
70 Ordered Regression Models
3.4 Partial Adjacent Category Models
Partial adjacent category models are an extension of the parallel adjacent category model
that relaxes the parallel regression assumption for a subset of variables. Additionally, the
partial adjacent category logit model is a constrained version of the more familiar multi-
nomial logit model. Partial adjacent category models do not have the same internal incon-
sistency found in partial cumulative models. However, like the stage models, the adjacent
models essentially select on the dependent variable. The cutpoint equations only involve
respondents in two adjacent categories. As a ...