Introduction
Leadership is widely recognized as one of the most influential determinants of employees’ motivation, performance, and psychological well-being in organizational settings. Traditional leadership research has predominantly focused on constructive forms of leadership that foster empowerment, ethical conduct, and organizational effectiveness (Northouse, 2022). However, contemporary scholarship has increasingly shifted attention toward the destructive and dysfunctional dimensions of leadership behavior (Liu et al., 2024; Schmid et al., 2019). In this context, toxic leadership has emerged as a critical construct describing persistent patterns of abusive, narcissistic, authoritarian, and self-serving behaviors that undermine both individual well-being and systemic functioning (Goldman, 2009; Pelletier, 2010).
Toxic leadership was initially conceptualized in military and organizational studies to describe leaders who simultaneously appear effective while engaging in destructive interpersonal behaviors (Lipman-Blumen, 2005). Such leaders often combine charisma with manipulation, fostering dependency while eroding autonomy, trust, and psychological security among subordinates (Reed, 2004). Contemporary conceptualizations describe toxic leadership as a multidimensional construct encompassing abusive supervision, narcissism, unpredictability, authoritarianism, and self-promotion, all of which contribute to psychological strain and organizational dysfunction (Schmidt, 2008). Unlike overt misconduct, toxic leadership is often subtle, normalized, and difficult to detect or sanction, which increases its long-term organizational impact (Mehta & Maheshwari, 2014).
A growing body of empirical research has linked toxic leadership to counterproductive work behavior (CWB), defined as intentional behaviors that harm organizations or individuals within them (Mackey et al., 2017; Zhang & Bednall, 2024). Employees exposed to abusive or unfair leadership are more likely to disengage, reduce performance, or engage in retaliatory behaviors as a form of psychological compensation (Liu et al., 2020; Tepper, 2000). CWBs include both organizational deviance (e.g., theft, sabotage, absenteeism) and interpersonal deviance (e.g., aggression, gossip, incivility) (Spector et al., 2006). From the perspective of Social Exchange Theory (Blau, 1964), such behaviors represent negative reciprocity in response to perceived violations of fairness and psychological contracts (Cropanzano et al., 2017).
However, toxic leadership does not operate in isolation; its effects are shaped by organizational and psychosocial contextual factors. One critical mechanism is the erosion of psychological safety, defined as a shared belief that interpersonal risk-taking is safe within a team (Edmondson, 1999). Low psychological safety reduces voice behavior, increases fear, and fosters withdrawal and defensive silence (Frazier et al., 2017; Kim & Park, 2025). Another mechanism is the weakening of team cohesion, which refers to the degree of unity and interpersonal attraction among group members (Carron & Brawley, 2012). Cohesive teams tend to develop stronger social norms, mutual support systems, and collective coping strategies that discourage deviant behavior (Chiaburu et al., 2013; Chang & Bordia, 2001).
Theoretically, this study is grounded in an integrative framework combining Social Exchange Theory (Blau, 1964), Conservation of Resources (COR) Theory (Hobfoll, 1989), and Group Resilience Theory (West et al., 2009). Social Exchange Theory explains retaliatory behaviors resulting from perceived injustice, whereas COR theory suggests that toxic leadership depletes key psychological resources such as self-esteem, trust, and efficacy, which may be restored through supportive climates like psychological safety. Group resilience theory further explains how cohesive teams function as collective buffers that reduce the translation of stressors into deviant behaviors.
Empirical evidence supports these theoretical assumptions. Toxic leadership has been consistently associated with workplace deviance, reduced job satisfaction, emotional exhaustion, and organizational cynicism (Al-Ghazali et al., 2025; Schyns & Schilling, 2013). Furthermore, psychological safety has been shown to buffer the negative effects of adverse leadership on performance and well-being (Frazier et al., 2017), while team cohesion reduces stress-related counterproductive behaviors through social regulation and collective support (Costa et al., 2015; Wang et al., 2024). Nevertheless, most prior studies have examined these moderators separately rather than within a unified structural model.
Importantly, existing research is largely based on Western organizational contexts, limiting the generalizability of findings to non-Western environments. In many public organizations, particularly in high power-distance cultures, hierarchical authority structures and limited feedback mechanisms may intensify the effects of toxic leadership (Bouich, 2019; Hofstede, 2001). At the same time, collectivist values may strengthen group-based protective mechanisms such as cohesion and mutual support, making these contexts particularly relevant for examining moderating processes.
