Different Worlds, Same Address: Social Voting Among Dissimilar Roommates

• In Event: Papers That Study Voting Behavior With Clever Identification Strategies

Abstract

We examine the extent to which people are mobilized or demobilized by *exposure to difference*—interpersonal interaction with someone who does not share their perspective or background. This topic has received significant attention from normative theories of deliberative democracy (Arendt 2013; Mill 1869) and empirical debates in political behavior (Huckfeldt, Johnson, and Sprague 2004; Key 1949; Matthews and Prothro 1963; Mutz 2006). Yet credible causal inferences about this topic have been difficult to obtain, in part, because measuring social interaction and social diversity is often challenging—see for example the ongoing debate about whether US citizens are sorting into increasingly homogenous neighborhoods (e.g., Abrams and Fiorina 2012; Bishop 2009; Mummolo and Nall 2016) —and few studies can disentangle the forces that shape the diversity of one’s social context from the effects of this context (Shalizi and Thomas 2011).

We address these observational challenges with fine-grained data and a research design that allow us to examine the effects of exposure to difference on political participation among first-year college students living in dormitories at Florida State University. Our rich archival records overcome the measurement challenges that plague many studies, allowing us to operationalize exposure to difference in three distinct ways that have each received extensive attention in the literature (Eveland, Appiah, and Beck 2017): the random assignment of roommates from different partisan backgrounds, racial and ethnic backgrounds, and socioeconomic backgrounds. Perhaps more importantly, the random assignment of roommates allows us to overcome observational challenges common to most studies of social influence including the inability to disentangle *selection into social interaction* from *the effects* of such interaction.

We have paired our roommate data with validated voter files, allowing us to examine the immediate effects on exposure to difference on turnout in the 2016 presidential election and the downstream effects on turnout in the 2018 midterm elections. Further, our university records provide students’ home addresses, allowing us to measure the partisan, racial/ethnic, and socioeconomic attributes in their childhood households, neighborhoods, and high schools. For instance, our records provide validated voting behavior of the students’ parents dating back to 2006, including participation in partisan primaries.

With these data, we examine differences in turnout based on whether students are randomly assigned roommates with similar or different partisan, racial, and socioeconomic upbringings. These data allow us to study a common setting in which people encounter difference at a time in their lives that can shape their political orientations for the rest of their lives (Holbein and Hillygus 2020).

Authors

• Brad T. Gomez, Florida State University

• Matthew Pietryka, Florida State University