Examining the Boundaries of Ethical Leadership: The Harmful Effect of Co-worker Social Undermining on Disengagement and Employee Attitudes

Ethical leadership involves “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). This style of leadership includes a moral person aspect and a moral manager aspect (Brown and Treviño 2006). The moral person aspect reflects leaders’ traits or qualities such as fairness, integrity, and trustworthiness. Comparatively, the moral manager aspect reflects how ethical leaders promote ethical behaviour in the workplace. Such behaviours include involving followers in decision-making, providing clarity about appropriate conduct and ethical expectations, incentivising ethical behaviours and punishing unethical conduct (Brown et al. 2005; Treviño et al. 2003; Vullinghs et al. 2018). In this respect, ethical leaders treat followers fairly and with respect by demonstrating people-oriented behaviours and care for their needs (Brown et al. 2005; Treviño et al. 2003; Vullinghs et al. 2018).

Ethical leadership is theoretically different from other forms of positive leadership, such as transformational and authentic leadership (Ng and Feldman 2015; Hoch et al. 2018). Contrary to transformational leadership, where the main focus is on role modelling, ethical leadership comprises a transactional component, which involves the use of punishment or discipline for unethical conduct or behaviour (Brown et al. 2005; Stouten et al. 2013). Ethical leadership is also distinct from authentic leadership, wherein leaders focus primarily on relational transparency and self-awareness rather than ethical behaviour (Stouten et al. 2013). Thus, rather than including ethics as an auxiliary or secondary dimension, ethical leadership explicitly emphasizes the moral aspects of leadership (Hoch et al. 2018; Mostafa 2018).

Recent research has shown that ethical leadership is negatively related to turnover intentions, and positively related to job satisfaction and organisational commitment (Ng and Feldman 2015; Demirtas and Akdogan 2015; Bedi et al. 2016; Hoch et al. 2018; Wang and Xu 2019). However, as noted before, a better understanding is still needed of the mediating psychological process through which ethical leadership induces its effects on these outcomes (Ng and Feldman 2015; Wang and Xu 2019). This study proposes that the relationship between ethical leadership and job satisfaction, organisational commitment, and turnover intentions is mediated by disengagement.

Although burnout has been conceptualised in different ways since its inception, there is general agreement that it consists of two main components: high levels of exhaustion and a cynical/distant reaction towards one’s work (Demerouti et al. 2019). The exhaustion component is known as emotional exhaustion, which involves “feelings of being overextended and depleted of one’s emotional and physical resources” (Maslach et al. 2001, p. 399). The cynicism component was originally known as depersonalisation, yet more recently it has been labelled disengagement (Sonnentag 2005). Disengagement is the degree to which a person withdraws or distances from all work aspects (Demerouti et al. 2010). Kahn (1990, p. 64) states that disengaged employees “withdraw and defend themselves physically, cognitively, or emotionally during role performances”. This perspective is echoed by Bakker et al. (2004, p. 84) who state that it involves an “emotional, cognitive, and behavioural rejection of the job”. As a result, disengaged employees not only perceive that their work tasks are routine, but they also conduct tasks in a mechanical manner or engage in withdrawal behaviours (Demerouti et al. 2001).

Limited attention has been directed towards disengagement and its antecedents (Pech and Slade 2006; Hejjas et al. 2019). This is relevant because recently, it has been suggested that engagement and disengagement are “not opposites” and that different factors could drive or inhibit each construct (Hejjas et al. 2019, p. 329). Indeed, as argued by Macey and Schneider (2008), the opposite of engagement is likely to be “non-engagement”, not disengagement. Therefore, whilst it has been documented that ethical leadership is positively linked to work engagement (Chughtai et al. 2015; Demirtas 2015), it cannot necessarily be inferred that it limits disengagement. Nor that the strategies ethical leaders use to promote work engagement are necessarily the same strategies that reduce disengagement. As a result, this study investigates whether ethical leadership is negatively related to disengagement and considers why such a relationship may exist.

