U-Shaped Conformity in Online Social Networks

We explore how people balance their needs to belong and to be different from their friends by studying their choices of wall color in a virtual house on a leading Chinese social networking site. The setting enables us to randomize both the popular color and the adoption rate at the individual level so that our experimental design minimizes informational social influence, homophily, and group-identity signaling to the general public. We find that there exists significant social influence within a user's friend circle. While learning about the most popular color among a user's friends generally increases the likelihood for the user to adopt that color, conformity first decreases and then increases with the adoption rate of that choice, which ranges from 50% to 100%. In addition, users who are of a minority or lower socioeconomic status or are newer are more likely to conform upon learning about the popular choice. Our findings are consistent with optimal distinctiveness and middle-status conformity theories and have implications for designing normative marketing campaigns.