The influence of abusive supervision on employee engagement, stress and turnover intention Effect of abusive supervision on employees

Purpose – The purpose of this research was to investigate the influence of abusive supervision on work engagement, stress and turnover intention of subordinates. It was also proposed that work engagement and stress mediate the relationship between abusive supervision and turnover intention. Self-determination theory and the job demands-resources (JD-R) model provide the theoretical framework for the hypotheses. Design/methodology/approach – The sample of this study consisted of 172 employees from public and private companiesoperating in different sectors of the economy.The hypotheses were tested through multiple regression analysis. Findings – The results show that abusive supervision negatively influences engagement and contributes to increasing subordinates ’ stress and turnover intention. The study also found that the relationship between abusive supervision and turnover intention is mediated by engagement, but not by stress. Research limitations/implications – Data was obtained from a convenience sample and cannot, therefore, be generalized. Practical implications – The study results suggest that employees are prone to leave an organization when they are subordinated to abusive leaders, corroborating the idea that workers choose organizations but leave their leaders. This, in turn, reinforces the importance of adopting proper leadership selection and training processes. Originality/value – This research addresses the dark side of the relationship between superiors and subordinates, which has been scarcely examined in the Brazilian literature. It also draws attention to different harmful consequences associated with dysfunctional behaviors of professionals that hold leadership positions in organizations.


Introduction
Leadership has historically attracted the attention of practitioners and researchers in the field of organizational behavior. The vast literature on the topic addresses both leaders' and subordinates' perspectives, as well as the positive or negative consequences of this relationship. In the international arena, evidence has pointed to several negative consequences associated with toxic or abusive leadership (De Hoogh, Den Hartog & Belschak, 2021;Dhanani & LaPalme, 2019;Fischer, Tian, Lee & Hughes, 2021;Tepper, Simon & Park, 2017). A literature review by Fischer et al. (2021) showed that abusive supervision influences subordinates' attitudes, behaviors, relationships and overall well-being.
In the Brazilian context, however, studies focusing on this aspect of the supervisory relationship are still scarce, despite its potentially harmful consequences for organizations  Ceribeli, Fernandes & Saraiva, 2021) in addition to a review of the international literature.
To contribute to filling this gap, the purpose of this study was to analyze the consequences of abusive supervision for subordinates. Specifically, we investigated the relationships between abusive supervision and employee engagement, stress and turnover intention among subordinates. Furthermore, we assessed the mediating role of engagement and stress in the relationship between abusive supervision and turnover intention. To this end, a quantitative survey was conducted with 172 employees of public and private Brazilian companies operating in different sectors of the economy.
Additionally, it seeks to fill a gap in the national literature, which lacks studies on dysfunctional leadership behaviors. From an applied perspective, we expect that the results of our study will promote the debate on abusive supervision and encourage organizations to adopt more careful processes of selection, evaluation and training of professionals holding such positions.

Theoretical framework
The theoretical framework was organized into four blocks that discuss abusive supervision, employee engagement, stress and turnover intention, respectively.

