In which circumstances is a narcissistic leader more
likely to engage in abusive leadership behaviour?
Master thesis final version
Karen Grootoonk, 11820489
MSc Business administration - leadership and management track Supervisor: A. De Hoogh
Second supervisor: D. Den Hartog Friday, 22nd of June 2018
Statement of originality
This document is written by Karen Grootoonk 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.
Table of contents
1. Introduction ... 1
2. Theoretical framework ... 5
2.1 Abusive leadership ... 5
2.1.1 Consequences of abusive leadership ... 5
2.1.2 Antecedents of abusive leadership ... 7
2.2 Narcissism ... 9
2.3 Rule environment ... 12
2.4 Competitive culture ... 14
2.5 The mediating role of abusive leadership ... 17
3. Method ... 18
3.1 Sample and procedure ... 19
3.2 Measures ... 21 3.2.1 Abusive leadership ... 21 3.2.2 OCB... 22 3.2.3 Narcissism ... 22 3.2.4 Rule environment ... 23 3.2.5 Competitive culture ... 23 3.2.6 Control variables ... 24 3.3 Statistical procedure ... 25 4. Results ... 26 4.1 Descriptive statistics ... 26 4.2 Hypotheses testing ... 29
4.2.1 Step one – Hypothesis 1 and 2 ... 29
4.2.2 Step two – Hypothesis 3 and 4 ... 30
4.2.3 Step three – Hypothesis 5 ... 33
5. Discussion ... 38 5.1 Limitations ... 41 5.2 Future research ... 42 5.3 Practical implications ... 44 6. Conclusion ... 46 7. References ... 47 8. Appendix ... 62
8.1 Appendix A – English survey items ... 62
8.1.2 OCB... 62
8.1.3 Narcissism ... 63
8.1.4 Rule environment ... 63
8.1.5 Competitive culture ... 63
8.2 Appendix B – Dutch survey items ... 64
8.2.1 Abusive leadership ... 64
8.2.2 OCB... 64
8.2.3 Narcissism ... 65
8.2.4 Rule environment ... 65
List of Tables
Table 1. Correlations matrix 28
Table 2.Simple linear regression of narcissism and abusive leadership 33
Table 3.Simple linear regression of Leadership/Authority (LA) 33 and abusive leadership
Table 4. Simple linear regression of Grandiose Exhibitionism (GE) 33 and abusive leadership
Table 5. Simple linear regression of Entitlement/Exploitativeness (EE) 34 and abusive leadership
Table 6. Simple linear regression of abusive leadership and OCB 34
Table 7. Moderating effect of Rule environment and competitive culture 34 on the relationship between Narcissism and Abusive leadership
Table 8. Moderating effect of Rule environment and competitive culture 35 on the relationship between LA Narcissism and Abusive leadership
Table 9. Moderating effect of Rule environment and competitive culture 35 on the relationship between GE Narcissism and Abusive leadership
Table 10. Moderating effect of Rule environment and competitive culture 36 on the relationship between EE Narcissism and Abusive leadership
List of Figures
Abstract
Considering the increase in levels of narcissism in society and of narcissists in leadership positions, there is a growing body of research into understanding narcissistic leadership. Scholars have theorized that narcissism is positively related to abusive leadership behaviour. However, until now, research findings have been inconsistent. Additionally, existing literature shows that organizational factors might inhibit or contribute to abusive leadership.
Therefore, this study identifies two organizational factors, organizational rule climate and competitive culture, and consequently examines the extent to which these factors
influence the relationship between a leader’s narcissism and abusive leadership behaviour as perceived by the followers.
This study proposes that narcissistic leaders are more likely to engage in abusive leadership behaviour, but less so when an organization has strict rules whereas more so when an organization has a competitive culture. In turn, higher levels of abusiveness are expected to be related to lower levels of a follower’s extra role behaviours, defined as OCB.
This research adopts a correlational research design for theory testing. Surveys were sent out to a number of leader-follower dyads. The results derived from the data of these surveys hint that narcissistic leaders are not more likely to engage in abusive leadership behaviour. Furthermore, no evidence was found for the moderating effect of rule environment and competitive culture on the relationship between leader narcissism and
abusive leadership. Lastly, this study did not support the claim that abusive leadership leads to lower follower OCB.
Even though findings of this study were non-significant, these contribute to a better understanding of narcissists in leadership positions. This paper discusses the implications for theory and practice.
1
1. Introduction
Leadership comprises a universal activity, present in both humans and animals (Antonakis, Cianciolo & Sternberg, 2004), with which we are confronted in our daily lives through politics, work, and social activities. Leaders build empires, can have an innovative business vision, or simply be your supervisor at work. Because of its omnipresence, leadership is of interest to both scholars as well as the general public (Den Hartog & Koopman, 2004). However, leadership does not always go well, since 65-75% of the employees feel that their supervisor was the worst part of their job (Hogan & Kaiser, 2005; Zhang & Bednall, 2015). Therefore, over the last decade, more attention has been paid to the dark side of leadership, such as the so-called abusive supervision or abusive leadership (Tepper, 2007). In this paper, both terms will be used interchangeably.
Abusive supervision provokes an image of a dictatorial boss, bullying, belittling, and undermining his or her subordinates (Ashforth, 1994). As such, it is the hostile behaviour of supervisors, as perceived by their subordinates (Tepper, 2000). Abusive leadership is a significant, widespread social problem, affecting approximately 13.6% of the U.S. workers (Tepper, 2007). Not only may employers be held liable for negative outcomes (Yamada, 2004), it also incurs substantial costs of roughly $23.8 billion annually in terms of potential withdrawal, increased healthcare costs, and loss of productivity (Tepper, Duffy, Henle & Lambert, 2006; Tepper, 2007). On top of that, abusive leadership has many consequences for both followers, and the organization (Mitchell & Ambrose 2007; Estes & Wang, 2008; Tepper, 2000; Tepper, Duffy, Hoobler & Ensley, 2004; Tepper, Henle, Lambert, Giacalone & Duffy, 2008). Amongst other things, abusive leadership has been found to decrease employees’ extra role behaviours, defined as Organizational Citizenship Behaviour (OCB) (Zhang & Liao, 2015). When experiencing abusive leadership behaviour, employees may engage in deviant behaviours to get back at the one who mistreated them (Bennett &
2 Robinson, 2003). This can be done by withdrawing positive behaviours, such as OCB (Aryee, Chen, Sun & Debrah, 2007; Zellars, Tepper & Duffy, 2002). This research attempts to confirm this finding.
The broad range of negative consequences make it an important concept. However, leaders may become abusive for a variety of reasons (Aasland, Skogstad, Notelaers, Nielsen & Einarsen, 2010), yet very little is known about the antecedents of abusive leadership (Aasland et al., 2010; Burton & Hoobler, 2011; Tepper, 2007; Zhang & Bednall, 2016). There have been many calls to fill this gap (Aasland et al., 2010; Burton & Hoobler, 2011; Tepper, 2007; Zhang & Bednall, 2016). This research identifies character traits as a possible antecedent for abusive leadership since researchers have also repeatedly argued for more integrative theories of leadership (Avolio, 2007; Derue, Nahrgang, Wellman & Humphrey, 2011), for example by combining leadership trait and behavioural theory (Johnson, Venus, Lanaj, Mao, & Chang, 2012). Furthermore, only a limited amount of studies has been conducted on personality traits as antecedents of abusive leadership behaviour (Zhang & Liao, 2015).
