Abstract
In the classic blockmodel formulation, a social network among members of a population with n actors and k relations (types of tie) is arrayed as k n X n matrices. Though this is a three‐dimensional data structure, it is typically reduced to a two‐way analysis. In this paper, a three‐way procedure for analyzing multigraph data is developed. Specifically, in addition to applying the rule of structural equivalence to collapse actors, it is also applied to the relations (the third dimension), and structurally equivalent sets of relations are collapsed. The result is a three‐dimensional blockmodel (image) of social structure that is a more parsimonious representation of social structure than the classic two‐dimensional blockmodel images. The three‐dimensional approach is illustrated by application to three case studies: Homan's Bank Wiring Room, Sampson's monastery, and a local economy of hospital services. The structural equivalence approach to relations is further explored by applying it to the individual‐level Liking and Antagonism relations and their compounds (of length four or less) in the Bank Wiring Room. This application demonstrates that the structural equivalence approach can be used to identify equality equations for primitive and compound relations.