Based on Henderson and Argyle’s (1986) social rules theory (SRT), we propose that ethical leadership leads to reduced levels of disengagement and consequently improved employee attitudes. SRT states that in working relationships, shared expectations exist regarding behaviour that should or should not be performed in specific situations (Argyle et al. 1981). These shared expectations are known as social rules, which Henderson and Argyle (1986, p. 260) describe as “behaviour which members of a group or subculture believe should or should not be performed, either in certain situations or in a range of situations”. The theory postulates that violation or non-compliance with these rules is more likely to be associated with reduced wellbeing and negative employee attitudes, whereas rule compliance will be associated with improved employee wellbeing and positive attitudes (Henderson and Argyle 1986).

SRT asserts that there are a number of universal social rules, which are appropriate in every social situation, such as being friendly, polite, respecting privacy, and maintaining eye contact (Argyle et al. 1981). However, some social rules also vary according to context. For example, Henderson and Argyle (1986) proposed a number of social rules that are unique to the relationship between leaders and their subordinates. These include giving clear guidance to subordinates, organising work efficiently, looking after subordinates’ welfare and acting with fairness. These social rules align with the nature of ethical leadership, which involves displaying normatively appropriate conduct in personal interactions to the benefit of subordinates. Ethical leaders care for their employees, support them, treat them fairly, and with respect (Brown et al. 2005). They consolidate “a general, consistent moral character with a focus on organizational or cultural norms, standards, and rule compliance” (Lemoine et al. 2019, p. 151). Therefore, they are more likely to enhance employee wellbeing and reduce disengagement. Previous research provides support for these assumptions. For example, Mo and Shi (2017) found that ethical leadership was negatively related to burnout. Similarly, Chughtai et al. (2015) found that ethical leadership reduced emotional exhaustion. We, therefore, expect ethical leadership to reduce disengagement.

As disengagement involves cognitively, emotionally, and behaviourally distancing oneself from work (Rathi and Lee 2016), reducing or limiting disengagement should alter how employees view their job and organisation. Specifically, we argue that when ethical leaders are able to reduce or limit disengagement, the extent to which employees wish to leave their organisation will also be limited. This is because when attempting to cope with the demands of a workplace, disengaged employees adopt avoidance based coping behaviours, such as withdrawal (Pienaar and Bester 2011; Rathi and Lee 2016). One of the cognitive manifestations of withdrawal involves thinking about quitting ones job (Lachman and Diamant 1987). Therefore leaders who are able to reduce disengagement should in turn limit the extent to which employees think about quitting.

We also contend that reducing or limiting disengagement should be associated with higher levels of job satisfaction and organisational commitment. When ethical leaders are able to maintain low levels of follower disengagement, employees should have more energy to devote to their tasks, which should result in higher levels of job satisfaction as tasks will not seem so monotonous. Disengagement should also be negatively associated with organisational commitment. Previous research has drawn on Conservation of Resources theory (Hobfoll 1989) to explain the relationship between disengagement and commitment. Thanacoody et al. (2014) found that disengagement mediated the relationship between emotional exhaustion and affective commitment. The authors suggested that disengaged employees have less resources and energy to commit to their organisations. As a result, when disengagement is limited, employees can invest energy in their job or exert effort on behalf of the organisation (Sun and Pan 2008; Rathi and Lee 2016). We, therefore, argue that by reducing disengagement, ethical leadership will have an indirect effect on job satisfaction, organisational commitment, and turnover intentions.

Co-worker social undermining refers to intentional behaviour from co-workers that reduces another employee’s ability to maintain good relationships, success, and a positive reputation at work (Duffy et al. 2002). In developing the construct for the workplace, Duffy et al. (2002) highlighted three core characteristics. Firstly, behaviour is only considered undermining if it is perceived as intentional by the target. This differentiates social undermining from other forms of workplace aggression, such as bullying, incivility, and abusive supervision, which can occur in the absence of ill-intent (Hershcovis 2011). Secondly, social undermining behaviours may not be harmful if they occur rarely, but their impact is cumulative in that they damage relationships in an iterative manner. Thirdly, social undermining is examined from the target’s perspective. Accordingly, there may be perceptual differences between the perpetrator and target over whether behaviour was intended to harm. The nature of undermining behaviour is varied, in that it may be direct (e.g. overt belittling), or indirect (e.g. withholding information). Furthermore, Duffy et al. (2002) differentiated co-worker social undermining from supervisor social undermining, with the former enacted by one’s colleagues and the latter by one’s supervisor.

