Master Thesis
The relationship between abusive leadership, stress and
commitment, moderated by self-efficacy and stress
mindset
Author: Manouk Rombout Student number: 11424109 Date of submission: 25/01/2018 Final version
University of Amsterdam, Faculty of Economics and Business MSc Business Administration – Leadership and Management Thesis supervisor: Dhr. Dr. M. Venus
2
Statement of Originality
This document is written by Manouk Rombout who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Table of content
Statement of Originality ... 2 Abstract ... 4 1. Introduction ... 5 2. Literature review ... 9 2.1 Stress ... 9 2.2 Abusive leadership ... 11 2.3 Organizational commitment ... 132.4 The moderating variable: Self-efficacy ... 15
2.5 The moderating variable: Stress mindset ... 17
2.6 Conceptual model ... 19 3. Methodology ... 21 3.1 Procedure ... 21 3.2 Sample ... 22 3.3 Measures ... 22 3.3.1 Abusive leadership ... 23 3.3.2 Stress ... 23 3.3.3 Organizational commitment ... 23 3.3.4 Self-efficacy ... 24 3.3.5 Stress mindset ... 24 3.3.6 Control variables ... 24 4. Results ... 25 4.1 Correlation analysis ... 25 4.2 Direct effects ... 28 4.3 Mediating effect ... 29 4.4 Moderation effect ... 30 4.5 Moderated mediation ... 32 4.6 Exploratory research... 32 5. Discussion ... 34 5.1 Summary ... 34 5.2 Theoretical implications ... 34
5.3 Limitations and future research ... 39
5.4 Practical implications ... 43
5.5 Conclusion ... 44
4
Abstract
Stress is a concept which is greatly known in organizations. Employees can feel a great deal of stress and this has consequences for the organization, for example low commitment of the employees. The leader plays an important part in the feeling of stress. This study investigates the relationships between abusive leadership and stress, stress and commitment and stress as a mediator between abusive leadership and commitment. Furthermore, self-efficacy is studied as a moderator on the relationship between abusive leadership and stress, and stress mindset is studied as a moderator on the relationship between stress and commitment. Data has been gathered through a survey. In total, 146 Dutch employees filled in the online questionnaire. This study found support for a mediating effect of stress between abusive leadership and commitment. However, no support was found for both of the moderating effects. The results contribute to the understanding of the relationship between abusive leadership, stress and commitment. The theoretical and practical implications are discussed as well as the limitations of this research. Moreover, directions for future research are given.
5
1. Introduction
The number one cause of work-related illness is stress: 36% of the absenteeism can be explained by stress and approximately one million Dutch individuals experience burnout complaints (Rijksoverheid, 2015). The absenteeism caused by stress costs the Dutch employers 1,8 billion yearly (TNO, 2014). These numbers show that stress is a subject that needs serious attention in the work field. To make managers and employees aware of the risks of work stress, companies in The Netherlands are campaigning ‘Check your work stress’. With this campaign, they hope to give a better insight in what causes stress and that there is not one solution to diminish work stress (Arboportaal, 2017).
Stress is generally seen as an unpleasant emotional experience and is associated with consequences such as irritation, annoyance, fear, anxiety, dread, grief, sadness, and depression (Motowidlo, Packard & Manning, 1986). The subject of stress has already been researched greatly and a lot of research has been done on stress in relation to leadership (Harms, Credé, Tynan, Leon, & Jeung, 2016). The meta-analysis of Harms et al. (2016) supports that a relationship between leadership style and stress exists. They found that the more transformational a leader is, the less stress an employee will experience, since transformational leaders are characterized by a positive outlook, a compelling vision, and show support towards their followers.
However, according to Kelloway, Sivanathan, Francis & Barling (2005) leadership is undoubtedly one of the most potential stressors at work when a leader is seen as a ‘poor leader’. The meta-analysis of Harms et al. (2016) looked at the poor leadership style ‘abusive leadership’ and explained that the more abusive a leader is, the more likely the employee will experience stress due to hostile behaviour – verbally or non-verbally (Tepper, 2000). In turn, stress can lead to undesired outcomes for an organization, for example low commitment of the employees (Montgomery et al. 1996; Chen, Silverthorne and Hung, 2006). The relationship between stress and commitment has received partial support, which is why it is important to research this relationship further. This research will focus on stress as a mediator between abusive leadership and commitment. It is expected that abusive leadership will lead to more stress, which will lead to a lower commitment and that stress can explain the relationship between abusive leadership and commitment.
6 The meta-analysis of Harms et al. (2016) shows the importance of the research on leadership and stress. They state that the leader plays a crucial role in the wellbeing of employees and therefore it is important that the relationship between abusive leadership and stress is researched more into depth. An interesting area of concern which deserves more attention is the role of moderators on the relationship between leadership and stress (Harms et al., 2016), since it could explain why individuals might experience different levels of stress when they are exposed to an abusive leader. According to Harms et al. (2016) traits such as secure attachment, hardiness, self-efficacy and resilience have been suggested as potential moderators of stress in the workplace, but the evidence for this is limited. The lack of knowledge about potential moderators can form a problem for companies when they try to understand why some employees feel more stressed than others when their leader is abusive. The aim of this research is to create a better understanding of the relationship between abusive leadership, stress and commitment and therefore, one of the gaps in the leadership and stress literature that will be researched further in this paper is the moderating effect of self-efficacy, which is the belief of an individual whether he or she can successfully execute a task (Bandura, 1977).
The reason this paper will focus on self-efficacy, is that within the organizational sciences, studies on self-beliefs and self-perceptions showed support that stressors are less detrimental when individuals are more positive about the self-beliefs (Jex and Bliese, 1999). Jex and Bliese (1999) and Jex, Bliese, Buzzell and Primeau (2001) researched self-efficacy as a moderator in the stressor-strain relationship. They expected that individuals with a higher level of self-efficacy are better able to deal with stressors, since they are likely to do something about the stressors (Jex and Bliese, 1999). However, there has been done little research on self-efficacy as a moderator when leadership forms the stressor in the stressor-strain relationship. Considering the lack of knowledge about the effect of self-efficacy on abusive leadership and stress, there is a need to research this concept more into depth. The expectation is that abusive leadership leads to stress, but when an individual has a high level of self-efficacy, the employee will experience less stress compared to an individual with a low level of self-efficacy, even when they are exposed to an abusive leader.
Besides self-efficacy, another gap in the literature on stress that will be researched is the moderating effect of stress-mindset, which is a rather new concept. The way an individual reacts on
7 stress can differ a lot between people, which depends on the stress mindset of the individual (Crum, Salovay and Achor, 2013). Here, the distinction can be made between a stress-is-enhancing mindset and a stress-is-debilitating mindset (Crum et al., 2013). The first mindset is the feeling that stress is positive and needed to attain goals and success, whereas the latter sees stress resulting in negative consequences; for example, a lower commitment to the organization.
