Analyzing Cross-Lag Effects: A Comparison of Different Cross-Lag Modeling Approaches

Developmental researchers often have research questions about cross-lag effects--the effect of one variable predicting a second variable at a subsequent time point. The cross-lag panel model (CLPM) is often fit to longitudinal panel data to examine cross-lag effects; however, its utility has recently been called into question because of its inability to distinguish between-person effects from within-person effects. This has led to alternative forms of the CLPM to be proposed to address these limitations, including the random-intercept CLPM and the latent curve model with structured residuals. We describe these models focusing on the interpretation of their model parameters, and apply them to examine cross-lag associations between reading and mathematics. The results from the various models suggest reading and mathematics are reciprocally related; however, the strength of these lagged associations was model dependent. We highlight the strengths and limitations of each approach and make recommendations regarding modeling choice.