<< more recent presentation << || view list || >> earlier presentation >>

“Actor-driven alternatives to exponential random graph models.”

Today, the most prominent tool for the analysis of complete networks is the family of exponential random graph (ERG, aka p*) models. While mathematically elegant because equipped with a transparent probability distribution function, these models are notoriously difficult to estimate and/or estimates, once obtained, are difficult to interpret. In this paper, alternatives to ERG models are proposed which are derived as equilibria of actor-driven models for longitudinal network data. The paper discusses the merits and problems of such an approach vis-à-vis ERG models as well as longitudinal models.

The slides were presented at the XXVI Sunbelt conference (24-30 April 2006, Vancouver)

download slides as PDF (184 KB)