Examining Moderation Effects in Latent Growth Model using A Bayesian Moderated Nonlinear Factor Analysis

• In Event: AERA Roundtable Session Friday 7:45 am - Gold 4
  In Roundtable Session: RT4: Modeling Complex Data and Patterns: Bayesian, Longitudinal, and Machine Learning Approaches

Abstract

This study examines the performance of Bayesian Moderated Nonlinear Factor Analysis (MNLFA) when applied to latent growth models (LGMs). Using a Monte Carlo simulation with 48 conditions and 1,000 replications per condition, we evaluated how well MNLFA recovers moderation effects on latent means, variances, and residual variances. Results show that the Full Moderation Model yields unbiased estimates and accurate coverage, while the Partial Moderation Model—excluding residual variance moderation—produces substantial bias and inflated Type I error. These findings underscore the importance of modeling measurement-level moderation, which may reflect meaningful differences in measurement precision. Ignoring such moderation risks misattributing measurement artifacts as structural effects, leading to inaccurate conclusions about developmental processes.

Authors

• Suyoung Kim, University of Chicago

• Jiwon Kim, Northwestern University

• Brian T. Keller, University of Missouri