We draw upon SRT to argue that when co-worker social undermining behaviour is experienced by followers of an ethical leader, the negative relationship between ethical leadership and disengagement is weakened. SRT states that besides social rules that guide the relationship between leaders and their subordinates, there are also rules which govern the relationships between co-workers, including accepting a fair share of the work, helping when asked and seeking to repay debts, favours and compliments. Henderson and Argyle (1986) also confirmed empirically that the common universal rules were highly endorsed for relationships between colleagues, such as not criticising each other publicly and standing up for a colleague when they are not present to defend themselves.

Co-worker social undermining is a particularly appropriate form of workplace aggression to examine as a rule-breaking behaviour for two reasons. Firstly, unlike other forms of workplace aggression, including bullying, incivility, and conflict, social undermining is perceived to be enacted intentionally (Hershcovis 2011). As a result, it is an overt form of rule-breaking behaviour, in that the target perceives that a co-worker has intentionally violated the rules that govern appropriate behaviour. Secondly, most other workplace aggression variables do not specify the source of the mistreatment within their measures (Hershcovis 2011), which means that they may have been enacted by a leader. Co-worker social undermining is explicitly enacted by one’s colleagues, which allows us to determine whether the social rules were broken by a source other than the ethical leader.

We contend that ethical leaders seek to establish work and group contexts where the social rules governing relationships between colleagues are upheld. Ethical leaders shape follower behaviour by using reward and punishment to clarify ethical expectations (Brown and Treviño 2006). Therefore, when social undermining occurs within their work group, one would expect an ethical leader to use transactional leadership principles by punishing the perpetrator. At the same time, ethical leaders will role model appropriate forms of behaviour and reward those that treat colleagues in an ethical manner (Den Hartog 2015). Theoretically, this should mean that social undermining rarely occurs under ethical leaders, and when it does, an ethical leader would intervene to prevent the behaviour reoccurring. Indeed, studies have shown that certain leadership styles, including transformational and transactional leadership are related to decreased levels of workplace aggression (Astrauskaite et al. 2015; Ertureten et al. 2013). However, research on the antecedents of social undermining has shown that there are a number of causal factors which are beyond a leader’s control, including bottom line mentality (Greenbaum et al. 2012) and moral disengagement (Duffy et al. 2012). Therefore it is possible that social undermining will occur under the supervision of ethical leaders (Taylor and Pattie 2014).

There are two potential ways in which social undermining can weaken the negative relationship between ethical leadership and disengagement. Firstly, social undermining can be perpetrated by other employees who are also supervised by the ethical leader. When this occurs, one would expect the ethical leader to intervene to prevent further undermining behaviour. Yet, despite this, the target may still experience heightened disengagement after the incident. This is because, when an act of social undermining occurs under an ethical leader, it is a more flagrant breach of social rules than when it occurs under other forms of leadership. Duffy et al. (2006) showed that the impact of social undermining is stronger when it only occurs rarely in the workgroup. They argued that this occurs partly due to the discordance between-group norms and personal experiences, as perceived violations arose less negative reactions when the group norm is more tolerant of violations (Leung and Tong 2003). This can be explained by fairness theory as employees cognitively compare their own treatment against that of their co-workers (Folger and Cropanzano 1998). If one’s co-workers are also experiencing social undermining, the victim is less likely to feel that the perpetrator could have acted differently, which affects how strongly a person reacts to the treatment. However, when one’s co-workers have not experienced social undermining, the victim is more likely to feel that the perpetrator could have acted according to the social rules set out by the ethical leader. Duffy et al. (2006, p. 108) state that this is important, because “being singled out creates a more mutable and feasible cognitive comparison and a more damaging context for social undermining”. Therefore the victim is much more likely to experience disengagement.

