*4.6. Mediation Model*

To test the first set of proposed hypotheses, PROCESS Macro (extension in SPSS) by Hayes [79] Model No. 4 was used. The results identified that SM has insignificant direct effect (B = −0.0641; *p* > 0.10) on SP. However, the indirect effect through SC is significant (B = 0.086; *p* < 0.10), hence identifying full mediation. Similarly, for safety climate as mediator, the SM has insignificant direct (B = −0.047; *p* > 0.10) and significant indirect effects (B = 0.069; *p* < 0.10) through safety climate, providing support for full mediation. The results of PROCESS Model 4 are presented in Tables 3 and 4.

**Table 3.** Five thousand bootstrap results for direct and indirect effects. PROCESS Model 4 (safety consciousness as mediator).


SM: safety management; SP: safety performance: SC: safety consciousness; \* *p* < 0.10.

**Table 4.** Five thousand bootstrap results for direct and indirect effects. PROCESS Model 4 (safety climate as mediator).


SM: safety management; SP: safety performance: SCL: safety climate; \* *p* < 0.10.

### *4.7. Moderation Analysis*

PROCESS Macro (extension in SPSS) by Hayes [79] Model No. 1 was used to test the proposed moderation hypotheses for responsible leadership. PROCESS Macro by Hayes [79] was preferred over simple regression analysis using interaction term and structural equation modeling because of its robustness. PROCESS Macro uses a bootstrapping approach with biased corrected at 95% confidence intervals that calculates the Johnson-Neyman outputs for the interaction term. The variables that define product term were first mean centered. Conditioning values at mean and ±1SD and Johnson-Neyman outputs for the interaction graph were also calculated. We have used a separate PROCESS Model No. 1 for safety consciousness and safety climate. The result of PROCESS Model 1 are presented in Table 5.


**Table 5.** Five thousand bootstrap results for PROCESS Model No. 1, simple moderation analysis.

SM: safety management; SP: safety performance: SCL: safety climate; SC: safety consciousness; RL: Responsible Leadership \* *p* < 0.01, \*\* *p* < 0.05.

The results identified that the interaction terms for both SC (B = −0.163 \*; *p* < 0.01) and SCL (B = −0.117 \*\*; *p* < 0.05) were significant and there was no zero in the lower and upper bound of the 95% confidence interval. Interaction graphs for low and high (Mean ± SD) values of SC and RL and SCL and RL were plotted. The interaction graph of the SC and RL relationship (shown in Figure 2) suggests that RL significantly enhances the relationship between safety consciousness and safety performance when safety consciousness is low. The role of RL as moderator is significant at low levels of safety consciousness, and it becomes insignificant when safety consciousness is high. Similarly, for safety climate, in the interaction graph of the RL and SCL relationship (shown in Figure 3), RL is significant at low levels of safety climate. The slope test shows that the presence of RL enhances the positive relationship of SP and SC and SP and SCL when SC and SCL are low.

**Figure 2.** Interaction plot for responsible leadership and safety consciousness.

**Figure 3.** Interaction plot for responsible leadership and safety climate.