Abusive supervision
While it is crucial to understand effective leadership behaviors, it seems equally necessary to recognize and investigate the behaviors of professionals in leadership positions that harm subordinates and organizations (Aryee, Sun, Chen & Debrah, 2007;Tepper, 2007;Watkins, Fehr & He, 2019). Over time, research has focused on positive and constructive ways (Almeida, 2018;Schyns & Schilling, 2013), whereas topics related to the negative side of leadership have been relatively less explored, despite the gradual growth observed in recent years (Dhanani & LaPalme, 2019;Schilling, 2009;Tepper et al., 2017). Padilla, Hogan and Kaiser (2007) point out the importance of acknowledging that leadership can produce results ranging from constructive to destructive. In studies on the dark side of leadership, the terms that have been used are abusive supervision (Tepper, 2000), as well as destructive leadership behavior (Einarsen, Aasland & Skogstad, 2007), toxic leadership (Lipman-Blumen, 2005) or destructive leadership (Erickson, Shaw, Murray & Branch, 2015).
This study addresses the concept of abusive supervision proposed by Tepper (2000, p. 178) and defined as "subordinates' perceptions of the extent to which supervisors engage in the sustained display of hostile verbal and non-verbal behaviors, excluding physical contact". In this same vein, Almeida (2018, p. 73) defines abusive supervision as "the continued demonstration of hostile and immoral verbal and non-verbal abusive behaviors by leaders aimed at influencing subordinates to pursue personal and/or organizational goals, punishing unwanted behaviors of subordinates to fulfill the leaders' personal projects or leaders' behaviors that disregard the well-being of subordinates".
Research shows that cultural differences can be relevant, as they influence subordinates' perceptions, attitudes and reactions (Li, Sun, Taris, Xing & Peeters, 2021;Mackey, Frieder, Brees & Martinko, 2017). According to Tepper (2007), abusive supervision is more common in countries with high power distance, where the unequal distribution of power is more widely accepted. In turn, Vogel et al. (2015) found that subordinates coming from Anglo-Saxon culture are less likely to accept abusive supervision than those from Asian culture. The metaanalysis by Li et al. (2021) pointed out that the relationship between abusive supervision and employee engagement is moderated by certain dimensions of national culture. In this sense, it is worth pointing out the importance of studies on abusive leadership in different national contexts and cultures.
Two meta-analyses have investigated the consequences of abusive supervision (Mackey et al., 2017;Zhang & Liao, 2015). Zhang and Liao (2015) showed that abusive supervisors negatively affect job satisfaction and organizational commitment and contribute to turnover intention and actual turnover of subordinates. Regarding well-being, the study has shown that abusive leadership contributes to anxiety, stress and emotional exhaustion among subordinates. Mackey et al. (2017) found comparable results, also pointing out differences between samples obtained in the USA compared to other countries.
More recently, a systematic literature review by Fischer et al. (2021) showed that abusive supervision has harmful effects not only on subordinates but also on work teams and the organization. In the case of subordinates, negative outcomes were categorized into affective, relational, motivational, behavioral and cognitive, in addition to those concerning their wellbeing. It is also worth noting the meta-analysis by Schyns and Schilling (2013), which included studies on different forms of destructive leadership, and identified harmful effects on subordinates and organizations.
Thus, based on the evidence found in the literature, it is possible to relate abusive supervision to employee engagement, stress and turnover intention of subordinates. These topics are discussed below.

Employee engagement
The concept of engagement was first proposed by Kahn (1990, p. 694), according to whom "in engagement, people employ and express themselves physically, cognitively, and emotionally during role performances". For Bakker et al. (2008), engagement is characterized by three psychological states, namely vigor, dedication and absorption. Vigor involves elevated levels of energy and mental resilience at work, dedication implies enthusiasm, inspiration and pride in one's work, whereas absorption is associated with a state of full concentration and immersion in the activity. Macey and Schneider (2008) point out that work engagement is different from job satisfaction or organizational commitment. Christian, Garza and Slaughter (2011) also suggest that, although there is a relationship between work engagement, job satisfaction, task involvement and organizational commitment, each construct refers to different aspects and, therefore, work engagement offers a relevant contribution to the organizational behavior literature.
Based on self-determination theory, we propose that abusive supervisors negatively affect the engagement of their subordinates. According to this theory, people are endowed with the psychological needs for autonomy, competence and relationships, and the satisfaction of these needs would have a positive impact on their well-being (Ryan & Deci, 2000.

Effect of abusive supervision on employees
Similarly, the lack of these elements would have deleterious effects, as pointed out by Ronena and Donia (2020). To the extent that abusive supervisors establish conflictive relationships with their subordinates, limiting their autonomy and compromising their sense of competence, work engagement will also tend to be negatively affected. Thus, the following hypothesis is proposed: H1. Supervisors perceived as abusive negatively influence the work engagement of their subordinates.

