Social Influence and Political Expression: Evidence from an Online Experiment

• In Event: From Deliberation to Participation

Abstract

Decades of social science research suggests that our expression can be influenced by those around us. Focusing on social media, we ask: How do people learn to signal their political allegiances through language? Building on a previous large-scale observational study that found suggestive evidence of social influence over political speech, we develop a randomized intervention in ideological network structure, combining surveys and online behavioral data from Twitter to track whether people model their own political speech on that of high-status peer exemplars. Specifically, we study the relationship between the ideological composition of people's follow networks and how they express themselves on Twitter. We design a real-time digital field experiment randomly encouraging participants to follow a handful of actual Twitter accounts that frequently tweet with a liberal perspective on U.S. politics. To collect our outcomes, we link participants' public activity with their survey responses and apply automated methods to summarize the ideological signals in their tweets. We allow treatment and collection of outcomes to continue for up to two months.

Authors

• William Schulz, Princeton University

• Andy Guess, Princeton University

• Pablo Barbera, University of Southern California

• Simon Munzert, Humboldt University of Berlin

• JungHwan Yang, University of Wisconsin-Madison

• Sono Shah, Pew Research Center

• Emma Remy, Pew Research Center

• Aaron Smith, Pew Research Center