Narcissism is a personality trait that has been defined as “a grandiose sense of self-importance” (Judge et al., 2006, p. 762). Levels of narcissism are rising (Twenge & Campbell, 2008; Twenge & Foster, 2008; Twenge, Konrath, Foster, Campbell & Bushman, 2008) and, on top of that, narcissists are likely to be in leadership positions (Judge, LePine & Rich, 2006; Grijalva, Harms, Newman, Gaddis & Fraley 2015; Nevicka, de Hoogh, Van Vianen, Beersema & Mcllwain, 2011; Rosenthal & Pittinsky, 2006). Furthermore, it is known that narcissists have aggressive tendencies (Lier, 2016; Penney & Spector, 2002). As a result, it is highly relevant to look at narcissism in the field of (abusive) leadership. Even though scholars have theorized that narcissism is positively related to abusive leadership behaviour, up to date, little is known about this relationship (Waldman, Wang, Hannah, Owens &
3 Balthazard, 2018; Wisse & Sleebos, 2016). Therefore, this research focuses on narcissism as a possible antecedent of abusive leadership.
Nevertheless, this does not imply that narcissism is the only cause of bad leadership (Higgs, 2009). Rather, the organization provides the setting in which the individual may or may not display specific behaviour (Van Fleet & Griffin, 2006). Therefore, it is interesting to see in which situations narcissism can lead to abusive behaviour, as the possibility exists that situational factors moderate the influence of narcissistic character traits on leadership. (Hoffman, Strang Kuhnert, Campbell, Kennedy, & LoPilato, 2013). Even though it is known that organizational factors might inhibit or contribute to abusive leadership, it has still received little direct study (Martinko, Harvey, Brees & Mackey, 2013). In this study we identify two organizational factors that may influence the extent to which the leader’s narcissistic traits are expressed in abusive leadership behaviour.
To start with, this research introduces organizational climate, which can be comprised of a rule environment. A rule environment establishes clear rules and regulations and decreases ambiguity (Victor & Cullen, 1988). When such an environment is present, it will reduce the freedom that leaders have in performing their tasks (Padilla, Kaiser & Hogan, 2007). In line with trait activation theory, in this situation, narcissistic traits are not encouraged by trait-relevant situational signals, and thus they will not be expressed in behaviour (Tett & Gutermann, 2000). As such, narcissists are less likely to show their narcissistic traits through engagement in abusive leadership behaviour.
Secondly, we identify the situational factor of organizational culture. One type of organizational culture is a competitive culture (Brown, Cron & Slocum, 1998; Schein 1990). In a highly competitive organizational culture, there is a focus on rewards and performance, and competition for recognition (Fletcher, Major & Davis, 2008). Narcissistic traits fit with this culture because narcissists have a need for power, recognition and domination (Rosenthal
4 & Pittinsky, 2006) and a desire to get ahead of others (Roberts & Robins, 2000). Therefore, we expect that these traits will prosper in such an environment and be expressed openly in abusive leadership behaviour.
To conclude, the purpose of this study is to link leader narcissism to abusive supervision and to examine under which circumstances narcissism is more strongly related to abusive leadership behaviour, because it is necessary to identify situations in which aggression as a result of leader narcissism can be excessive, constant, or even stopped (Burton & Hoobler, 2011). In order to identify effective solutions, it is important to identify the real origins, the antecedents of abusive leadership (Lawrence & Robinson, 2007).
The present study adds to the existing literature by examining narcissism as a possible antecedent of abusive leadership behaviour, and by examining the role of organizational climate and culture in relation to the manifestation of leader narcissism through abusive leadership. As such, this research aims to fill two literature gaps in the field of abusive leadership. Firstly, this paper contributes to knowledge regarding antecedents of abusive behaviour. Additionally, contributions are made to research surrounding situational and organizational factors by looking at the importance of the context in abusive leadership.
To be able to answer the research question: in which conditions is a narcissistic leader
more likely to engage in abusive leadership, we developed the conceptual model as depicted
in Figure 1 on the following page. The model will be further explained in the following section, in which relevant literature regarding the theoretical concepts is reviewed.
5
2. Theoretical framework
In this section, we review the theoretical concepts included in our conceptual model to provide an overview of where research stands regarding the specific concepts, and to present an explanation for the expected relationships included in our model.
2.1 Abusive leadership
Abusive supervision provokes a vivid image of a dictatorial boss (Ashforth, 1994). In literature, abusive leadership is defined as “the subordinates' perceptions of the extent to which supervisors engage in the sustained display of hostile verbal and nonverbal behaviours, excluding physical contact” (Tepper, 2000, p. 178). An example of abusive leadership is a leader telling his or her subordinates that their feelings are stupid or putting them down in front of others (Mawritz, Mayer, Hoobler, Wayne & Marinova, 2012). Abusive supervision can be manifested in multiple ways: ridiculing, loud and angry outbursts, rudeness, intimidation, taking credit for subordinates’ work or inappropriately blaming them for failure (Ashforth, 1994; Tepper et al., 2006).
2.1.1 Consequences of abusive leadership
Abusive leadership is considered an important concept, mainly because of its negative consequences. To start with, amongst others, abusive supervision leads to employee
6 frustration and alienation of work (Ashforth, 1994). Additionally, abusive leadership has been found to be related to increased strain (Mackey, Frieder, Brees & Martinko, 2017), such as emotional exhaustion (Wheeler, Halbesleven & Whitman, 2013). Moreover, it can decrease employee satisfaction (Tepper et al., 2004), as well as increasing intentions to leave (Tepper, 2000). Furthermore, it has been found to decrease an employee’s family well-being (Tepper, 2000).
Not only for the employee abusive supervision produces negative outcomes, also the organization suffers. Abusive supervision can decrease individual and organizational performance (Estes & Wang, 2008). Furthermore, it may produce employee deviance (Mitchell & Ambrose, 2007; Tepper et al, 2008). In addition to increasing negative behaviours, such as workplace deviance, abusive leadership also decreases positive employee behaviours, such as OCB (Mitchell & Ambrose, 2007; Xu, Huang & Miao, 2012). OCB is defined as “individual behaviour that is discretionary, not directly or explicitly recognized by the formal reward system and that in the aggregate promotes effective functioning of the organization” (Organ, 1988, p. 4). In other words, OCB is comprised of voluntary, positive employee behaviours that help the organization achieve its goals.
Multiple explanations have been given for the negative relationship between abusive supervision and OCB. Social exchange theory is used as a building block for these explanations. According to social exchange theory, individuals will return favours to maintain the relationship and equalize exchange (Gouldner, 1960). In this context, subordinates will reciprocate beneficial, supportive leadership behaviours by offering a favourable return, such as OCB (Xu et al., 2012; Zellars et al., 2002). On the other hand, when employees experience negative supervisor behaviour, such as abusive leadership, they will respond by engaging in unfavourable behaviours or withdrawing positive behaviours (Xu et al., 2012; Mitchell & Ambrose, 2007).