Stress
Stress has attracted the attention of researchers from different fields of knowledge (Sonnentag & Frese, 2013) and can be defined as the process of assessing and responding to a potentially threatening or challenging event. Therefore, any situation that threatens an individual's well-being and requires some form of coping may cause stress (Myers & Dewall, 2019). One of the most influential theoretical approaches was proposed by Lazarus in the 1960s and later expanded in the 1980s by Lazarus and Folkman (Bliese, Edwards & Sonnentag, 2017). According to this approach, the experience of stress depends on the individual's appraisal of the situation, which is a two-stage process. In the primary appraisal stage, the individual determines whether the event is harmful, threatening or challenging, whereas in the secondary stage he/she considers how to react to the situation (Bliese et al., 2017).
Regarding the factors that cause stress in the organizational setting, Sonnentag and Frese (2013) categorized them into physical, role and social stressorsincluding relationship problems with supervisors and coworkers -, as well as related to the task, work schedule, career and change. Abusive supervisors can be sources of stress to the extent that they pose a threat to the employees' integrity (Harms, Cred e, Tynan, Leon & Jeung, 2017;Schyns & Schilling, 2013;Tepper, 2007;Zhang & Liao, 2015).
According to the job demands-resources (JD-R) model (Bakker & Demerouti, 2017;Bakker, Demerouti & Sanz-Vergel, 2014), leaders can represent both a resource, when they offer adequate support to subordinates (Schaufeli, 2015), and a demand, when their attitudes and behaviors drain the energy of subordinates (Harms et al., 2017;Scheuer et al., 2016). In this regard, Vogel and Bolino (2020) point out that abusive supervision can have long-term deleterious effects on subordinates, even after the supervisory relationship has ended.
In line with the JD-R model (Bakker & Demerouti, 2017;Bakker et al., 2014), we argue that abusive supervisors contribute to the stress of their subordinates, since being part of a conflicted relationship represents a demand that requires constant cognitive and emotional efforts. Thus the following hypothesis is proposed: H2. Supervisors perceived as abusive positively influence the stress of their subordinates.

Turnover intention
Voluntary turnover generates significant costs for organizations, including time and resources for hiring and training new employees, as well as expenses associated with the interruption of services and even the loss of customers. Many of these costs are significant and difficult to measure, so problems associated with voluntary turnover are often underestimated (Hancock et al., 2011;Richer, Blanchard & Vallerand, 2002).
Regarding the antecedents of voluntary turnover, Griffeth, Hom and Gaertner (2000) identified the influence of employee's individual characteristics, situational factors and the macro-environment, which includes labor market conditions. Among the situational factors identified in the literature, the relationship between supervisors and subordinates can be highlighted. A survey conducted in a Hong Kong hotel chain has shown that turnover intention is related to the quality of the relationship between leaders and subordinates (Kim, Lee & Carlson, 2010). Aquino, Griffeth, Allen and Hom (1997) identified a negative relationship between satisfaction with supervisors and turnover intention. Finally, evidence shows that abusive supervisors can drive their subordinates to leave the organization (Lyu, Ji, Zheng, Yu & Fan, 2019;Moin, Wei, Khan, Ali & Chang, 2021;Pradhan & Jena, 2018;Tepper, 2000). Given this evidence, we propose the following hypothesis: H3. Supervisors perceived as abusive positively influence their subordinates' turnover intention.
Furthermore, based on the evidence of the close negative relationship between work engagement and turnover intention (Rubenstein, Eberly, Lee & Mitchell, 2018) and between stress and turnover intention (Dai, Zhuang & Huan, 2019;Rubenstein et al., 2018), we propose that the influence of abusive supervision on turnover intention is mediated by engagement and stress. Research conducted by Pradhan and Jena (2018) showed that abusive leaders influence turnover intention, and that this relationship is mediated by stress. Thus the following hypothesis is proposed: H4. The positive relationship between abusive supervision and turnover intention is mediated by (a) work engagement and (b) stress. Figure 1 presents the hypotheses proposed in this study.

