Abstract

Some versions of structuralism consider actors to be necessary for structural analysis; others argue that they are not. All versions of structuralism consider social structure to be analytically independent of actors. I show through examples and subsequently through deduction that this position is wrong. That is, any conceptualization of social structure necessarily involves a conception of its constituent actors. Moreover, I generalize this point to argue that the structure of scientific knowledge follows a multilevel modeling approach: theory at every level of aggregation is based on actor models of appropriate units. Several practical benefits result from using actor models explicitly in structural analysis in particular, and from following the multilevel modeling approach in general