14. Hayes Process Macro Model 4 with Multiple Mediators and Covariates

The core takeaway is that “culture” affects “commitment” both directly and indirectly through two parallel mediators—“reliability” and “Assurance”—even after accounting for covariates like age and gender. In the final specification (after dropping gender), both indirect pathways remain statistically significant, producing a partial and complementary mediation pattern.

The model treats culture as the independent variable (X) and commitment as the outcome (Y), with reliability and Assurance acting as mediators (M1 and M2). Age and gender are included as covariates that can influence the mediators and/or the outcome, even though neither is the main focus. The analysis uses Hayes Process Macro Model 4, which supports multiple mediators operating in parallel.

In the initial run (including gender), culture significantly predicts both mediators: reliability shows a significant effect from culture (beta ≈ 0.5485), and Assurance also shows a significant effect from culture (beta ≈ 0.5322). Age and gender do not significantly predict either mediator, meaning the covariates do not explain variability in reliability or Assurance.

When predicting commitment, culture is significant (beta ≈ 0.3733), and both mediators are significant predictors of commitment as well. Reliability and Assurance each contribute positively to commitment, with p-values below 0.05 and t-statistics exceeding the typical two-tailed threshold (about 1.96). Age also significantly predicts commitment (positive coefficient; beta ≈ 0.210), while gender is insignificant—commitment does not differ meaningfully between male and female respondents once culture and the mediators are in the model.

The total effect of culture on commitment (including age and gender) is reported as beta ≈ 0.5996, with R ≈ 0.6223 and R² ≈ 0.3873, indicating that culture, age, and gender explain about 38.7% of the variance in commitment. The decomposition into direct and indirect effects shows that the direct effect of culture on commitment remains significant in the presence of the mediators. That matters because it rules out full mediation.

Bootstrapped indirect effects indicate mediation through both reliability and Assurance. In the first model, one indirect pathway is described as borderline when confidence intervals cross zero, but the bootstrap logic (dividing the indirect effect by the bootstrap standard error to obtain a t-statistic) supports significance. After removing gender—since it is consistently insignificant—the indirect effects through reliability and Assurance both show confidence intervals that no longer include zero, and both indirect effects become clearly significant.

Because the direct effect remains significant alongside significant indirect effects, the mediation is partial. The sign pattern is also consistent: the direct effect and the indirect effects share the same positive direction, so the mediation is complementary rather than competitive. The results are then framed as two mediation hypotheses—reliability mediates culture → commitment, and Assurance mediates culture → commitment—supported by significant indirect effects, plus a significant direct effect indicating additional influence beyond the mediators.