Page 198 - 33-3

Shrewd Calculation or Beautiful Illusion? Exploring the Relationship between Abusive Supervision, Follower
Perceived Leadership Effectiveness, and Supervisor-Directed Citizenship Behaviors: The Moderating Roles of
Supervisor Political Skills and Physical Attractiveness

and supervisor-directed helping as the dependent variable. We also include subordinates’
negative emotions (i.e., the affective mechanism) and LMX (i.e., the relational mechanism)
as alternative mediators and control for their effects at the Level-1. At the Level-2 (i.e.,
the leader-level), we specify leader political skills as the moderator, whereas leaders’

gender and tenure are included as the Level-2 control variables. Moreover, we follow the
suggestions to group-mean center the Level-1 predictor and moderator and grand-mean
center the Level-2 moderator and control variables (Zhang, Zyphur, and Preacher, 2009).

Finally, we use the Monte Carlo Simulated Confidence Intervals for Moderated Mediating
Effects (MCCIMM) program developed by Cheung (2007) to test the indirect effects and
conditional indirect effects using Monte Carlo simulations.

3. Findings

The results of the multilevel path-analysis show that leader political skills negatively
moderate the negative relationship between abusive supervision and leader effectiveness

(estimate = -0.70, p < 0.05). This negative relationship is stronger for politically skilled
leaders (simple slope = -0.63, p < 0.01), failing to support Hypothesis 1. Moreover, leader
physical attractiveness positively moderates the negative relationship between abusive
supervision and leader effectiveness (estimate = 0.66, p < 0.01). This relationship becomes
marginally significant and positive when the leaders are physically attractive (simple slope

= 0.29, p < 0.10). Therefore, Hypothesis 2 is supported.
We further test Hypotheses 3 and 4 using MCCIMM program (Cheung, 2007). The
results indicate that the negative indirect effect of abusive supervision on supervisor-
directed helping via reduced leader effectiveness is stronger for leaders with better political

skills (indirect effect = -0.066, p < 0.01; 95% CI = [-0.151, -0.016]), which is inconsistent
with our expectations. Hence, Hypothesis 3 was not supported. Finally, the results reveal
that the negative indirect effect of abusive supervision on supervisor-directed helping
via reduced leader effectiveness becomes marginally significant and positive when the

leaders are physically attractive (indirect effect = 0.29, p < 0.10; 90% CI = [0.001, 0.085]),
supporting Hypothesis 4.
We also include negative emotions and LMX as additional mediators to rule
out alternative explanations. These analyses did not change the research conclusions,

190