Reevaluating R²med as an effect size measure for indirect effects

Psychol Methods. 2025 Jul 10. doi: 10.1037/met0000771. Online ahead of print.

Abstract

med quantifies the size of the indirect effect using a combination of R² values from multiple regression equations. While it exhibits many characteristics as an effect measure, prior research has identified potential limitations. Specifically, it has been argued that R²med may yield nonzero values even when no indirect effect exists and that it can produce negative values in certain cases. In response, alternative R² measures have been proposed. This article clarifies that, contrary to previous critiques, R²med consistently returns zero when the indirect effect is truly absent. Additionally, R²med distinguishes between mediation and suppression effects, assigning positive values to mediation effects and negative values to suppression effects. Therefore, rather than being a limitation, this distinguishing property is a strength. Overlooking this distinction may lead researchers to misinterpret unintended indirect effects ("flipped indirect effects") or unexpected direct effects ("mysterious direct effects") when suppression effects are present. (PsycInfo Database Record (c) 2025 APA, all rights reserved).