Matthew T. Pietryka, Florida State University, firstname.lastname@example.org
Jack Lyons Reilly, New College of Florida, email@example.com
Daniel M. Maliniak, College of William & Mary, firstname.lastname@example.org
Patrick R. Miller, University of Kansas, email@example.com
Robert Huckfeldt, University of California, Davis, firstname.lastname@example.org
Ronald B. Rapoport, College of William & Mary, email@example.com
Donald Trump’s victory in the 2016 U.S. Presidential election left many journalists scrambling to explain how he won—and why so few liberals saw it coming. Many outlets advanced a common explanation: we are all living in political bubbles. We are surrounded by people who look and think like us. Social media algorithms accentuate that similarity by highlighting posts we are likely to enjoy and concealing messages that might draw our ire. As a result, we only see our views expressed by people we respect and we encounter opposing views only when they are shared derisively by our friends.
The assumption underpinning this logic is that is that many of us are susceptible to influence by a wide range of associates extending well beyond our closest connections. Had liberals only interacted with more Trump supporters, the logic goes, they would have been less surprised in the outcome and, perhaps, less aggrieved. This assumption may hold in online social networks, but previous research provides little insight into whether we are influenced by individuals beyond our few closest friends and family. In our article in Political Behavior, we examine this assumption.
Most previous research has been unable to address this question because real-world social networks are so difficult to measure. Typically, scholars measure social networks using a survey battery, which asks respondents to describe the handful of people they talk to most frequently. To avoid exhausting respondents, the surveys rarely ask them to describe more than five associates. As a result, we have learned much about how these closest associates influence people’s political attitudes and behavior, but we know little about how these outcomes are shaped by their more casual acquaintances.
To overcome this problem, we invited all students at College of William & Mary to respond to a survey asking about their political views and asking them to name their closest friends at the school. By focusing on this small, closed community, we were able to explore the role of these immediate friends as well as second-order contacts—associates identified by their friends. Thus we are able to examine how the views expressed in the broader social network help us understand and predict students’ awareness of local issues and participation in politics.
We show that, indeed, the broader social network may act as a filter, insulating students from important campus issues. But it can also act as a megaphone, exposing individuals to issues they and their closest friends have not experienced firsthand. The network also seems to condition the potential influence of individual friends, magnifying some voices while muting others. And while we show that immediate friends have the strongest relationship with individual attitudes, less immediate associates adds further explanatory power to our models.
We hope our work provides insight for scholars pondering how political bubbles and social interaction contribute to individual votes and beliefs. Together, our results suggest that models of political attitudes and behavior will gain additional traction if they can focus not only on immediate relationships, but also the broader network. Scholars should consider not just who individuals are connected to, but how they are connected and where those connections are situated in broader social environments.