Random Effects Models for Personal Networks

We propose analyzing personal or ego-centered network data by means of two-level generalized linear models. The approach is illustrated with an example in which we assess whether personal networks are homogenous with respect to marital status after controlling for age homogeneity. In this example, the outcome variable is a bivariate categorical response variable (alter’s marital status and age category). We apply both factor-analytic parametric and latent-class-based nonparametric random effects models and compare the results obtained with the two approaches. The proposed models can be estimated with the Latent GOLD program for latent class analysis.