Leader prohibitive voice behavior and its effects on followers through leader identification and political skill

The study addresses how leaders' prohibitive voice behavior—raising concerns to prevent harm and pointing out existing problems—translates to followers' own prohibitive voice. In highly relational, high power-distance cultures (e.g., China), prohibitive voice is especially risky and often avoided due to potential negative social consequences, yet it is vital for error prevention, safety, and innovation. Drawing on social learning theory, the authors propose that followers observe and imitate leaders who model prohibitive voice, and that this process is facilitated by leader identification (a cognitive-emotional connection with the leader). They also examine whether followers' political skill conditions this process. The purpose is to clarify whether leaders’ prohibitive voice encourages followers’ prohibitive voice, by what mechanism (leader identification), and under what boundary condition (follower political skill). This work is important because prohibitive voice, while critical for avoiding organizational failures, is underutilized and under-studied in terms of leader modeling pathways.

The review distinguishes promotive voice (future-oriented suggestions) from prohibitive voice (problem-focused, risk-averse, past/present-oriented). Although both aim at constructive change, antecedents and outcomes differ. Prohibitive voice often evokes defensiveness, negative affect, and interpersonal risks, leading to lower evaluations and reluctance to speak up. Yet it is crucial for safety and avoiding unnecessary failures, sometimes more immediately impactful than promotive voice. Prior research links transformational, ethical, inclusive, servant leadership, leader openness, and high LMX to employee voice, but has not directly examined leaders’ own prohibitive voice as a modeled behavior. Using social learning theory, the authors argue that leaders’ behavior serves as a salient standard; followers attend to, retain, and reproduce observed leader prohibitive voice, especially when leaders solicit input and support voice. They theorize leader identification as a key mediating cognitive process and political skill as a moderator that may strengthen the learning and enactment of prohibitive voice.

Design and sample: Cross-sectional multi-source survey of 317 matched leader–follower dyads across 59 Chinese organizations spanning manufacturing, finance, IT, services, and government. Targeted departments included general management roles to ensure observable leader behavior. A total of 400 dyads were approached; 360 responded (90% response), and 317 usable dyads remained after screening. Demographics (followers): 45.7% male, 54.3% female; age distribution concentrated in 30s (55.6%); education ranged from high school (18.9%) to graduate degrees (13.9%); tenure with leader varied (<1 year to >15 years).

Data collection and CMB mitigation: Leaders rated follower prohibitive voice (outcome). Followers rated leader prohibitive voice (independent variable), leader identification (mediator), and follower political skill (moderator). Surveys were anonymous; participation was voluntary; consent obtained. English-origin scales were translated into Chinese with back-translation procedures.

Measures (5-point Likert, 1=strongly disagree to 5=strongly agree):

• Prohibitive voice behavior: 5-item scale (Liang et al., 2012) used for both leader PVB (wording adapted, "My leader") and follower PVB; alphas: follower PVB α=0.891; leader PVB α=0.889.
• Follower political skill: 6-item scale (Ferris et al., 2005); α=0.829.
• Leader identification: 7-item scale (Shamir et al., 1998); α=0.891.
• Controls: gender (0=male, 1=female), age (years), tenure with leader (years), education (1–4), and leader promotive voice (5-item Liang et al., 2012; α=0.844).

Analytic strategy: Confirmatory factor analysis (CFA) tested construct distinctiveness. Hypotheses tested via hierarchical regression (SPSS) and PROCESS macro (Hayes, 2013) for mediation, moderation, and moderated mediation. All predictors were mean-centered; multicollinearity checks showed VIF ≤ 1.80. Simple slopes probed interactions at ±1 SD of political skill; bootstrapping (1,000 samples) estimated conditional indirect effects.

Measurement model fit: 4-factor model (Leader PVB, Leader Identification, Follower Political Skill, Follower PVB) fit well and exceeded alternatives: χ²(224)=333.278, CFI=0.971, TLI=0.966, IFI=0.970, RMSEA=0.039.