Secondly, social undermining can be perpetrated by other organisational members who are not under the supervision of ethical leaders, or by others more senior than the ethical leader in the organisational hierarchy. This may particularly be the case in large organisations where there are many layers of management. Other organisational members may be supervised by leaders who do not encourage norms of appropriate conduct towards others. Therefore, they may be more inclined to engage in social undermining behaviour towards colleagues, and particularly towards those who they see as out group members (Ramsay et al. 2011). Nonetheless, all organisational members remain bound by the social rules that govern relationships between co-workers, and targets will still perceive social undermining as rule-breaking behaviour. In such situations, the relationship between ethical leadership and disengagement is still likely to be weakened by social undermining, as the ethical leader has not been able to protect their subordinate’s welfare. Therefore the leader/subordinate specific social rule “look after subordinates’ welfare” will have been broken. Indeed, several studies suggest that ethical leaders act to protect their followers (Bhal and Dadhich 2011; Kalshoven et al. 2013b), which means that failure to protect followers from harm will be particularly salient.

When social rules are broken it results in dissatisfaction for the target (Henderson and Argyle 1986).This dissatisfaction is likely to be felt especially strongly by followers of ethical leaders, because these leaders create conditions where followers expect social rules to be upheld. As a result, when rules are violated, the negative relationship between ethical leadership and disengagement is likely to be weakened, because rule-breaking behaviour serves to undermine the leader’s influence and ability to prevent disengagement. In this respect, co-worker social undermining and rule breaking more generally may be seen as leadership neutralizers, which counteract the effectiveness of leadership by making it impossible for the leader to exert influence (Kerr and Jermier 1978). We, therefore, pose the following hypotheses:

Based on hypotheses 1 and 2, we posit that when co-worker undermining is high, the mediated relationships between ethical leadership and employee attitudes via disengagement will be weaker.

Data for this study was collected from nurses working in 49 hospitals in Romania. Access to hospitals was obtained through personal contacts. A representative of each hospital was responsible for distributing the questionnaires to nurses. To minimise the likelihood of common method bias, the survey was administered across three time points. At time 1, nurses rated their direct supervisor’s ethical leadership behaviours, as well as the extent to which they had experienced co-worker undermining from their colleagues. Two weeks later, at time 2, they completed questionnaires measuring disengagement. Subsequently, after another two weeks, at time 3, they completed questionnaires on turnover intentions, job satisfaction, and organisational commitment. As noted by Dormann and Griffin (2015), shorter time lags are methodologically advantageous compared to longer ones. The two-week time lag helped to reduce respondent attrition as well as minimise the influence of contaminating factors that may cover or conceal relationships between variables (Kovjanic et al. 2012a, b).

Out of the 540 questionnaires distributed, 511 were returned at Time 1. At Time 2, 484 nurses took part in the survey (89.6% response rate), whereas 460 completed the survey at Time 3 (96% response rate). Most of the nurses in the final sample were female (77.7%), with an average age of 37.4 years old, and were working in their hospital for an average of 8.8 years. Almost half of them (45%) had a higher education degree.

The questionnaire was administered in Romanian. Therefore, all items were initially translated from English into Romanian and then translated back into English by a bilingual researcher (Brislin 1980). The original and translated English versions were then compared by another bilingual academic to ensure equivalence of meaning. A pilot study was then administered on eight nurses to test the questionnaire and no major issues were identified. Responses to all items were on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree).

Ethical leadership. Brown et al.’s (2005) 10-item scale was used to measure ethical leadership. Sample items include “My supervisor disciplines employees who violate ethical standards” and “My supervisor sets an example of how to do things the right way in terms of ethics”. Alpha for the scale was 0.94.

Co-worker undermining. Duffy et al.’s (2002) 13-item scale was used to measure co-worker undermining. Sample items include “Other nurses in the hospital spread rumours about me” and “Other nurses in the hospital talk bad about me behind my back”. Cronbach’s alpha was 0.95.

