The Effects of Online Video Personalization Algorithms

• In Event: Computational Social Science: New Approaches

Abstract

An explosion of internet media outlets and the ubiquity of personalized recommendation algorithms has led to concern about partisan “information bubbles,” which are thought to increasingly contribute to an under-informed and politically polarized public. A vast body of work examines the nature, extent, and effects of these bubbles on individuals’ political attitudes and behavior in the context of traditional cable TV news or textual news. But little is known about online video platforms, which can personalize recommendations based on past viewing histories and are often associated with extremist ideas. We study the effects of online video using a controlled design that randomly manipulates the algorithm used to recommend videos to subjects. By experimentally injecting diversity into the stream of naturalistic suggestions on our own online video platform, we simultaneously consider interacting and self-reinforcing processes in the supply, demand, and persuasive effects of media consumption. In particular, we examine the extent to which individuals take up counter-attitudinal viewpoints when given the option; whether initial exposure to diverse perspectives begets more willingness to listen or a retreat to partisan silos; if it persuades individuals to moderate their opinions or instead produces a “backlash”; and how all of the above is moderated by partisanship and preexisting beliefs.

Authors

• Dean Knox, University of Pennsylvania

• Andy Guess, Princeton University

• Brandon Michael Stewart, Princeton University

• Jason Anastasopoulos, University of Georgia

• Adam J. Berinsky

• Matthew A. Baum, Harvard University

• Justin de Benedictis-Kessner, Harvard University

• Christopher Lucas, Washington University in St. Louis

• Naijia Liu, Princeton University