• Correlations: Leader prohibitive voice correlated positively with leader identification (r=0.317, p<0.001), follower political skill (r=0.391, p<0.001), and follower prohibitive voice (r=0.499, p<0.001). Leader identification correlated with follower political skill (r=0.456, p<0.001) and follower prohibitive voice (r=0.611, p<0.001).
• H1 (direct effect): Leader prohibitive voice positively predicted follower prohibitive voice beyond controls and leader promotive voice (β=0.25, p<0.001 in Model 2; effects remained positive though attenuated with additional variables).
• H2 (mediation via leader identification): Indirect effect significant; PROCESS estimate=0.11, SE=0.05, 95% CI [0.023, 0.209].
• H3 (moderated mediation by follower political skill): Contrary to the hypothesized positive moderation, the interaction between leader identification and follower political skill on follower prohibitive voice was negative and significant (β=−0.15, p<0.001). Simple slopes: at low political skill, the relation was stronger and positive (slope=0.427, p<0.001); at high political skill, non-significant (slope=−0.213, ns). Conditional indirect effects: low political skill b=0.340, SE=0.048, 95% CI [0.017, 0.204]; high political skill b=−0.160, SE=0.032, 95% CI [−0.014, 0.142], ns. Thus, the mediated effect via leader identification was more pronounced at lower, not higher, levels of political skill.
• Overall: H1 and H2 supported; H3 refuted with evidence of negative moderation.
• Measurement validity supported by excellent CFA fit indices (χ²(224)=333.278, CFI=0.971, TLI=0.966, IFI=0.970, RMSEA=0.039).

Findings show that leaders' own prohibitive voice serves as a salient modeled behavior that followers observe and emulate, aligning with social learning theory. Leader identification emerged as the cognitive mechanism linking leader and follower prohibitive voice: followers who identify with leaders internalize their values and behaviors and are more willing to accept the risks of speaking up about problems. Contrary to expectations, follower political skill weakened (rather than strengthened) the link between leader identification and follower prohibitive voice, suggesting that lower-skill followers rely more on identification cues from leaders to manage the risks of prohibitive voice. The cultural context (Chinese organizations with high power distance and harmony norms) may intensify the role of leader modeling and identification in facilitating risky voice behaviors, potentially overshadowing individual differences in political skill. The results extend voice and leadership literature by specifying a direct leader-behavior modeling pathway, clarifying a cognitive mediator (leader identification), and revealing a boundary condition that operates in the opposite direction of initial theorizing. Practically, the study emphasizes leader role modeling, building identification, and recognizing varying needs based on followers’ political skill.

The study demonstrates that leader prohibitive voice behavior promotes follower prohibitive voice and that leader identification mediates this effect. It underscores the leader’s role as a model within social learning processes for risky, problem-focused voice. The moderated mediation analysis revealed that the indirect effect through leader identification is stronger for followers low in political skill, challenging assumptions about the uniformly facilitating role of political skill. Contributions include: highlighting prohibitive voice as distinct and vital for organizational safety and error prevention; establishing leader identification as a mechanism through which leader behavior translates to follower behavior; and identifying a counterintuitive boundary condition. Future research should replicate across cultures, use longitudinal designs to establish causality, incorporate additional mediators (e.g., affect, justice, LMX), consider the interplay with promotive voice, integrate social exchange perspectives, and test additional boundary conditions (e.g., proactive personality, achievement motivation, organizational structure and openness).

• Cross-sectional design limits causal inference despite multi-source data; longitudinal or experimental designs are recommended.
• Chinese sample may limit cultural generalizability; replication across cultural contexts is needed.
• Unmeasured mediators (e.g., emotional states, perceived fairness, LMX) may further explain effects.
• Focused solely on prohibitive voice; parallel modeling with promotive voice could clarify distinct processes and outcomes.
• Additional theoretical integration (e.g., social exchange theory) may enrich understanding of identification and reciprocity mechanisms.
• Boundary conditions beyond political skill (e.g., proactive personality, achievement motivation, decentralization, openness) warrant examination.