In this paper, the expectation is that there exists a negative relationship between stress and commitment. The more stress one experiences, the less likely the individual will feel committed to the organization, due to the negative experience with the organization (Tepper, 2000). Stress mindset is expected to form a moderator which influences the relationship between stress and the stress response. The reasoning is that when an employee entails a stress-is-enhancing mindset, the stress an employee experiences will lead to positive outcomes such as more commitment to the organization, and when someone has a debilitating mindset, then stress will lead to a lower commitment of the employee. This is based on the literature of challenge and hindrance stressors, where the enhancing mindset is based on challenge stressors which are seen as positive as they can lead to personal development, achievement and higher commitment, whereas the debilitating mindset is based on hindrance stressors which are seen as negative since it can hinder development and lead to lower commitment (Podsakoff, LePine and LePine, 2007).
Thus, the relationship between abusive leadership, stress and commitment will be examined further. The following research question will be answered: what is the relationship between abusive
leadership, stress and commitment and what are the effects of self-efficacy and stress mindset on this relation?
This study will create a better understanding of the relationship between abusive leadership, stress and commitment. Furthermore, the study contributes to the literature in two ways. First, the focus on self-efficacy expands the literature on self-belief and the stressor-strain literature. Self-efficacy has been researched as a moderator with different stressors and strains, but the literature falls short on self-efficacy as a moderator when leadership forms the stressors. The focus on self-self-efficacy could explain why some individuals experience more stress than others when they are inferior to an abusive leader.
8 Second, the findings on the moderating role of stress mindset on stress and commitment, contribute to the literature since it is a relatively new concept, certainly in the leadership literature. With a closer focus on stress mindset, this research could contribute to the explanation why some individuals don’t experience a negative stress response following from the stress caused by abusive leadership, but use the stress to their advantage to attain goals and to eventually feel more committed to the organization than individuals who have a debilitating stress mindset.
In the next chapter, the literature review will give the reader a better understanding of the concepts and it elaborates on what is known about each concept in the literature. In this part, expected relationships between the concepts are explained. Then, in the method section the procedure of this research and which sample and measures have been used are explained. The findings of the gathered data will be presented in the result section. In the discussion, the implications and limitations of this research will be discussed and ideas for further research are given. The paper ends with a conclusion.
9
2. Literature review
2.1 Stress
Stress is a concept that is, as reflected by researchers’ beliefs, a major factor that affects people’s lives and is linked with mental and physical health (Hobfoll, 1989). It is a concept that has been researched greatly. Stress can be described as a transactional process that occurs when an individual perceives an event relevant to well-being, and this event has the potential for harm or loss, which requires psychological, physiological or behavioural efforts to manage the event and its outcomes (Lazarus & Folkman, 1984). Thus, stress is an individual’s psychological or/and physiological response to a situation where something is at stake for the individual and that situation exceeds the individual’s resources or capacity (Hobfoll, 1989; Le Pine, LePine & Jackson, 2004).
Stress is generally assumed to be bad and therefore organizations need to find ways to reduce or even prevent feelings of stress (Boswell, Olson-Buchanan & LePine, 2004). Montgomery et al. (1996) argue that job stress is dysfunctional and decreases the commitment and productivity of the employee (as cited in Chen et al., 2006).
However, stress can sometimes have desirable consequences (LePine et al., 2004). Cavanaugh, Boswell, Roehling and Boudreau (2000) and LePine et al. (2004) make the distinction between feelings of stress associated with challenging or rewarding job experience and feelings of stress associated with constraining or hindering job experiences. Cavanaugh et al. (2000) define challenge stressors as “work-related demands or circumstances that, although potentially stressful, have associated potential gains for individuals” (p.68). Hindrance stressors were defined as “work-related demands or circumstances that tend to constrain or interfere with an individual's work achievement and that do not tend to be associated with potential gains for the individual” (p.68). This division between the two types of stressors illustrates that stress isn’t necessarily harmful for an organization. However, the stress level must not become too high so that it affects the individual’s mental health or even physical health (Harms et al., 2016). An example of a negative effect of stress is burnout, which is a psychological response to chronic work stress were emotional exhaustion, reduced personal accomplishment and depersonalization are combined (De Hoogh and Den Hartog, 2009).
10 As Cavanaugh et al. (2000) suggested, it depends on the stressors whether an individual will experience the stress to be challenging or to be hindering. A stressor is a stimulus that leads to emotional upset, psychological and physical stress (Hobfoll, 1989). In the literature, a lot of research has been done on different type of stressors (Cavanaugh et al, 2000; Gilboa, Shirom, Fried & Cooper, 2008; Jex & Bliese, 1999). Gilboa et al. (2008) describe stressors such as: role ambiguity, role conflict, role overload, work-family conflict, job insecurity, environmental uncertainty and situational constraints. Other stressors can be work hours and task significance (Jex & Bliese, 1999). Harms et al. (2016) make a distinction between two types of stressors. On the one hand, stress can take the form of job stress, which is caused by the nature of the task. Here, stressors are for example complexity or difficulty of the task (Fiedler, 1992). Other job stress related stressors are time pressure and working conditions. On the other hand, stress can take the form of interpersonal stress, which is caused by being in conflict with others or the feeling that one must meet the demands of others, for example the leader (Harms et al., 2016).
According to a sample of 5000 managers, the most common sources of stress are inadequate support by supervisors and ineffective performance by supervisors (Pearse, 1977, as cited in Schuler, 1982). Hogan and Kaiser (2005) found that many employees think of their supervisors as the worst aspect of their jobs and the relationships between employees and supervisors are often seen as stressors for individuals in organizations (Schuler, 1982). Furthermore, the style of the leader can lead to whether an individual experiences stress or not (Lyons & Schneider, 2009). This shows the important role of the leader during stressful situations and that is why this research will focus more on the interpersonal stress, caused by the leader.
Hence, when the concept of stress is being researched, it is almost impossible to separate leadership and stress from each other. It has been implicated that stress is a determinant of leadership functioning, but that the leaders’ behaviours also can be a determinant of the stress level among the followers (Harms et al., 2016). For this research the focus will be on the effect of leadership style on stress; thus, leadership as a determinant or cause of the stress level. However, there are different leadership behaviours which can lead to different stress levels. In this paper, a leadership style expecting to cause a higher stress level among the employees will be researched: abusive leadership. In the next paragraph, abusive leadership and its expected relationship to stress will be further explained.