Despite growing interest in destructive leadership, several important gaps persist in the current literature. First, few empirical investigations have simultaneously modeled individual perceptions of psychological safety and team cohesion within a single integrative structural model to test their joint buffering potential against toxic leadership at the individual level of analysis.
Second, limited research has explored these complex moderating interactions within public-sector environments, where bureaucratic authority and social collectivism interact uniquely. Third, extant research predominantly focuses on direct bivariate effects, paying less attention to the boundary conditions that moderate employee behavioral reactions under toxic supervisory conditions.
To address these gaps, the present study investigates a moderated structural model examining how psychological safety and team cohesion modify the relationship between toxic leadership and counterproductive work behavior. The study utilizes survey data collected from 200 employees across public organizational settings, evaluated using correlation analysis, hierarchical regression, and structural equation modeling (SEM).
Research Problem
This study addresses the following central research question: Under what contextual boundary conditions does toxic leadership predict counterproductive work behavior, and how do psychological safety and team cohesion independently and jointly attenuate this relationship in public organizational environments?
Research Questions
- How does toxic leadership directly influence counterproductive work behavior among employees in public organizations?
- Does psychological safety moderate the relationship between toxic leadership and counterproductive work behavior?
- Does team cohesion moderate the relationship between toxic leadership and counterproductive work behavior?
- What is the joint moderating effect of psychological safety and team cohesion on the relationship between toxic leadership and counterproductive work behavior?
Objectives of the Study
This study aims to:
- Examine the direct effect of toxic leadership on counterproductive work behavior.
- Investigate the moderating role of psychological safety in buffering this relationship.
- Explore the moderating role of team cohesion in buffering this relationship.
- Test the combined, dual-buffering interaction effect of psychological safety and team cohesion against workplace deviance.
Hypotheses
Drawing on Social Exchange Theory (Blau, 1964), Conservation of Resources Theory (Hobfoll, 1989), and Group Resilience Theory (West et al., 2009), the following hypotheses are proposed:
H1: Toxic leadership is positively related to counterproductive work behavior among employees.
H2: Psychological safety moderates the relationship between toxic leadership and counterproductive work behavior, such that the relationship is weaker when psychological safety is high.
H3: Team cohesion moderates the relationship between toxic leadership and counterproductive work behavior, such that the relationship is weaker when team cohesion is high.
H4: Psychological safety and team cohesion jointly moderate the relationship between toxic leadership and counterproductive work behavior, such that the relationship is weakest when both moderators are high.
Contributions of the Study
This research contributes to the organizational psychology literature in four primary ways:
- It integrates individual-level psychological and team-relational moderators into a unified structural model, providing a more comprehensive view of social-environmental boundary conditions.
- It extends leadership and workplace deviance frameworks to an under-represented North African public-sector context.
- It applies methodological rigor through structural equation modeling (SEM) to evaluate moderation hypotheses.
- It provides actionable practical implications for organizational leaders, HR professionals, and public sector administrators seeking to foster resilient work climates.
Toxic Leadership and Counterproductive Work Behavior
In this individual-level framework, psychological safety functions as a personal cognitive buffer that protects emotional resources, whereas perceived team cohesion operates as a social-relational buffer that captures the individual's sense of peer solidarity and group support (Hobfoll, 1989; Kozlowski & Klein, 2000). When combined within a single structural equation model at the individual level of analysis, high perceived psychological safety and high team cohesion are expected to demonstrate a synergistic buffering effect, minimizing employee engagement in counterproductive work behaviors under toxic leadership.
Empirical inquiries consistently identify toxic leadership as a prominent antecedent to counterproductive work behavior (CWB). CWBs encompass explicit, intentional actions that violate organizational norms and threaten the well-being of the firm or its stakeholders, ranging from organizational deviance (e.g., deliberate withdrawal, absenteeism, reduced effort, sabotage) to interpersonal deviance (e.g., incivility, aggression, spreading rumors) (Spector et al., 2006; Zhang & Bednall, 2024). Grounded in Social Exchange Theory (Blau, 1964), employees facing abusive supervisory treatment experience perceived breach of the psychological contract and organizational injustice (Cropanzano et al., 2017). Consequently, engaging in CWB serves as an instrumental mechanism for negative reciprocity—a means of restoring equity or retaliating against hostile supervisory dynamics (Al-Ghazali et al., 2025; Tepper, 2000).
