Generalized Estimation of the Treatment Effects in Mixed-Effects Model: A Comparison with ML and REML

• Activity Number: 522
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Ching-Ray Yu*+ and Kelly Zou and Martin M Carlsson and Samaradasa Weerahandi
• Companies: Pfizer Inc. and Pfizer Inc. and Pfizer Inc. and Pfizer Inc.
• Keywords: Mixed Models ; Best Linear Unbiased Predictor ; Generalized Estimate ; Maximum Likelihood ; Restricted Maximum Likelihood

Abstract

In applications of Mixed Models as multicenter studies, the covariates can be mixed with fixed (e.g. treatment) and random (e.g. age ) effects. The best linear unbiased predictor (BLUP) in Mixed Models is a function of the variance components and they are typically estimated using conventional maximum likelihood (ML) or restricted ML (REML) methods. In practice, frequently non-convergence of BLUP would occur due to a drawback of the standard Likelihood based approaches. In such situations, ML and REML either do not provide any BLUPs or random effects all become practically zero. This is because there is no guarantee that such methods lead to positive variance components even in the simple case of a one-way balanced layout involving one factor variance and the error variance. To overcome this drawback by taking the generalized approach to inference ( Tsui and Weerahandi, 1989; Weerahandi, 1993; Weerahandi,2012), we provide a Generalized Estimate (GE) of BLUP that does not suffer from the problem of negative or zero variance components, and compare its performance against the ML and REML estimates of BLUP. Simulated and published data are used to compare BLUP under various variance.