Searching for the Accurate Growth Trajectory in Latent Growth Models With Covariates

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

Abstract

The current study examines the performance of different starting models in latent growth models to search for the accurate growth shape. This study extends the previous simulation study conducted by Kim et al. (2016) by including the covariates of the linear and quadratic growth model. Findings show that the model specified with the most complex mean structure (i.e., highest polynomial growth model) and the saturated residual variance structure (i.e., unstructured [UN] variance-covariance) outperforms other starting models to search for the accurate growth shape. Implications will be discussed.

Authors

• Minjung Kim, The Ohio State University

• Hsien-Yuan Hsu, University of Texas Health Science Center at Houston

• Oiman Kwok, Texas A&M University - College Station