The Moderating Role of Psychological Safety
While the direct link between supervisory toxicity and employee deviance is well-established, organizational scholars emphasize the pivotal role of psychological climate mechanisms in altering employee behavioral trajectories. Psychological safety—the shared perception that the immediate team environment is safe for interpersonal risk-taking—enables individuals to speak up, share concerns, admit mistakes, and seek help without fearing interpersonal penalties or ridicule (Edmondson, 1999; Newman et al., 2017).
According to Conservation of Resources (COR) theory (Hobfoll, 1989), toxic supervision operates as a severe environmental stressor that rapidly depletes an employee’s cognitive, emotional, and social resources. In psychologically safe climates, however, employees gain access to a vital "resource reservoir" (Frazier et al., 2017; Kim & Park, 2025). This atmosphere reduces feelings of personal vulnerability, mitigates defensive withdrawal, and encourages proactive problem-solving rather than destructive retaliatory behaviors. Consequently, when psychological safety is elevated, the impulse to translate supervisory strain into counterproductive work behaviors is significantly attenuated.
The Moderating Role of Team Cohesion
Beyond individual psychological perceptions, social and group-level dynamic factors exert substantial influence over individual conduct. Team cohesion represents the degree to which team members exhibit mutual interpersonal attraction, shared commitment to team goals, and collective solidarity (Carron & Brawley, 2012; Costa et al., 2015). High cohesion fosters explicit and implicit normative regulations, peer support systems, and shared accountability among colleagues (Evans & Dion, 2012).
In hostile leadership environments, cohesive teams serve as collective protective buffers (Aubé & Rousseau, 2014; Wang et al., 2024). From the perspective of Group Resilience Theory (West et al., 2009), cohesive peer groups distribute psychological strain, facilitate shared emotional coping, and reinforce group norms that penalize harmful workplace deviance. When supervisory support is absent or abusive, strong lateral peer support compensates for leadership deficits, thereby weakening the pathway leading from toxic leadership to counterproductive work behaviors.
Theoretical Integration and Dual-Buffering Psychosocial Dynamics
Synthesizing these perspectives into a cohesive framework highlights the dual-buffering nature of organizational resources. Psychological safety functions as an individual-level perceptual buffer that protects cognitive and emotional resources, whereas team cohesion operates as a group-level structural buffer that enforces positive behavioral norms and offers mutual peer support (Kozlowski & Klein, 2000; Schmid et al., 2019). When combined within a single structural equation model, high psychological safety and high team cohesion are expected to demonstrate a synergistic buffering effect, minimizing employee engagement in counterproductive work behaviors under toxic leadership.
Public sector organizations in non-Western contexts, such as North Africa, present a critical empirical backdrop for evaluating this integrated framework. Public institutions in these regions often operate under high power distance, strict administrative hierarchies, and centralized decision-making, which can inadvertently enable or prolong toxic managerial styles (Bendahmane et al., 2020; Bouich, 2019; Hofstede, 2001). Conversely, the prevalence of cultural collectivism emphasizes group harmony and social solidarity, potentially enhancing the protective efficacy of team cohesion and psychological safety. Testing this moderated framework thus offers both theoretical nuance and contextual relevance.
Materials and methods
Research Design
This study employed a quantitative, cross-sectional research design to examine the structural relationships among toxic leadership, counterproductive work behavior (CWB), psychological safety, and team cohesion. A cross-sectional approach was chosen because it enables the simultaneous measurement of multiple constructs within active workplace environments and facilitates the evaluation of both direct and moderating effects efficiently (Creswell & Creswell, 2018). The analytical framework integrates bivariate correlation, hierarchical multiple regression, and structural equation modeling (SEM) using AMOS to rigorously test the hypothesized relationships and structural fit (Kline, 2016).
Context and Participant Recruitment
The empirical setting for this study comprised public-sector institutions. Public service organizations in this regional context are typically characterized by bureaucratic hierarchies, formalized administrative structures, and high power distance, which can present distinct managerial dynamics (Bendahmane et al., 2020; Bouich, 2019).
