#SmartPLS4 Series 22 - How to Run a Simple Mediation Model?

Mediation analysis in SmartPLS4 hinges on a simple idea: an exogenous construct affects a mediator, which then affects an endogenous construct. That chain matters because it lets researchers distinguish whether an observed relationship is transmitted indirectly (through the mediator), directly (without the mediator), or through both routes. In practice, the indirect effect is computed as the product of two path coefficients—P1 (independent → mediator) and P2 (mediator → dependent)—and the total effect is the sum of the direct effect (P3: independent → dependent) and the indirect effect (P1×P2). This framework is the basis for deciding whether mediation is full or partial and whether it is complementary or competitive, depending on the signs and significance of the paths.

Before running mediation tests, the analysis must clear measurement and structural hurdles. For reflective mediator constructs, reliability and validity are non-negotiable: weak reliability can shrink estimated indirect effects and lead to underestimating mediation. The same discipline applies to the structural model. Collinearity must be checked because high collinearity can bias path coefficients—sometimes making a direct effect look insignificant even when mediation exists, or producing unexpected sign changes that blur the difference between mediation types. Discriminant validity also matters: if the mediator is not sufficiently distinct from the exogenous or endogenous constructs, indirect effects can appear significant yet be substantially biased, producing incorrect conclusions about whether mediation exists and what kind it is.

Once measurement quality and structural assumptions are satisfied, the mediation procedure becomes a significance test sequence using bootstrapping in SmartPLS4. The workflow starts by estimating P1 and P2 and confirming that their product (the indirect effect) is significant. In the worked example, the indirect effect from collaborative culture (CC) to organizational commitment (OC) to organizational performance (OP) is significant because the indirect effect equals 0.616×0.438 ≈ 0.270 (reported as 0.270), with a p-value below 0.05. The analysis then checks P3—the direct path from CC to OP—using the bootstrapped results.

Here, P3 is also significant, which signals partial mediation: some of CC’s influence on OP runs through OC, while some remains direct. The final classification depends on the signs of the paths. Because P1×P2×P3 is positive in this case, the mediation is complementary partial mediation. Had P3 been insignificant while the indirect effect remained significant, the result would have been full mediation (only the indirect route). If P1 and P2 were insignificant, even a significant P3 would imply no mediation effect through the mediator. If P3 were insignificant and the indirect effect were also insignificant, there would be no effect and no mediation.

The session ends by positioning this as a basic one-mediator model and previewing more complex mediation with multiple mediators in a later session, along with moderation analysis afterward.