Fitting mixed-effects models for repeated ordinal outcomes with the NLMIXED procedure

Abstract

This paper presents an analysis of repeated ordinal outcomes arising from two psychological studies. The first case is a repeated measures analysis of variance; the second is a mixed-effects regression in a longitudinal design. In both, the subject-specific variation is modeled by including random effects in the linear predictor (inside a link function) of a generalized linear model. The NLMIXED procedure in SAS is used to fit the mixed-effects models for the categorical response data. The presentation emphasizes the parallel between the model specifications and the SAS statements. The purpose of this paper is to facilitate the use of mixed-effects models in the analysis of repeated ordinal outcomes.