Methods of Covariate Inclusion in Latent Transition Analysis: A Monte Carlo Simulation Study

• In Event: 32.082 - Roundtable Session 11
  In Roundtable Session: 32.082-16 - Structural Equation Modeling With Difficult Data

Abstract

Latent transition analysis (LTA) is a longitudinal variant of latent class analysis that allows researchers to analyze how individuals change between latent classes over time. In LTA models, one is often interested in relating latent class membership to auxiliary variables. However, empirical investigations of covariate inclusion methods for LTA are scant. In the present study, we conducted a Monto Carlo simulation to evaluate several recent bias-corrected three-step approaches with the traditional three-step and one-step methods for incorporating covariates into LTA under various conditions, including class separation, covarite effect size, type of parameters estimated, and sample size. Preliminary results suggest that the newer three-step approaches yield consistently less biased parameter estimates and more accurate standard errors than the uncorrected three-step method.

Authors

• Ai Ye, University of North Carolina - Chapel Hill

• Jeffrey R. Harring, University of Maryland

• Luke Rinne, University of Delaware