Evaluation of Goodness-of-Fit Tests in Random Intercept Cross-Lagged Panel Model: Implications for Small Samples

Abstract

The development of the random intercept cross-lagged panel model (RI-CLPM) is an extension of the traditional cross-lagged panel model (CLPM), which aims to study between and within person variances in longitudinal data. Despite its growing popularity in behavioral and social sciences, our understanding of goodness-of-fit tests of RI-CLPMs is limited. Using Monte Carlo simulations across different sample sizes and model complexity, this study evaluates goodness-of-fit tests applied to RI-CLPMs by comparing the test statistics of the maximum likelihood (ML), generalized least squares (GLS), and reweighted least squares (RLS), as well as their corresponding NFI, CFI, and RMSEA. Our results showed that when (Formula presented.) was significantly large; ML, GLS, and RLS tended to have similar performances. When (Formula presented.) was small relative to the model complexity, RLS outperformed ML and GLS and produced consistent (Formula presented.) test statistics and fit indices. These results have implications for fitting RI-CLPM with finite data.