11 2.2 Abusive leadership
In the literature on leadership, several leadership styles have been discussed. In this paper, the term leader refers to individuals who have assumed a formal leadership role, such as managers or supervisors. One of the most researched leadership styles is transformational leadership (Judge and Bono, 2000; Judge and Piccolo, 2004; Harms et al., 2016). Transformational leadership can be defined as a leadership style that creates valuable and positive changes among its followers (Chou, Lin, Chang & Chuang, 2013). Also, the transformational leader focuses mostly on short-term goals and on intrinsic needs of the subordinates (Judge and Piccolo, 2004). Harms et al. (2016) found that transformational leadership is negatively related to stress. The more transformational a leader is, the less stress the subordinate will experience, since transformational leaders are able to project a positive outlook and vision. This could lead to more reassurance during a stressful period. Other leadership styles are for example charismatic leadership, transactional leadership and laissez-fair leadership (Judge & Piccolo, 2004). Charismatic leaders are seen by their followers as individuals with extraordinary qualities (Weber, 1947; Yukl, 1999) and this theory has a lot in common with transformational leadership. Transactional leadership is focused on the exchange of resources between leader and employee (Kuhnert and Lewis, 1987). Laissez-fair leadership is seen as a non-leadership, or the absence of leadership (Judge and Piccolo, 2004). Furthermore, some leaders are considered to have ‘dark traits’, for example narcissistic or Machiavellian leaders (Judge, Piccolo & Kosalka, 2009).
The meta-analysis of Harms et al. (2016) researched stress in combination with transformational leadership, leader-member exchange, and abusive leadership. Abusive leadership – in contrast to transformational leadership and leader-member exchange – is a leadership style which is expected to have a negative influence upon stress. The meta-analysis of Harms et al. (2016) showed support for a relationship between abusive leadership and stress, but the effect of moderators on the relationship deserves more attention. Therefore, this research will focus on the relationship between abusive leadership and stress and the effect of a possible moderator.
Abusive supervision is “the sustained display of hostile verbal and nonverbal behaviours, excluding physical contact” (Tepper, 2000, p.179). It occurs when individuals who are in a formal leadership role, behave aggressively toward their employee (Kelloway et al., 2005). Examples of
12 abusive leadership are: shouting to the employees, using threats such as losing the job, purposely withholding information, criticizing employees publicly or neglect employees by giving them the silent treatment (Tepper, 2000, as cited in Aryee, Sun, Chen & Debrah, 2008). The assessment of abusive supervision is subjective, since one individual could view a leader as abusive in one context, but non-abusive in another context and the experience might differ between individuals. Overcoming abuse often is a difficult situation, since the target of abuse feels powerless or is economically dependent on the abuser. Also, abusive individuals often fail to take responsibility for their abusive behaviour and therefore few try to modify it (Tepper, 2000).
Abusive leadership can lead to frustration, helplessness and alienation from work among employees (Tepper, 2000). Zhang and Liao (2015) did a meta-analysis of the consequences of abusive supervision and found that abusive supervision is related to employees’ well-beings, attitudes, workplace behaviours, organizational justice perceptions, performance, and family-related outcomes. Moreover, abusive supervision can lead to low job satisfaction and organizational commitment (Tepper, 2000). Support was found for the relationship between abusive leadership and psychological distress and emotional exhaustion (Harvey, Stoner, Hochwarter & Kacmar, 2007). In this case, abusive supervision is a stressor causing an unpleasant feeling of stress amongst employees. This feeling of stress is caused by the interpersonal conflict and continuous assault on the feelings and self-esteem of the employee (Harvey et al., 2007).
A good leader shows social support; he or she gives the employee the ability to rely on social resources. This will help the individual cope with stressful events (Kelloway et al., 2005). Also, a good leader shows instrumental support (task-specific help), emotional support (affect, empathy and comfort), informational support (the leader gives advice and shows awareness) and appraisal support (suggestions, feedback and encouragement) (Kelloway et al., 2005). An abusive leader often fails to show social support, but also fails to help an overworked employee by reducing the task load, which is the lack of instrumental support. Furthermore, abusive leaders lack needed empathy (emotional support), don’t provide the employee with necessary directives (informational support), and fail to engage and motivate employees (appraisal support) (Kelloway et al., 2005). This shows that abusive leaders don’t take the required action whenever an employee experiences stressful situations.
13 The meta-analysis of Harms et al. (2016) provides the reader with an overview of all the research that has been done regarding abusive leadership and stress. The authors show that higher levels of abusive leadership are related to higher levels of subordinate stress. Based on this information, the first hypothesis can be formulated. The conceptual model can be found at the end of this chapter (figure 1).
Hypothesis 1: Abusive leadership is positively related to employee stress.
In this first hypothesis, stress is a consequence of abusive leadership. However, these stressful situations could turn into causes leading to negative consequences for the individual or for the organization. A negative consequence for the individual could be burnout. Consequences for the organization could be low job satisfaction, low performance (Schuler, 1982) and low organizational commitment (Tepper, 2000), which is the lack of the individual’s involvement in the organization (Steers, 1977). However, the relationship between stress and commitment has received partial support. Therefore, the mechanism of organizational commitment and its relation to stress and abusive leadership will be elaborated on further in the next paragraph.
2.3 Organizational commitment
Organizational commitment can be defined as the individual’s involvement in an organization (Steers, 1977). Three factors characterize commitment: an individual’s beliefs in the goals and values of the organization, a willingness to put effort into the work, and a desire to maintain the job in the organization (Steers, 1977). Commitment can be further conceptualized in three themes: affective, normative and continuance commitment (Meyer & Allen, 1991). Affective commitment is the emotional attachment towards the organization. The individual feels an affective and emotional identification with the organization. Normative commitment occurs when the employee feels obligated to remain within the organization. In this situation, employees believe it is the right thing to do, even if it gives them no satisfaction. The last form of commitment, continuance commitment, is based on the costs of leaving the organization. For example, an employee won’t leave the organization, simply because then he will not have a job anymore and therefore not have an income.
14 Before the indirect effect of stress will be discussed, the direct relationship between abusive leadership and commitment will be described. In the literature it is stated that when an individual is exposed to an abusive leader, he or she will experience less organizational commitment, since they don’t belief in the goals of the leader, have low willingness to put in more effort and are not motivated to maintain their job (Tepper, 2000). When an individual assesses the goals and values of an organization, they will choose others as exemplars (Duffy, Ganster and Pagon, 2002). When this exemplar is an abusive leader that causes personal and professional hindrances, the employee chooses not to feel committed to the organization, since the behaviour of the leader is often perceived as a reflection of the organizational values. In that case, the employee is not willing to identify himself with these organizational values.
The mechanism described above explains the basic relationship between abusive leadership and commitment, but it doesn’t involve the mechanism of stress. The relationship between stress and commitment has received partial support. Montgomery et al. (1996) found a negative relationship between the two concepts. Shirom (1989, as cited in Van Dierendonk, Haynes, Borrill and Stride, 2004) states that increases in stress levels are associated with reduced commitment. Michael, Court and Petal (2009) researched the three different types of commitment (affective, continuance and normative) and their relationship to stress. They only found support for a negative relationship between stress and affective commitment. However, the study focussed on a particular group of mentors and is not generalizable. Therefore, the question whether stress leads to less commitment deserves more attention. Based on the known literature, it is expected that when an employee experiences stress, he or she will feel less committed to the organization. This leads to the second hypothesis.
Hypothesis 2: the stress level of the employee is negatively related to the commitment of the employee.
