In 2017-2022 we have studied patterns of homophily, i.e. mechanism that describes when similar individuals to engage or form relationship with similar individuals, using various data sources. Our initial investigation focused on using WIFI data to determine if stuents on-campus would be spending time with students of similar socio-demographic characteristics and of similar performance. We used WIFI data collected at the University of Michigan to observe that this was, in fact, the case. In the follow-up work, we examined if such grade-based homophily (peers with similar grades) would be observed in online discussion forums in university settings. We have examined this using data from over 14 thousand students from the University of South Australia, National University of Singapore, and the University of Michigan, In this first comparative study of three universities, we show that grade-based homophily is prevalent in online discussions. We examine causal mechanisms explaining this observed homophily, with both social selection (similar peers choose to talk to similar peers) and social influence (peers influence their peers, resulting in similar performance levels) hypothesis gaining little traction. In the follow up work, we examined causal mechanisms explaining this observed discussion-based homophily, particularly focusing on the elements of pedagogical design and structures of discussion forums due to which students with similar performance end up interacting with one another. 

Research outputs characterizing this line of research: