Why a*b + C' ≠ C ? Addressing Conservation Anomalies In Mediation Analysis through Simulations and an R Shiny Tool

This study identifies two anomalies of mediation analysis in psychological research, in which the total effects and explained variance are not conserved. To safeguard the validity of mediation results, a three‑step correction procedure is proposed, supported by analytical proof, validated through simulations, and supplemented by an R Shiny web application.