iSchool Research Symposium: Sabina Tomkins
Title: Can predictive models help us understand political behavior?
Abstract: This talk will discuss projects which use predictive modeling to address political science questions. The bulk of the talk will be about political partisanship in the United States. Partisanship is an essential aspect of political behavior, especially in the United States. One view of partisanship is as a calculated vote towards a party platform, the more a party's platform aligns with your policy preferences, the more likely you are to support that party. Alternatively, parties may be viewed as compositions of one's social group identities. That is, the more a party represents you as an amalgamation of social identities, the more likely you are to support that party. As partisanship is increasingly employed across theories of political behavior, it is essential to understand how one's party identification is influenced by issues vs. identity. We conduct two comprehensive studies, one of political identity and one of issue positions. We show that for the same respondents incorporating issue stance is much more predictive of partisanship than social group identity. Social group identity adds little predictive power beyond issue positions. Next, we explore the extent to which social group identity is useful in predicting issue position and importance. Here we see that even when accounting for party identification, for a wide variety of issues, some subset of social group identities provides meaningful and statistically significant predictive power to where people stand on issues. And, this is even more extreme for prioritization of issues. Thus, while social group identity is not significant to understanding partisan identification in the presence of issue positions, it is significant in predicting issue positions in the presence of partisan identification.
Speaker Bio: Sabina Tomkins is an Assistant Professor at the University of Michigan School of Information. Her research is in the area of computational public policy; she uses computational tools in order to both understand and intervene within social systems and focuses on the substantive areas of educational access, environmental sustainability and political participation.