Abstract:
|
Subgroup structure is an inherent characteristic of most types of social networks in which ties are more common among actors within a subgroup than between actors in any two different subgroups. Subgroup integration describes the degree of separation across subgroups; more ties for actors in different subgroups indicates higher subgroup integration. In this paper, we incorporate network level covariates in the hierarchical mixed membership stochastic block model (HMMSBM) to estimate the effects of network attributes on subgroup integration. In addition, we present several simulation studies to evaluate the performance of HMMSBM with network-level covariates as well as a real-world data example to relate teacher practices on students' friendship network integration and provide insight of how this model can be used in practice.
|