Investigation of covariance structures in modelling longitudinal ordinal responses with skew normal random effect

摘要

In this article, a mixed ordinal responses is presented with fundamental skew normal random effects and a link function based on fundamental skew normal is used due to the greater flexibility. An appropriate random effect variance covariance (VC) structure is determined to obtain reliable statistical inferences. Therefore, various types of random effects VC structures are considered. We use simulation to find parameters estimates properties and the model is applied for analyzing the Schizophrenia Collaborative Study data. Our results show that AR(1) VC structure improves parameters estimates substantially and the drugs effect significantly on the schizophrenia treatment over time.