The Intersection of Sample Size, Number of Indicators, and Class Enumeration in Latent Class Analysis: A Monte Carlo Study

• In Event: Estimation Accuracy and Conditions for Complex Latent Variable Models

Abstract

This Monte Carlo study examined the performance of commonly used fit indices in selecting the “correct” latent class model while varying the number and nature of latent classes, class prevalences, the number of indicators, and sample size. In addition to the common fit indices used, we included the approximate Bayes Factor (BF) and the correct model probability (cmP), which have not been included in LCA simulation studies before. The adjusted Bayesian Information Criterion (ABIC) and the bootstrap likelihood ratio test (BLRT) emerged as the best performing fit indices. Findings indicate interplay between sample size, number of indicators, and class enumeration in LCA models. Practical implication related to sample size and the number of indicators used in LCA will be discussed.

Authors

• Diane Morovati

• Ryan Grimm, University of California - Davis

• Karen L. Nylund-Gibson, University of California - Santa Barbara