A Unified Framework for Inference in Network Models with Degree Heterogeneity and Homophily

Abstract

The degree heterogeneity and homophily are two typical features in network data. In this paper, we formulate a general model for undirected networks with these two features and present the moment estimation for inferring the degree and homophily parameters. The binary or nonbinary network edges are simultaneously considered. We establish a unified theoretical framework under which the consistency of the moment estimator holds as the size of networks goes to infinity. We also derive the asymptotic representation of the moment estimator that can be used to characterize its limiting distribution. The asymptotic representation of the moment estimator of the homophily parameter contains a bias term. Two applications are provided to illustrate the theoretical result. Numerical studies and a real data analysis demonstrate our theoretical findings.

arXiv comment33 pages. Major revision. Add simulations. Changed the original title to "A Unified Framework for Inference in Network Models with Degree Heterogeneity and Homophily"

References in corpus (8)

• Stochastic blockmodels and community structure in networks
• The statistical mechanics of networks
• A central limit theorem in the $\beta$-model for undirected random graphs with a diverging number of vertices
• Asymptotics in directed exponential random graph models with an increasing bi-degree sequence
• Null models for network data
• Maximum entropy distributions on graphs
• Asymptotic normality in the maximum entropy models on graphs with an increasing number of parameters
• A Probit Network Model with Arbitrary Dependence

Cited by in corpus (1)

• A network Poisson model for weighted directed networks with covariates