Peer Effects and Peer Avoidance: The Diffusion of Behavior in Coevolving Networks
研究动态网络中个体行为（好或坏）的长期演化，发现网络随行为变化（好个体间连接更持久）对好行为的维持有显著影响，为理解同伴压力和社会行为扩散提供新视角。
We study the long-run emergence of behavioral patterns in dynamic complex networks. Individuals can display two kinds of behavior: G (“good”) or B (“bad”). We assume that the exposure of a G agent to bad behavior on the part of peers/neighbors triggers her own switch to B behavior, but only temporarily. We model the implications of such peer effects as an epidemic process in the standard SIS (Susceptible-Infected-Susceptible) framework. The key novelty of our model is that, unlike in the received literature, the network is taken to change over time within the same time scale as behavior. Specifically, we posit that links connecting two G agents last longer, reflecting the idea that B agents tend to be avoided. The main concern of the paper is to understand the extent to which such biased network turnover may play a significant role in supporting G behavior in a social system. And indeed we find that network coevolution has nontrivial and interesting effects on long-run behavior. This yields fresh insights on the role of (endogenous) peer pressure on the diffusion of social behavior and also has some bearing on the traditional study of disease epidemics.