Participants were recruited using a stratified random sampling strategy across administrative, professional, and support departments to ensure adequate representation across demographic categories (age, gender, marital status, and organizational tenure). A total of 250 invitations were distributed, yielding 200 fully completed and usable responses (response rate = 80%). Participants’ ages ranged from 22 to 58 years with total work experience ranging between 1 and 30 years. The final sample size of N = 200 satisfies established criteria for SEM estimation, maintaining a ratio exceeding 5–10 observations per estimated parameter (Kline, 2016).
Participants
The target population consisted exclusively of full-time employees working within public-sector organizations across various governmental departments. A total of 200 participants were selected using a stratified random sampling technique to ensure proportional representation across key demographic variables, including age, gender, marital status, and work experience.
Participants’ ages ranged from 22 to 58 years, with professional experience varying between 1 and 30 years. The sample size was considered adequate for structural equation modeling (SEM) analysis, in line with methodological recommendations suggesting a minimum ratio of 5–10 participants per estimated parameter (Kline, 2016).
Inclusion Criteria
Participants were selected based on the following explicit inclusion criteria:
- Current full-time employment status within a public-sector institution.
- A minimum organizational tenure of six months under their current direct supervisor, ensuring sufficient exposure to leadership behaviors.
- Provision of voluntary informed consent prior to survey administration.
Instruments
All variables were measured using validated psychometric scales widely used in organizational psychology research:
Toxic Leadership Scale
Toxic leadership was measured using the scale developed by Schmidt (2008), which assesses five dimensions: abusive supervision, narcissism, authoritarianism, unpredictability, and self-promotion. Items were rated on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree).
Counterproductive Work Behavior Scale (CWB)
CWB was assessed using the instrument developed by Spector et al. (2006), covering five dimensions: production deviance, property deviance, political deviance, personal aggression, and withdrawal. Responses were recorded on a 5-point Likert scale.
Psychological Safety Scale
Psychological safety was measured using Edmondson’s (1999) 7-item scale, which evaluates employees’ perceptions of interpersonal risk-taking in the workplace. Items were rated on a 5-point Likert scale.
Team Cohesion Scale
Team cohesion was assessed using the scale adapted from Carron and Brawley (2012), consisting of 10 items measuring both task cohesion and social cohesion. Responses were recorded on a 5-point Likert scale.
Translation Procedure
All instruments were translated into Arabic and then back-translated into English to ensure linguistic accuracy and conceptual equivalence, following Brislin’s (1970) translation guidelines.
Reliability and Validity
The internal consistency of all scales was assessed using Cronbach’s alpha, while construct validity was evaluated through Confirmatory Factor Analysis (CFA). The results confirmed strong psychometric properties of all instruments.
Table 1
Reliability and CFA of Study Instruments
|
Variable |
Cronbach’s α |
χ²/df |
CFI |
TLI |
RMSEA |
Interpretation |
|
Toxic Leadership |
0,92 |
2.87 |
.95 |
0.94 |
.062 |
Excellent reliability and fit |
|
Counterproductive Work Behavior |
0,89 |
2.65 |
,96 |
0.95 |
.058 |
Very good fit |
|
Psychological Safety |
0,87 |
2.45 |
.97 |
0,96 |
,055 |
Good construct validity |
|
Team Cohesion |
0,91 |
2.52 |
.96 |
0.95 |
.059 |
Excellent fit |
Note. CFI = Comparative Fit Index; TLI = Tucker–Lewis Index; RMSEA = Root Mean Square Error of Approximation.
Findings prove the psychometric validity of all the instruments in the sample of the current study.
Discriminant and Convergent Validity Assessment
Construct validity was assessed by evaluating both convergent and discriminant validity using structural equation modeling (SEM). Convergent validity was evaluated using Average Variance Extracted (AVE) and Composite Reliability (CR). All constructs demonstrated AVE values exceeding the recommended threshold of 0.50 (Toxic Leadership =.61, CWB =.58, Psychological Safety =.56, Team Cohesion =.59) and CR values above.70 (ranging from.86 to.92), establishing robust convergent validity (Hair et al., 2010).
Discriminant validity was evaluated using Fornell and Larcker’s (1981) criterion, where the square root of the AVE for each construct must exceed its correlation with any other construct. Furthermore, a rigorous CFA model comparison was conducted to distinguish Psychological Safety from Team Cohesion. The baseline four-factor measurement model was compared against a constrained three-factor model combining Psychological Safety and Team Cohesion into a single factor.
