|
Activity Number:
|
72
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 29, 2007 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Social Statistics Section
|
| Abstract - #310352 |
|
Title:
|
Multilevel Mixture Modeling Applications
|
|
Author(s):
|
Tihomir Asparouhov and Bengt Muthen and Shaunna Clark*+
|
|
Companies:
|
Muthen & Muthen and University of California, Los Angeles and University of California, Los Angeles
|
|
Address:
|
2023 Moore Hall, Los Angeles, CA, 90095-1521,
|
|
Keywords:
|
Mixture Models ; Multilevel Models ; Latent Class Models ; Non-Parametric Random Effects ; Mplus
|
|
Abstract:
|
New developments are discussed for the analysis of multilevel data where the latent classes appear not only as level-1 (individual-level) variables but also as cluster-level variables. With categorical and count outcomes, cluster-level latent class variables also simplify maximum-likelihood computations and avoid normality assumptions for random effects by a non-parametric representation of the random effects. Several applications will be presented using the latest version of the Mplus program. Examples include multilevel analysis of achievement data with classification of both students and schools, and longitudinal analysis of mental health data using non-parametric random effects modeling. Results from standard multilevel regression and growth analysis will be contrasted with the new techniques.
|
