Using latent variables to account for heterogeneity in exponential family random graph models

We consider relaxing the homogeneity assumption in exponential family random graph models (ERGMs) using binary latent class indicators. This may be interpreted as combining a posteriori blockmodelling with ERGMs, relaxing the independence assumptions of the former and the homogeneity assumptions of the latter. We propose a Markov chain Monte Carlo al- gorithm for drawing from the joint posterior of the model parameters and latent class indicator