Causal assumptions of the two-wave longitudinal mediation model

This study provides a clearer understanding of causal estimands and assumptions in the two-wave longitudinal mediation model. Using causal graphical models, we demonstrate that while the ANCOVA and change score methods yield the same causal interpretation, they rely on different assumptions to produce unbiased results. By comparing these approaches across various realistic scenarios, we hope to guide researchers in selecting the most appropriate method, thereby enhancing the accuracy and reliability of mediation analysis in longitudinal data.