Galperin (2012) defines constructive deviance as a voluntary behavior that violates significant norms with the intent of improving the wellbeing of an organization, its members, or both. For example, employees are usually required to follow certain organizational rules and standards when interacting with customers, but sometimes must deviate from these established processes or standards to adequately respond to customer needs. Previous studies have stressed that constructive deviance is characterized by the following. (a) Deviations from the norms of the reference group. This characteristic reflects the differences between constructive deviance and organizational citizenship behaviors, because the latter do not involve rules-violating and risk-taking (Vadera et al., 2013). (b) Benefits to the reference group. As Galperin (2012) notes, distinguished from the construct of destructive deviance, the intent of constructive deviance is to benefit the organization. (c) Conformation to hypernorms which are “globally held beliefs and values” (Vadera et al., 2013). Constructive deviance conforms to moral norms and standards (Vadera et al., 2013), and so is theoretically different from unethical pro-organizational behavior. Although both constructs are pro-organizational, unethical pro-organizational behavior is morally problematic because it violates core societal values or norms and potentially damages the organization’s sustainability (Bryant and Merritt, 2021).

Based on social learning theory, Brown et al. (2005) propose the concept of ethical leadership, defined as “the demonstration of normatively appropriate conduct through personal actions and interpersonal relationships, and the promotion of such conduct to followers through two-way communication, reinforcement, and decision-making” (Brown et al., 2005, p. 120). As Brown et al. (2005) highlight, two crucial dimensions of ethical leadership are being a moral person and a moral manager. The former refers to the leaders’ desirable personal characteristics such as integrity, altruism, and trustworthiness. The latter refers to how a leader utilizes moral managerial measures to influence followers’ ethical decision-making and guide their ethical behaviors. Researchers on behavioral ethics have built on this work and demonstrated the positive influence of ethical leadership on many outcomes related to employees’ ethics, such as moral voice (Lee et al., 2017), knowledge sharing (Bavik et al., 2018), and internal whistleblowing (Cheng et al., 2019).

Constructive deviance—deviating from organizational norms but conforming to moral norms and standards (Warren, 2003) is increasingly regarded as an ethical decision by employees (Monin et al., 2008; Jetten and Hornsey, 2014; Vardaman et al., 2014; Zhang et al., 2021). Because ethical leadership is important in influencing employees’ ethical decision-making, we posit that ethical leadership may positively affect employees’ constructive deviance. First, ethical leaders value organizational wellbeing and speak out publicly against inappropriate organizational norms and actions that harm the interests of the organization and its members (Walumbwa and Schaubroeck, 2009; Islam et al., 2021). From a social learning perspective (Bandura, 1977, 1986), employees proactively learn and emulate their leaders’ behaviors. Consequently, employees may learn their ethical leaders’ behaviors and challenge inappropriate organizational rules, which triggers constructive deviance.

Second, at its core, ethical leadership is about contributing to the organization and positively helping others (organizational members and other stakeholders), without expecting personal gain or reward (Kanungo, 2009). Social learning theory holds that employees observe and learn their supervisors’ behavioral norms to understand how to behave at work (Bandura, 1977, 1986). As a result, ethical leaders can shape employees’ moral altruistic attitudes through the process of social learning. Such moral altruistic attitudes enhance employees’ concerns for the organization, thus, these employees will not only seek to satisfy their personal interests but also strive to benefit their organization, which may entail constructive deviance.

Finally, ethical leaders show genuine concern and care for their followers, and fully respect their interests (Brown et al., 2005). When employees feel respected and supported by their leaders, they exhibit more favorable attitudes and reciprocate with pro-organizational behaviors (Qian et al., 2017). Accordingly, constructive deviance may arise from employees’ willingness to make extra effort beyond their ordinary work duties to realize the organizational goals. In line with this reasoning, we hypothesize that ethical leadership promotes the emergence of employees’ constructive deviance.