Method
To achieve the proposed objectives, a quantitative study was conducted with Brazilian workers employed in public or private companies from different economic sectors. The link to the questionnaire, developed on the Qualtrics platform, was shared on the researchers' social media. The questionnaire contained a consent form informing about the confidentiality of the collected data, as well as the participants' anonymity.
Out of the 303 participants who accessed the questionnaire, 202 (67% on total) were eligible to participate. The remaining 101 were not considered because they were either unemployed, self-employed or owned a business, and therefore were not subject to a direct supervisory relationship. Out of the 202 eligible participants, 30 were excluded for not completing the survey, resulting in a final sample of 172 participants.
The scales that composed the questionnaire are detailed below.

Effect of abusive supervision on employees
Abusive supervision: Since there was no previously validated scale in Portuguese, we chose to adopt the abusive supervision scale proposed by Tepper (2000). A back-translation procedure was performed, in which the scale was translated into Portuguese by two professionals who are fluent in both languages, and later retranslated into English by two other professionals with the same qualifications. The complete analysis of the material resulted in the final scale used in the study. It consists of 15 items, assessed through a fivepoint scale ranging from 1 (he/she never behaves this way) to 5 (he/she very often behaves this way).
Employee engagement: We used the Utrecht Work Engagement Scale, developed by Schaufeli and Bakker (2003), and consisting of nine items assessed through a seven-point scale, ranging from 0 (never) to 6 (every day). The scale measures three aspects of engagement, namely vigor, absorption and dedication. For the purposes of this study, we used the Brazilian version validated by Vazquez, Magnan, Pacico, Hutz and Schaufeli (2015).
Stress: To measure stress levels, we adopted the scale developed by Anderson, Coffey and Byerley (2002), translated to Portuguese by Oliveira, Cavazotte and Paciello (2013). This scale consists of seven items, measured through a five-point scale, ranging from 1 (never) to 5 (often).
Turnover intention: We used the scale proposed by Jensen, Patel and Messersmith (2013), consisting of four items measured through a seven-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree).
The proposed hypotheses were tested through multiple regression analysis, a multivariate technique that measures the relationship between a dependent variable and other independent variables, and also allows the test of mediations and moderations (Tabachnick & Fidell, 2019). The analyses were performed with the support of the software SPSS (Statistical Package for the Social Sciences) and the PROCESS tool was used to test the mediations (Hayes, 2018). Table 1 presents the descriptive statistics and correlations between variables. Abusive supervision is negatively correlated with engagement and positively correlated with stress and turnover intention, in line with the study hypotheses. Furthermore, abusive supervision was significantly correlated with the control variables gender, age, level of education, tenure and type of company. These results suggest that female, younger and less educated participants perceive their leaders as more abusive, when compared to male, older and more educated ones. Additionally, participants with fewer years of tenure and who work for private organizations also perceive their leaders as more abusive, compared to the ones with longer tenure and who work for public organizations.

