|
Activity Number:
|
53
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 8, 2004 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
General Methodology
|
| Abstract - #300700 |
|
Title:
|
A Unified Approach to Studying Two-level Structural Equation Models and Linear Mixed Effects Models
|
|
Author(s):
|
Jiajuan Liang*+ and Peter M. Bentler
|
|
Companies:
|
University of New Haven and University of California, Los Angeles
|
|
Address:
|
300 Orange Ave., West Haven, CT, 06516,
|
|
Keywords:
|
covariance structure ; EM algorithm ; linear mixed effects model ; mean structure ; two-level structural equation model
|
|
Abstract:
|
Two-level structural equation models and linear mixed effects models are two different types of statistical models. Existing publications usually study these two types of models separately. We will develop a unified approach to studying the two types of models simultaneously. The idea is based on the similar structure of the two types of models. A two-level structural equation model consists of level-1 effect and level-2 effect. A linear mixed effects model consists of a mixed effect and a random effect. So there are some common points in the structure of these two types of statistical models. In our unified approach, we will consider the mixed effect as the mean structure and the random effect as the covariance structure. Based on this restructuring of linear mixed effects models, we can obtain the same formulation of these two different types of models. An EM algorithm will be developed to fit the same formulation for both types of models. Numerical examples will be given to illustrate the effectiveness of our EM approach applied to linear mixed effects models.
|
