Quantifying uncertainty in PLS-SEM-based mediation analyses

Abstract

To identify potential mediating effects, researchers applying partial least squares structural equation modeling (PLS-SEM) typically contrast specific indirect and direct effects in a sequence of steps. Extending this standard procedure, we conceive mediation analysis as a type of model comparison, which facilitates quantifying the degree of the model effects’ uncertainty induced by the introduction of the mediator. By introducing a new dimension of quality assessment, the procedure offers a new means for deciding whether or not to introduce a mediator in a PLS path model, and improves the replicability of research results.