Multivariate t Linear Mixed Models with AR(p) Errors for Multiple Longitudinal Data

• Activity Number: 523
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Computing
• Author(s): Wan-Lun Wang*+ and Tsai-Hung Fan
• Companies: National Central University, Taiwan and National Central University, Taiwan
• Keywords: ECME algorithm ; MLMM ; MtLMM ; Outliers ; Random effects ; Score test

Abstract

The multivariate linear mixed model (MLMM) has been a widely used tool for a joint analysis of more than one series of longitudinal data. Motivated by a concern of sensitivity to potential outliers and data with longer-than-normal tails or possible serial correlation, we aim to develop a robust generalization of the MLMM, which is constructed by using the multivariate t distribution on the random effects and the within-subject errors with AR(p)dependence structure. A score test for inspection of the existence of autocorrelation among within-subject errors is derived. An efficient ECME procedure is developed for computing the maximum likelihood estimates. We demonstrate the proposed methodology on a dataset of the ACTG 175 study.