In the second hypothesis, the variable of abusive leadership isn’t included. However, for this research, the aim is to find out whether stress can explain the relationship between abusive leadership and commitment, so that an indirect effect exists. The variables of the first hypothesis (abusive leadership and stress) and second hypothesis (stress and commitment) are combined. This lead to the assumption
15 that abusive leadership will lead to more stress, which will then lead to lower commitment of the employees. Here the mediating effect is described, which leads to the third hypothesis.
Hypothesis 3: The negative relationship between abusive leadership and commitment is mediated by
stress.
Besides the mediating effect that is going to be researched, two moderating effects will also be researched. First, the moderating effect of self-efficacy on the relation abusive leadership and stress will be described. This moderating effect is interesting to elaborate on, since it can weaken or strengthen the relation between abusive leadership and stress.
2.4 The moderating variable: Self-efficacy
As was described in the introduction, the effect of moderators on leadership and stress deserves more attention. This research focusses on self-efficacy as a moderator, which is a form of self-belief. Previous studies on self-beliefs showed support that positive perceptions about the self could make stressors less detrimental (Jex and Bliese, 1999). Whether an individual is sensitive to an abusive leadership style, could depend on the person and not everyone experiences abuse in the same way (Tepper, 2000). One concept that could explain this reaction is self-efficacy. Self-efficacy is defined as an individual’s belief one can successfully execute a task (Bandura, 1977). The perceived self-efficacy influences the behaviour of an individual. When people believe a situation exceeds their coping skills, people try to avoid this situation. However, when people believe in their own capability of handling a situation, they get involved in activities that could otherwise be challenging or intimidating.
It is expected that an abusive leadership style will lead to more stress among the employees. It is interesting to investigate if self-efficacy as a moderator has an effect. According to Leiter (1991) and Stumpf, Brief & Hartman (1987), self-efficacy has an impact on the way how employees cope with stressors (as cited in Jex and Bliese, 1999). Individuals who have a high level of self-efficacy are more likely to use problem-focused coping strategies, whereas individuals with a low level of self-efficacy use a more emotion-focused coping strategy (Jex and Bliese, 1999). The authors explain this by stating
16 that individuals with a high level of self-efficacy are more willing to do something about stressors, whereas the opposites tend to worry about them more.
Abusive leadership can form a stressor for individuals, leading to more stress. However, when an individual has a high level of self-efficacy, a problem-focused coping strategy and therefore tends to do something about stressors, this will lead to a less stressed individual when he or she is exposed to an abusive leader. Individuals who use an emotion-focused coping strategy and have a low level of self-efficacy tend to react more heavily on and worry more about stressors such as abusive leadership and therefore it is expected they will experience a higher level of stress.
To summarize, self-efficacy is expected to have a moderating effect on the relationship between abusive leadership and stress. An individual high on self-efficacy is confident about his or her own skills and is less likely to be intimidated by stressors such as abusive leadership. This will lead to a weakened positive relationship. Thus, a high level of self-efficacy causes an individual to feel less stressed, even if they have an abusive leader. An individual low on self-efficacy is more prone to stressors, and is therefore expected to experience more stress when they are supervised by an abusive leader. This leads to the fourth hypothesis.
Hypothesis 4: the positive relationship between abusive leadership and stress is moderated by
self-efficacy, so that this relationship is weakened for high levels of self-efficacy.
Thus, it is expected that self-efficacy influences the relationship between abusive leadership and the level of stress among the employees. As was described before, the effect of stress on commitment will be researched as well. Here, the expectation is that stress will lead to less commitment of the employee. However, it would be interesting to find out if this relation can be affected by a moderator, so that the relationship is strengthened or weakened. The moderator that will be researched and elaborated on in the next paragraph is stress mindset; the mindset an individual holds about stress.
17 2.5 The moderating variable: Stress mindset
As was written in the previous paragraph, it is expected that self-efficacy can have an influence on whether an individual experiences high levels of stress or not when he or she is exposed to an abusive leader. Most people experiencing stress see it as a negative construct. The fact that people feel stressed leads to them worrying even more about the feeling of stress (Crum et al., 2013). This is a negative spiral, which makes it hard to see the possible advantages of stress. Even though a lot of people think stress is discouraging, some individuals need stress to perform better. Crum et al. (2013) describe this stress paradox and introduce the concept of stress mindset. Stress mindset is the mindset someone holds whether stress has enhancing or debilitating consequences for stress related outcomes such as wellbeing and health, performance and productivity, and learning and growth (Crum et al., 2013). When individuals hold a stress-is-enhancing mindset they experience positive stress related outcomes. For these individuals stress is enhancing for the development of mental toughness, new perspectives, increased sense of awareness and meaningfulness, stronger relationships, and a greater appreciation for life (Park & Helgeson, 2006; Tedeschi & Calhoun, 2004; as cited in Crum et al., 2013). Individuals who hold a stress-is-debilitating mindset belief stress has negative and debilitating consequences for those outcomes, such as lower productivity, causes of death, depression, conflict and burnout.
Thus, a mindset of someone can determine how it psychologically, behaviourally or physiologically influences the outcomes in different domains of their life, for example the stress domain (Crum et al., 2013). When faced with a stressful situation, individuals with a stress-is-enhancing mindset will choose behaviours that can help with dealing with the situation. The stress is then actively utilized to a positive outcome. Thus, stress mindset has an impact on the way an individual behaviourally approaches stress and how they psychologically experience stress. The study of Crum et al. (2013) shows that participants who developed a stress-is-enhancing mindset, also reported better health condition and work performance in contrast to the ones with a debilitating mindset.
Stress mindset hasn’t been researched greatly in combination with leadership and commitment. The concept of stress mindset is based on the concepts of eustress and distress (Crum et al., 2013). Eustress is the appraisal of stressors as opportunities and challenges, since the individual feels confident about overcoming these by using the right coping strategies (Lazarus, 1993). Distress is the appraisal of
18 stressors as harming or threatening (Lazarus, 1993). In the paragraph about stress the concepts of challenge and hindrance stressors were explained. The challenge stressors can be compared with eustress, whereas the hindrance stressors can be compared with distress. The meta-analysis of Podsakoff et al. (2007) found support for the fact that hindrance stressors were negatively related to organizational commitment. In contrast, challenge stressors where positively related to organizational commitment. This can be explained by the fact that challenge stressors are opportunities for task accomplishment and opportunities for personal growth.
Although stress mindset isn’t the same concept as challenge or hindrance stressors and is expected to have a moderating effect instead of a direct effect, it is based on the same literature (Crum et al., 2013). Challenge stressors and a stress-is-enhancing mindset are both related to the positive side of stress, whereas hindrance stressors and a stress-is-debilitating mindset are related to the negative side. Based on the information about stressors and commitment, it is expected that an enhancing stress mindset can influence the relationship between stress and commitment, so that the employee feels more committed to the organization, since they see the stressors as a challenge. A debilitating stress-mindset is expected to influence the relationship between stress and commitment, so that the employee feels less committed to the organization, since they see the stressors as a threat.