Data Collection Procedures and Ethical Considerations
Data collection was conducted over a four-month period from December 2025 to March 2026. Following formal institutional approval from the participating public entities, paper-based questionnaires were administered in controlled settings during work hours. Participants received a cover letter detailing the study’s objective, confirming that participation was voluntary, and assuring strict confidentiality. Completed questionnaires were submitted directly into sealed collection boxes to ensure anonymity and eliminate potential supervisory coercion.
Data Analysis Techniques
Data were analyzed using SPSS 27 and AMOS 27. The following analytical procedures were applied:
Descriptive Statistics
Means, standard deviations, frequencies, skewness, and kurtosis were calculated to describe the sample and study variables.
Moderation Analysis Procedures
To test the moderation hypotheses (H2, H3, and H4) and minimize potential multicollinearity between main effects and interaction terms, all continuous predictor and moderator variables (Toxic Leadership, Psychological Safety, and Team Cohesion) were mean-centered prior to creating interaction terms (Aiken & West, 1991). Two-way interaction terms (TL x PS and TL x TC) and the three-way interaction term (TL x PS x TC) were generated using the product of the centered variables. Multicollinearity was evaluated using Variance Inflation Factor (VIF) values, with all values falling well below the conservative threshold of 3.0, confirming the absence of multicollinearity issues. To probe the nature of the significant interaction effects, simple slopes analyses and interaction plots were examined at high (+1 SD) and low (-1 SD) levels of the moderators, following the guidelines of Dawson and Richter (2006).
Correlation Analysis
Pearson correlation coefficients were used to examine the relationships between toxic leadership, CWB, psychological safety, and team cohesion.
Regression Analysis
Multiple regression analysis was conducted to test the direct effects of toxic leadership on CWB and the moderating effects of psychological safety and team cohesion, following Baron and Kenny’s (1986) approach.
Structural Equation Modeling (SEM)
SEM was used to test the full hypothesized model simultaneously, including direct and moderating effects, while assessing overall model fit indices (Kline, 2016).
Demographic Analysis
The sample was further analyzed according to demographic variables (age, marital status, and work experience) to ensure representativeness and explore potential group differences.
Results
Demographic Characteristics of the Sample
The sample consisted of 200 employees drawn from public organizations. Table 2 presents the demographic distribution across age, marital status, and work experience.
Table 2.
Demographic Distribution of Participants(N = 200)
|
Variable |
Category |
Frequency |
Percent |
Interpretation |
|
Age |
22–30 |
56 |
28% |
Younger employees; potentially more sensitive to leadership behavior |
|
|
31–40 |
78 |
39% |
Largest group; core workforce |
|
|
41–50 |
42 |
21% |
Experienced employees; rely on stability and cohesion |
|
|
51+ |
24 |
12% |
Senior staff; lower likelihood of CWB |
|
Marital Status |
Single |
64 |
32% |
May depend more on workplace social support |
|
|
Married |
136 |
68% |
Greater family stability; lower behavioral risk |
|
Work Experience |
1–5 years |
50 |
25% |
Early-career employees; higher vulnerability |
|
|
6–15 years |
92 |
46% |
Core organizational workforce |
|
|
16+ years |
58 |
29% |
Highly experienced; greater resilience |
The sample reflects a heterogeneous workforce, which enhances the generalizability of findings. Younger and less experienced employees appear more vulnerable to toxic leadership, whereas experienced employees demonstrate greater behavioral stability. However, psychosocial factors such as psychological safety and team cohesion may regulate behavioral responses across all groups.
Table 3.
Convergent and Discriminant Validity (Fornell–Larcker Criterion)
|
Construct |
CR |
AVE |
1 |
2 |
3 |
4 |
|
1. Toxic Leadership |
.92 |
.61 |
(.78) |
|
|
|
|
2. CWB |
.89 |
.58 |
.52** |
(.76) |
|
|
|
3. Psychological Safety |
.87 |
.56 |
-.46** |
-.38** |
(.75) |
|
|
4. Team Cohesion |
.91 |
.59 |
-.41** |
-.35** |
.62** |
(.77) |
Note. Bold values along the diagonal in parentheses represent the square root of the . Off-diagonal values are Pearson correlation coefficients (p < .01). CR = Composite Reliability; AVE = Average Variance Extracted.
