated for allowing Spencer to speak on campus: 'it is an important conversation to have, and bringing him in would spark that.' Thinking about broader patterns of American political culture, Aria commented: 'this is America and…we base everything on freedom, you should be able to say what you want, the government shouldn't be able to censor what you say.' It was in presenting evidence of other approaches to managing public speech that we saw the greatest number of arguments in favor of hosting Spencer on campus. These types of evidence were not used any more or less by members of a particular demographic group or by the conflict-avoidant or approaching, suggesting an effective deliberative environment.

Discussion
Prior research suggests that conflict orientation has the potential to undermine engagement in contentious political discussions, to the detriment of democratic society. If only those who seek out conflict are sharing their opinions and talking about politics, we cannot fully identify consensus positions or accurately represent public opinion. Moreover, when discourse is conducted primarily between people who enjoy confrontation and argument, it becomes more divisive, uncivil, and in many cases, less focused on the substantive argument than on winning the battle against the other side. Deliberation is designed to counteract these tendencies, encouraging mutual respect and substantive engagement within heterogenous groups in order to hear everyone's perspective and find points of consensus from which to govern. Because it is still inherently conflictual, however, deliberative approaches may not erase the privileged position of the conflict-approaching when they engage in conversation. Understanding conflict orientation, therefore, also helps us understand the successes and limitations in deliberative political talk.
While we expected to see differences in the way that the conflict-avoidant and conflict-approaching engaged in political discussion, we ultimately saw few differencesand almost no differences that undercut the deliberative potential of these discussions. The conflict-approaching and conflict-avoidant talked about the same amount, were prompted similarly, and were both equally willing to start the conversation. There was minimal incivility in the discussions, even after participants had been exposed to comments that were uncivil in both tone and substance. The conflict-approaching were more likely to use their (mostly negative) emotional reactions as evidence in their arguments for or against invited white nationalist Richard Spencer to speak on campus, while the conflict-avoidant were more likely to use personal stories in the same manner. Looking at a different set of characteristics, we found that women and those further along in their university career drew on their knowledge and expectations of the university community in making their case. These forms of evidence were mixed in with a variety of others that were used equally across groups; assertion of facts, cause-and-effect statements, and broader arguments about American societal expectations or political culture were deployed both in support of and against Spencer's presence on campus.
While some of our findings ran counter to our expectations, they offer important insight into the possibilities for deliberation on college campuses and the ways students talk about contentious politics. Some deliberative theorists (Cohen 1989;Habermas 1996;Saam 2018) are skeptical about the roles of storytelling and emotion in deliberation, yet it is clear that these were some of the most powerful forms of evidence and experience that students brought to their discussions about the impact of having a white nationalist visit their campus. And these approaches can have a positive impact on individuals' understanding of the issues and their relationships with one another. As Black notes, 'the common themes that arise as participants tell and respond to stories can forge connections among group members and illuminate the tensions among these multiple voices ' (2008: 111). Storytelling can help people identify their own preferences and understand how another person's values and experiences shape their political opinions (Polletta & Lee 2006). From this perspective, the conflict avoidant might not be disadvantaged in their political discussions but instead hold an important place in deliberative theory, offering alternative ways to connect with the other members of the conversation and facilitate understanding.
This practical finding reinforces the argument by Black and others that deliberation needs to be inclusive of other forms of expression-the more traditional, narrow take on what is valuable in the deliberative context would exclude these voices by default. This also highlights an avenue by which the conflict-avoidant can be incorporated into American democracy. While previous research demonstrates that conflict orientation shapes how people respond to disagreement and incivility, these findings suggest that it does not prevent them from engaging in small-group conversations about tough issues. They might be less likely to hop online and share their feelings, but when prompted from within a community that supports them or in a controlled environment like a social science lab they can reason and reflect on the implications of political events at the same level as their conflict-approaching peers. Simply promising civil rhetoric is not enough to bring the conflict-avoidant into the political arena, but asking specific questions about their opinions and giving them space to express themselves using whatever form of evidence they feel comfortable using may be a viable alternative for bringing them into the conversation.
We would be remiss if we didn't acknowledge that this research suffers from several limitations. First, the statistical analyses suffer from the small sample size; there are a few cases where the patterns are in the expected direction but fail to reach statistical significance, and even those that are deemed significant are based on a relatively small number of people. In addition, while we have framed this paper as focusing on deliberative conversation, the discussions were not set up in the exact style of empirical deliberative discussions like the Deliberative Poll. Rather than a conversation among four to six individuals of different backgrounds, we had two students discussing questions offered by a student researcher in a relatively brief amount of time (relative to full deliberative exercises). Many of the participants knew the confederate and her political views, and this likely altered the way in which they engaged relative to those who are participating in conversations with people they do not know at all. 4 Among other things, it is possible that conflict orientation shapes interactions with peers and strangers differently. However, given that most in-person political conversations are with a participant's 'strong ties'-their friends and family-this set-up may actually better reflect the ways in which students talk about politics outside of the laboratory.
