|
Activity Number:
|
379
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #309261 |
|
Title:
|
Modeling Covariance Structure in Unbalanced Longitudinal Data
|
|
Author(s):
|
Min Chen*+ and Jianhua Z. Huang
|
|
Companies:
|
Texas A&M University and Texas A&M University
|
|
Address:
|
Statistics Department, 1100 Hensel Dr., Apt. X2H, College Station, TX, 77840,
|
|
Keywords:
|
logitudinal ; unbalanced ; covariance ; reparameterization ; EM algorithm ; decomposition
|
|
Abstract:
|
Modeling covariance structure is important in efficient estimation for longitudinal data models. Pourahmadi (1999,2000) promoted to use modified Cholesky decomposition as an unconstrained reparameterization of the covariance matrix. The new parameters have transparent statistical interpretations and are easily modeled using covariates. However, this approach is not directly applicable when the longitudinal data are unbalanced, because a Cholesky factorization for observed data coherent across all subjects usually does not exist. We overcome the difficulty by treating the problem as a missing data problem and employing a generalized EM algorithm to compute the ML estimators. We illustrate our method by reanalyzing Kenward's (1987) cattle data and conducting a simulation study.
|