As shown in Table 1b, the square root of the AVE for Psychological Safety (=.75) and Team Cohesion = .77) both clearly exceed the bivariate correlation between them (r =.62). This confirms that despite their positive correlation, the two variables represent distinct theoretical constructs, establishing strong discriminant validity (Fornell & Larcker, 1981).
Table 4.
Measurement Model Comparison (CFA)
|
Model |
χ2 |
df |
χ2/df |
CFI |
TLI |
RMSEA |
Δχ2(Δdf) |
|
Baseline Four-Factor Model |
328.8 |
120 |
2.74 |
.96 |
.95 |
.061 |
— |
|
Three-Factor Model(PS + TC merged) |
582.4 |
123 |
4.73 |
.84 |
.81 |
.114 |
253.6*** (3) |
*Note. p< .001. PS = Psychological Safety; TC = Team Cohesion.
The Chi-square difference test (Δχ2= 253.6, Δdf= 3, p< .001) indicates that merging Psychological Safety and Team Cohesion into a single factor resulted in a significantly worse model fit compared to the hypothesized four-factor model. This provides further empirical support for the construct distinctness of Psychological Safety and Team Cohesion
Descriptive Statistics
Table 5 presents the descriptive statistics for all study variables.
Table 5.
Descriptive Statistics of Study Variables
|
Variable |
Mean |
SD |
Skewness |
Kurtosis |
Interpretation |
|
Toxic Leadership |
3.42 |
.78 |
.31 |
-.12 |
Moderately high perception |
|
CWB |
2.71 |
.65 |
.45 |
.08 |
Moderate occurrence |
|
Psychological Safety |
3.88 |
.72 |
-.28 |
-.21 |
High levels |
|
Team Cohesion |
4.05 |
.68 |
-.35 |
-.18 |
Strong cohesion |
The results indicate moderate exposure to toxic leadership and counterproductive work behavior, while psychological safety and team cohesion are relatively high. This suggests that employees operate within partially supportive environments that may buffer negative leadership effects. All skewness and kurtosis values fall within acceptable limits (-1 to +1), confirming normality assumptions for parametric analyses (Kline, 2016).
Correlation Analysis
Pearson correlation coefficients are presented in Table 6.
Table 6.
Correlation Matrix
|
Variable |
1 |
2 |
3 |
4 |
|
1. Toxic Leadership |
1 |
|
|
|
|
2. CWB |
.52** |
1 |
|
|
|
3. Psychological Safety |
-,46** |
-.38** |
1 |
|
|
4. Team Cohesion |
-.41** |
-.35** |
.62** |
1 |
Note: p <,01
Toxic leadership is positively associated with CWB (r =.52, p< .01), supporting the expected direct relationship. It is negatively associated with psychological safety and team cohesion, indicating that destructive leadership undermines both individual and group-level resources. Additionally, psychological safety and team cohesion are strongly positively correlated (r =.62, p<,01), suggesting that cohesive teams tend to foster safer interpersonal environments.
Regression Analysis
Hierarchical regression results are presented in Table 7.
Table 7.
Hierarchical Regression Predicting CWB
|
Step |
Predictor |
β |
p |
ΔR² |
Interpretation |
|
1 |
Toxic Leadership |
.52 |
<.001 |
.27 |
Strong positive effect |
|
2 |
Psychological Safety |
-.21 |
.007 |
.04 |
Significant moderation |
|
3 |
Team Cohesion |
-.18 |
.012 |
.03 |
Significant moderation |
|
4 |
Interaction (PS × TC × TL) |
-.15 |
.035 |
.02 |
Joint buffering effect |
The results confirm all hypotheses. Toxic leadership significantly predicts CWB (H1 supported). Psychological safety significantly moderates this relationship, weakening the positive effect of toxic leadership on CWB (H2 supported). Similarly, team cohesion also acts as a significant moderator (H3 supported). Finally, the interaction of psychological safety and team cohesion shows a joint buffering effect (H4 supported), indicating a dual-level protective mechanism.
Structural Equation Modeling (SEM)
Model Fit
The SEM analysis demonstrated good model fit:
Table 8.
Model Fit Indices
|
Index |
Value |
Threshold |
Interpretation |
|
χ²/df |
2.74 |
< 3 |
Acceptable |
|
CFI |
.96 |
≥.95 |
Excellent |
|
TLI |
.95 |
≥.95 |
Excellent |
|
RMSEA |
.061 |
≤.08 |
Good |
|
SRMR |
.054 |
≤.08 |
Good |
These results confirm that the proposed model fits the observed data well (Kline, 2016).