7 First, researchers have used justice theory to explain the withdrawal of OCB as a result of abusive supervision. Aryee et al. (2007) argue that abusive leadership behaviour causes a lower quality of social exchange between leader and follower, which creates a perception of a lack of interpersonal justice. As a result, employees do not feel obliged to engage in positive behaviours, such as OCB (Aryee et al., 2007).
Secondly, based on reactance theory, individuals aim to maintain a sense of autonomy and personal control (Brehm & Brehm, 1981). When employees experience abusive leadership behaviour, they feel little to no control (Mitchell & Ambrose, 2007). One way to restore this is by exercising autonomy over your behaviour (Wright & Brehm, 1982). This means that employees will choose whether they want to display specific behaviours. As a result, employees may decide to reduce work efforts or OCB (Mitchell & Ambrose, 2007; Xu et al., 2012; Zellars et al., 2002)
Finally, retrieved from aggression theory, researchers have argued that employees experiencing abusive leadership will likely try to get even (Bennett & Robinson, 2003; Mitchell & Ambrose, 2007). Since the relationship between the subordinate and supervisor is hierarchical, it may be costly for one’s self to do so (Decoster, Camps, Stouten, Vandevyvere & Tripp, 2013). Withholding effort may be a way to get back without risking punishment (Burris, Detert & Chiaburu, 2008). As such, employees may reduce their intended helping behaviours, which are part of OCB (Aryee et al., 2007; Xu et al., 2012; Zellars et al., 2002).
In this study, we attempt to confirm this relationship and previous research finding by testing the following Hypothesis:
Hypothesis 1: Abusive leadership is negatively related to OCB.
2.1.2 Antecedents of abusive leadership
Given the fact that abusive leadership has so many negative consequences, the question remains: why do people engage in abusive leadership behaviour? According to
8 Aasland et al. (2010), there are various reasons for leaders to become abusive. Nevertheless, up to date, only a handful of antecedents have been identified. This study attempts to advance the field of abusive leadership by studying a possible important antecedent to fill this gap. Therefore, we will now divert our focus from abusive leadership consequences to its antecedents.
Amongst others, when the leader experiences injustice within the firm, they may get feelings of depression, which makes them more likely to engage in abusive leadership behaviour towards their vulnerable subordinates (Tepper et al., 2006). Furthermore, leaders experiencing abusive behaviour, be it in their family life during their childhood (Kiewitz, Restubog, Zagenczyk, Scott, Garcia & Tang, 2012) or in the workplace (Aryee, Chen, Sun & Debrah, 2007; Mawritz et al., 2012), are more likely to become abusive themselves. This is explained by the social learning theory, dictating that individuals learn behaviour by observing others, or by directly experiencing it (Bandura, 1978). As such, abusive behaviour trickles down the organization (Aryee et al., 2007). Additionally, when the leader and the subordinate experience a conflicting relationship, the leader is more likely to experience displaced aggression which can lead to abusive behaviour (Harris, Harvey & Kacmar, 2011). More specifically, when the leader perceives the subordinate to be dissimilar, this may lead to conflict (Tepper, Moss & Duffy, 2011). Lastly, supervisors experiencing stress or dissatisfaction at work may also become abusive (Burton, Hoobler & Scheuer, 2012; Tepper, 2007).
Nevertheless, abusive behaviour may also stem from personal factors (Zhang & Bednall, 2015). Perhaps some people are more likely to exhibit abusive leadership behaviour than others. This idea combines leadership trait with behavioural theory (Johnson et al., 2012). Research has found evidence that supervisor Machiavellianism can trigger abusive behaviour (Kiazad, Restubog, Zegenczyk, Kjewitz & Tang, 2010) since Machiavellianism
9 involves a tendency to manipulate and exploit others (House & Howell, 1992). The trait Machiavellianism involves engagement in amoral, manipulative behaviour, and a search for status for oneself and control over others (Dahling, Whitaker & Levy, 2009; Greenbaum, Hill, Mawritz & Quade, 2017), in line with abusive leadership behaviour. Furthermore, psychopathy is another character trait that has been found to be an antecedent of abusive leadership (Mathieu & Babiak, 2016). Psychopathy refers to impulsiveness and a lack of empathy and feelings of guilt (Wisse & Sleebos, 2016), which leads to aggressiveness that is expressed through abusive leadership (Mathieu & Babiak, 2016).
These two character traits, together with narcissism form the dark triad of personality (Paulhus & Williams, 2002). Dark personality traits can predict dysfunctional leader performance, such as abusive leadership because they involve aggressive thoughts, hostility, as well as tendencies to engage in counterproductive work behaviours (Benson & Campbell, 2007; Burton & Hoobler, 2011; Kiazad et al., 2010). Therefore, narcissism may also be considered as being related to abusive supervision (Tepper, 2007). Narcissism is a very relevant and important trait to consider, because levels of narcissism are rising (Twenge & Campbell, 2008; Twenge & Foster, 2008; Twenge et al., 2008). Additionally, narcissists are prone to be in leadership positions, which makes it important to study in the context of (abusive) leadership (Grijalva et al., 2015; Nevicka et al., 2011).
2.2 Narcissism
Narcissism stems from an old Greek myth, where Narcissus, out of vanity, fell in love with his own reflection (Judge et al., 2006; Nevicka et al., 2011; Rosenthal & Pittinsky, 2006). In social-personality literature, narcissism is conceptualized as a personality trait that is normally distributed in the population and related to other traits, such as entitlement (Campbell, Bonacci, Shelton, Exline & Bushman, 2004; Foster & Campbell, 2007). Broadly stated, narcissism is “a grandiose sense of self-importance” (Judge et al., 2006, p. 762). It
10 encompasses a strong self-importance, inflated self-views, self-love, a lack of interest in warm interpersonal relationships, and hostility (Campbell, Hoffman, Campbell & Marchisio, 2011; Foster & Campbell, 2007; Grijalva et al., 2015; Rosenthal & Pittinsky, 2006) As such, narcissism contains features regarding three components: a positive self, interpersonal relations, and self-regulatory strategies (Campbell et al., 2011; Foster & Campbell, 2007).
Narcissists are likely to be in leadership positions, because of self-selection and nomination resulting from their desire for power, and because others tend to pick narcissists as leaders as they fit leader prototypes (Judge et al., 2006; Grijalva et al., 2015; Nevicka et al., 2011; Rosenthal & Pittinsky, 2006). In these positions narcissists have a bright and a dark side (Campbell et al., 2010; Grijalva et al., 2015; Nevicka et al., 2011; Rosenthal & Pittinsky, 2006), with positive traits such as charm and charisma, and negative traits such as a lack of empathy and arrogance (Rosenthal & Pittinsky, 2006). As such, it is argued that narcissism is a mixed blessing (Campbell & Campbell, 2009).
Literature focuses mainly on the downside of narcissistic leaders (Rosenthal & Pittinsky, 2006). According to Penney and Spector (2002), narcissists are more likely to engage in counterproductive work behaviours, since they display trait anger, which they release through aggression, a form of counterproductive work behaviour. Therefore, it has been proposed that narcissistic leaders are more likely to display abusive leadership behaviour (Martinko et al., 2013; Tepper, 2007; Wisse & Sleebos, 2016). However, Wisse and Sleebos (2016) found no significant results to support this claim. Due to time constraints their study used a short version to measure dark trait characteristics, which may have affected the results. As such, they proposed that future research replicates their study, using other scales, such as the NPI to measure narcissism (Wisse & Sleebos, 2016).
