Reinflating the Filter Bubble: Domain versus URL Level Analyses of News Sharing

• In Event: Crises in Digital Journalism

Abstract

Observers of American politics have frequently expressed concerns about the possibility of “filter bubbles” to contribute to polarization. Recent work has begun to alleviate this concern, finding that partisans are exposed to sources shared by political opponents at relatively high rates. However, these studies are primarily conducted at the level of the source domain, as opposed to individual news stories at the level of the URL. It may be the case that partisans are particularly likely to share information from sources associated with their political opponents precisely when that information is favorable to them, as partisan-affirming information from unexpected sources may be perceived as more credible (Baum and Groeling 2009). Using a panel of over 1 million Twitter users matched to demographic and voter registration information, we test this dynamic by examining within-domain sharing patterns by user partisanship. If partisans do tend to share specific identity-confirming information from opposing sources, this would provide new evidence for “filter bubbles”. It would suggest that domain-level analyses of information sharing gives a false impression of exposure to politically diverse content, raising new concerns regarding polarization in the consumption and sharing of digital media.

Authors

• Jon Green, Northeastern University

• David Lazer, Northeastern University

• Sarah Shugars, New York University

• Stefan McCabe, Northeastern University