Structural Cohesion and Structural Diversity as Opposing Mechanisms of Social Influence

• In Event: 562 - Regular Session. Social Networks, Identities, and Inequality

Abstract

Although past work on social networks has studied how one person’s behavior can influence another’s, much of the social influence we experience comes from groups of people rather than singular individuals. In this paper, we make use of a novel dataset that records the online activity of over 43 million users in 40,000 online communities to demonstrate how the relationship between the network structure of an actor’s alters and the actor’s tendency to adopt the alters’ behavior depends on the nature of the behavior. Specifically, we show that a user is more likely to join an online group if the user’s friends in the group themselves lack friendship ties with one another. However, by contrast, we find that a user is more likely to attend an offline event if a user’s friends who have committed to attending the event possess friendship ties with one another. This sheds light on an important debate in social network research, which has produced contradictory results that support both connectedness – or structural cohesion – and disconnectedness – or structural diversity – among one’s alters as sources of group influence. Broadly, our results suggest that the diffusion of different phenomena, such as social protest or clothing fashions, is most effective when tailored to different social network structures.

Authors

• Dan Wang, Columbia University

• Jure Leskovec, Stanford University