Therefore, following this suggestion, this study identifies narcissism as a potential antecedent of abusive leadership behaviour. In the leadership position, narcissists have the
11 freedom to adopt their own leadership style, consistent with their characteristics. One of the narcissistic character traits is a lack of empathy, a key aspect of emotional intelligence (Rosenthal & Pittinsky, 2006). Empathy is needed to regulate negative emotions, using regulation strategies (Wong & Law, 2002). Narcissists lacking this crucial trait are, therefore, more likely to unleash negative emotions (Zhang & Bednall, 2016). This can result in, per example, loud and angry outbursts towards subordinates, consistent with abusive leadership behaviour. In addition, due to their lack of empathy, narcissists tend to make decisions by themselves, guided by their self-centred view of the world, thereby ignoring any advice that contradicts their view (Rosenthal & Pittinsky, 2006). A narcissistic leader who ignores advice limits the subordinates’ voice. It can also be experienced as belittling as if the subordinates’ opinion or advice is not of value. Limiting follower voice, and belittling subordinates are in line with abusive leadership behaviour.
Furthermore, narcissists have a need for power, recognition, and domination (Morf & Rhodewalt, 2001). They use a variety of means to prove their influence. Therefore, we can expect narcissistic leaders to demand unquestioning loyalty from followers (Harwood, 2003). This might give followers the feeling that their privacy is invaded, while at the same time putting them under enormous pressure. Simultaneously, narcissistic leaders might abuse the power they already have in their quest for ever more domination (Rosenthal & Pittinsky, 2006). On top of that, the narcissistic quest for power might lead narcissistic leaders to attribute success entirely to themselves, ignoring the importance of their followers. All these behaviours can be characterized as abusive leadership behaviour.
Lastly, narcissists possess an inflated self-confidence, overestimating themselves (John & Robins, 1994; Judge et al., 2006). This can lead to self-enhancement, which makes it harder for them to build interpersonal relationships (Paulhus, 1998). As a result, they may come across as rude. Furthermore, narcissists will act to protect their inflated self-image
12 (Twenge & Campbell, 2003). When their confidence is threatened or endangered, they may use aggression towards a third party, such as their subordinate (Baumeister, Bushman & Campbell, 2000; Twenge & Campbell, 2003). Likewise, they tend to attribute success to themselves, while not recognizing the importance of the follower’s contributions (Rosenthal & Pittinsky, 2006). All these behaviours resulting from the narcissist’s inflated self-confidence are consistent with abusive leadership behaviour.
Following the line of characteristics that narcissists tend to exhibit, and their resulting actions, we predict that narcissists are more likely to engage in abusive leadership behaviour. This results in the following Hypothesis:
Hypothesis 2: Narcissism is positively related to abusive leadership behaviour
2.3 Rule environment
Nonetheless, it is more likely that abusive leadership may stem from a combination of situational and personal factors (Zhang & Bednall, 2015). Situations may exist in which narcissists display less abusive behaviour than in others. People that score high on aggression are also not always aggressive, but they only exhibit aggressive behaviour in specific situations (Tett & Gutermann, 2000). This highlights the importance of the situational context for displaying a specific type of behaviour. According to trait activation theory, for a trait to be expressed in behaviour, it needs to be encouraged by trait-relevant situational signals (Tett & Gutermann, 2000). In other words, expression of personality traits in behaviour depends on the relevance of the situation to the trait. For (narcissistic) leaders, the organization provides the setting in which they may or may not display specific behaviour (Van Fleet & Griffin, 2006). As such, situational factors moderate the influence of narcissistic character traits on leadership (Hoffman, Strang Kuhnert, Campbell, Kennedy, & LoPilato, 2013).
One of the situational factors within an organization is its work climate, the perceptions of what is correct behaviour, as interpreted from organizational policies,
13 procedures, and reward systems (Schneider & Rentsch, 1988; Victor & Cullen, 1988). There is not only one type of work climate (Schneider, 1975). Rather, Victor and Cullen (1988) conducted a factor analysis, and identified five different climates, amongst which a caring climate, a law and code climate and a rules climate.
According to their research, a rule environment is a work climate that has a local locus of analysis, and ethical criterion defined by principles (Victor & Cullen, 1988). This means that ethical reasoning originates from sources within the organization, and that right and wrong is determined by principles (Victor & Cullen, 1988). As such, a rule environment includes clear policies, which provide meaning and clarity (Finkelstein & Hambrick, 1990; Victor & Cullen 1988). Thereby, a rule environment decreases ambiguity (Victor & Cullen, 1998). Unless employees know how management wants them to behave and how to meet these expectations, they cannot behave in this way (Cullen, Victor & Stephens, 1989). Therefore, the primary function of a rule environment is to influence and guide behaviour (Pelaez & Moreno, 1998). As a result, within the organization, it leaves employees with less discretion to do their jobs (Bosson, Johnson, Niederhoffer & Swann, 2006; Davis & McLeod, 2003; Finkelstein & Hambrick, 1990). In other words, a rule environment will provide employees with less freedom to engage in specific behaviours.
According to Higgs (2009), the ability of leaders to engage in bad behaviour rises from their positional power. Leaders need a certain amount of discretion to perform their tasks (Kaiser & Hogan, 2007). However, when there is an absence of clear rules, leaders can take on authority (Vroom & Jago, 1974), undermine existing rules and take over (Padilla, Hogan & Kaiser, 2007). Therefore, we argue that these unclear environments would suit narcissists well in their search for indefinite power.
Nevertheless, when a rule environment is established, checks and balances are in place (Padilla et al., 2007). As a result, narcissistic leaders need to adhere to the organization’s rules
14 (Victor & Cullen, 1988), as this is perceived to be the correct or ethical way of working in the organization (Pelaez & Moreno, 1998; Victor & Cullen, 1988). More specifically, in order to be perceived as successful, employees will have to go by the book or strictly obey the company policies (Victor & Cullen, 1988). Because of the fact that a rule environment informs members how they should and should not behave and interact (Victor and Cullen, 1988), a rule environment limits a leader’s discretion, power, and freedom to engage in specific behaviours (Pelaez & Moreno, 1998).
More specifically, a rule environment establishes ethical or correct ways of working, which influences employees’ ethics-related attitudes and behaviours (Treviño, Weaver & Reynolds, 2006). Thereby, adherence to the organization’s rules will limit a leader’s ability to engage in unethical behaviour, since unethical behaviour stems from individual, organizational, and relationship characteristics (Brass, Butterfield & Skaggs, 1998; Kish-Gephart, Harrison & Treviño, 2010; Treviño, Weaver & Reynolds, 2006). Abusive supervision is argued to be a form of unethical leadership behaviour (Liu, Liao & Loi, 2012; Brown & Mitchell, 2010). Arguably, a rule environment will diminish the narcissistic leader’s ability to engage in abusive leadership behaviour.
As a result, we predict that a rule environment influences the extent to which the narcissistic leader can become abusive. This moderation is captured in Hypothesis 3, which states that:
Hypothesis 3: a rule environment moderates the relationship between narcissism and abusive leadership, such that the relationship is weaker when a rule environment is higher.
