Homophily in Social Media and News Polarization

Abstract

We study how media bias is affected by the structure of social networks on social media. We consider an ad-financed media firm which chooses the ideological location of its news and targets consumers who can share the news with their followers on an online social media. After studying how a targeted consumer’s incentive to share the news is shaped by the network structure of her followers, we study the firm’s strategy to maximize the breadth of news sharing and find that when the mean (respectively, the variance) of followers’ ideological locations is a convex (respectively, concave) function of a direct consumer’s location, the media firm is likely to produce polarized news. The analysis of the case in which consumers are uniformly distributed reveals that news polarization is more likely to occur as the degree of homophily increases. We also find that media competition makes polarization more likely.