ABSTRACT
This chapter discusses the extension of latent growth curve modeling: growth mixture modeling (GMM). It also discusses the possible existence of heterogeneity that may be characterized into distinct sub-populations of individual trajectories. The chapter focuses on social-psychological and health attributes, but such clustering of attributes could extend to other study areas as well. It explains why fitting a conventional latent growth curve model (LGCM) is problematic when trajectory heterogeneity exists. The chapter shows how the heterogeneity of individual trajectories is taken into account in a GMM by incorporating not only continuous latent variables that reflect growth parameters but also categorical latent variables that capture the heterogeneity of developmental trajectories. It summarizes the necessary steps for successfully estimating a GMM. For each model, figures and Mplus syntax are presented. The chapter elaborates on several common issues that arise when estimating a GMM, including model convergence problems and choosing the optimal number of classes.
