A Finite Mixture Model for Working Correlation Matrices in Generalized Estimating Equations

• Activity Number: 348
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: International Chinese Statistical Association
• Author(s): Nan Lin* and Lili Xu and Baoxue Zhang+
• Companies: Washington University in St. Louis and Northeast Normal University, China and Northeast Normal University, China
• Keywords: Longitudinal data ; GEE ; Working correlation ; Finite mixture ; Pseudo-likelihood ; EM algorithm

Abstract

It is well known that the efficiency of the generalized estimating equations (GEE) estimator can be seriously affected by if the working correlation matrix is misspecified. To address the misspecification issue, we propose a finite mixture model for the working correlation called mix-GEE. Under mild regularity conditions, the mix-GEE estimator is consistent and asymptotically normal. And it is also asymptotically efficient if data are from a Guassian mixture model. An important feature of the mix-GEE method is that it guarantees the positive definiteness of the estimated working correlation matrix if either the AR(1) or exchangeable structure is included. So, it is numerically more stable and displays a better finite sample efficiency than the hybrid GEE method (Leung, Wang and Zhu, 2009). The value of our method is further demonstrated by simulation studies and real data examples.