Frequentist and Bayesian Approaches to Latent Class Modeling with a Distal Outcome: One- Versus Three-Step Approaches

Methodologists have largely recommended the use of three-step approaches to estimating LCA models with distal outcomes. The present simulation study examined the efficacy of frequentist one- and three-step approaches and Bayesian estimation under different modeling conditions, including poor class separation, small sample sizes, unequal class sizes, and class misspecification.