Networks and Neighborhoods: Modelling generalised network structure

In this talk I propose that the task of constructing models for social networks is assisted by recognising that the social processes underlying network formation are interactive, dynamic, and socially-situated. (They are dynamic and interactive in that the formation of a tie between a pair of actors may both depend on and, in turn, influence the formation of ties among other pairs of actors; they are socially-situated in that network ties depend on, and also, in turn, influence, a multi-layered complex of social entities that includes not only other social relations, but also group affiliations, sociocultural and psychological orientations, and social and geographical settings.) I briefly review how exponential random graph models represent networks as the outcome of such interactive processes, and I argue that the problem of model specification is the problem of specifying plausible neighbourhood assumptions for the models from our knowledge of these interactive processes. I then present two extensions of the general modelling approach that emphasise, in turn, the socially-situated and dynamic nature of the underlying network processes. I conclude with some comments on the modelling problems and prospects raised by these developments.