To summarize, it is expected that abusive leadership will lead to more stress, which then will lead to lower commitment of the employee. In this case, stress has a negative impact on the employee and therefore feels less committed to the organization. However, the stress mindset causes individuals to feel either positive or negative about stress which leads to different outcomes, and therefore one could expect a moderating effect of stress mindset. If an individual has a stress-is-enhancing mindset, thus feels that stress gives opportunities for personal growth and task accomplishment, it is expected that this will lead to a more positive outcome; it is more likely this individual will feel committed to the organization. If an individual has a stress-is-debilitating mindset, then stress will lead to a lower feeling of commitment. The stress caused by the abusive leader can therefore result in either enhancing or debilitating outcomes, thus more or less commitment. The fifth hypothesis is stated:
19
Hypothesis 5: the negative relationship between stress and commitment is moderated by stress mindset
so that this relationship becomes positive for a stress-is-enhancing mindset.
The overall model presumes moderated mediation on the first and second path of the mediation. Abusive leadership will lead to a higher level of stress, which in turn will lead to a lower commitment. When people have a high level of self-efficacy, they are expected to experience less stress than individuals with a low level of self-efficacy. Furthermore, when they do think that stress is enhancing, this could lead to more commitment even when they are stressed, since they feel that stress is positive for task accomplishment and personal growth. Based on this information the overall moderated mediation hypothesis is stated.
Hypothesis 6: The relationship between abusive leadership and commitment is mediated by stress and
this relationship can change based on different levels of self-efficacy and stress mindset, where high levels of self-efficacy weaken the relationship between abusive leadership and stress and where a stress-is-enhancing mindset changes the negative relationship between stress and commitment to a positive relationship.
2.6 Conceptual model
The described concepts and hypotheses are presented in a conceptual model (figure 1). To summarize, it is expected that abusive leadership will lead to more stress amongst the employees and that employee stress will lead to less commitment in the organization. The relationship between abusive leadership and commitment is expected to be mediated by stress. In this research the moderators self-efficacy and stress mindset are involved. Self-efficacy is expected to moderate the relationship between abusive leadership and stress so that this relationship becomes weaker for a high level of self-efficacy. Stress mindset is expected to moderate the relationship between employee stress and commitment, so that this relationship becomes positive for a stress-is-enhancing mindset.
20
Abusive leadership Employee stress Organizational
commitment
Self-efficacy Stress mindset
21
3. Methodology
3.1 Procedure
This research has an explanatory objective, which means the goal is to find a causal relationship between variables (Saunders, Lewis and Thornhill, 2012). The research has a cross-sectional design. With a cross sectional design, more than one case will be researched, at one moment in time, where quantitative data will be collected and then analysed, to hopefully find patterns between variables (Bryman, 2012). The data was gathered through a quantitative research method: a survey. The program Qualtrics was used to make the online survey. The survey was filled in digitally by the respondents, since this is a fast and convenient way to reach a lot of respondents.
The data collection was done with another student who also researched within the field of leadership, stress and commitment. However, we both researched another type of leadership and the influence of different moderators. The survey consisted of seven different constructs to gather the needed data. One of the constructs of the other student needed to be measured by the leader rate. Therefore, two surveys have been designed: one survey for the employee and one for the leader. For the first question, respondents were asked to fill in a code so that the surveys of the employee and the leader could be connected. At the end of the survey, the respondents were asked some demographic questions which were used for control variables. For my research, I will only use the data that has been collected via the employee.
The data has been collected from the 18th of October 2017 till the 21st of November 2017. Before
the survey was spread, a short text was written to introduce the respondents to our research and to ask permission to send them the survey. The survey was spread via an anonymous link. All the respondents received the link via e-mail. The e-mail addresses were obtained through our network. Since the response of the leader was also needed, it was important to make sure the leader was willing to fill in the survey as well. It was possible for employees to forward the survey to colleagues and other people in their network. Therefore, it is unknown how many individuals have been asked to fill in the survey, which makes it difficult to estimate the response rate. Next to our network, we had access to a company
22 who was willing to spread the survey to all their employees. The survey was added to the monthly newsletter and a link to the survey was posted on their intranet page.
3.2 Sample
For this research, a non-probability sample was used. The sampling method used is convenience sampling. This means that a sample is relatively easy available to the researcher since he or she has accessibility (Bryman, 2012). Next to convenience sampling, snowball sampling was used. We asked all the respondents to send the survey to their network. The sample consisted of employees and supervisors working in Dutch organizations. In total, 146 respondents (N=146) filled in the employee survey and 47 respondents filled in the leader survey. Of the 146 employee respondents, 113 respondents were gathered through our network and 33 employees where gathered via the company. Multiple employees under the span of one manager were able to fill in the survey. According to the rule of thumb of Green (1991), this is a representative sample size. To test multiple correlation, Green (1991) suggests a minimum of N > 50 + 8m (m is the number of independent variables).
For this particular research, only the data of the employees is used. Of the 146 respondents, 39 respondents are men (26,7%), 99 respondents are women (67,8%), and of 8 respondents the gender is unknown. The average age of the respondents is 41 years old, ranging from 19 to 65 years. The respondents work 29,21 hours per week on average. The average tenure with the organization is 12,21 years. The average tenure with the leader is 2,53 years. On average, the employee is in contact with the leader 2,38 days of the week.
3.3 Measures
In the survey, the following concepts were measured: abusive leadership, stress, commitment, self-efficacy and stress-mindset. Commitment is the dependent variable and abusive leadership is the independent variable. Stress is a mediator, which is expected to explain the relationship between abusive leadership and commitment. Self-efficacy is the moderator which is expected to influence the relation between abusive leadership and stress. Stress mindset is the moderator which is expected to influence the relation between stress and commitment. The survey is based on validated measures designed by
23 others to measure the variables. All the items, except for abusive leadership, were measured using a 5-point Likert scale, ranging from strongly disagree to strongly agree.
3.3.1 Abusive leadership
Abusive leadership was measured by the 15 items developed by Tepper (2000) to test whether a leader showed abusive behaviour. The reliability analysis of this research showed that the 15 items have an internal consistency of α = .87. Examples of items are: “My boss ridicules me”, “My boss puts me down in front of others” and “My boss is rude to me”. For the item “My boss breaks promises he or she makes”, the corrected item-total correlation is 0.28. However, the Cronbach’s alfa doesn’t significantly raise if the item is deleted. Therefore, the item will remain in the dataset. The items were measured using a 5-point response scale, where 1 is ‘I cannot remember him/her ever using this behaviour with me’ and 5 is ‘He/she uses this behaviour very often with me’.
3.3.2 Stress
The concept of stress was measured by the scale developed by Motowidlo, Packard and Manning (1986). The scale consists of 4 items which showed a high internal consistency of α = .84. The corrected item-total correlation of all the items indicates that the items have a good correlation with the item-total score of the scale, since they are all above 0.30. Two of the four items were stated in the reversed direction. An example of an item is “I feel a great deal of stress because of my job” and “Very few stressful things happen to me at work” which is a reversed item.
