The Effect of Conversation with an Outparty Member on Outparty Animosity

• In Event: Conversations with the Enemy

Abstract

Conventional wisdom dictates that we must talk about our political differences in order to heal America's bitter partisan divide. Not only do media, politicians, and pundits cite talking politics as a way to heal our hostile attitudes toward one another, but nonprofits spend a great deal of money promoting this philosophy. But because anticipating political discussion makes people anxious and they prefer to avoid it, especially when it is with an outparty member, conversations across party lines are likely to feature non-political topics as well. In this paper, I am interested in the interparty conversations---both political and non-political---that occur in our daily lives. Of particular interest is how interparty conversation affects the strong, negative affect American's hold for the outparty. Americans think outparty members are hypocritical, selfish, and mean; Americans do not want their children to marry members of the outparty; and Americans discriminate against outparty members in non-political settings such as the allocation of college scholarships, in the review of job applications, and more. What is the effect of everyday conversations---those that touch on politics and those that do not---on the heightened outparty animosity that characterizes today's political climate?

Because partisanship in America is a strong social and group identity, theory on intergroup relations sheds light on this question. Social identity theory assumes that individuals need to maintain a positive distinction for the ingroup relative to the outgroup. However, this ingroup favoritism need not imply outgroup hostility or discrimination. How can outparty animosity be improved while maintaining intergroup distinctiveness? Allport's contact hypothesis states that positive intergroup contact, under certain conditions, can promote tolerance and acceptance of the outgroup. Because outparty affect is at heightened negative levels, even significant improvement in outgroup affect can occur without threatening ingroup favoritism. In the main, I expect positive interparty conversation to improve negative affect toward that outgroup member, regardless of topic of conversation. Furthermore, I expect conversation with an outgroup member to generalize to other members of the outgroup because it features personalization, whereby the interaction is understood more at the interpersonal level than the intergroup level. What’s more, I expect personalization to moderate the effects of contact. Because political conversation is less personalized than non-political conversations, I expect political conversations to dampen the effect of interparty contact on outgroup animus.

In order to test my hypotheses regarding interparty conversations, I use two novel methodological approaches. First, I designed a software called "Chatter" by which participants can have real-time, written conversations online. Chatter allows researchers to relatively easily emulate a real social experience when experimentally studying conversation. Additionally, via online conversation, Chatter facilitates the isolation of the causal effect of engaging in conversation on important political outcomes from other phenomena that may influence how a conversation unfolds. Second, I use a blocked-cluster experimental design which I adapt for experiments involving group conversation. Blocking is useful because it can improve the efficiency of these experiments by ensuring balance on relevant covariates. I block with clusters because inherent in the study of actual political conversation are groups, or clusters, of people. The typical blocked-cluster design features pre-existing clusters like cities or classrooms. Critically, in a lab experiment, the researcher is tasked with creating the relevant clusters (i.e., discussion groups). While guidance and tools exist for blocking with pre-existing clusters, it is not clear how to simultaneously block and cluster units. I outline an algorithm to create a blocked-cluster design when the researcher also controls creating the clusters themselves.

Using this software and experimental design, I find support for my hypotheses. Using an experiment involving conversations amongst Amazon Mechanical Turk workers, I find that both political and non-political conversations can decrease outparty animosity, but that non-political conversations do so to a greater extent. In all, this paper helps us understand how interparty conversations unfold given today's heightened animosity across party lines and how the effects of these interpersonal conversations generalize to outparty as a whole.

Author

• Erin Rossiter