| In many longitudinal studies, repeated observations of clustered ordinal outcomes along with several covariates are taken from a sample of subjects at irregular time points, resulting in clustered longitudinal ordinal data. When the outcomes fluctuate over time, it is often of interest to investigate the relationship between the trend of the underlying outcome processes and the covariates. One of the proposed models available for trend analysis of this kind is the univariate local equilibrium distribution model of Kosorok and Chao (1996). In this talk, we extend their model to a multivariate model for clustered outcomes. Global odds ratios are used to measure the association among clustered outcomes. Extenstion feasibility is examined from the estimationpoint of view. Real data analysis will be provided for illustration. |