6. Hayes Process Macro Model 4 with Multiple Mediators

Q: How does Model 4 structure the mediation when there are two mediators in parallel?

Model 4 treats commitment and reliability as two mediators (M1 and M2) that both sit between the IV (culture) and the DV (organizational performance). Culture has two “a paths”: A1 (culture → commitment) and A2 (culture → reliability). Each mediator then has its own “b path” to the DV: B1 (commitment → performance) and B2 (reliability → performance). The DV also receives a direct path from culture, labeled C′ (culture → performance in the presence of the mediators).

Q: What formulas determine the indirect effects in this setup?

The indirect effect through commitment is A1 × B1, where A1 is the culture → commitment coefficient and B1 is the commitment → performance coefficient. The indirect effect through reliability is A2 × B2, where A2 is the culture → reliability coefficient and B2 is the reliability → performance coefficient. The total effect equals C′ plus both indirect effects.

Q: What evidence indicates that the mediators carry the IV’s influence to the DV?

Mediation is supported when the indirect effects are significant. In practice, that means the computed indirect effect (A1×B1 for commitment; A2×B2 for reliability) has a confidence interval that does not include zero and is accompanied by significance testing (the transcript notes indirect effects are significant and “no zero” appears between bounds).

Q: How do you decide between full and partial mediation here?

Full mediation would require the direct effect C′ (culture → performance controlling for both mediators) to be insignificant. Partial mediation occurs when C′ remains significant while the indirect effects are also significant. In this example, the direct effect is significant, so the conclusion is partial mediation.

Q: What does “complementary” versus “competitive” mediation mean in this context?

The transcript uses the sign of the combined effects to classify mediation. If A1×B1×C′ is positive (and similarly for A2×B2×C′), the mediation is “complementary,” meaning the indirect and direct influences align in direction. A negative sign would indicate “competitive” mediation.

Q: What does the SPSS/Process output require attention to when multiple mediators are used?

Because commitment and reliability are endogenous variables (each is affected by culture), SPSS provides separate outcome-focused tables for each endogenous variable. Commitment is assessed as an outcome affected by culture (and reliability is assessed separately as an outcome affected by culture). Organizational performance is then assessed as an outcome affected by culture, commitment, and reliability, with coefficients for the direct effect (C′) and the b paths (B1, B2).

TL;DR

Use Hayes Process Model 4 to test parallel multiple mediation with one IV and one DV.

Briefing Cornell Notes

Briefing

Hayes Process Model 4 is used to test multiple-mediator relationships in a single mediation framework—here, “collaborative culture” (IV) predicts “organizational performance” (DV) through two parallel mediators: “commitment” and “reliability.” The core idea is to estimate separate indirect pathways from the IV to the DV via each mediator, while also estimating the direct effect of the IV on the DV after accounting for both mediators. This matters because it clarifies whether culture influences performance only through these mechanisms (full mediation) or also directly (partial mediation), and whether the mediators operate in parallel.

In the example, the path from culture to commitment is labeled A1, and the path from commitment to performance is labeled B1. The indirect effect through commitment is computed as A1 × B1. Similarly, the path from culture to reliability is A2, and the path from reliability to performance is B2, with the indirect effect through reliability computed as A2 × B2. The direct effect of culture on performance, controlling for both mediators, is labeled C′ (the “direct effect in presence of the mediator”). The total effect of culture on performance is then the sum of the direct effect (C′) plus both indirect effects.

Running the analysis in SPSS uses Process Macro via Analyze → Regression → Process, selecting Model 4. The setup specifies culture as the independent variable, organizational performance as the dependent variable, and commitment and reliability as the mediators. The output is organized by endogenous variables: commitment and reliability are each assessed as outcomes influenced by culture, while organizational performance is assessed as an outcome influenced by culture, commitment, and reliability.

Interpretation begins with the “a paths” (culture → commitment and culture → reliability). Culture shows a significant effect on commitment (p < .05), and also a significant effect on reliability (p < .001). Next come the “b paths” (commitment → performance and reliability → performance). Commitment significantly predicts performance (p < .01), and reliability also significantly predicts performance (p < .01). With both sets of paths significant, the analysis moves to the indirect effects.

Indirect effects are tested by multiplying the relevant coefficients: A1 × B1 for the commitment pathway and A2 × B2 for the reliability pathway. Both indirect effects are significant, with confidence intervals reported as not crossing zero. The direct effect (C′) is also significant, which leads to a conclusion of partial mediation rather than full mediation. Because the indirect-effect components and the direct effect are positive, the mediation is described as “complementary” (not competitive). Finally, the results are reported as evidence that collaborative culture affects organizational performance through commitment and through reliability, while still retaining a significant direct relationship between culture and performance after accounting for both mediators.

Cornell Notes

Hayes Process Model 4 estimates how an independent variable (collaborative culture) affects a dependent variable (organizational performance) through multiple mediators operating in parallel (commitment and reliability). The method computes two indirect effects: culture → commitment → performance (A1×B1) and culture → reliability → performance (A2×B2), alongside a direct effect of culture on performance controlling for both mediators (C′). Significance is assessed using p-values and confidence intervals that do not include zero for the relevant paths and indirect effects. Because the direct effect remains significant while both indirect effects are significant, the mediation is classified as partial and complementary (all effects positive).

How does Model 4 structure the mediation when there are two mediators in parallel?

Model 4 treats commitment and reliability as two mediators (M1 and M2) that both sit between the IV (culture) and the DV (organizational performance). Culture has two “a paths”: A1 (culture → commitment) and A2 (culture → reliability). Each mediator then has its own “b path” to the DV: B1 (commitment → performance) and B2 (reliability → performance). The DV also receives a direct path from culture, labeled C′ (culture → performance in the presence of the mediators).

What formulas determine the indirect effects in this setup?

The indirect effect through commitment is A1 × B1, where A1 is the culture → commitment coefficient and B1 is the commitment → performance coefficient. The indirect effect through reliability is A2 × B2, where A2 is the culture → reliability coefficient and B2 is the reliability → performance coefficient. The total effect equals C′ plus both indirect effects.

What evidence indicates that the mediators carry the IV’s influence to the DV?

Mediation is supported when the indirect effects are significant. In practice, that means the computed indirect effect (A1×B1 for commitment; A2×B2 for reliability) has a confidence interval that does not include zero and is accompanied by significance testing (the transcript notes indirect effects are significant and “no zero” appears between bounds).

How do you decide between full and partial mediation here?

Full mediation would require the direct effect C′ (culture → performance controlling for both mediators) to be insignificant. Partial mediation occurs when C′ remains significant while the indirect effects are also significant. In this example, the direct effect is significant, so the conclusion is partial mediation.

What does “complementary” versus “competitive” mediation mean in this context?

The transcript uses the sign of the combined effects to classify mediation. If A1×B1×C′ is positive (and similarly for A2×B2×C′), the mediation is “complementary,” meaning the indirect and direct influences align in direction. A negative sign would indicate “competitive” mediation.

What does the SPSS/Process output require attention to when multiple mediators are used?