“Closure, constraint and homophily:
Joint determinants of network segregation.”

Since the seminal work by Schelling, agent-based models of network segregation have largely focused on effects on the level of dyads, such as homophile selection (or heterophobe de-selection) of network neighbors. In this paper, it is shown how the empirically well-corroborated effects of transitive closure (on the level of triads) and constraint (on the level of single actors) can crucially determine a variety of network segregation measures, and interact with tendencies towards homophile selection. The SIENA software is first used to obtain comparatively realistic models for network evolution from an empirical data set. Based on these analyses, a meaningful region in the parameter space is identifed. By investigating simulated network evolution processes in this region (also obtained with SIENA), the impact of transitive closure, homophily, and their interaction on various indices of network segregation is studied, while controlling for the constraint dimensions out-degree and reciprocity.

The slides were presented at the XXVII Sunbelt Social Network Conference (1-6 May 2007, Corfu)

download slides as PDF (922 KB)