Multilevel Factor Mixture Modeling for Within-Level Constructs: A Simulation Study

Multilevel factor mixture modeling (ML FMM) allows researchers to identify unobserved heterogeneous groups at both within and between levels. However, most FMM applications using multilevel data ignore nesting, using single-level FMM. We investigate the appropriateness of single-level FMM when the multilevel construct is conceptualized at the within level. Furthermore, the efficacy of ML FMM in detecting heterogeneity at different levels is examined through Monte Carlo simulation. The single-level FMM can be used for within constructs, but ML FMM offers higher class enumeration and classification accuracy. We discuss the dynamics of within and between classes of multilevel data and provide practical guidelines.