LEADERSHIP AND MANAGEMENT TRACK THESIS
The effects of interaction frequency on the perceptions of narcissistic
leaders.
MSc in Business Administration: Leadership and Management track MASTERS THESIS FINAL VERSION
Author: Darta Rozentale 11364424
Supervisor: Dr. Barbara Nevicka
Statement of originality
This document is written by Student Darta Rozentale 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 ... 2 2. Theoretical framework ... 5 2.1 Leader narcissism ... 5 2.1.1 The impact of narcissistic leaders ... 7 2.2 Abusive leadership and the moderating role of interaction frequency ... 9 2.3 Abusive leadership and follower well-being ... 14 2.4 Abusive leadership and perceived leader effectiveness ... 16 3. Method ... 19 3.1 Participants ... 19 3.2 Procedure ... 19 3.3 Measures ... 20 4. Results ... 22 4.1 Descriptive statistics ... 23 4.2 Test of the Hypotheses ... 25 4.2.1 Testing Hypothesis 1 ... 25 4.2.2 Testing Hypothesis 2 ... 27 4.2.3 Testing Hypothesis 3 ... 30 4.3 Exploratory Analyses ... 33 5. Discussion ... 37 5.1 Strengths, limitations and suggestions for future research ... 42 5.2 Practical implications ... 44 5.3 Conclusion ... 46 6. References ... 47List of tables
Table 1. Correlations matrix 28
Table 2. Hierarchical regression model of narcissistic leadership and abusive
leadership, and direct interaction frequency as a potential moderator
30
Table 3. Hierarchical regression model of narcissistic and abusive leadership,
and indirect interaction frequency as a potential moderator
31 Table 4. The relationship between abusive leadership and employee wellbeing 32 Table 5. The moderating effect of direct interaction frequency between the leader
and follower on the relationship between leader narcissism and employee wellbeing
33
Table 6. The moderating effect of indirect interaction frequency between the
leader and follower on the relationship between leader narcissism and employee wellbeing
34
Table 7. The relationship between abusive leadership and perceived leader
effectiveness
35
Table 8. The moderating effect of direct interaction frequency between the leader
and follower on the relationship between leader narcissism and perceived leader effectiveness
36
Table 9. The moderating effect of indirect interaction frequency between the
leader and follower on the relationship between leader narcissism and perceived leader effectiveness
List of figures
Figure 1. Research conceptual model. 23
Figure 2. Interaction effect of the moderating effect of low self-esteem and leader
narcissism on abusive leadership.
Acknowledgements
First, I would like to thank my supervisor Barbara Nevicka for all the support, and constructive feedback throughout the research process. Secondly, I would like to thank the students in my thesis group that each put their best efforts in collecting the data, and were always available for advice throughout the writing process.
Abstract
Given the rise of narcissism in the society, and the rise of narcissists to leadership positions, there is a growing body of research into understanding the effectiveness of narcissistic leaders and the effect they have on their followers. However, until now the findings have been inconsistent. Additionally, existing literature shows that the darker characteristic of narcissists become more evident after establishing closer relationships with others. Therefore, this study examines extent to which interaction frequency between a follower and a leader influences the relationship between leader’s narcissism and abusive leadership as perceived by the followers. It proposes that followers who interact more often with their narcissistic leader would be more likely experience their leader as abusive, which leads to them being perceived as less effective and decreases follower wellbeing. The research adapted quantitative approach, using a survey with 128 leader-follower dyads. The results indicated that interaction frequency did not moderate the relationship between leader narcissism, abusive leadership, employee wellbeing and perceived leadership effectiveness. However, additional analysis found that self-esteem, could provide an explanation for the currently inconsistent findings as only followers with low self-esteem perceive narcissistic leaders as more abusive, less effective, and have a lower wellbeing. Implications for theory and practice are discussed.
