Applying the Frailty ERGM (FERGM) Through six substantive applications (MesaHigh, Florentine Marriage, Military Alliances, Militarized Interstate Disputes, Regional Planning, Brain Complexity), we illustrate how unobserved heterogeneity and confounding can lead to degenerate inferential network model specifications and errors, as well as poor model fit. The figure shows the improvements made by accounting for unit-level heterogeneity with the FERGM. The percent improvement is calculated as the difference in the mean accuracy of tie prediction for the FERGM relative to the ERGM across 500 simulations. Across most of these networks, we find a fairly large increase in the percent of ties accurately predicted by using the FERGM.
Loading...















