What are P * Models?

The p* family of models developed by Wasserman and Pattison (1996), Pattison and Wasserman (1999), and Robins, Pattison, and Wasserman (1999) are based on the pathbreaking Markov spatial interaction models for random graphs of Frank and Strauss (1986) and Strauss and Ikeda (1990). These models allow researchers to break free of the severe independence assumptions of earlier statistical models for social networks, permitting a very general dependence structure for the network quantities. Further, the p* formulation allows network measurements to be viewed in a standard response/explanatory variables setting in which the response variable is the log odds of the probability that a relational tie is present. The explanatory variables can be quite general, including network structural properties like the tendency towards mutuality or transitivity; nominal, discrete or continuous actor attributes, as well as the interactions between these elements.

p* Home Page