1
1. Introduction
Recent research shows a steadily growing trend of narcissism – an overinflated self-image - in the society (W. K. Campbell, Hoffman, Campbell, & Marchisio, 2011; Twenge & Campbell, 2009). Generation Y, which is increasingly becoming an integrated part of the labour market, is known to be characterised by high individualism and an inflated sense of self-importance (W. K. Campbell et al., 2011; Twenge, Konrath, Foster, Keith Campbell, & Bushman, 2008). Moreover, research shows that narcissists often emerge as leaders because they possess prototypical leadership characteristics (e.g. dominance, confidence, extraversion) (Brunell et al., 2008; W. K. Campbell & Campbell, 2009; W. K. Campbell et al., 2011; T. A. Judge, Piccolo, & Kosalka, 2009; Nevicka, De Hoogh, Van Vianen, Beersma, & McIlwain, 2011). Thus, narcissists are likely to find themselves in positions of power, as can be seen from the president of United States, Donald Trump.
Narcissists’ rise to powerful positions could be problematic to organisations and followers because, even though they possess some positive characteristics (e.g. charisma, confidence) they also possess many “dark side” characteristics such as aggression, a lack of empathy and exploitative behaviours (Rosenthal & Pittinsky, 2006). Due to narcissists’ disagreeableness, arrogance and lack of empathy, it has been argued that leader narcissism might negatively impact follower wellbeing (Maccoby, 2000; Rosenthal & Pittinsky, 2006). Moreover, the dark characteristics of narcissists, such as, aggression, and arrogance, are reflective of behaviours typically associated with abusive leadership - “followers’ perceptions of the extent to which their supervisors engage in the sustained display of hostile verbal and nonverbal behaviours, excluding physical contact” (Zellars, Tepper, & Duffy, 2002, p. 1068). Abusive leadership in turn, has shown to have a consistently negative effect on follower wellbeing – psychological health of the followers, and leader effectiveness (Mitchell &
2 Ambrose, 2007; Tepper, 2007; Zellars et al., 2002). Thus, indicating the potentially detrimental effects of narcissistic leaders not only on the followers, but also organisations.
However, despite the often-theorised negative effects of narcissistic leaders, research still shows inconsistent findings. As previously mentioned, it is argued that leader narcissism has negative effects on follower wellbeing (Maccoby, 2000; Rosenthal & Pittinsky, 2006). Nonetheless, this relationship has not been yet empirically proven. In addition, there is also evidence suggesting that leader narcissism has no adverse effects on follower wellbeing (Volmer, Koch, & Goritz, 2016). The same contradiction is evident in the academic discussion on narcissism and perceived leader effectiveness, where some research has found positive effects (Deluga, 1997), while others argued that there is in fact no effect of leader narcissism on the perceived leader effectiveness as the positive and negative characteristics counterbalance each other (Resick, Whitman, Weingarden, & Hiller, 2009). This, suggests that there might be other variables influencing the relationship between leader narcissism and perceived effectiveness.
One potential explanation could be the duality of narcissism, where in the emergent zone (i.e. starting or short-term) of relationships they shine, because of their charisma, dominance, and extraversion, while in the enduring zone (i.e. more intimate, long-term) of relationships, their lack of empathy, and arrogance becomes more evident, thus leading to more negative perceptions (W. K. Campbell & Campbell, 2009; Leckelt, Küfner, Nestler, & Back, 2015). This is also supported by research in student groups, which shows that the popularity (positive image) of narcissists in their peer groups changes over time, due to the exposure of their antagonistic behaviours, such as aggression, thus indicating that they might be perceived as more abusive (Leckelt et al., 2015; Ong et al., 2016). Extending this to an organizational context where leaders would hold legitimate positions of power, follower familiarity with their leader (i.e. how well a follower knows their leader) could similarly
3 determine the extent to which followers perceive narcissistic leaders as abusive. However, it is important to distinguish between simply the length of acquaintance and the frequency of interaction between a leader and a follower, because the length of the relationship itself might not necessarily be associated with closer and more intimate relationships. Instead it is the repeated exposure to someone’s behaviours that allows us to get to know the other person better and make more accurate perceptions regarding this person’s characteristics (Biesanz, West, & Millevoi, 2007; Leckelt et al., 2015; Ong et al., 2016). Therefore, the current research will examine extent to which interaction frequency between a follower and a leader influences the relationship between leader’s narcissism and abusive leadership as perceived by the followers. I propose that followers who interact more often with their narcissistic leader would be more likely experience their leader as abusive. Secondly, this research will examine how abusive leadership in turn influences the perceived leader effectiveness as rated by followers and follower well-being. In line with prior research (Bono et al., 2007; Keashly et al., 1994; Kuoppala et al., 2008; Mitchell & Ambrose, 2007; Schyns & Schilling, 2013; Wu & Hu, 2009; Zellars et al., 2002) it is expected that abusive leadership has a negative effect on follower well-being, and perceived leader effectiveness.
