A One-Stage Approach for Estimating a Meta-Analytic Continuous-Time Random-Intercept Cross-Lagged Panel Model

This article describes, evaluates, and applies a meta-analytic random-intercept cross-lagged panel model (RI-CLPM). The model enables meta-analytic study of temporal relations between two variables while separating stable between-person differences from within-person fluctuations. The RI-CLPM has been proposed as an alternative to the cross-lagged panel model and has recently motivated many reanalyzes of longitudinal datasets. To support flexible, integrative reanalysis across multiple datasets, we introduce a Bayesian one-stage approach for fitting a meta-analytic RI-CLPM using correlation matrices from longitudinal primary studies. The method combines Bayesian meta-analytic structural equation modeling with continuous-time modeling. The continuous-time framework allows us to account for different time lags across studies, a common challenge in longitudinal meta-analysis. We propose four model variants that differ in their random-effects structures for capturing between-study heterogeneity. A simulation study compares their performance, and an empirical example from educational psychology examines reciprocal relations between academic self-concept and achievement, illustrating their application.