25. SEMinR Lecture Series - Mediation Analysis with Multiple Mediators

A mediation model with two parallel mediators shows that Vision’s impact on organizational performance runs through both collaborative culture and organizational commitment—and the indirect paths are statistically significant. In the SEMinR setup, Vision is treated as the independent variable, organizational performance as the dependent variable, and the two mediators are collaborative culture and commitment. The analysis first estimates the measurement and structural models, then uses bootstrapping to test the significance of indirect effects for each mediator-specific pathway.

After confirming the model estimates cleanly (all 341 observations valid and bootstrap iterations within range), the total indirect effect from Vision to organizational performance is reported as 0.270. But significance is assessed at the mediator-specific level rather than relying on the total indirect effect alone. Using the specific_effect_significance function with the bootstrapped model, the pathway from Vision → collaborative culture → organizational performance is found significant (T statistic exceeds the 1.96 threshold). The pathway from Vision → commitment → organizational performance is also significant by the same criterion. Taken together, both mediators carry a meaningful portion of Vision’s effect on organizational performance.

The analysis then determines the mediation type by checking whether the direct effect remains significant once mediators are included. The direct effect of Vision on organizational performance is 0.373, and its T statistic is again above 1.96, meaning the direct relationship does not disappear. With significant indirect effects for both mediators and a still-significant direct effect, the model indicates partial mediation.

Finally, the model classifies whether the mediation is complementary or competitive by examining the sign of the combined mediated effects. The workflow multiplies the relevant path coefficients—Vision → collaborative culture → organizational performance and Vision → commitment → organizational performance—using the paths stored inside the summary_simple object. The resulting mediated products are positive, leading to a complementary mediation conclusion for the two-mediator structure.

A path plot and a bootstrap output are generated to visualize the model and the significance results, and the overall procedure is presented as a repeatable SEMinR workflow for multiple mediators: specify formative measurement blocks, define the structural paths, estimate and bootstrap, test mediator-specific indirect effects, evaluate direct effects for partial vs. full mediation, and compute mediated-effect products to label complementary vs. competitive mediation. The practical takeaway is that SEMinR can test not only whether mediation exists, but also which mediator-specific channels matter and how the mediators interact in shaping the dependent variable.