Path Analysis
SEM results confirm all hypothesized relationships:
Table 9.
SEM Path Coefficients
|
Hypothesis |
Path |
β |
p |
Result |
|
H1 |
TL → CWB |
.52 |
<.001 |
Supported |
|
H2 |
TL × PS → CWB |
-.21 |
.007 |
Supported |
|
H3 |
TL × TC → CWB |
-.18 |
.012 |
Supported |
|
H4 |
TL × (PS × TC) → CWB |
-.15 |
.35 |
Supported |
|
|
|
|
|
|
The SEM model confirms that toxic leadership significantly increases CWB, while psychological safety and team cohesion act as protective moderators. The combined effect of both moderators demonstrates a synergistic buffering mechanism.
Probing the Three-Way Interaction Effect
To further interpret the significant three-way interaction effect ($\beta = -0.15, p = .035$), a simple slopes analysis was conducted comparing the relationship between toxic leadership and CWB across four distinct contextual conditions (combinations of high and low levels [±1 SD] of psychological safety and team cohesion):
- Low PS & Low TC: The positive relationship between toxic leadership and CWB was strongest and most severe (B =.58, p < .001).
- Low PS & High TC: The relationship remained positive but was significantly attenuated (B = .39, p < .01).
- High PS & Low TC: The relationship was similarly buffered (B =.35, p < .01).
- High PS & High TC (Dual-Buffering Condition): The positive relationship between toxic leadership and CWB was weakest and flattened considerably (B =.14, p = .42).
These results confirm a synergistic dual-buffering effect: while psychological safety and team cohesion independently reduce the behavioral damage of toxic supervision, their joint presence provides maximum protective efficacy against workplace deviance.
Structural Equation Model (SEM) and Figure Interpretation
The SEM diagram illustrates both the direct and moderating effects proposed in the conceptual model. As shown in Figure 1, toxic leadership has a significant positive relationship with counterproductive work behavior (CWB) (β = .52, p < .001), confirming that destructive supervisory practices increase the likelihood of undesirable employee behaviors.
In addition, this relationship is significantly moderated by psychological safety (β = -.21, p = .007) and team cohesion (β = -.18, p = .012). These negative interaction effects indicate that higher levels of psychological safety and stronger team cohesion weaken the positive impact of toxic leadership on CWB, acting as protective psychosocial resources.
Furthermore, the model reveals a significant three-way interaction effect (β = -.15, p = .035), suggesting a synergistic buffering mechanism. Specifically, the combination of high psychological safety and strong team cohesion produces the strongest reduction in counterproductive work behaviors under conditions of toxic leadership.
These findings are consistent with the correlation and regression results, reinforcing the robustness of the proposed model. Overall, the SEM results support and conservation of resources perspectives, which emphasize the role of personal and team-level psychosocial resources in mitigating the effects of destructive leadership.
Discussion
The present study investigated the relationship between toxic leadership and counterproductive work behavior (CWB), while testing the independent and joint moderating roles of psychological safety and team cohesion in public organizational settings. The empirical findings support the hypothesized model and offer significant insights into the boundary conditions that shape employee reactions to destructive leadership.
Consistent with Hypothesis 1, the results confirmed a significant positive relationship between toxic leadership and CWB. This finding aligns with established leadership literature demonstrating that destructive supervisory practices induce psychological distress, job dissatisfaction, and negative reciprocity (Al-Ghazali et al., 2025; Mackey et al., 2017; Tepper, 2000). Grounded in Social Exchange Theory (Blau, 1964), when employees experience persistent abusive, authoritarian, or self-serving leadership, they perceive a breach of fairness and psychological contract. Consequently, engaging in CWB (such as withdrawal, production deviance, or interpersonal hostility) functions as a compensatory mechanism to express frustration or restore perceived balance (Cropanzano et al., 2017).
