Model Assessment and Comparison on Longitudinal Correlated Data Analysis

• Activity Number: 177
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): Tan Li* and Wensong Wu
• Companies: Florida International University and Florida International University
• Keywords: Longitudinal Data ; Dyad ; Multilevel Mixed Model ; Repeated Measure SEM

Abstract

Longitudinal data with complex correlations is usually seen in social and behavioral sciences. The potential correlations may be caused by repeated measurements at individual level, family members at family level, and dyads defined by behavior type. Such data usually also contains correlated independent variables. Multilevel mixed model is capable to address the correlations at different levels, but unable to model the correlations among independent variables. Repeated measures structural equation modeling is more flexible and able to allow most types of correlations. The purpose of this paper is to assess performance of these two classes of models based on simulation study, as well as on a real dataset from a social work study on Latino women substance abuse.