臺大管理論叢第31卷第3期

65 NTU Management Review Vol. 31 No. 3 Dec. 2021 using Mplus software. For the theoretical model we test, we control for employees’ sex, age, and organizational tenure in every path. We also control for leaders’ PIFTs in predicting abusive supervision and LMX as a comparison with the influences of NIFTs on the first-stage mediators. Mplus results show that leaders’ NIFTs had a positive relationship with employees’ perception of abusive supervision (γ = .27, p < .001), providing support for Hypothesis 1. Results also show that abusive supervision has a positive relationship with negative mood (γ = .35, p < .001), which further has a significant and negative relationship with service performance (γ = -.12, p < .05) but does not have a significant relationship with altruistic behavior toward colleagues (γ = -.12, p > .05). These results provide a preliminary support for Hypothesis 2 and Hypothesis 3a, but do not support Hypothesis 3b. To more precisely examine the mediating effects proposed in these hypotheses, we estimate the strengths and the 95% CIs of the indirect effects of leaders’ NIFTs on employees’ negative mood and two outcome variables. Table 4 presents the Monte Carlo simulation results. As expected, through the mediation of abusive supervision, leaders’ NIFTs have a significant indirect effect on employees’ negative mood (estimate = .097, 95% CI [.046, .158]). These results provide support for Hypothesis 2. Furthermore, through the serial mediation of abusive supervision and negative mood, leaders’ NIFTs have a significant indirect effect on employees’ service performance (estimate = -.012, 95% CI [-.027, -.001]), but do not have a significant indirect effect on employees’ altruistic behavior toward colleagues (estimate = -.011, 95% CI [-.027, .000]). These results support Hypothesis 3a but not Hypothesis 3b. Our Mplus results reveal that leaders’ NIFTs have a negative relationship with employees’ perception of LMX (γ = -.13, p < .05), providing support for Hypothesis 4. Results also show that LMX has a positive relationship with psychological empowerment (γ = .46, p < .001), which further has a positive relationship with service performance and altruistic behavior toward colleagues (γ = .34, p < .01; γ = .40, p < .01). These results provide a preliminary support for Hypotheses 5, 6a, and 6b. To more precisely examine the mediating effects proposed in these hypotheses, we estimate the strengths and the 95% CIs of the indirect effects of leaders’ NIFTs on employees’ psychological empowerment and two outcome variables. The Monte Carlo simulation results reveal that through the mediation of LMX, leaders’ NIFTs have a significant indirect effect on employees’ perception of psychological empowerment (estimate = -.058, 95% CI [-.112, -.011]). These results provide a support for Hypothesis 5. Furthermore, through the serial mediation of LMX and psychological empowerment, leaders’ NIFTs have a significant indirect effect

RkJQdWJsaXNoZXIy ODg3MDU=