Disengagement. Four items from the Oldenburg Burnout Inventory (OLBI; Demerouti and Bakker 2008) were used to measure disengagement. The OLBI is composed of two subscales: exhaustion and disengagement. Both subscales include four positively worded items and four negatively worded items. The positively worded items in the disengagement subscale are generally viewed as markers for work engagement (Halbesleben and Demerouti 2005). Accordingly, and since the focus in this study is on disengagement rather than engagement, we used the four negatively worded items. Sample items include “Lately, I tend to think less at work and do my job almost mechanically” and “Sometimes I feel sickened by my work tasks”. Cronbach’s alpha for this scale was 0.77.

Turnover intentions. O’Reilly et al.’s (1991) 4-item scale was used to measure turnover intentions. Sample items include “I would prefer another more ideal job to the one I have now” and “I have seriously thought about leaving this hospital”. Alpha for the scale was 0.93.

Job satisfaction. Seashore et al.’s (1982) 3-item scale was used to measure job satisfaction. Sample items include “In general, I like working here” and “Overall, I am satisfied with my job”. Cronbach’s alpha was 0.92.

Organisational commitment. Six items developed by Allen and Meyer (1990) were used to measure affective organisational commitment. Sample items include “I feel 'emotionally attached' to this hospital” and “This hospital has a great deal of personal meaning for me”. Alpha for the scale was 0.95.

Controls. Transformational leadership and neuroticism were controlled for as previous research suggests that both variables are important predictors of employee burnout and work attitudes (e.g. Judge et al. 2002; Walumbwa and Lawler 2003; Langelaan et al. 2006; Castro et al. 2008; Eckhardt et al. 2016; Hildenbrand et al. 2018). Transformational leadership was measured using four items developed by Podsakoff et al. (1990). These items measured four different facets of transformational leadership: idealised influence, inspirational motivation, individualised consideration, and intellectual stimulation (Bass 1985). Neuroticism was measured using three items from Judge et al.’s (2003) core self-evaluations scale. Cronbach’s alpha for transformational leadership was 0.90 and for neuroticism was 0.81. Nurses’ gender, age, and tenure were not significantly related to disengagement or employee attitudes and were, therefore, not included in the analysis.

A series of confirmatory factor analyses (CFAs) were conducted to assess convergent and discriminant validity. Three indices were used to assess model fit: the comparative fit index (CFI), the Tucker Lewis index (TLI), and the root mean square error of approximation (RMSEA). Good fit is achieved when the CFI and the TLI are 0.90 or above, and when the RMSEA is 0.08 or below (Hu and Bentler 1999; Williams et al. 2009).

Nurses were grouped by hospitals and the intraclass correlation coefficient (ICC) for disengagement, turnover intentions, job satisfaction, and organisational commitment was 0.30, 0.30, 0.20 and 0.29, respectively. This suggests the presence of significant between-group variance. Therefore, the model and proposed hypotheses were tested using generalised multilevel structural equation modeling (GSEM) with STATA. GSEM is a technique that combines the flexibility and power of both Generalized Linear Models and Structural Equation Models in an integrated modeling framework (Lombardi et al. 2017). It simultaneously considers direct and indirect effects of several interacting factors and is ideal for addressing hypotheses with nested data (Preacher et al. 2010; Lombardi et al. 2017). The use of GSEM in this study enabled us account for the clustering or nesting of observations within hospitals.

All relationships were tested simultaneously. Disengagement (i.e. the mediator variable) was regressed on the controls (i.e. transformational leadership and neuroticism), ethical leadership, co-worker undermining and their interaction term (ethical leadership × co-worker undermining). The outcome variables (i.e. turnover intentions, job satisfaction and organisational commitment) were regressed on the control variables, ethical leadership, co-worker undermining, their interaction term and disengagement (Preacher et al. 2007; Hayes 2013). Following the recommendations of Hofmann and Gavin (1998) and Hofmann et al. (2000), all variables were grand mean centred.

Eight latent factors (i.e. ethical leadership, co-worker undermining, disengagement, turnover intentions, job satisfaction, organisational commitment, transformational leadership and neuroticism) with their 47 items were included in the analyses. The hypothesised eight-factor model had an acceptable fit (χ² = 3259.01, df = 1006, p < 0.01; CFI = 0.88, TLI = 0.87, RMSEA = 0.070; Hu and Bentler 1999). All factor loadings were greater than 0.60 and significant at p < 0.01 level, suggesting convergent validity (Anderson and Gerbing 1988).

