Intersectional homophily: Accounting for Multiple Dimensions of Homophily

• In Event: 40416 - New approaches to social network theory and methods

Abstract

Scholars frequently report the presence of homophily – that network ties are more likely between nodes with one or more shared characteristics – in interpersonal networks. In inferential network modeling, theses homophily findings are generated through an ubiquitous strategy of estimating independent homophily effects for each sociodemographic characteristic under study. More recently, several scholars have proposed strategies for accounting for multiple dimensions of characteristics and their roles in homophily, but theoretical and empirical work on this issue remains limited. In this paper, we develop a theory of intersectional homophily as an alternative to social distance, by-product, and consolidation arguments and estimate exponential random graph models (ERGM) with homophily interactions to examine expectations related to intersectional homophily arguments. We utilize Wave I of the Student Experiences in Law School Study, which surveyed first-year JD students in the fall of 2019 at three law schools. We analyze these data using ERGMs, which include main effects for homophily based on race, gender, status, sexual identity, and political orientation as well as two-way interactions. Results show distinct patterns of significant main and interaction effects, which cannot be fully explained by the consolidation between attributes. The findings support the notion that homophily research should examine how two forms of homophily intersect, above and beyond the main effects of homophily.

Authors

• Chen-Shuo Hong, National Taiwan University

• Anthony Paik, University of Massachusetts-Amherst