Do I Follow My Friends or the Crowd? Information Cascades in Online Movie Ratings

Abstract

Online product ratings are widely available on the Internet and are known to influence prospective buyers. An emerging literature has started to look at how ratings are generated and in particular how they are influenced by prior ratings. We study the social influence of prior ratings and, in particular, investigate any differential impact of prior ratings by strangers (“crowd”) versus friends. We find evidence of both herding and differentiation behavior in crowd ratings wherein users’ ratings are influenced positively or negatively by prior ratings depending on movie popularity. In contrast, friends’ ratings always induce herding. Further, the presence of social networking reduces the likelihood of herding on prior ratings by the crowd. Finally, we find that an increase in the number of friends who can potentially observe a user’s rating (“audience size”) has a positive impact on ratings. These findings raise questions about the reliability of ratings as unbiased indicators of quality and advocate the need for techniques to de-bias rating systems.

Keywords: online word of mouth, ratings, social influences, informational cascades, latent variables, multilevel models, online social media