How Small Is Too Small? Investigating Small Class Sizes in Growth Mixture Modeling Through Simulation

• In Event: Innovations for Complexity in Models and Design: Inference, Estimation, Classification, and Longitudinal Issues

Abstract

Growth mixture modeling (GMM) allows researchers to explore heterogenous growth trajectories across individuals. The unknown subgroup in GMM could be small in the population, but unestablished minimum class size rules (e.g., 1%, 5%, or 10%) may exclude small classes that may need unique educational support. There has been no systematic investigation on the behaviors of GMM specifically for the identification of small latent classes. We investigate the conditions under which small classes can be accurately detected and recovered using Monte Carlo simulations across various research settings. These findings inform the development of evidence-based guidelines for identifying small classes. Keywords: growth mixture, latent class, small class

Authors

• Courtney Howard-Kirby, University of South Florida

• Eunsook Kim, University of South Florida

• Emma Elísa Evudóttir, University of South Florida - Tampa

• Xilong Jing, University of South Florida

• Yan Wang, University of Massachusetts Lowell

• Xin Tong, University of Virginia

• David Goretzko, Utrecht University

• John M. Ferron, University of South Florida