Partisan Misperception Drives Self-Segregation in Cooperative Dilemmas

• In Event: Grassroots Politics

Abstract

Partisans frequently take costly actions on issues in which they have no direct stake. In doing so, they are effectively paying a cost to help other party members who actually do have a stake in the issue at hand. This type of interparty cooperation can be sustained through a number of mechanisms, such as indirect reciprocity or the punishment of non-cooperators within a party. Importantly, equilibrium concepts from evolutionary game theory show that these mechanisms can work even if individuals do not fully understand their logic. This is because any partisan who tries to deviate from a cooperative equilibrium may still learn, through negative feedback, that there’s no benefit to doing so.

However, because partisans can cooperate with one another without fully understanding the basis for their cooperation, they are also free to form some false beliefs about the origins of interparty cooperation. In particular, they might hold the simple, but false belief that co-partisans are unconditionally more cooperative than other individuals. If so, partisans should choose to self-segregate in any environment where there is a benefit to interacting with other cooperators. This, in turn, could help drive the highly segregated partisan communities observed in recent studies of social networks.

To test this hypothesis, we conducted two conjoint experiments where participants are able to use multiple demographics, including partisanship, to choose their partner in a monetarily incentivized Prisoner’s Dilemma game. Doing so, we show that partisans choose to interact more with co-partisans in a Prisoner’s Dilemma game, and the effect of shared partisanship is substantially larger than any other measured demographic trait. Furthermore, we use an incentivized belief elicitation procedure to show that this choice is driven by the false belief that co-partisans are unconditionally more likely to cooperate in this game.

Authors

• Jung Chen, University of California, Merced

• Matthew V. Hibbing, University of California, Merced

• Brad L. LeVeck, University of California, Merced