Self-efficacy refers to an individual’s perception of their capability to accomplish tasks (Bandura, 1997). Hannah and Avolio (2010) expand this construct to the domain of behavioral ethics and defined moral self-efficacy as “an individual’s belief in his or her capabilities to organize and mobilize the motivation, cognitive resources, means, and courses of action needed to attain moral performance, within a given moral domain, while persisting in the face of moral adversity” (Hannah and Avolio, 2010, p. 297). They further describe moral self-efficacy as “state-like vs. trait-based,” and thus it is context-dependents. Ethical leadership can shape employees’ moral self-efficacy through the social learning processes. Social learning theory states that individuals look to their leaders’ behaviors for cues to form or develop their own attitudes and behaviors (Bandura, 1977, 1986). By observing and learning ethical leaders’ values and norms concerning moral issues, employees not only gain meaningful experience relevant to solving moral issues (Kim and Vandenberghe, 2020) but also enhance their moral capabilities (Ruedy and Schweitzer, 2011). Such moral capabilities help to increase employees’ moral self-efficacy because they develop “potential response repertoires” toward moral issues at work (Hannah and Avolio, 2010).

A high sense of moral self-efficacy implies that employees believe they can implement ethical behaviors successfully in accordance with their ethical beliefs (Ogunfowora et al., 2021). Although they may experience adversity in performing moral behaviors, employees with moral self-efficacy will remain resilient to adversity, and attempt to utilize relevant resources to overcome obstacles (May et al., 2013; Afsar and Shahjehan, 2018). Previous studies have also shown that moral self-efficacy promotes such ethical risk-taking behaviors as moral voice (Lee et al., 2017; Afsar and Shahjehan, 2018) and team organizational citizenship behaviors (Kim and Vandenberghe, 2020). Given that constructive deviance involves moral relevance and risk-taking, we argue that employees with moral self-efficacy are more likely to persist when resolving ethical dilemmas and make ethical decisions, which inspires constructive deviance. Thus, we hypothesize:

Uncertainty reduction theory suggests that uncertainty is an adverse experience; therefore, individuals rely on information from their work environment to mitigate concerns and anxiety brought about by uncertainty (Lind and Bos, 2002). A psychological safety climate is a supportive environment state characterized by “the shared belief among team members that the team is safe for interpersonal risk taking” (Edmondson, 1999, p. 354). Edmondson (1999) stresses that a psychologically safe climate is a work climate characterized by interpersonal trust and mutual respect. As such, a psychological safety climate suppresses distracting interpersonal concerns and relieves team members’ fear of uncertainty. Because constructive deviance is associated with uncertainty and risk, employees may have a higher intention to conduct constructive deviance when the team climate reduces uncertainty. Drawing on uncertainty reduction theory, we expect that ethical leadership positively predicts employees’ constructive deviance by providing a psychological safety climate.

According to uncertainty reduction theory, leaders are essential in helping employees cope with uncertainty (Lind and Bos, 2002; Tu et al., 2018). Ethical leaders are moral managers who set ethical standards and utilize reward and punishment mechanisms to uniformly enforce these standards in the work context (Brown and Treviño, 2006). In this way, ethical leaders’ ethical managerial measures convey to their followers what constitutes appropriate behavior and thus clarify followers’ behavioral roles. By doing so, ethical leaders reduce subordinates’ feelings of uncertainty and increase their psychological safety. Moreover, ethical leaders care for the interests of employees, provide them fair treatment, and use ethical values to integrate teams (Brown et al., 2005). Therefore, ethical leaders promote interpersonal trust and mutual respect among members, thereby creating a psychologically safe team climate. In a psychologically safe environment, team members believe that they share mutual concern and respect with their coworkers (Edmondson, 1999). Thus, members of psychologically safe teams focus on constructive problem-solving without fear of being excluded or humiliated by other members (Bradley et al., 2012; Liang et al., 2012). Furthermore, in a team climate of strong psychological safety, members feel safe taking risks, thus, spurring their moral risk-taking behaviors. Indeed, extensive research supports the relationship between psychological safety climate and employees’ moral risk-taking behavior. For instance, Walumbwa and Schaubroeck (2009) suggest that members working in psychologically safe teams willingly engage in constructive voice behavior. Wadei et al. (2021) also find that in psychologically safe teams, members are likely to challenge the status quo. On this basis, we argue that psychologically safe climates promote employees’ constructive deviance. We thus hypothesize:

Prior studies suggest that people who work in a low team psychological safety climate exhibit worse task performance than people in a high team psychological safety climate (Edmondson, 1999; Koopmann et al., 2016). When in teams with weaker psychological safety climate, team members consider the potential risks of engaging in constructive deviance because they focus on self-protection and risk prevention. For instance, employees may be hesitant to conduct constructive deviance because they are anxious about ostracism from their coworkers or criticism for breaking organizational rules. Thus, team members in a weak psychological safety climate may regard constructive deviance as an irrational behavior with high risk and low return; even if they feel a high sense of moral self-efficacy, they may avoid constructive deviance to reduce the risks of uncertainty. In contrast, a stronger psychological safety climate usually indicates beneficial interpersonal interactions (Koopmann et al., 2016). In such a climate, members are likely to trust their coworkers and not be embarrassed to express themselves (Zhang et al., 2010), which reduces their concerns regarding constructive deviance. Therefore, we argue that a stronger psychological safety climate may increase the effect of employees’ moral self-efficacy on constructive deviance. Thus, we hypothesize:

Based on the above conjecture, we predict that team psychological safety climate moderates the mediating effect of moral self-efficacy. Specifically, we predict that the association between ethical leadership on employees’ moral self-efficacy is stronger for higher levels of psychological safety climate, which improves moral self-efficacy and so reinforces team members’ beliefs that they are capable of enacting constructive deviance. In contrast, in a climate with a lower level of psychological safety, the association between ethical leadership and employees’ moral self-efficacy may be weaker, which may lead to lower confidence in their ability to enact constructive deviance, reducing their enactment of constructive deviance. Therefore, we hypothesize:

We issued questionnaires to 69 supervisors and 405 of their direct subordinates who engage in R&D (research and development) or customer services employed by 15 different companies in high technology, finance, and service industries in eastern China. These participants were chosen for two reasons. First, R&D and customer service jobs are flexible and members of such teams are more likely to solve problems creatively. Second, R&D and customer service jobs require members to collaborate, implying that these jobs have a high degree of task correlation, and so individual variables will generally have small between-group differences, thus making systemic between-group differences are more prominent, facilitating cross-level analysis. Before distributing questionnaires, the anonymity of responses and the definition and examples of constructive deviance were communicated to participants. To ensure that supervisor-subordinate data are well matched, we sought participants’ permission to use the initials of each subordinate’s name and the last 4 digits of their employee identification number to code the questionnaires (e.g., LN6890, etc.).

To alleviate possible common method bias, we adopted the methods recommended by Podsakoff et al. (2003) to collect survey data at two different times. At time 1, all participants were asked to respond to measures of demographic variables, and 405 subordinate employees were asked to complete questionnaires on ethical leadership and psychological safety climate in a separate survey. Four weeks later (time 2), 69 team supervisors were asked to rate their direct subordinates’ moral self-efficacy and constructive deviance. After excluding incomplete questionnaires, a total of 313 valid matched supervisor-subordinate dyads were obtained in 52 teams from 15 companies, including 52 supervisor surveys and 313 subordinate surveys, giving an overall response rate of 77.3%. Of the 52 supervisors, 44 were male (84.62%) and 8 were female (15.38%); 67.9% were in the age group 41–50 and 15.1% were above 50 years old; the overall level of education was high, with 32% having a graduate degree. Of the 313 subordinates, 62.5% were female; 44.7% were in the age group 31–40; and the average tenure was 8.89 years.

Questionnaire items used a five-point Likert scale (1 = strongly disagree to 5 = strongly agree). To ensure clarity and consistency of the survey instrument in the Chinese context, a back-translation process was applied to the survey, as recommended by Brislin (1986). All questionnaire items are presented in the Appendix.

A ten-item scale (Brown et al., 2005) was used to measure ethical leadership. One sample item is “Our team leader disciplines employees who violate ethical standards” Cronbach’s alpha was 0.933. Following previous studies (Bai et al., 2017; Christensen-Salem et al., 2020), we conceptualize ethical leadership at the team level. The results of data aggregation testing show that the values for r_wg and intraclass correlation index ICC (1) and ICC (2) are 0.954, 0.746, and 0.945, respectively. All three values are above the recommended thresholds (Schneider et al., 1998), indicating that aggregation to the team level is justified.

A five-item scale developed by Hannah and Avolio (2010) was used to assess moral self-efficacy. One sample item is “My subordinate can fight against people who use unethical behavior to solve problems.” Cronbach’s alpha was 0.908.

Liang et al.’s (2012) five-item scale was used to assess psychological safety climate. One sample item is “In my work unit, I can freely express my thoughts.” Cronbach′salpha was 0.854. This scale indicated r_wg, ICC (1) and ICC (2) values of 0.905, 0.321 and 0.739, respectively, thus supporting team-level aggregation.

