How does socioeconomic homophily emerge? Testing for the contribution of different processes to socioeconomic segregation in adolescent friendships

Abstract

Homophily – the fact that friendships happen at a higher rate among similar individuals – does not necessarily imply homophilic selection – the tendency to look for similar friends. This is particularly true for socioeconomic homophily: because individuals’ social class impacts most aspects of their lives, there are several ways in which it can favor homogeneity in friendship networks. Applying this view to the relationships of French middle-school students, the present article tries to unravel the contribution of various relational processes to the emergence of socioeconomic homophily. Stochastic Actor-Oriented Models, a class of generative models designed for network panel data, are applied to the friendship networks of 820 students surveyed over a year and a half. Simulations derived from the estimated models are then used to assess the impact of different processes on aggregated levels of homophily. To that aim, a new metric is proposed that help researchers decompose an observed property of a network into a set of contributions from low-order processes, called “contribution scores”. Results suggest that direct homophilic selection can be important in explaining socioeconomic homophily, but not in all cases. Indirect inducers, such as residential propinquity or ethnic selection, also play a significant role. Moreover, endogenous network processes – namely reciprocation and transitive closure – strongly contribute to homophily by reinforcing other homophily-inducing processes.