Online Communities and Social Network Structure

Abstract

This paper aims to understand the effects of online communities on the network structure and their marketing implications. To account for the fact that communities are not formed exogenously, we model and estimate how they co-evolve with the friendship network. We control for unobserved heterogeneity and homophily by modeling individuals’ latent positions and similarities. Using data from a worldwide Massively Multiplayer Online (MMO) Game, we find that individuals’ friendship formation decision and community membership decision reinforce each other. Moreover, model simulation reveals that, under some circumstances, online communities may promote not only within-community ties but also cross-community ties. Here, one important moderating factor is the community stickiness, which is captured by the positive effects of an individual’s number of friends in a community on her probability of staying in that community (instead of switching to a different community). As shown by our simulated diffusion exercises, only a moderately sticky community system promotes cross-community ties and consequently improves information reach. This insight helps social platforms manage online communities more effectively to improve the marketing value of social networks.