Do Online Video Recommendation Algorithms Increase Polarization?

• In Event: Social Media and Polarization

Abstract

Do personalization algorithms serve up ever more extreme videos to users, resulting in attitudinal or affective polarization? Academic researchers have struggled to add evidence to this debate due to both data limitations and serious research design challenges. We propose a novel experimental approach that seeks to resolve the impasse on this critical question by randomizing the recommendations themselves. Our team has built an interface that serves sequences of videos using naturalistic recommendations drawn directly from YouTube's API. We create our own variations on the existing algorithm by injecting differing proportions of cross-cutting video content into experimentally modified recommendation systems, then randomly assign these modified algorithms to subjects. This approach enables an attractive combination of experimental control and external validity without the need for reverse-engineering YouTube's black-box recommendations. Subjects in our study choose to watch a sequence of short videos and follow recommendations while the interface tracks watch time and other measures of engagement. This is a three-wave design, with a pre and post survey one week before and after the viewing session. Furthermore, right after the viewing session, participants are surveyed on issue opinions, affective polarization, and perceptions of media hostility to their preferred viewpoint. We present results from studies of gun control and other issue areas, providing rigorous evidence of the extent and magnitude of the effect of recommendation algorithms on polarization.

Authors

• Naijia Liu, Princeton University

• Matthew A. Baum, Harvard University

• Adam J. Berinsky

• Justin de Benedictis-Kessner, Harvard University

• Andy Guess, Princeton University

• Dean Knox, University of Pennsylvania

• Christopher Lucas, Washington University in St. Louis

• Brandon Michael Stewart, Princeton University

• Jason Anastasopoulos, University of Georgia