Comparing Estimation Methods for Network Models
The paper:
A Framework for the Comparison of Maximum Pseudo Likelihood and Maximum Likelihood Estimation of Exponential Family Random Graph
Models
by Marijtje A. van Duijn, Krista J. Gile, Mark S. Handcock in Social Networks, Volume 31, Issue 1, 2009, Pages 52-62
presents methodology to enable estimators of Exponential Family Random Graph model parameters to be compared.
We use this methodology to compare
the bias, standard errors, coverage rates and efficiency of maximum likelihood and maximum pseudolikelihood estimators. We also propose an
improved pseudo-likelihood estimation method aimed at reducing bias. The comparison is performed using simulated social network data based on
two versions of an empirically realistic network model, the first representing Lazega’s law firm data and the second a modified version with
increased transitivity. The framework considers estimation of both the natural and the mean-value parameters.
The software to reproduce the results in this paper are
here.