As shown in Table 1, the fit of the eight-factor model (i.e. the baseline model) was significantly better than other alternate models with fewer factors. In particular, the hypothesised model fitted the data significantly better than a seven-factor model which combined ethical and transformational leadership on one latent factor (Δχ² = 404.32, Δdf = 7, p < 0.01), and a six-factor model which combined job satisfaction, commitment and turnover intentions on one factor (Δχ² = 1760.33, Δdf = 13, p < 0.01). The baseline model was also significantly better than a five-factor model which combined job satisfaction, commitment, turnover intentions, and disengagement on one factor (Δχ² = 2221.36, Δdf = 18, p < 0.01) as well as a four-factor model which combined ethical and transformational leadership on one latent factor, and combined job satisfaction, commitment, turnover intentions and disengagement onto another factor (Δχ² = 2618.32.32, Δdf = 22, p < 0.01).Thus, discriminant validity was evidenced.

Table 2 shows the correlations between variables as well as the means and standard deviations. As shown in the table, ethical leadership was negatively correlated with disengagement (r = − 0.23, p < 0.01), while co-worker undermining was positively correlated with disengagement (r = 0.43, p < 0.01). Disengagement was positively correlated with turnover intentions (r = 0.55, p < 0.01), and negatively correlated with both job satisfaction (r = − 0.44, p < 0.01) and commitment (r = − 0.30, p < 0.01). None of the correlation estimates exceeded 0.80, which suggests that multicollinearity is unlikely (Kline 2005). Nonetheless, consistent with prior research (Mayer et al. 2012; Kalshoven et al. 2013a; Hoch et al. 2018), transformational leadership and ethical leadership were highly correlated (r = 0.79, p < 0.01).Therefore, to provide further evidence on the distinctiveness of both constructs, the square root of the average variance extracted (AVE) of each construct was compared with the correlation between them (Fornell and Larcker 1981). For both constructs, the square root of the AVE was higher than the corresponding interconstruct correlation estimate (0.84 for transformational leadership and 0.80 for ethical leadership). This confirms that both constructs are conceptually distinct from each other.

Table 3 presents the full results of the mediated moderation model. Hypothesis 1 predicted that disengagement mediates the relationship between ethical leadership and employee attitudes. Ethical leadership was negatively related to disengagement (β = − 0.16, p < 0.05). Disengagement was also positively related to turnover intentions (β = 0.37, p < 0.01) and negatively related to job satisfaction (β = − 0.22, p < 0.01) and organisational commitment (β = − 0.35, p < 0.01). Moreover, the indirect effect of ethical leadership on employee attitudes via disengagement was significant. The indirect effect was negative for turnover intentions (β = − 0.06, p < 0.05, 95% CI − 0.11 to − 0.01) and positive for job satisfaction (β = 0.041, p < 0.05, 95% CI 0.003–0.07) and commitment (β = 0.06, p < 0.05, 95% CI 0.007–0.10). Together, these results suggest that disengagement mediates the relationship between ethical leadership and employee attitudes, providing support for Hypothesis 1. However, since the direct relationship between ethical leadership and employee attitudes was significant (as shown in Table 3), this mediation is partial rather than full or complete.

Hypothesis 2 stated that co-worker undermining moderates the relationship between ethical leadership and disengagement. The interaction term of ethical leadership and co-worker undermining was significant and positive (β = 0.11, p < 0.01). Figure 2 shows this interaction’s simple slope plot following Aiken and West’s (1991) approach. The negative relationship between ethical leadership and disengagement was significant when co-worker undermining was low (β = − 0.299, SE = 0.083, t = − 3.60, p < 0.01) and was non-significant when undermining was high (β = − 0.023, SE = 0.069, t = − 0.34, p > 0.10). This provides support for Hypothesis 2.

