CMStatistics 2015: Start Registration
View Submission - CMStatistics
B0151
Title: Analysis of longitudinal ordinal data with general random effects covariance matrix Authors:  Keunbaik Lee - Sungkyunkwan University (Korea, South) [presenting]
Jiyeong Kim - Sungkyunkwan University (Korea, South)
Abstract: To analyze the longitudinal categorical data, we typically use generalized linear mixed models (GLMMs). In the models, the random effects covariance matrix is used to demonstrate both the subject-specific and time variations, and the covariance matrix may also be heterogenous. However, the structure of the covariance matrix is assumed to be homogeneous and restricted because of the high dimension and the positive definiteness of the matrix. To release these assumptions two Cholesky decomposition methods were proposed in linear mixed models: the modified Cholesky (Pourahmadi, 1999) and the moving average Cholesky (Zhang and Leng, 2012) decompositions. In this paper we propose a cumulative logit random effects model with heterogeneous random effects covariance matrix for longitudinal ordinal data. We also exploit the two decompositions to model the random effects covariance matrix, and compare estimated parameters using the two decompositions. Methods are illustrated with a lung cancer data set.