3.3.3 Organizational commitment
Commitment was measured with the Organizational Commitment Questionnaire developed by Mowday, Steers and Porter (1979). They developed a scale of 15 items. Six of these items were asked in the other direction. Mowday et al. (1979) have also developed a shortened scale excluding the six reversed items. For this research, the shortened scale with only the positive items is used. These 9 items showed an internal consistency of α = .87. An example item is: “I find that my values and the
24 organization’s values are very similar”. For one item (“I am willing to put in a great deal of effort beyond that is normally expected in order to help this organization be successful”), the corrected item-total correlation has a value of 0.26. However, the value of the Cronbach’s alfa will not raise significantly if the item would be deleted. Therefore, the item will not be deleted.
3.3.4 Self-efficacy
Self-efficacy was measured using the New General Self-Efficacy Scale (Chen, Gully and Eden, 2001). The scale consists of 8 items, which showed an internal consistency of α = .81. Example items are “I will be able to achieve most of the goals that I have set for myself” and “When facing difficult tasks, I am certain that I will accomplish them”.
3.3.5 Stress mindset
This concept was measured with the Stress Mindset Measure as developed by Crum, Salovey and Achor (2013). The 8 items showed an internal consistency of α = 0.87. Four of the eight items are asked in the opposite direction. Example items are “The effects of stress are negative and should be avoided” and “Experiencing stress facilitates my learning and growth”.
3.3.6 Control variables
At the end of the survey, some demographic questions were asked. These questions asked for the age and gender of the employee, but also how long they work in the organization, how long they work with their leader, how many hours they work per week, and how often they have direct contact with their leader. For this research the information of the demographic questions will be used as control variables. It would be interesting to see whether male and female tend to respond differently to stress. Therefore, gender will be used as a control variable. Furthermore, when commitment is measured, it is interesting to control for the tenure with the organization and the workhours per week. Lastly, since the employee has to answer questions about the leader, the contact with the leader will also be used as a control variable.
25
4. Results
4.1 Correlation analysis
After the collection of the data, the data was analysed using IBM SPSS 25. Before the data was ready to be analysed, the dataset has been cleaned and organised. The data doesn’t show missing data, since it was mandatory to answer every question before the questionnaire could be finished. Therefore, no data had to be excluded. However, for the control variables gender, age, workhours per week, tenure, tenure with the leader, and contact with the leader, there are some missing values, but this won’t be a problem for the analysis. The frequencies analysis indicates that all the answers were answered correctly and that they were all answered within the response range. Within two scales, for stress and stress mindset, some items were stated negatively to avoid response bias. Therefore, before the analysis of the data, these items (two items of stress and four items of stress mindset) have been recoded, so that every item indicates the same direction.
Table 1 shows the results of the descriptive statistics (mean and standard deviation), the correlations and the reliability analysis. To run the analysis, the items of each scale were computed into scale mean variables (Aiken & West, 1991). The correlations were measured with a Pearson Correlation test. The five variables of the conceptual model indicate two significant correlations. Abusive leadership is positively related to stress, which was expected (r = .22, p < 0.01). Stress and commitment are also significantly correlated. The correlation indicates a negative relationship between stress and commitment which was expected as well (r = -.18, p < 0.05). Furthermore, the demographic variables have some correlations with the other variables as well. Gender is negatively related to stress mindset (r = -.23, p < 0.01). This indicates that men have a more enhancing stress mindset than woman. Stress mindset also correlates with workhour (r = .21, p < 0.05). The positive relationship indicates that people who work more hours per week, think more positive about stress than people who work less hours per week. The tenure with the organization has a positive relationship with self-efficacy, which indicates that people who work longer in the organization have a higher level of self-efficacy (r = .18, p < 0.05). Although the correlations are significant, they are not very strong correlations. According to the rule of thumb of Cohen (1988; Field, 2003) a correlation of 0.1 indicates a small effect, a correlation of
26 0.3 a medium effect, and a correlation of 0.5 a high effect. The correlation between abusive leadership and stress indicate a higher correlation than stress and commitment, but they both do not meet the criterium for a medium effect. Therefore, it can be concluded that the correlations that do exist, are small correlations. The absence of many correlations and the presence of small correlations should be kept in mind when analysing the hypotheses.
Table 1. Correlation Analysis Variables M SD 1 2 3 4 5 6 7 8 9 1. Self-efficacy 3.74 0.42 (0.81) 2. Stress mindset 2.54 0.66 .11 (0.87) 3. Stress 3.17 0.76 -.14 -.11 (0.84) 4. Abusive leadership 1.19 0.32 .04 -.06 .22** (0.87) 5. Commitment 3.38 0.55 .14 -.06 -.18* -.09 (0.87) 6. Gender 1.74 0.48 -.01 -.23** .04 -.19* .14 - 7. Workhour 29.21 9.48 .07 .21* .10 .19* -.08 -.54** - 8. Tenure organization 12.22 11.79 .18* .05 -.05 -.01 -.11 -.18* .21* - 9. Contact leader 2.38 1.55 .03 .14 -.06 .02 -.01 -.23** .48** .06 - Note: N=146
*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed).
4.2 Direct effects
First, hypothesis 1 will be tested which predicted that abusive leadership is positively related to stress; the more abusive a leader is, the more stress an employee will experience. To test hypothesis 1, a hierarchical regression analysis has been executed. Stress was entered as the dependent variable and in the first step of hierarchical regression analysis, the control variables gender, tenure, workhour and contact with the leader were entered as predictors. Table 2 provides an overview of the analysis. After controlling for these four variables, the model was not statistically significant F (4, 129) = 1,80; p = 0.133 and explained 5,3% of variance in stress. The introduction of abusive leadership at step 2 explained an extra 5% variance in stress, after controlling for gender, tenure, workhour and contact (R2 Change = .05; F (5, 128) = 2,94; p = 0.015). When abusive leadership is added as a predictor, the model becomes statistically significant (p < .05). In the final model, two out of five predictor variables were statistically significant. Workhour is statistically significant (β = .26; p < .05), which indicates that if a person’s workhour increases for one, their stress level will increase for 0.26. Furthermore, abusive leadership is statistically significant (β = .23; p < .01). If abusive leadership increases for one, the stress level of the employee will increase for 0.23. Based on the results, hypothesis 1 is supported; a positive relationship exists, where more abusive leadership leads to more stress.
Table 2. Hierarchical regression model of the relationship abusive leadership and stress
R R² R² Change B SE β t Step 1 .23 .05 Gender .18 .16 .13 1.25 Contact -.08 .05 -.17 -1.70 Tenure -.01 .01 -.10 -1.02 Workhour .02 .01 .29 2.50 Step 2 .32 .10* 0.05* Gender .25 .16 .16 1.61 Contact -.08 .05 -.15 -1.59 Tenure -.01 .01 -.08 -.95 Workhour .02 .01 .26** 2.29 Abusive leadership .56 .21 .23* 2.68
Note. Statistical significance: *p = .01, **p = .05. Dependent variable = stress
The second hypothesis will also be tested by using a hierarchical regression analysis. This hypothesis predicts a negative relationship between stress and commitment. Commitment is the dependent variable.
