|
Activity Number:
|
284
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #305507 |
|
Title:
|
Model Selection in Linear Mixed Effects Models
|
|
Author(s):
|
Ying Lu*+ and Heng Peng
|
|
Companies:
|
University of Colorado at Boulder and The Hong Kong Baptist University
|
|
Address:
|
Ketchum Hall, Boulder, CO, 80309,
|
|
Keywords:
|
model selection ; penalized least squares ; oracle property ; group selection ; random effects
|
|
Abstract:
|
Mixed effect models are fundamental tools for the analysis of longitudinal data, panel data and cross-sectional data. However, the complex nature of these models have made variable selection and parameter estimation a challenging problem. In this paper, we propose a simple iterative procedure that estimates and selects fixed and random effects for linear mixed models. In particular, this method utilizes the partial consistency property of the random effects so it selects the group of random effect coefficients simultaneously via a data-oriented penalty function (the smoothly clipped absolute deviation function). Simulation studies and data analysis are conducted to empirically examine the performance of this procedure under different sample sizes. We show that the proposed method is a consistent variable selection procedure with Oracle properties.
|
