Cross-lagged panel model 2026-10-25
A model relating repeated measures of variables across occasions. Conventional cross-lagged paths can mix stable between-person differences with within-person change and do not automatically identify causal effects.