The methodology described here is designed for social networks and is
based on the research of Holland and Leinhardt, Wasserman and Iacobucc
i, and many others. Holland and Leinhardt termed the simplest model fo
rm their family of log-linear models pt. The models presented in this
article are not log-linear-rather they are log-multiplicative, in the
spirit of other models described in this special issue. Our models gen
eralize the p(1) family of models for social networks by introducing m
ultiplicative interaction parameters. These generalizations are applic
able to a much wider range of data, particularly valued relations.