2.4 Competitive culture
Another situational factor present in organizations is culture. At first sight, organizational culture and climate might seem very similar. However, they should be treated
15 as two distinct, yet related constructs (Tesluk, Farr & Klein, 1997). Climate refers to a situation, so it is subjective, temporal, and may be manipulated by people with power. Culture, on the other hand, refers to an evolved context, meaning that it is rooted in history (Denison, 1996). As such, it is generally argued that culture represents the underlying context, while climate refers to the manifestation of this underlying context (Denison, 1996; Tesluk et al., 1997). Therefore, it is important to incorporate both concepts, as they complement each other (Tesluk et al., 1997).
In literature, culture has been used to refer freely to anything related to principles, values, norms, philosophy, and management style. However, it is defined as “the accumulated learning that a given group has acquired during its history” (Schein, 1990, p. 63). Social identity theory argues that people differ in their tendencies to align behaviour with their personality, and one critical contextual factor enhancing or weakening the alignment of identity with behaviour is the group culture (Hogg & Terry, 2000). As a result, culture is a highly relevant situational factor to include in this research.
Organizations are seen as ‘little societies’, displaying their own culture (Allaire & Firsitoru, 1984). Evidence is found that differences between employees within the same firm across borders can be used as an estimate for the differences in national culture because they share the same corporate values (Hofstede, 1980). Culture can thus be seen as a characteristic of the organization but is displayed in and measured by looking at (non)verbal behaviour of individuals (Hofstede, 1998). In other words, culture results in an automatic pattern of perceiving, feeling, thinking, and behaving (Schein, 1990). As such, organizational culture can influence individual behaviour.
Several different types of organizational cultures exist (Brown, Cron & Slocum, 1998; Schein, 1990), one of which is a competitive culture. In literature, a competitive culture is referred to as “the degree to which employees perceive organizational rewards to be
16 contingent on comparisons of their performance against that of their peers” (Brown, et al., 1998, p. 89). Excellence in such a culture is defined as someone’s performance in relation to the performance of others (Stanne, Johnson & Johnson, 1999). A competitive culture is characterized by perceptions of differences in reward distribution, while at the same time making performance and status comparisons to others (Fletcher, Major & Davis, 2008). Moreover, it represents a win-lose perspective, meaning that only one can win, at the cost of others (Stanne et al.,1999). According to Fletcher, Major and Davis (2008), when an organization has a highly competitive culture, employees will value outperforming their colleagues before they can be recognized or rewarded. As such, there will be an emphasis on performance and a competition to be the best to receive rewards, recognition, and power.
Since narcissistic leaders have to operate within the organizational culture of the company they work for, this environment is likely to determine whether the leaders’ narcissistic traits can flourish through (abusive) behaviour or might be more hidden (Van Fleet & Griffin, 2006). For the reason that narcissistic traits fit with the competitive culture, in line with trait activation theory, we expect that these traits will prosper in such an environment and be activated to be expressed in behaviour (Tett & Gutermann, 2000).
Narcissists’ have a great need for power, recognition and domination (Rosenthal & Pittinsky, 2006), as well as a desire to get ahead of others (Roberts & Robins, 2000). This fits a competitive culture, as this culture encourages competition for recognition (Fletcher et al., 2008). As such, it is likely that the narcissistic leader participates in the competition, in order to get recognition and maintain their inflated self-image (Baumeister et al., 2000). For example, in order to achieve a higher status in the company, the narcissistic leader will have to compete with fellow supervisors to achieve a high performance ranking. As a result, the narcissistic leader is more likely to engage in behaviours such as externally blaming failure, as
17 well as inappropriately taking credit for subordinates’ work. Hereby they openly expose their narcissistic character, engaging in abusive leadership behaviour.
In line with the argumentation above, we argue that competitive culture has a moderation effect on the relationship between narcissism and abusive leadership behaviour. This moderation is captured in the following Hypothesis:
Hypothesis 4: a competitive culture moderates the relationship between narcissism and abusive leadership, such that, the relationship is stronger when a competitive culture is higher.
2.5 The mediating role of abusive leadership
Following the argumentation in the previous sections, combining all the information, we propose that the interaction between narcissism and rule environment, as well as the interaction between narcissism and competitive culture affects abusive leadership, which in turn affects OCB.
When a rule environment establishes clear rules and regulations, to which leaders need to adhere (Pelaez & Moreno, 1998; Victor & Cullen, 1988), narcissistic leaders are limited in their ability to engage in abusive leadership behaviour. This is due to the fact that rules limit the narcissistic leader’s discretion to engage in various abusive leadership behaviours (Bosson, Johnson, Niederhoffer & Swann, 2006; Davis & McLeod, 2003; Finkelstein & Hambrick, 1990).
In contrast, when a competitive culture is present, the organization encourages competition for recognition (Fletcher et al., 2008). This fits the narcissistic character traits, such as the need for power, recognition and domination (Rosenthal & Pittinsky, 2006), as well as a desire to get ahead of others (Roberts & Robins, 2000). Therefore, in accordance with trait activation theory, these traits are more likely to be openly expressed in abusive leadership behaviour (Tett & Gutermann, 2000). In other words, in organizations with a rule
18 environment, narcissistic leaders are less likely to engage in abusive leadership, whereas in organizations with a competitive culture, narcissistic leaders are more likely to engage in abusive leadership.
In those situations in which the narcissistic leaders have the ability to express their character traits openly and engage in abusive leadership behaviour, the level of abusive leadership will be inflated. Followers will respond to this by withdrawing OCB, because they feel no obligation to respond favourably to the leader, try to restore their sense of autonomy or to get even (Aryee et al., 2007; Wright & Brehm, 1982; Xu et al., 2012; Zellars et al., 2002).
Thus, as reflected in our overall theoretical model illustrated in Figure 1, we propose an indirect moderated mediation (Edwards & Lambert 2007; Preacher, Rucker & Hayes, 2007). This relationship is captured in Hypothesis 5.
Hypothesis 5: Narcissism is related to OCB via conditional indirect effects, such that the interaction between narcissism and rule environment, and the interaction between narcissism and competitive culture, is related to abusive leadership, which in turn is related to OCB.
3. Method
Having introduced the theoretical background of all the concepts included in the conceptual model, the following chapter marks the beginning of the empirical part of this study. This study was conducted to gain insight into the relationship between narcissism, abusive leadership behaviour and OCB and the potential moderating effect of rule environment and competitive culture on the relationship between narcissism and abusive leadership. In order to test this, this study used a cross-sectional, correlational research design for theory testing. In the following section, the methods used will be explained thoroughly. More specifically, the data collection procedure, the sample, and the measurement of the
19 variables is extensively discussed. Hereafter, a brief description is given of the statistical approach taken to test for the hypothesized relationships.
3.1 Sample and procedure
Data for this study was acquired as part of a larger project designed to examine dark and bright leader traits and their influence. For this project, two different surveys are used to collect cross-sectional data: one for supervisors and one for subordinates. Since subordinates or supervisors may not have all the information by themselves, by collecting information from both sources, more information can be obtained. Furthermore, common method bias is reduced by measuring from multiple perspectives (Podsakoff, MacKenzie & Podsakoff, 2012). The surveys were administered digitally via the online survey software Qualtrics. The survey was administered in Dutch only, to rule out cultural influences (Martinko et al., 2013).
