The crux of my dissertation interests lie in exploring how people currently engage others in political conversation and using those insights to inform the critique and design of computer-mediated systems for future discourse. At the heart of this work is an assumption of polarization—we Americans are in a political rut that grows deeper the further we travel.
The first part of my research is about understanding how people are talking politics right now. This will require a combination of examining existing artifacts—Twitter streams and the archived posts from two WebLab forums—and a inquiry built around new conversations. This summary will detail the plan for the conversation experiment.
Some inspirational music from Leonard Cohen: “Democracy“
“Lunch With Other” Experiment
Inspired by an activity suggested by Elizabeth Lesser—in which two polar opposites sit down over a public lunch with the intent of being curious, conversational, and authentic—I propose to explore the strategies and language used when two people engage in a discussion. This experiment will include four parts: a screening survey, a series of 1:1 conversations, pre- and post-conversation mood assessment, and follow-up interviews with participants.
There are two key reasons for this initial survey. First, I can include a much larger participant group (in theory) than I’ll be working with during the later stages of this inquiry. This may permit some generalizations I won’t be able to get through the conversations I’ll analyze. Second, this serves as my primary means of slotting respondents into pair-able conversations.
The survey will include questions about:
- Self-declared affinity to political labels (e.g., “Conservative”)
- Indication of support for specific political issues (e.g., “How much do you support expansion of I-69?”)
- A short description of current personal politics
- A short description of a key political influence leading to current political philosophy
Since this initial survey is meant to help filter respondents for inclusion in the later stages, the questions here will have to reflect the dimensions being examined in those conversations (see below).
Recruiting for participation in this project will likely target local online forums (MoB Talk, Bloomington Online, local Twitter, and possibly the Herald-Times online comments) to focus on accessible participants who can easily be observed and interviewed. Because one key dimension for this work will be the effect of computer mediation, recruitment outside of Indiana is important, too. General social media, established political forums, and blog posts will spread interest in this initial survey across a wider demographic.
From the participant group from the initial survey, a few dozen people will be invited to join the second phase of this study. Selection will specifically look for pairs of people showing polarized ideology and support of single issues. This will filter out moderate politics to allow me to focus on extremes during the conversations.
There are a number of possible dimensions to explore with each pairing:
- Affinity to ideological labels (liberal, moderate, conservative)
- Spectrum of support for a single issue
- Scope of single issues (local, regional, national, global)
- Age of participants (18-29, 30-44, 45-64, 65+)
- Locality (Indiana, Not Indiana)
- Mediation of discourse (face-to-face or computer mediated)
- Prior knowledge of Other’s affinity (i.e., does a liberal know she is talking to a conservative?)
- Deception (i.e., misinform people about the other person’s ideology)
- Purpose of the conversation (political issue, non-political topic, seeking understanding)
- Preparatory information (i.e., talk around a particular news article)
My initial strategy will look to capture information about eight kinds of conversations across three key variants: ideology, mediation, and purpose.
Each participant in this phase of the study will be asked to have four conversations over a span of a few weeks. In two of the conversations, they will be matched with someone considered a polar opposite, by virtue of ideology and perhaps other factors, such as stance on a key issue or age. In the other two conversations, they will talk with someone who is similar. Their conversations will either be primed—we’ll provide some article to discuss and ask them to match a picture with their presumed ideology and stance on the issue—or the pair will be simply asked to get to know one another. Future iterations of this study could delve further the nuances of presumption and the nature of the task.
The start of each paired session will be recorded, with a transcript used for content analysis. Conversations can last longer—I want to allow them to reach comfortable closure—but I’m most interested in comparing the first thirty minutes of discourse. For those in primed conversations, an article about the topic of conversation will be sent to them in advance.
Due to the desire to assess mood and preserve as pristine the initial stages of discourse, participants will be separated prior to their conversation. As part of the paired conversations, a short survey will be administered to each participant—at the start and immediately following the discussion—to gauge each person’s mood and the perceived mood of their partner. Mood will be determined by selecting among eight faces, arranged in a circle, representing possible expressions and avoiding issues with semantic interpretation of terminology.
Those in primed conversations will also be asked to select from four possible descriptions of the politics of their partner, guessing which one is accurate. The options will derive from the mix of ideology and support of the single issue being discussed. This will be repeated at the conclusion of the conversation as well.
After the conversation and survey data has been analyzed, specific participants may be invited to individual or group interviews to explain their strategic decisions during the discussions and provide more insight into their own political philosophies and activities. The purpose of this part of the inquiry would be to clarify and deepen the understanding of key findings from the previous data.
The findings from the survey will be most useful in trying to answer three important questions:
- How strongly are single issues aligned with political labels?—My assumption is that, in our current political landscape, positions are defined by the affinity to a label, rather than contributing to a flexible definition of that label (e.g., conservatives oppose gun control).
- What common influences lead to shared political views?—I expect to find some words, ideas and experiences expressed by participants will show strong associations with descriptions of one’s personal politics. This also includes finding links between political descriptions and labels.
I would love to add a third question, looking at potential shifts between political background and current politics, but I think that is a rabbit hole better answered by existing research that points to family as a key factor in social learning. That said, I do expect to find some examples of a negative experience leading to a political shift.
Analysis of the conversations will look for differences in the eight types of conversations (primed and not primed, similar and different ideologies, face-2-face and computer mediation) to see if any conclusions can be made about the effect these dimensions have on discourse. In particular, I’m asking:
- What distinguishes polarized from kindred discourse?—I expect to see more aggressive and defensive language, a quicker declaration of position, and less willingness to change in discussions between polarized individuals than kindred ones.
- What distinguishes computer-mediate discourse from face-to-face conversation?—My hypothesis is that the mediation has much less impact on political discourse strategies and language than polarized participants, but people will prove to be bolder and more task-oriented through the computer than in person.
- What impact does focused purpose have on political discourse?—The conversations that prepare by reading a specific article to act as a catalyst for their exchange will be less diverse in the topics they discuss and less empathic in how they converse.
- How is discourse affected by knowing the politics of your partner?—If polarization is the dominant factor, then being accurate in guessing the ideology and single-issue position of the other person will not dictate discourse strategy as much as which politics you assign. I expect that the conversations that aren’t primed will be more exploratory and adaptive than those where the participants are asked to assign a political view to their partners.
At this level, the expectations for polarized discourse will be tempered with the awkwardness of personal exchange between two strangers. The characteristics of the exchange won’t rise to the level of systemic polarization, as described by conflict theorists Dean Pruitt and Jeffrey Rubin. However, I expect that some of the seeds of systemic conflict will be observable, such as increased sarcasm, more generalizations, and diversity of tactics to prolong engagement.
The end outcome of this inquiry will be the ability to articulate some of the dynamics of political discourse, as they relate to the political labels, advance preparation, and relational language. Understanding these factors may help guide development of a framework for the design and critique of political forums.Tags: conversations, discussion, dissertation, experiment, plan, polarization, Politics, research