An extended class of RC association models: definition, properties and estimation
share
The extended RC association models introduced in this paper allow the user to select the type of logit (local, global, continuation, reverse continuation) suitable for the row and column variables and the form of the divergence measure which, as in Kateri and Papaioannou (1994) may be used to define an extended class of bivariate interactions. An algorithm for computing maximum likelihood estimates is described; it exploits a reduced rank property without the need for identifiability constraints. Linear constraints that define parsimonious models are allowed on marginal logits and generalized interactions. An application to social mobility data is presented and discussed. It is also shown that the models proposed here are determined by an optimality property which is analogous to the one described by Gilula et al. (1988) for ordinary RC association models and extended by Kateri and Papaioannou (1994).
READ FULL TEXT