The population of interest in this study is Dutch supervisors and subordinates. To make sure that relationships are strongly present, only subordinates who work more than 32 hours per week are selected to participate in this study. Since the population is very large and there is no sampling frame available, non-probability convenience sampling is used. In accordance with network and snowball sampling, participants are obtained by making use of the researchers’ personal networks and by referrals (Biernacki & Waldorf, 1981; Nadler, Bartels, Naumann, Morr, Locke, Beurskens & Ginder, 2015). Every researcher involved in the project gathers dyads of supervisors and subordinates. This should ensure a diverse sample (Field, 2009).
All participants were approached in a two-step way. First, they received an email with a request for help and some general information regarding the study. After they had agreed on filling in the survey, an official invitation was sent via email including the link to the survey and a unique dyadic code. These codes were later used to link the supervisor and subordinate
20 answers to form a dyad. During the administration of the survey, the response rate is closely kept track of. To ensure a high response rate, the survey has been sent out mentioning sponsorship of the University of Amsterdam (Fox, Crask & Kim, 1988). Moreover, respondents received several reminders (Cook, Heath & Thompson, 2000; Fan & Yan, 2010). Out of the 155 dyads approached for participation in this study, 123 followers and 114 leaders completed the survey. This results in a follower response rate of 79.35% and a leader response rate of 73.55%. In the end, 102 dyads were complete, giving an overall response rate of 65.81%. This is fairly high, as web surveys tend to yield a low response rate (Cook et al., 2000). The average response rate of 39.6% found by Cook et al. (2000) may generalize to web-based surveys when internet penetration in the population is high, which is the case in our sample (Dillman, 2000).
Nonetheless, due to the removal of 5 outliers and listwise deletion of missing values, in the end, our correlation table includes 84 dyads, whereas our statistical analysis includes 96 dyads. Even though listwise deletion of missing data is not an optimal strategy, because it can result in a significantly smaller sample size (Acock, 2005; Schlomer, Bauman & Card, 2010), we choose this method to equalize sample sizes across the analyses to enhance comparability.
In the end, our sample contains a diverse set of subordinates and leaders working in different sectors. The majority of the participants work in education, healthcare, or the financial sector. The leaders are 21 to 62 years old (M = 41.53; SD = 10.58), while the followers’ age ranged from 18 to 61 years (M = 33.10; SD = 12.49). A small majority of the followers were female (51%), while the leader sample contains more male participants (52.9%). All in all, the gender division of the sample is balanced. Additionally, most of the participants have obtained a diploma at the University of applied sciences (‘Hoger
beroepsonderwijs’, HBO, HTS, HEAO). Moreover, the leaders had spent 4 months up to 40
21 tenure of 1 month to almost 40 years (M = 6.40; SD = 9.18). Finally, the leader and follower, on average, had spent 5 months working together (SD = 3.03), and have frequent contact with 56.7% reporting daily, 38.5% weekly, and only 4.8% reporting monthly contact.
3.2 Measures
Multiple constructs were included in the broad surveys, but this paper focuses on five variable constructs derived from the survey: abusive leadership, OCB, narcissism, rule environment, and competitive culture. Of these constructs, OCB and narcissism were rated by the leader, while abusive leadership, rule environment, and competitive culture were measured from the subordinate’s perspective. All variables were measured using existing, validated 7-point Likert scales (1 = Strongly disagree; 2 = Mostly Disagree; 3 = Somewhat Agree; 4 = Neither Agree nor Disagree; 5 = Somewhat Agree; 6 = Mostly Agree, 7 = Strongly Agree), as this is the preferred scale format (Dawes, 2012).
All scales were directly translated into Dutch to ensure better participant understanding, and avoid bias resulting from interpretation. The original English and the translated Dutch items are included in Appendix A and B respectively.
3.2.1 Abusive leadership
Mitchell and Ambrose (2007) developed a shortened, 5-item scale based on the original 15-item abusive supervision scale by Tepper (2000). They conducted exploratory and confirmatory factor analysis on two data sets using the original scale by Tepper (2000), discovering two factors: active and passive abusiveness. Mitchell and Ambrose’s (2007) scale measures the active interpersonal acts of abuse. This active scale reflects best our definition since this study looks at expressions of abusive leadership behaviour. Therefore, this scale is used. Example items are my supervisor ridicules me, and my supervisor makes negative
comments about me to others. The scale proved reliable (Cronbach’s α = 0.862), therefore, the
22
3.2.2 OCB
OCB is measured using the combined 13-item scale of Van Dyne and Le Pine (1998) and MacKenzie, Podsakoff and Fetter (1991). It reflects the degree to which the employee engages in voluntary, positive behaviours that help the organization to achieve its goals. Seven items measure challenging OCB, which includes behaviour that emphasizes ideas and issues. An example item is this particular co-worker communicates his/her opinions about
work issues to others in this group even if his/her opinion is different and others in the group disagree with him/her. The remaining six items reflect affiliative OCB, that is, interpersonal
and cooperative behaviours. An example reads this particular co-worker considers the impact
of his/her actions on others. The combination scale proved reliable (Cronbach’s α = 0.767),
thus answers were averaged into one score.
3.2.3 Narcissism
Narcissism is assessed using the NPI-13 by Gentile, Miller, Hoffman, Reidy, Zeichner and Campbell (2013), which includes 13 items. The NPI-13 was developed as a shorter variant of the larger NPI-40 scale, with the goal to preserve the three-factor structure underlying the NPI-40 (Gentile et al., 2013; Lee & Sung, 2016). The subfactors include leadership/authority (LA), grandiose exhibitionism (GE), and entitlement/exploitativeness (EE).
LA reflects the narcissist’s belief that he or she has an extraordinary leadership ability, and his/her preference for roles that involve authority (Hansbrough & Jones, 2015). As such, this factor incorporates social influence, self-perceived leadership ability and, to some extent, dominance (Ackerman, Witt, Donnellan, Trzespiewski, Robins & Kashy, 2011). Secondly, GE is comprised of self-love, superiority, exhibitionistic tendencies, and vanity (Ackerman et al., 2011). Furthermore, it involves a constant need to be the centre of attention (Carpenter, 2012). Lastly, the EE subscale captures beliefs and behaviours related to social contexts
23 (Ackerman et al., 2011; Brown, Budzek & Tamborski, 2009). This interpersonal subdimension includes manipulative intent, taking advantage of others and the sense of entitlement to respect, the so-called socially toxic elements of narcissism (Ackerman et al., 2011).
As such, the NPI-13 allows for the computation of both total and subscale scores of narcissism (Gentile et al., 2013). In addition, this study applies the scale, because the NPI-13 is a brief measure (Lee & Sung, 2016). Furthermore, we follow the recommendation of Wisse and Sleebos (2016) to repeat their research investigating narcissism as an antecedent of abusive leadership behaviour using the NPI scale to measure narcissism. The scale is reliable, (Cronbach’s α = 0.893), and thus answers were averaged across the 13 items. An example item is I am a born leader.
3.2.4 Rule environment
Rule environment was measured using Victor and Cullen’s (1988) 4-item measure. This measure reflects the degree to which ethical reasoning stems from sources within the organizations, and the extent to which right and wrong are determined by principles. Examples include successful people in this company go by the book, and everyone is expected
to stick by company rules and procedures. The Cronbach’s alpha equals 0.838, thus answers
were averaged into one score.
