Classification and Estimation Accuracy of Growth Mixture Models with Class-Predicting Covariates: Frequentist and Bayesian Approaches

Growth mixture models classify individuals into unobserved latent classes based on their developmental trajectories. This study demonstrates that classification accuracy is remarkably low across existing frequentist and Bayesian approaches. However, growth trajectory estimates of the latent classes were relatively robust to classification inaccuracy. Implications and recommendations for researchers are discussed.