The Impact of Incomplete Data on Two-Level Linear Growth Models: A Monte Carlo Examination

• In Event: 69.046 - Latent Growth/Transition Modeling and Model Selection

Abstract

Obtaining accurate and precise parameter estimates in two-level linear models can be quite complex. Interestingly, in terms of two-level models, most of the published methodological literature has focused on organizational models; much less has been published on the functioning of two-level growth models under different study conditions. Given the benefits of using multilevel models for the analysis of longitudinal data, more methodological research focused on how well growth models work under various design conditions is warranted. To address this gap, we examined a variety of sample size combinations and different proportions of incomplete series data, in two-level growth models with both continuous and binary predictors, and the impact these factors have on numerous statistical outcomes.

Authors

• Bethany A. Bell, University of South Carolina

• Jason Schoeneberger, ICF International

• Zhaoxia Guo, University of South Carolina - Columbia

• Mary Ann Priester, University of South Carolina

• Kristina Webber, University of North Carolina

• Whitney Smiley, American Board of Internal Medicine

• John M. Ferron, University of South Florida