A 10-item scale developed by Galperin (2012) was used to measure constructive deviance. Sample items are “My subordinate sought to bend or break the rules in order to perform your job” and “My subordinate departed from dysfunctional organizational policies or procedures to solve a problem.” Cronbach′salpha was 0.872.

Individual-level control variables include employees’ gender, age, and tenure were controlled, because these variables potentially impact constructive deviance. Previous studies suggest that male employees are more likely than females to exhibit constructive deviance (Dahling and Gutworth, 2017). Age and tenure may also influence constructive deviance, because employees who have a longer-term relationship with their leader (reflected in age and tenure) may be more willing to make extra effort for their organization (Tangirala and Ramanujam, 2008; Duan et al., 2017). Further, following prior studies (e.g., Bai et al., 2016; Liu and Li, 2018), leaders’ age, gender, and education, as team-level control variables, were also controlled.

We first conduct confirmatory factor analyses using Mplus7.4 to examine the discriminant validity of our four main variables ethical leadership, psychological safety climate, moral self-efficacy, and constructive deviance. Table 1 shows that the four-factor model (χ²/df = 2.033, RMSEA = 0.058, SRMR = 0.0422, CFI = 0.945) has a better fit with the data than the alternative models, indicating that the discriminant validity of the four constructs was good.

Following previous studies’ suggestions regarding checking common method bias (Podsakoff et al., 2003; Liang et al., 2021), the unmeasured latent method construct approach is used to analyze our collected data to further examine whether common method variance could bias our results. Specifically, we add the common method bias factor to the baseline model (i.e., the four-factor model) and contrast the fit indices of the two models. The fit indices of the five-factor model do not show a significant improvement (ΔSRMR = 0.0046, ΔRMSEA = 0.002, ΔCFI = 0.003), indicating that common method variance should not be a serious concern.

Descriptive statistics and correlations of all studied variables are presented in Table 2. Correlation analyses show that moral self-efficacy is positively associated with constructive deviance (r = 0.653, p < 0.01), and ethical leadership is positively associated with psychological safety climate (r = 0.674, p < 0.01).

Using Mplus7.4, we first examine the null model with no predictors to provide support for further multilevel analyses. The test results showed that the intragroup variance, intergroup variance, and ICC values of constructive deviance were 0.197, 0.348, and 0.644, respectively, supporting the use of the data in cross-level analysis.

First, our findings reveal a cross-level positive effect of ethical leadership on employees’ constructive deviance. Therefore, our empirical analysis enhances our understanding of the outcomes of team ethical leadership, and provides evidence for the impact of ethical leadership on employees’ behaviors at the team-level in China. In addition, prior studies mainly focus on the effect of leadership style on employees’ constructive deviance at the individual level, and less attention has been given to this relationship at the team level, particularly in a Chinese context. Vadera et al. (2013) have called for more studies on the facilitators and inhibitors of constructive deviance using cross-level analysis. Wang (2022) has also stressed the imperative of exploring team-level antecedents of constructive deviance. This study responds to these calls by exploring the influence of team-level contextual factors (i.e., ethical leadership) on employees’ constructive deviance.

Second, most prior studies rely on theoretical perspectives such as social identity theory or social exchange theory to explain how leadership style influences employees’ constructive deviance. Our study contributes to the existing research from two perspectives. First, by integrating social learning and uncertainty reduction theories, we provide additional theoretical lenses to uncover the influencing mechanisms of leadership style on employees’ constructive deviance. Second, we identify psychological safety climate and moral self-efficacy as two important mediating mechanisms that link ethical leadership and employees’ constructive deviance. Our study therefore offers a more complete understanding of how team-level ethical leadership influences employees’ pro-organizational behavior (e.g., constructive deviance) by introducing different mediating mechanisms into our cross-level process model.

Third, this study reveals the key role of external contextual factors as boundary conditions in the link between leadership style and constructive deviance by demonstrating the moderating effect of psychological safety climate. Previous studies on boundary conditions of constructive deviance focus on individual traits, such as employees’ traditionality or positive personality, and studies on the impact of external contextual factors as moderators of constructive deviance are scarce. Our study enhances our understanding of the boundary effects of external contextual factors on constructive deviance by empirically examining the moderating role of team psychological safety climate in employees’ moral self-efficacy-constructive deviance relation. Hence, we more completely discuss when and why employees are willing to engage in constructive deviance.