Results
Regarding the sample profile, out of 172 participants, 55% were men, aged between 20 and 65 years (mean age 39.4 years). Of this total, 75% have a graduate degree, and most worked in public companies (59%). Regarding tenure, 51% have worked in the same organization for 11 years or more.
After the descriptive analyses, the data was submitted to factor analysis through the principal components method and Varimax rotation. The purpose was to evaluate the behavior of the four scales used in the study: abusive supervision, engagement, stress and turnover intention. The first round of analysis resulted in five factors, and pointed to problems in the abusive supervision scale. After three items were removed, a new factor analysis resulted in four factors, as expected. Regarding the reliability of the scales, Cronbach's alphas   (Tabachnick & Fidell, 2019). As data was obtained from the same source and at the same point in time, we conducted Harman's one-factor test (Podsakoff & Organ, 1986). A principal components factor analysis including all items in the survey resulted in four factors, with the first accounting for less than 50% of the total variance. Therefore, common method bias did not seem to be a problem in this case.
Multiple linear regression analyses were conducted to test the hypotheses. Table 2 presents the results of the relationship between abusive supervision and employee engagement. According to Model 1, with only control variables, age was significant, suggesting that older workers are more engaged than younger ones (β 5 0.031, p < 0.05). In Model 2, which included abusive supervision, this variable was statistically significant, indicating that leaders perceived as abusive negatively affect their subordinates' engagement (β 5 À0.698, p < 0.001), thus corroborating H1. Table 3 presents the results of the relationship between abusive supervision and stress. Model 1, with only control variables, was not significant. In Model 2, which included abusive supervision, this variable was statistically significant, indicating that leaders perceived as abusive contribute to increased levels of stress of their subordinates (β 5 0.374, p < 0.001), thus corroborating H2. In addition, education level was also significant (β 5 0.153, p < 0.05), suggesting that more skilled workers experience higher levels of stress than less skilled ones. Table 4 presents the results of the relationship between abusive supervision and turnover intention. Model 1, with only control variables, was statistically significant and indicates that workers who have been working for longer in the company are less likely to leave their organizations. In Model 2, which included abusive supervision, this variable was statistically significant, indicating that leaders perceived as abusive positively affect their subordinates' turnover intention (β 5 0.862, p < 0.001), thus corroborating H3. In addition, tenure (β 5 À0.054, p < 0.05) and type of company (β 5 0.645, p < 0.05) were also significant, showing that employees with more years of tenure and those working in public companies are less likely to leave their organizations.
Finally, we used the PROCESS tool, developed by Hayes (2018), to test the mediating effects of engagement (H4a) and stress (H4b) on the relationship between abusive supervision and turnover intention. The results presented in Table 5 show that the effect of abusive leadership on turnover intention is mediated by employee engagement, but not by stress, since the value zero is part of the confidence interval for this variable.
In other words, these results indicate that abusive leaders contribute to subordinates' turnover intention and that this influence predominantly happens through a lower degree of engagement, but not higher stress. Therefore, all the proposed hypotheses were confirmed for this sample, except H4b. 0.158*** 0.158*** Note(s): *p < 0.05; **p < 0.01; ***p < 0.001 Table 2. Regression analysisemployee engagement