Hypothesis 3 predicted that co-worker undermining moderates the indirect relationship between ethical leadership and employee attitudes through disengagement. The indirect relationship between ethical le adership and quit intentions via disengagement was significant and negative when undermining was low (β = − 0.11, p < 0.01, 95% CI − 0.18 to − 0.04) but non-significant when undermining was high (β = − 0.009, p ˃ 0.10, 95% CI − 0.059 to 0.041). The indirect relationship between ethical leadership and job satisfaction via disengagement was significant and positive when undermining was low (β = 0.07, p < 0.01, 95% CI 0.02–0.11) but non-significant when undermining was high (β = 0.005, p ˃ 0.10, 95% CI −0.02 to 0.04). Similarly, the indirect relationship between ethical leadership and commitment via disengagement was significant and positive when undermining was low (β = 0.10, p < 0.01, 95% CI 0.04–0.17) but non-significant when undermining was high (β = 0.008, p > 0.10, 95% CI − 0.04 to 0.06). These results provide support for Hypothesis 3.

The results showing a negative relationship between ethical leadership and disengagement suggest that organisations could benefit from employing human resource practices that help stimulate ethical conduct of leaders. For example, integrity tests could be used alongside questions on ethics during the interview process to assess a managerial candidate’s morality and integrity. Organisations could also train leaders on the ethical requirements of their job and how to deal with ethical issues in the workplace (Mostafa 2018). Rewarding leaders for acting proactively with ethical matters and addressing ethical issues could also help further enhance ethical leader behaviours. However, it is also important for organisations to be aware that the occurrence of social undermining within the workplace can undermine the efficacy of ethical leadership. Therefore, employing and training ethical leaders alone will not be effective when toxic cultures exist within organisations, as the benefits of ethical leadership transpire at low levels of co-worker social undermining. As a result, organisations should look to support ethical leaders by building an ethical infrastructure, which can limit social undermining. An ethical infrastructure involves formal ethical systems, such as “codes of ethics, written procedures for handling complaints, formal training programs, the use of formal sanctions against unethical behaviour, recurrent communication of policies, and formal surveillance of the social work environment” (Einarsen et al. 2017, p. 39). However, they also involve informal ethical signals (e.g., conversations, observations, and socialisation) that demonstrate how employees should behave when unethical behaviour could occur (Einarsen et al. 2017).

This study is not without limitations. First, despite the time-lagged nature of the study design, causal inferences cannot be made. For example, it is possible that employees who are not satisfied with their jobs, or attached and committed to their organisations, are more likely to be disengaged and distance themselves cognitively, behaviourally and emotionally from their work. The study’s theoretical rationale and previous empirical research provide support for the proposed causal direction. Nevertheless, future research adopting experimental or longitudinal designs could help better gauge casual effects. The second limitation relates to the use of self-reports, which makes the findings susceptible to single source bias. However, it is important to note that recent research has shown that there is no difference between other and self-ratings of leadership (Lee and Carpenter 2018). In addition, most leadership studies rely on follower reports of leader behaviour (Walsh and Arnold 2018). Self-reports are also usually regarded as the best means for assessing social undermining and individual attitudes (Solberg and Olweus 2003; Hildenbrand et al. 2018; Walsh and Arnold 2018). Importantly, the temporal separation of measurements provides additional confidence in the results. Scholars argue that disengagement is usually associated with counterproductive employee behaviours, such as absenteeism (Carter and Baghurst 2014). Therefore, future research may wish to focus on behavioural rather than attitudinal outcomes of disengagement and collect data from multiple sources, such as supervisory ratings of employee behaviours. This could help further allay common method bias concerns. The final limitation relates to the limited generalisability of the results where all data came from nurses working in Romanian hospitals. Future work in other countries and contexts is needed to establish the generalisability of the study findings.

Drawing on SRT, this study examined the mediating role of disengagement on the relationship between ethical leadership and employee attitudes of job satisfaction, organisational commitment, and turnover intentions. The study also examined the moderating role of co-worker undermining on these mediated relationships. All the study hypotheses were supported. Ethical leadership was indirectly related to employee attitudes via disengagement, and these mediated relationships were conditional on co-worker undermining. These results extend our understanding of ethical leadership and demonstrate that it is less likely to be effective when co-worker undermining occurs in the workplace. Thus, organisations should look to support ethical leaders by building an ethical infrastructure, which can limit social undermining.