This article presents the results of a controlled experiment on the influence of social network structure on the health behaviors of individuals. The author examines the spread of behavior through a network where connections are clustered and another network where connections are dispersed. Previous research on infectious disease and social networks has suggested that behavior, like simple contagion, spreads faster through a widely dispersed network.

The author created an online health community where individuals were assigned “health buddies.” In one community, health buddies were created in clusters; in the other, health buddy connections were randomized. The creators of the experiment then introduced a health behavior to the communities and analyzed the spread of the behavior through the communities.

Key Findings:

• Health behaviors spread faster in highly clustered networks than in randomized networks. Approximately half of individuals in the clustered communities adopted the health behavior, while 38 percent of individuals in the randomized networks adopted the health behavior.
• Individuals who had multiple health buddies who adopted the health behavior were significantly more likely to adopt the health behavior themselves. They were also significantly more likely to repeat the health behavior more than once.

This research suggests that highly clustered social networks spread health behaviors more quickly than randomized networks. Public health interventions that make use of social network models may benefit from targeting clustered social networks.