29 The four control variables (gender, contact, tenure and workhour) will form the predictors in step 1. Stress will be added as a predictor in step 2. Table 3 provides an overview of the results of the analysis. After controlling for the four variables in step 1, the model was not statistically significant F (4, 129) = 0.91; p = 0.46 and explained 2,7% of variance in commitment. When stress is added at step 2, an extra 7,6% variance in commitment can be explained (R² = 10.4), after controlling for gender, tenure, workhour and contact (R² Change = .076; F (5, 128) = 2,96; p = 0.02). When stress is added as a predictor, the model becomes statistically significant (p < .05). In the final model, only stress is statistically significant (β = -.28; p < .01). The Beta value indicates that if a person’s stress level increases for one, his or her commitment decreases for 0.28; this indicates a negative relationship. According to the results, hypothesis 2 is supported.
Table 3. Hierarchical regression model of the relationship stress and commitment
R R² R² Change B SE β t Step 1 .17 .03 Gender .16 .16 .14 1.34 Contact -.00 .04 -.01 -.12 Tenure -.00 .00 -.10 -.90 Workhour .00 .01 .02 .17 Step 2 .32 .10* 0.08* Gender .20 .11 .17 1.74 Contact -.02 .03 -.06 -.61 Tenure -.01 .00 -.11 -1.22 Workhour .01 .01 .10 0.88 Stress -.21 .06 -.28* -3.30
Note. Statistical significance: *p = .01. Dependent variable = commitment
4.3 Mediating effect
To test the third hypothesis, the simple mediation model of Hayes (2012) was used. According to the causal steps approach of Baron and Kenny (1986) before testing for a mediating effect, one of the four criteriums is that a direct relationship exists between the independent variable (X) and dependent variable (Y). The correlation matrix shows that abusive leadership (X) and commitment (Y) do not significantly correlate with each other. Also, a simple regression analysis shows no significant effect between abusive leadership and commitment (β = -.08; p = .35). However, according to Hayes (2009)
30 and Zhao, Lynch and Chen (2010), the theory of Baron and Kenny (1986) needs to be reconsidered. Therefore, the mediation effect will still be tested.
In SPSS, PROCESS model 4 was used. Gender, tenure, contact and workhour were added as control variables. Since twelve cases had missing data in the control variables, they were excluded from the analysis, which led to a sample size of 134. The results in table 4 show that the effects of abusive leadership and stress and stress and commitment are significant. The direct effect of abusive leadership and commitment is not significant. The overall model can explain 10,39% of the variance of commitment (Y) and this is statistically significant (p < 0.05). The results show that a significant indirect effect exists of abusive leadership on commitment, via stress. The indirect effect of -0.11 means that when two workers differ by one unit of the reported abusive leadership, their commitment is expected to differ -0.11 as a result of experiencing stress. The negative indirect effect is statistically significant, with a 95% bootstrap confidence interval excluding zero (-0.414 to -.010). Therefore, hypothesis 3 can be supported.
Table 4. Mediation analysis
Consequent
M (Stress) Commitment (Y)
Antedent Coeff. SE p Coeff. SE p
Abusive leadership (X) a1 .56 .21 <.01 c1' -.04 .16 .82 Stress (M) --- --- --- b1 -.20 .06 <.01 constant i1 1.70 0.52 <.01 i2 3.65 .06 <.001 R² = .10 R² = 0.10 F (5, 128) = 2.94; p < 0.05 F (6, 127) = 2.45; p < 0.05 4.4 Moderation effect
The conceptual model entails two moderators: self-efficacy and stress mindset. First, the interaction effect of self-efficacy on the relationship between abusive leadership and stress is tested. The variables gender, tenure, workhour and contact with the leader were entered as control variables again. To test for a moderation effect, PROCESS model 1 is used. For the moderation analysis, the variables are mean-centered by using the option ‘mean center for products’. The results of the analysis are presented in table 5. It was expected that self-efficacy would weaken the relationship between abusive leadership and stress when the level of self-efficacy is high. However, the results show that there is no significant effect
31 of self-efficacy on this relationship. The regression coefficient for the interaction effect (abusive leadership x self-efficacy) is 0.003, but this is not statistically different from zero, t (126) = 0.003; p = 0.998. Thus, the effect of abusive leadership on stress is not influenced by the level of self-efficacy of the employee. Therefore, hypothesis 4 is not supported by the data.
Table 5. Moderation analysis of self-efficacy on abusive leadership and stress
Variable Coeff. SE t p Constant i1 2.26 0.41 5.56 <0.001 Abusive leadership (X) c1 0.60 0.57 1.05 0.30 Self-efficacy (SE) (M) c2 -.35 0.15 -2.31 0.02 Abusive leadership x SE (XM) c3 0.003 1.05 0.003 0.99 Gender 0.27 0.14 1.96 0.05 Tenure -0.003 0.01 -.48 0.64 Workhour 0.02 0.01 2.32 0.02 Contact -0.07 0.06 -1.18 0.24 R² = .14 F (7, 126) = 1.95; p = 0.07
The fifth hypothesis also contains a moderator and it predicted that the relationship between stress and commitment would be influenced by stress mindset, so that the negative relationship becomes positive. The second moderator also has been tested using PROCESS model 1. Table 6 shows the results of the moderation analysis. There is no significant interaction effect of stress mindset on the relation between stress and commitment. The regression coefficient for the interaction effect (stress x stress mindset) is -0.006, but this is not statistically different from zero, t (126) = -0.06; p = 0.95. Thus, the effect of stress on commitment is not influenced by the stress mindset of the employee. The fifth hypothesis is therefore not supported by the data.
32
Table 6. Moderation analysis of stress mindset on stress and commitment
Variable Coeff. SE t p
Constant i1 2.97 0.36 8.27 <0.001
Stress (X) c1 -0.20 0.07 -2.90 0.004
Stress mindset (M) c2 -0.02 0.08 -0.27 0.79
Stress x Stress mindset (XM) c3 -0.01 0.10 -0.06 0.95
Gender 0.19 0.13 1.44 0.15 Tenure -0.01 0.004 -1.21 0.23 Workhour 0.01 0.01 0.89 0.38 Contact -0.02 0.04 -0.57 0.57 R² = .10 F (7, 126) = 1.80; p = 0.09 4.5 Moderated mediation
The final hypothesis includes the complete model and it suggests a moderated mediation. Since the model contains two moderators in two different stages, there is a dual-stage moderated mediation. This model can be analysed by using PROCESS model 21. However, since there is no evidence for both of the moderation effects, there is no reason to test the final hypothesis. Running the analysis indeed show that no significant effects exists. Interaction effect 1, which is self-efficacy on abusive leadership and stress, has an insignificant effect of 0.003, t (126) = 0.003; p = 0.998. Interaction effect 2, which is stress mindset on stress and commitment, has an insignificant effect of -0.06, t (122) = -0.50; p = 0.61. Hypothesis 6 is therefore rejected.