3.2.5 Competitive culture
Competitive culture is measured using the measure as developed by Brown, Cron and Slocum, 1998, which was developed specifically for use in a sales industry. To make it more general and applicable to multiple industries we adapted it consistent with the research of Fletcher, Major and David (2008). The original scale for example includes the item the
amount of recognition you get in this company depends on how your sales rank compared to other salespeople, while our research uses the adaptation of Fletcher, Major and David
24 (2008), where the item states the amount of recognition you get in this company depends on
how you perform compared to others. Another example of the adaptation is everybody is concerned being the top performer, which in the original scale is phrased as everybody is concerned with finishing at the top of the sales rankings. The adapted scale had a reliability of
α=0.792, so a competitive culture scale was calculated by averaging the answers across the five items into one score.
3.2.6 Control variables
The surveys also included a number of questions concerning the participants’ demographics, such as gender, age, tenure and education level. Out of these demographics, based on theory, two potential control variables were identified: duo tenure, and leader gender. Narcissists’ popularity is said to decrease over time (Leckelt, Küfner, Nestler & Back, 2015). As a result, the abusive behaviours of a narcissistic leader may only become apparent after a certain period of time. Therefore, dyads with a short tenure may not display the relationship between narcissism and abusive leadership yet. However, incorporating the variable duo tenure as a control variable in our analyses made only a marginal contribution of 1% increase in variance explained in our model. Therefore, analyses were continued without the incorporation of duo tenure as a control variable.
Furthermore, leader gender has also become a common control variable in research on narcissism (Grijalva et al., 2015; Nevicka et al., 2011), since men tend to be more narcissistic than women (Tschanz, Morf & Turner, 1998). In this study, leader gender does add variance explained to our model. Therefore, this variable is included as a control variable throughout our analyses.
25 3.3 Statistical procedure
The statistical procedure consists of multiple steps. First, we checked the data set for all the standard requirements, such as normality, reliability, and multicollinearity. All VIF values were below the threshold of 5, which indicates that there are no problems with multicollinearity (Craney & Surles, 2002). However, not all variables were normally distributed, so we trimmed the data by removing 5 outliers based on the principle of Z-scores higher than 3 or lower than minus 3 (Sincich 1986). As a result, all variables were normally distributed, except for the variable Abusive leadership. This variable reported a low mean, consistent with other studies (Mitchell & Ambrose, 2007; Tepper, 2000). This is can be explained through literature, as abusive supervision is said to be a low base-rate phenomenon (Mitchell & Ambrose, 2007; Tepper, 2000). We conducted transformations for positively skewed data using log, square root and inverse computations. However, none of the transformations solved the skewness problem. Therefore, we proceeded with the original variable, but used bias-corrected confidence intervals from bootstrapping for Hypotheses testing as a means to avoid a biasing problem (Tepper, Carr, Breaux, Geider, Hu & Hua, 2009; Vogel, Mitchell, Tepper, Restubog, Hu, Hua & Huang, 2015), since the Bootstrapping method allows for violation of the normality assumption (Field, 2013).
Hereafter, the Hypotheses were tested using hierarchical regression analysis. The first step involved testing only the direct effects as proposed in the first two Hypotheses, where after both moderators were added to the equation, consistent with Hypothesis 3 and 4. Lastly, the indirect conditional effect was tested as proposed in Hypothesis 5.
26
4. Results
The following section presents the results of the data analysis. First of all, the general data requirements and correlation matrix are discussed. Hereafter, the steps taken to test the proposed Hypotheses will be explained thoroughly. They will be tested in chronological order using linear regression analysis, and the SPSS macro PROCESS to probe further into the interaction effects (Hayes, 2013).
4.1 Descriptive statistics
Means, standard deviations, and correlations among the variables are presented in Table 1. We see a significant, positive correlation between follower age and follower tenure (r= 0.753; p < 0.01) This correlation makes sense, as older employees will report a longer working life, and thus have a higher tenure. Contrastingly, there is no significant correlation between leader age and leader tenure.
Interestingly, the table also shows some unexpected significant correlations. Follower gender and leader gender show a significant, positive correlation (r= 0.367; p = 0.001). This could imply that when the leader of the dyad is a woman, the follower is also more likely to be female. Furthermore, leader age and follower age show a positive, significant correlation (r= 0.496; p < 0.01). Arguably, when the leader is older, the follower is also older. As a result, we also see a significant positive correlation between leader tenure and follower tenure (r = 0.251; p = 0.021). Moreover, the table shows a significant positive correlation between leader tenure and follower age (r = 0.269; p = 0.013), and between follower tenure and leader age (r = 0.455; p < 0.01).
However, we must not forget that spurious correlations can occur, where two constructs appear to be related but are in fact unrelated (Brett, 2014). Spurious correlations do not reflect properties of the underlying data but are a result of data transformations (Pearson, 1987). In this case, it could be the result of selection mechanisms, as for this study either the
27 leader or the follower was approached, and they themselves selected the other half of their dyad. In this sample, female participants were more likely to ask a female to complete their dyad. Additionally, older participants were more likely to select an older subordinate or supervisor.
In addition, it is important to look at correlations with the variables included in our model. In line with previous research, Table 1 shows a significant, negative correlation between narcissism and leader age (r = -0.275, p = 0.011) (Wisse & Sleebos, 2016). However, in contrast to previous studies, no significant negative relationship between narcissism and leader gender was found (r = 0.000, p = 0.997) (Foster, Campbell & Twenge, 2003; Nevicka, et al., 2011; Owens, Wallace & Waldman, 2015).
Nonetheless, surprisingly, a significant, positive correlation was found between narcissism and competitive culture (r= 0.220; p = 0.045). This could imply that narcissists are attracted to organizations with a competitive culture, as this culture suits their characteristics. Person-Environment (PE) fit theory supports this claim. This theory argues that individuals tend to seek and create work environments compatible with their personalities (Chatman, 1989; Holland, 1997; Su, Murdock & Rounds, 2015). As a result, there are more narcissistic employees in a competitive culture environment.
28
Table 1 – Correlation matrix
Variables M SD 1 2 3 4 5 6 7 8 9 10 11 1. Follower Gender ¹ 1.54 0.50 - 2. Follower Age ² 33.10 12.49 -0.114 - 3. Follower Tenure ² 6.44 9.26 -0.166 0.753** - 4. Leader Gender ¹ 1.45 0.50 0.367** -0.038 -0.001 - 5. Leader Age ² 41.73 10.58 0.024 0.496** 0.455** -0.151 - 6. Leader Tenure ² 0.31 0.26 -0.090 0.269* 0.251* 0.071 0.167 - 7. Narcissism 4.02 0.89 -0.038 -0.170 -0.200 0.000 -0.275* -0.135 (0.89) 8. Abusive leadership 1.29 0.49 -0.135 -0.050 -0.060 -0.046 -0.196 -0.105 0.164 (0.86) 9. OCB 5.14 0.60 0.209 -0.080 -0.073 0.119 0.202 -0.040 0.001 0.055 (0.77) 10. Rule environment 4.44 1.27 -0.041 -0.017 -0.008 0.028 -0.092 -0.007 0.035 0.097 0.149 (0.84) 11. Competitive culture 3.57 1.09 -0.040 0.062 0.044 -0.179 -0.009 -0.037 0.220* 0.157 0.056 0.163 (0.79) Note: Listwise N=84
** Correlation is significant at the 0.01 level (2-tailed) * Correlation is significant at the 0.05 level (2-tailed)
¹ Gender (both for leader and follower) was coded as (1) = male, (2) = female ² Age and tenure (both for leader and follower) in years
29 4.2 Hypotheses testing
After careful examination of the correlation table, we turned to Hypotheses testing by means of hierarchical regression analysis in three steps.