While some research has been done investigating the declining popularity perceptions of narcissists as a function of time (Leckelt et al., 2015; Ong et al., 2016), the focus in theese articles has been on peer/student- groups and not on narcissists in legitimate leadership positions within organizations. Thus, the current research will add to the existing knowledge not only by proposing a possible explanation for the inconsistent findings regarding the impact of narcissistic leaders on follower well-being and perceived leader effectiveness, but also by doing so in organizational setting. Up to 60% of all lost working days in European Union are caused by work-related stress, the cost of which reaches up to EUR 20,000 million each year (Skakon, Nielsen, Borg, & Guzman, 2010). In addition,
4 research shows that leaders have a significant effect on follower wellbeing (Kuoppala, Lamminpää, Liira, & Vainio, 2008; Skakon et al., 2010; Volmer et al., 2016; Wu & Hu, 2009). Which clearly shows the practical importance of being able to understand the potential variables influencing relationship between narcissistic leaders and follower wellbeing. First, narcissism and prior research thereof will be discussed as well as leader narcissism in particular. Next, the relationship between leader narcissism and abusive leadership will be explained through the effects of interaction frequency, which will be followed by the link between abusive leadership and follower well-being, and abusive leadership and perceived leader effectiveness.
5
2. Theoretical framework
2.1
Leader narcissism
Narcissism was first introduced in Greek mythology, where a young man called Narcissus fell in love with his self-reflection, and eventually perished from being completely self-consumed. Narcissism is generally characterised by inflated-self views, grandiosity and self-love (W. K. Campbell et al., 2011; Deluga, 1997; Grijalva, Harms, Newman, Gaddis, & Fraley, 2015; Rosenthal & Pittinsky, 2006). It appears both in the clinical psychology and social psychology and personality literature. The clinical psychology literature sees narcissism as a personality disorder (NPD), associated with constant desire for admiration, grandiosity and lack of empathy (W. K. Campbell et al., 2011). For someone to be diagnosed with NPD they have to have 5 out of 9 symptoms, and be unable to perform at work or establish relationships (W. K. Campbell et al., 2011). Consequently, there is only a small percentage (6,2% of adults in US (Stinson et al., 2008)) of people diagnosed with clinical narcissism. Therefore, social psychology and personality literature focuses on narcissism from the trait perspective, where it is normally distributed in the population, and is often related to other traits, such as, Machiavellianism and self- esteem (Foster & Campbell, 2007). This research will focus on narcissism from the subclinical perspective, as it offers a possibility to detect more occurrences of narcissistic individuals within organizations.