4.6 Exploratory research
The results show no significant effects of the moderators. An exploratory analysis has been executed to research whether self-efficacy and stress mindset do have a moderating effect when the dependent or independent variable changes. For all the exploratory analyses, PROCESS model 1 has been used.
First, abusive leadership was entered as an independent variable and commitment as the dependent variable and stress as a mediator was left out. Self-efficacy was added as a moderator. The regression coefficient for the interaction effect is -.12 (abusive leadership x self-efficacy). However, this
33 effect is not significant t (126) = -.25; p = 0.80. Therefore, there is no reason to expect a moderation effect of self-efficacy between abusive leadership and commitment.
Second, instead of self-efficacy as a moderator, stress mindset was used as a moderator between abusive leadership and commitment. The results show a regression coefficient of .27 (abusive leadership x stress mindset) which is not statistically significant t (126) = .57, p = 0.57. Therefore, no moderation effect of stress mindset can be expected between abusive leadership and commitment.
Finally, the moderation effect of self-efficacy has been tested when stress is the independent variable and commitment the dependent variable (stress x self-efficacy). The regression coefficient of 0.05 is not statistically significant either with t (126) = .29, p = 0.77. These exploratory analyses show that the moderators do not have an effect when testing alternative models.
34
5. Discussion
5.1 Summary
The purpose of this study was to expand the already rich literature on leadership and stress. The aim was to create a better understanding of the relationship between leadership, stress and commitment. Furthermore, two moderators, self-efficacy and stress mindset, were added to research what the influence of these variable was on this relationship.
Leadership consists of many forms, but for this research abusive leadership was used. It was expected that the presence of abusive leadership would lead to more stress among the employee (hypothesis 1) and that stress would lead to a lower level of commitment (hypothesis 2). The third hypothesis stated that stress was expected to form a mediator between abusive leadership and stress. Next to a mediator, this research also looked at two moderators: self-efficacy and stress mindset. It was expected that the first path, between abusive leadership and stress, would be moderated by self-efficacy, so that a higher level of self-efficacy would lead to a weaker relationship between abusive leadership and stress (hypothesis 4). Stress mindset was expected to influence the second path, so that individuals with a stress-is-enhancing mindset would be more committed to the organization even when they experience stress (hypothesis 5).
After testing the hypotheses, it was found that the first three hypotheses can be supported. On the contrary, no results were found for the two hypotheses testing the moderators. Self-efficacy had no significant effect on the relationship between abusive leadership and stress and stress mindset had no significant effect on the relationship between stress and commitment. Therefore, the last hypothesis testing the whole model had to be rejected as well (hypothesis 6).
5.2 Theoretical implications
The findings of this research provide support for the direct relationships that were tested. The results show that when an employee identifies their leader as abusive, this will lead to a higher feeling of stress. This finding confirms the findings from the literature (Harvey et al., 2007; Harms et al., 2016; Tepper, 2000, 2007; Chi and Liang, 2013) that abusive leadership is positively related to stress. Abusive leaders
35 cause interpersonal conflict and continuously attack the feelings and self-esteem of the employee which causes stress (Harvey et al., 2007), and they lack the required skills that are needed when an employee experiences a stressful situation (Kelloway et al., 2005). When testing this hypothesis, after controlling for gender, contact, tenure and workhour, it was also found that workhour leads to more stress among employees. The positive significant relationship indicates that the more hours an employee works per week, the more stress he or she experiences. This shows that work hour can be seen as a stressor as well which wasn’t expected at first.
Furthermore, support was found for a negative relationship between stress and commitment. This contributes to the literature, since this relationship has received partial support in previous studies (Chen et al., 2006; Michael et al., 2009). It was expected that when people experience an unpleasant feeling of stress caused by the organization, they wouldn’t feel a desire to stay longer or put more effort into the organization (Steers, 1977). The findings of this research indeed indicate that when an employee experiences stress, this will then lead to the employee feeling less committed to the organization. In this research, the concept of commitment was researched as organizational commitment. However, commitment can be divided into three types of commitment: affective, continuance and normative commitment (Meyer and Allen, 1991). Michael et al. (2009) researched these three types of commitment and they only found a negative relationship between stress and affective commitment. This could be explained by the fact that affective commitment indicates an emotional identification with the organization. When someone experiences stress, the emotional identification is very low which indicates a low affective commitment of the employee. Normative commitment on the other hand, indicates that an employee feels obligated to remain with the organization. When an individual experiences stress, the individual may still feel obligated to stay in the organization and therefore is still committed. It is interesting to see that the findings of this research are significant and that a negative relationship exists, as was expected. However, it is important to keep in mind that these results could have differed when commitment was measured using a different scale that makes a distinction between the three types of commitment.
In the described relationships above, stress acted first as a dependent variable and in the second as an independent variable. For the third hypothesis, stress was tested as a mediator. The results show
36 support for the expectation that stress mediates the relationship between abusive leadership and commitment. The research shows that the presence of abusive leadership leads to more stress, which then leads to less commitment and this explains the negative mediating effect. In the literature the relation between leadership and stress has been tested, but stress as a mediator with commitment as an outcome, hasn’t been researched before. Therefore, the effect that has been found is an important contribution to the literature, since it shows that stress can explain why employees are less committed to the organization when they are exposed to an abusive leader.
Furthermore, even though it isn’t the aim of the research, it is interesting to have a look at the direct relationship between abusive leadership and commitment. Although this wasn’t a hypothesis, the relationship has been tested. No relationship was found based on the results. This is quite a remarkable finding, since this relationship was found in previous research (Tepper, 2000; Duffy et al., 2002; Mackey, Frieder, Brees and Martinko, 2017; Zhang and Liao, 2015). They found that abusive leadership leads to more anger, anxiety, depression, emotional exhaustion, distress, and to lower organizational identification, job satisfaction and organizational commitment. The results of this research are in contrast with the findings of the literature, since there is no support for a significant relationship between abusive leadership and commitment, but only for an indirect effect when stress is added as a mediator. An explanation for the absence of a direct relationship could be because no distinction between the types of commitment has been made in the scale I used from Mowday et al. (1979). As was described before, commitment can be divided into three types of commitment. Zhang and Liao (2015) found a significant negative relationship between abusive leadership and affective commitment. Tepper (2000) found a positive relationship between abusive leadership and continuous commitment and a negative relationship between abusive leadership and affective and normative commitment. The authors who did find a significant relationship between abusive leadership and the different types of commitment used a different scale to measure commitment. This could explain the difference in results and why no relationship has been found in this research.
The aim of this research was to create a better understanding of the relationship between leadership, stress and commitment by investigating whether two moderators had an influence on the relationships described above. First, self-efficacy was tested as a moderator to see whether it would
