Formal methods for dealing with change in social networks have been the subject of increasing interest among social scientists (e.g., Doreian, 1979, 1980, 1986; Hallinan, 1978; Holland & Leinhardt, 1977; Hunter, 1978; Iacobucci & Wasserman, 1988; Killworth & Bernard, 1976b; Runger & Wasserman, 1979; Wasserman, 1980; Wasserman & Iacobucci, 1988). Two distinct approaches to the development of these methods have emerged. In one, a social network is represented as a single relation matrix, and explicit “process” models for change in the constituent social relationships have been sought; for example, Wasserman's (1980) reciprocity model represents a relationship between any pair of network members as a stochastic function of characteristics of the relationship at an earlier time. In the other approach, a more complex structural representation of the social network is constructed, and the more modest goal in relation to change has been a language for the description of structural “evolution”; Doreian (1980), for example, presents an analysis of the changing structure in a group that is implied by Atkin's (1977) representation of structure using combinatorial topology.

The coupling of more complex structural representations for social network data and more descriptive approaches to accounts of their changing structure is not accidental. Not only are the analytic difficulties in developing complex structural models for change much more forbidding – because, in general, potential models are more numerous and undoubtedly less tractable – but also the structural models are usually constructed on the basis of some continuity in time.