4.2.1 Step one – Hypothesis 1 and 2
In step one, we investigated the direct effects between narcissism and abusive leadership, and between abusive leadership and OCB as proposed in Hypotheses 1 and 2 respectively. To test these proposed relationships, a simple linear regression analysis was conducted, including leader gender as a control variable. Results are included in Table 2 to 6.
Table 2 indicates that narcissism (β = 0.064, p = 0.291) and leader gender (β = -0.099, p = 0.355) are not significantly related to abusive leadership, and together do not explain a significant amount of variance in abusive leadership (R² = 0.021, p = 0.367). Thus, Hypothesis 1 is not supported, which means that narcissism is not positively related to abusive leadership.
In order to find out whether this insignificant relationship between narcissism and abusive leadership is true for all three subcomponents of narcissism, we probed further into this proposed relationship by conducting regression analysis for the three subscales of narcissism: Leadership/Authority (LA), Grandiose Exhibitionism (GE), and Entitlement/ Exploitativeness (EE). Results are included in Table 3 to 5 respectively.
First of all, Table 3 indicates that for the model including LA narcissism, LA narcissism (β = -0.092, p = 0.217) and leader gender (β = -0.099, p = 0.355) do not explain any significant variance in abusive supervision (R² = 0.021, p = 0.362). Additionally, this subcomponent does not show an increased amount of variance explained for abusive leadership compared to the total scale of narcissism. Thus, the subcomponent LA narcissism is not positively related to abusive leadership.
30 Secondly, the model including GE narcissism is not statistically significant (F (2.428) = 1147, p = 0.322), as indicated in Table 4. Together, GE narcissism (β = 0.076; p = 0.186) and leader gender (β = -0.094; p = 0.450) do not significantly explain variance in abusive supervision (R² = 0.023). Consequently, GE narcissism is not related to abusive leadership.
Lastly, turning to EE narcissism, Table 5 shows that EE narcissism (β = 0.058, p = 0.353) and leader gender (β = -0.087, p = 0.383) together explain the insignificant amount of 1.4% of variance in abusive leadership (R² = 0.014, p = 0.505). This is the lowest of all three models regarding subcomponents of narcissism. Thus, EE narcissism is not positively related to abusive leadership. Neither the global model incorporating the total scale of narcissism nor the three models including the subscales of narcissism were significant. Therefore, we conclude that narcissism is not positively related to abusive leadership.
Hereafter, we turned to Hypothesis two, which proposed that abusive supervision is negatively related to OCB. Results in Table 6 show that abusive leadership (β = -0.081, p = 0.477) and leader gender (β = 0.095, p = 0.355) are not significantly related to OCB. Together they do not explain any significant variance in OCB (R² = 0.01, p = 0.623). Accordingly, Hypothesis 2 is rejected, which implies that abusive leadership is not negatively related to follower OCB.
4.2.2 Step two – Hypothesis 3 and 4
In step two, the moderators rule environment and competitive culture were included, as argued by Hypotheses 3 and 4. Hypothesis 3 proposed that a rule environment moderates the relationship between narcissism and abusive leadership, such that the relationship is weaker when a rule environment is higher, whilst Hypothesis 4 argued for a moderating effect of competitive culture on the relationship between narcissism and abusive leadership, such that when a competitive culture is higher, the relationship is stronger. To test these two
31 Hypotheses, the SPSS macro Process was used (template 2). All independent variables were mean centred before conducting the analyses (Aiken & West, 1991).
The results in Table 7 indicate that the model including both rule environment and competitive culture as moderators does not explain any significant variance in abusive leadership (R² = 0.04, F (6.89) = 0.54, P = 0.697). Even though adding the moderators helps to explain an additional 1% of the variance in abusive supervision compared to the model without moderators (ΔR² = 0.010), this increase is not significant (p = 0.415). The moderation model displays a constant of 1.39 (p < 0.01), and an insignificant direct effect of narcissism on abusive leadership (β = 0.08, p = 0.32), consistent with our previous findings.
Turning to our first moderator, rule environment, the results indicate that there is no significant interaction effect between narcissism and rule environment on abusive leadership (β = -0.0039; p = 0.94). Since the model is not statistically significant, Hypothesis 3, stating that a rule environment moderates the relationship between narcissism and abusive
leadership, such that the relationship is weaker when a rule environment is higher is rejected.
Moreover, looking at the second moderator, competitive culture, Table 4 indicates a non-significant interaction effect between narcissism and competitive culture on abusive leadership (β = -0.07; p = 0.19). As a result, Hypothesis 4 is not supported, meaning that a competitive culture does not moderate the relationship between narcissism and abusive leadership, such that the relationship is stronger when a competitive culture is higher.
Probing further into the moderating effect of rule environment and competitive culture on the relationship between a narcissistic leader’s character traits and his or her abusive leadership behaviour, we repeated the analysis in Process as explained above for all three subscales of narcissism. Results of these three additional analyses are included in Tables 8 to 10.
32 To start with, the moderation model incorporating LA narcissism as an independent variable is statistically not significant in explaining variance in abusive supervision (R² = 0.05, p = 0.54). In this model, neither rule environment (β = 0.06, p = 0.26) nor competitive culture (β = 0.00, p = 0.93) show a significant moderating effect on the relationship between LA narcissism and abusive leadership. Thus, rule environment and competitive culture do not moderate the relationship between LA narcissism and abusive leadership.
Secondly, turning to GE narcissism, Table 9 indicates that neither rule environment (β = 0.04, p = 0.35) nor competitive culture (β = -0.02, p = 0.67) show a significant interaction effect on the relationship between GE narcissism and abusive leadership. Overall, the model is not significant in explaining variance in abusive supervision (R² = 0.05, p = 0.55, F(6.91) = 0.84). Again, rule environment does not have a negative moderating effect on the relationship between GE narcissism and abusive leadership, whilst competitive culture does not display a positive moderating effect on this particular relationship.
Lastly, results in Table 10 indicate that the model including EE Narcissism is also not statistically significant in explaining variance in abusive supervision (R² = 0.05, P = 0.54). Rule environment (β = 0.06, p = 0.27) and competitive culture (β = 0.03, p 0.65) do not show significant interaction effects. Thus, rule environment does not moderate the relationship between EE narcissism and abusive leadership negatively, whilst competitive culture does not have a positive moderating effect on the relationship between EE narcissism and abusive supervision.
All in all, both the model incorporating the total narcissism scale and the three models looking at the narcissism subscales show insignificant results. Therefore, both Hypothesis 3 and 4, arguing for the moderating effects of rule environment and competitive culture respectively, are rejected.