Similarly, the educational context of both the study and the event about which the students were deliberating may have also eased their discomfort in the face of conflict. The American education system is inherently adversarial, training and socializing students to overcome discomfort around conflict through an emphasis on debate, critique, and formal (rather than relational) knowledge (Tannen 1998). At the same time, it is possible that at a small school where no one can hide their actions or opinions, the educational environment makes it more difficult for students to share their true opinions and insights. A spate of articles in major newspapers have highlighted the extent to which young people in the United States hesitate to share their opinions for fear of being judged by their peers (see for example Camp 2022; Healy & Garcia-Navarro 2022). From this perspective, students' familiarity with their surroundings and their conversation partners might help them overcome their desire to avoid conflict while their awareness of their own positionality and identity within the school's social network could heighten their sensitivity to the same.
The study's emphasis on the micro-level foundations of political communication and behavior is rare in studies of deliberation but opens up several potential avenues for future research. While the focus on discussion presents researchers with a wealth of rich qualitative data, the conversation framework could also be experimentally manipulated to assess the importance of different elements-the inclusion or exclusion of the confederate, the closeness of the relationship between the confederate and participant, and the controversial nature or personal salience of the topic are just a few elements that could be randomly assigned across participants. We focused on holistic assessment of the conversation, coding the use of evidence and argument at the discussion level, rather than the topical level. While our findings suggest that, overall, the conflict approaching and conflict avoidant draw on different types of evidence, it is also possible they do so more when talking about certain topics (events on campus, for example) than others. Regardless of the road researchers take next, our findings highlight the importance of focusing on the specific contours of the conversation itself and evaluating mannerisms within the deliberative conversation, not just aggregate behaviors or attitudes that are changed in the wake of discussion. If we want to encourage greater political talk as a means of generating greater respect and compromise, we have to understand more about the dynamics that make conversation difficult for so many people and create environments in which they can feel comfortable communicating.
As the public narrative in the United States around politics and contentious speech on college campuses becomes increasingly dismissive of students' willingness to engage with a range of ideas, our findings offer an alternative story. Students, even those who dislike conflict and avoid confrontation, participated in conversations about politics, incivility, and offensive rhetoric. They were able to offer arguments-backed by a range of types of evidence from their own experiences of harassment and discomfort to understandings of First Amendment rights and American political culture-for why they did or did not support the invitation of a controversial speaker onto their campus. The deliberative process of political talk did not disadvantage the conflict-avoidant; instead, it led to a deeper understanding among participants of their own political beliefs and relationship to the university community. While the study described here is grounded in the U.S. context, we have little reason to doubt the same dynamics would be at play in other countries, particularly those in Western democracies, where conflict orientation tends to manifest in similar ways (Gudykunst & Ting Toomey 1988).
The premise behind deliberation is simple: to improve democracy, we need to strip the veneer, to talk about ideas-even ideas that we find disgusting, discouraging or counter to democratic goals-sometimes in direct, face-to-face conversation with the people holding these beliefs. We need to face the conflict endemic to politics head on, an experience that is profoundly uncomfortable for many people and entertaining for others. Deliberative democracy scholars have theorized about and empirically tested dozens of structures and rules designed to make conflict more comfortable and productive, many of which also reduce the entertainment value of participation. These rules emphasize substantive matters over conflict for conflict's sake, a tendency that helps integrate the conflict-avoidant into deliberation in a way they are not involved in other forms of political participation. As we refine our understanding of conflict orientation as an individual psychological trait, political psychologists can offer further insight into what contexts, modes of argument, and rhetorical commitments must be deployed (or relaxed) to improve deliberation among everyone, not just those with the power or inclination to express their opinions.

Notes
1 While not being used as such for this paper, the comments served as an experimental treatment to assess the effects of incivility in online comments on individuals' stress levels, as measured through their salivary cortisol. Both sets of comments were reactions to the video about Spencer's visit to Auburn, but one set talked about the visit and his ideas in a more uncivil tone than the other. 2 The study was approved by the Institutional Review Board at the authors' institution. Participants' names have been changed in this paper to maintain confidentiality. 3 We were not able to calculate the conflict orientation of seven of our respondents because they either experienced computer issues that made it impossible for them to complete the online questionnaire (see Appendix B for a description of the process) or they did not get through the questionnaire in their allotted time. Because we were measuring participants cortisol levels at various points throughout the study, they were kept on a strict time schedule, leading to some missing data that we would have collected without the time constraint. 4 If we were to run the study again, we would ask participants to assess their closeness with the confederate in order to better tease apart this dynamic. However, given the size of the university at which the study was run and the nature of the recruitment process, it is unlikely that the participant and the confederate were complete strangers; at least, they shared common identities as university students who were likely to have some sort of future, iterative interaction outside of the lab space.