Studying Trajectories With Clustered Longitudinal Data: A Comparison Between Mixed-Effect Modeling and Multilevel Latent Growth Curve Modeling

• In Event: 45.075 - Multilevel Models, Multiple Imputation, and Latent Models: Unrelated but Equally Important Topics

Abstract

The purpose of our paper is to conduct a Monte Carlo study comparing the performance of ME and LC approaches in terms of model convergence as well as the accuracy and efficiency of parameter estimates in the analysis of three-level longitudinal data. The design factors of simulation included number of cluster, cluster size, R2 of measured variables, and ICC (intraclass correlation coefficient) of repeated measures. We used Mplus 7.4 to generate simulated replications and estimate each of the models. Upon completion of our full simulation work, recommendations for choosing between ME and LC approaches will be made to applied education researchers.

Authors

• Hsien-Yuan Hsu, University of Massachusetts Lowell

• Jr-Hung Lin, National Chiao Tung University

• Minjung Kim, The Ohio State University

• Sandra T. Acosta, Texas A&M University - College Station