Discussion and final remarks
This study aimed to shed light on the negative side of leadership, with a focus on the consequences of abusive supervision on subordinates. Therefore, we sought to contribute to the literature on a relatively scarce research topic, particularly in the Brazilian context. The literature review has shown that few studies have been published in the country and, among these, only one addressed the consequences of abusive supervision (Almeida, 2018). Furthermore, considering the possible cultural influences on this phenomenon (Li et al., 2021;Mackey et al., 2017;Tepper, 2007;Vogel et al., 2015), this study also contributes to the literature by revealing the perception of Brazilian workers on these issues.
Hypothesis H1, which proposed that leaders perceived as abusive negatively influence the engagement of their subordinates, was confirmed, corroborating international evidence on the subject (Li et al., 2021;Scheuer et al., 2016;Wang et al., 2020;Zhang & Liao, 2015). Selfdetermination theory, according to which motivation results from the need for autonomy, competence and caring and reciprocal relationships (Ryan & Deci, 2000), seems to explain this  Table 3. Regression analysisstress Table 4. Regression analysisturnover intention  Ronena and Donia (2020). By subjecting subordinates to a conflictual relationship, with little freedom and autonomy, and which tends to adversely affect their sense of competence, abusive supervisors end up reducing their subordinates' motivation and, consequently, their work engagement. This result brings an important contribution to the literature on leadership, given the scarcity of studies on the relationship between leader behavior and employee engagement (Carasco-Saul et al., 2015;Xu & Thomas, 2011). Hypothesis H2, which proposed that leaders perceived as abusive trigger stress in their subordinates, was also confirmed, in line with evidence found in the international literature on the topic (Harms et al., 2017;Scheuer et al., 2016;Schyns & Schilling, 2013;Tepper, 2007;Zhang & Liao, 2015). In Brazil, Almeida's (2018) research with employees of a financial institution pointed to a positive relationship between abusive supervision and absenteeism due to illness, corroborating the damaging effects of abusive leaders on their subordinates. The JD-R model (Bakker & Demerouti, 2017;Bakker et al., 2014) provides the theoretical ground for this phenomenon. In other words, relationships with abusive leaders seem to demand cognitive and emotional efforts from subordinates, contributing to increasing their stress and their risk of falling ill. Considering the negative effects of stress on employees' wellbeing and performance, this result represents yet further evidence of the importance of those who occupy leadership positions in organizations.
A positive and significant relationship between education levels and stress (β 5 0.153, p < 0.05) was also identified, indicating that more qualified workers experience higher levels of stress. This result suggests that more qualified professionals may be dealing with higher levels of stress due to their greater degree of responsibility and the constant challenges they face in their activity, compared to less-qualified employees. In this sense, this finding is a warning call for human resource managers since stress can impair these workers' performance and decision-making ability.
Regarding turnover intention, this study found that abusive supervision positively influences the subordinates' desire to leave the organization (H3). This result corroborates other evidence found in the literature (Lyu et al., 2019;Moin et al., 2021;Pradhan & Jena, 2018;Reina et al., 2018;Tepper, 2000) and is also in line with the idea that professionals choose organizations but leave their leaders, which is often highlighted in the management literature. In this sense, in addition to its negative effect on subordinates, abusive supervision also harms the organization, considering the loss of knowledge stemming from employees who choose to leave, and to operational problems and financial costs associated with hiring replacements (Hancock et al., 2011).
The variables tenure and type of company also influenced turnover intention, so that longtenured employees and those working in public companies are less likely to leave their organizations. Indeed, the negative relationship between tenure and turnover intention finds ample support in the literature (Griffeth et al., 2000;Rubenstein et al., 2018). The lower propensity of public employees to leave their organizations when compared to private ones was also expected, if we consider that the search for stability is usually a crucial factor in the choice for public organizations (Silva, Balassiano & Silva, 2014). Finally, the hypothesis that the relationship between abusive supervision and turnover intention is mediated by engagement (H4a) was also confirmed, although it is not mediated by stress (H4b), contrary to evidence found by Pradhan and Jena (2018).
In summary, the results of our study show that abusive leaders reduce employees' engagement and increase their stress and turnover intention. They also show that work engagement mediates the influence of abusive supervision on turnover intention. Therefore, if organizations are to minimize the risks of losing employees, they should invest in actions to select and train leaders, as well as in promoting an environment that strengthens values such as collaboration, consistency, transparency, respect and ethics (Almeida & Porto, 2019;Erickson et al., 2015). Almeida and Porto (2019), for example, identified a negative correlation between abusive leadership and the organization's ethical climate.
Furthermore, given the difficulty of eliminating the problem, organizations should create support mechanisms for subordinates, including emotional support as well as the creation of communication channels that reassure workers that choose to expose their abusive leaders (Lipman-Blumen, 2005). Additionally, training programs can also be designed to develop skills that enable subordinates to deal with the pressures from their leaders.
This research counted on the participation of a group of employees from public and private companies operating in different sectors of the economy. The findings reveal that women and employees of private companies are more subject to abusive leadershipor perceive it more intenselythan men and employees of public organizations. In this sense, future studies could investigate these issues more thoroughly. Additionally, since the study involved a convenience sample and cannot be generalized, we suggest its replication with larger and more representative samples. Furthermore, it is worthwhile mentioning that all variables were obtained from the participants and at a single point in time. Although tests indicated that common method bias was not an issue, we suggest more sophisticated methodological approaches, including the collection of data at different points in time.
Note 1. The search was performed in July 2021 in the Spell and Scielo databases, with the following terms in Portuguese and English: "abusive supervision", "abusive leadership", "toxic leadership" and "destructive leadership". On the same date, a search with these expressions in Portuguese was performed in Google Scholar, which allowed us to find a doctoral thesis on the subject (Almeida, 2018). Works in Portuguese language whose field research had been conducted in other countries, most notably Portugal, were not considered.