Major barriers to the stochastic modeling of social networks are the specification of realistic models, the computational difficulties of the inferential methods, and assessment of the goodness-of-fit. Exponential-family Random Graph Models (ERGM) have shown themselves to be a useful class of models for representing complex social phenomena. However, they suffer from these and other deficiencies.
In this talk we discuss alternatives to ERGM that retain many of their desirable properties while addressing their deficiencies.
|