|
Activity Number:
|
431
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
IMS
|
| Abstract - #305703 |
|
Title:
|
Robust Covariance Structure Estimation in Linear Mixed Models
|
|
Author(s):
|
Xueliang Pan*+ and David Jarjoura
|
|
Companies:
|
The Ohio State University and The Ohio State University
|
|
Address:
|
Center for Biostatistics, Columbus, OH, 43220,
|
|
Keywords:
|
covariance structure ; repeated measure ; contrast ; mixed model
|
|
Abstract:
|
With repeated conditions or measurements over time unrestricted covariance structures may involve too many parameters when faced with a limited number of replications. We provide some examples to identify methods to ensure that an adopted simplified covariance structure will be adequate for testing fixed effects with little or no bias. In particular, we elucidate how fixed effects contrasts, that answer the primary question of the studies, combines the covariance parameters for hypothesis testing, and how to avoid over-simplifying the covariance structure. An example of over-simplification is when not all interactions between random effects and these fixed effect contrasts are included in the model. We use simulations and a few examples to illustrate approaches.
|