Finally, psychological safety climate is found to moderate the association between employee moral self-efficacy and constructive deviance. However, our findings did not support the moderated mediation effect, which suggests that other plausible intervening variables may exist. For example, guanxi, a type of norms of interaction grounded in the Chinese Confucian ethical system, profoundly influences Chinese employees’ behavior (Chen and Bedford, 2022). For Chinese employees, establishment and maintenance of relationships with their supervisors are not limited to work, but also occur in a large amounts of non-work-related social activities. Such personal relationships with supervisors even influence employees’ work behaviors (Ko et al., 2017). Ding et al. (2017) show that guanxi orientation may weaken the positive impact of employee intrinsic motivation on voluntary behavior intention because individuals with high guanxi orientation value relationship harmony and pay little attention to self-actualization. Given that constructive deviance involves potential interpersonal conflicts, guanxi orientation may negatively influence employee constructive deviance, and thus could statistically offset the positive effect of team psychological safety climate on employees’ constructive deviance.

Our findings have several practical implications for organizations. First, we show that team-level ethical leadership plays a vital role in motivating constructive deviance among team members. Thus, ethical leadership behaviors should be rewarded and developed. Relevant measures include promoting leaders who possess high ethical standards, linking the reward and punishment system to assessment of ethical leadership behaviors, and developing ethical leadership training programs that cultivate team leaders’ ethical leadership abilities.

Second, given the positive influence of employee moral self-efficacy on constructive deviance, leaders should develop this positive psychological state in their employees. Leaders’ primary efforts may concentrate on raising employees’ confidence in their ability to participate in ethical behaviors. For example, leaders may develop employees’ knowledge and understanding of ethical decision making by involving them in moral discourse and discussing moral issues with them. Also, using case studies and role-taking are both effective ways to enhance employees’ moral efficacy. When employees’ moral efficacy has been developed, they may become more confident in their ability to successfully implement constructive deviance.

Finally, to create a climate of team psychological safety, leaders must consider potential interpersonal risks and uncertainty in the workplace. They should recognize that they can function as organizers and coordinators to guide employees to trust and respect each other and reduce employees’ concerns regarding potential interpersonal conflicts. By doing so, leaders can provide a psychologically safe climate for their subordinates, which in turn, helps to facilitate the emergence of constructive deviance in the team.

Our research has some limitations that warrant consideration. First, although we obtain data from multiple sources, the problem of social desirability may still influence our research results. Specifically, when employees report the level of their supervisor’s ethical leadership, they may conceal their true perceptions out of consideration of social desirability (Loo and Thorpe, 2000; Spector, 2006; Feng et al., 2016). Therefore, in accordance with Feng et al. (2016), we encourage future research to use random experiments and larger longitudinal samples to alleviate contamination due to social desirability.

Second, the construct of ethical leadership is proposed and developed based on Western business ethics. As stressed by Resick et al. (2006), the content of ethical leadership is varies across culture.

Influenced by Chinese collectivistic culture, Chinese employees are more concerned about whether ethical leaders’ behaviors meet the highest moral standards than are employees who are influenced by individualistic culture (Resick et al., 2006; Zhu et al., 2019). This difference may lead employees in different cultures to have different perceptions and reactions to the behavioral manifestations of ethical leaders, which may influence their job outcomes (e.g., constructive deviance). In this study, we do not control for the impact of this cultural effect because our participants were all from similar cultural settings. However, considering the question of cultural differences, we hope that future research can replicate or extend our model in different cultural contexts, which will help to attain a more nuanced understanding of ethical leadership’s impact on employee behaviors.

Finally, although we examine two different mediating mechanisms (i.e., psychological safety climate and moral self-efficacy), other potential linkage mechanisms have not yet been fully explored. For example, moral ownership is defined as a sense of responsibility that individuals feel for themselves, others, and organizational ethical actions. Ogunfowora et al. (2021) demonstrate that moral ownership encourages employees to conduct morally courageous behaviors. Likewise, intrinsic motivation is an individual’s vital motivational force, and it may cognitively stimulate employees’ constructive deviance. These possibilities indicate that other linkage mechanisms may also influence the relationship between ethical leadership and employee constructive deviance. Therefore, future studies may continue investigating other potential mechanisms and further enhance our understanding of how ethical leadership influences employees’ constructive deviance.