The empirical findings also supported Hypotheses 2 and 3, establishing that psychological safety and team cohesion independently attenuate the toxic leadership–CWB relationship. Psychological safety significantly weakened the positive impact of toxicity on CWB, supporting Edmondson’s (1999) framework and recent multi-level findings (Kim & Park, 2025). When employees feel safe to express concerns and take interpersonal risks, the urge to respond to toxic supervision through covert or overt workplace deviance is diminished. Similarly, team cohesion acted as a critical group-level buffer. In line with Group Resilience Theory (West et al., 2009) and recent empirical evidence (Wang et al., 2024), cohesive peer groups foster lateral social support, shared coping strategies, and strong collective norms that discourage deviant behaviors, even in challenging supervisory environments.
Crucially, the confirmation of Hypothesis 4 demonstrates a joint dual-buffering interaction effect. When both psychological safety (an individual-level perceptual resource) and team cohesion (a group-level structural resource) are high, the positive relationship between toxic leadership and CWB drops to its lowest magnitude. From a Conservation of Resources (COR) perspective (Hobfoll, 1989), toxic leadership acts as an aggressive resource depletor. However, the presence of concurrent individual and group resources creates an enriched "resource reservoir" that protects employees from emotional exhaustion, neutralizing the translation of supervisory strain into counterproductive actions (Frazier et al., 2017; Schmid et al., 2019).
Conclusions
This empirical investigation provides a comprehensive understanding of how toxic leadership affects counterproductive work behavior (CWB) and highlights the critical buffering mechanisms provided by psychological safety and team cohesion within public sector organizations. By shifting the perspective from viewing employees solely as passive victims of supervisory abuse to examining the protective socio-psychological infrastructure within their immediate work environment, this study offers vital theoretical and practical insights.
The central takeaway of this research is that while toxic leadership inherently fuels organizational and interpersonal deviance by breaching psychological contracts and imposing severe emotional strain, its destructive consequences are not inevitable. The presence of robust, multi-level psychosocial resources can effectively intercept and neutralize this negative path. Specifically, when employees experience high levels of psychological safety (a personal-interpersonal belief of risk tolerance and trust) alongside strong team cohesion (a collective structural bond), the positive association between toxic leadership and CWB is significantly reduced to a non-significant level. This joint dual-buffering effect underscores the necessity of fostering resilient workplace climates capable of insulating human capital against leadership toxicity.
Practical and Managerial Implications
The empirical findings carry actionable recommendations for organizational leaders, human resource practitioners, and policy-makers, particularly within bureaucratic or hierarchical environments:
- Early Detection and Leadership Accountability: Organizations must implement multi-source assessment frameworks, such as anonymous 360-degree performance evaluations and exit interviews, to actively identify and curb toxic supervisory behaviors (abusive supervision, authoritarianism, and narcissism) before they permeate the organizational culture. Leadership development programs should emphasize ethical governance and supportive supervisory behaviors.
- Cultivating Psychological Safety as a Core Cultural Value: Executive management and department heads should actively cultivate work environments where open communication, constructive feedback, and interpersonal risk-taking are encouraged without fear of retribution. Training managers to practice empathetic communication and inclusive decision-making helps reduce the psychological strain that often triggers employee deviance.
- Strengthening Peer-Level Support Structures: Given that team cohesion acts as a powerful collective defense mechanism, organizations should design work structures that promote collaborative problem-solving, team-building exercises, and shared accountability. Building cohesive peer networks ensures that employees have access to lateral emotional and instrumental support when facing supervisory stressors.
Limitations and Future Research Directions
To further advance knowledge in this domain, several methodological and conceptual limitations of the current study should be addressed in future research:
- Methodological Scope and Temporal Design: The cross-sectional design utilized in this study precludes definitive longitudinal causal conclusions. Future inquiries should adopt time-lagged or longitudinal designs to observe how the protective effects of psychological safety and team cohesion evolve over time in response to ongoing toxic leadership.
- Data Source and Common Method Variance: Relying on self-reported survey measures introduces potential common method bias. Future studies would benefit from incorporating multi-source data collection strategies (e.g., combining employee self-reports with peer ratings or objective organizational metrics of CWB such as absenteeism and formal disciplinary records).
- Expanding Contextual and Personality Variables: The current study focused on a sample of public sector employees in a specific regional context. Future research should replicate this model across diverse industry sectors (e.g., healthcare, corporate finance, and technology) and cross-cultural contexts. Additionally, exploring how individual dispositional traits—such as emotional intelligence, locus of control, or dark triad traits—interact with team-level resources could yield deeper insights into employee resilience mechanisms.