In addition, most studies on narcissism typically distinguish between two types of narcissism – grandiose and vulnerable. Vulnerable narcissists are characterised by being hostile, anxious and depressive. They possess a mix of high entitlement and low self-esteem (W. K. Campbell et al., 2011). Whereas grandiose narcissism is associated with overconfidence, assertiveness, interpersonal dominance, extraversion, lack of empathy and aggression (W. K. Campbell et al., 2011). As the characteristics of grandiose narcissism overlap with some of the traits typically associated with leaders, it is known to be the
6 primarily focus in organizational context (Watts et al., 2013), while vulnerable narcissism is mainly studied in the clinical context (W. K. Campbell et al., 2011). Furthermore, based on a study on U.S. presidents Watts et al. (2013) also confirms this notion, further arguing that, grandiose narcissists possess traits (e. g. extraversion, assertiveness, dominance) which are often associated with leader emergence. This research will focus primarily on grandiose rather than vulnerable narcissism.
As previously mentioned grandiose narcissists are known to be very charismatic, extraverted and dominant. They have an urge for power, sense of entitlement and self enhancement (W. K. Campbell & Campbell, 2009; W. K. Campbell et al., 2011; Rosenthal & Pittinsky, 2006). Leckelt et al. (2015) note that being liked by others is a fundamental factor in the emergence of relationships, and that positive social interactions correlate with acquiring a high status, and being popular. In these early social interactions, the need for admiration and attention drives narcissists to perform at their best and shine by using their strengths (e.g. charm, assertiveness and dominance). Thus, matching the prototypical leader stereotype and being perceived positively. It is argued that individuals have implicit expectations, ideas and perceptions on the types of traits and behaviours that are inherent in a leader (i.e. leader prototypes), and that these prototypes guide their responses and perceptions of one’s leadership potential (Offermann, Kennedy, & Wirtz, 1994). Narcissistic leaders are known to possess these prototypical leader traits (e.g. extraversion, dominance, and assertiveness), thus they tend to emerge as leaders (W. K. Campbell et al., 2011; Deluga, 1997; Nevicka et al., 2011; Rosenthal & Pittinsky, 2006). With their charisma, and ability to inspire, at least initially, narcissists are able to attract devoted followers, which, in turn feeds the narcissists need for admiration, and boosts their confidence and belief in their own vision (Deluga, 1997; Rosenthal & Pittinsky, 2006). Given that narcissists are more likely to reach
7 positions of power it is important to understand how narcissists function within an organisation, and the effects they have on their followers.
2.1.1 The impact of narcissistic leaders
Narcissistic individuals are known to be very complex, due to the duality of their personality, where positive traits such as vision, charisma, charm, and assertiveness could be beneficial for followers, while aggression, arrogance, and lack of empathy, can have potentially detrimental effects. This is often referred to as the “bright” and “dark” side of narcissistic leaders (Grijalva et al., 2015; Rosenthal & Pittinsky, 2006; Volmer et al., 2016). Maccoby (2000) refers to individuals who possess the “bright side” characteristics as productive narcissists. They possess compelling visions, and have excellent oratory capabilities to communicate these visions, thus inspiring and arousing enthusiasm in their followers (Maccoby, 2000; Rosenthal & Pittinsky, 2006). Furthermore, narcissists’ creativity, innovativeness, vision and inspiration. They are hands on risk takers, willing to do what it takes to get the job done, while focusing on the bigger picture (W. K. Campbell et al., 2011). Inspired by their personal need for power, legacy and glory, the ideas of narcissistic leaders are visionary and intricate (Sedikides, C., & Campbell, W. K, 2017).
On the contrary, the “dark” side of narcissistic leaders is linked to their lack of empathy, arrogance, aggression, and exploitative behaviours (W. K. Campbell et al., 2011; Deluga, 1997; Rosenthal & Pittinsky, 2006; Volmer et al., 2016). They are typically known to only listen to the information they want to hear, and to believe that their ideas and solutions are the best (Maccoby, 2000). Research shows due to their belief that their arguments and ideas are better than others, narcissists tend to discount advice from others, and even more so when they find themselves in positions of power (Kausel, Culbertson, Leiva, Slaughter, & Jackson, 2015). This can be detrimental to organizational performance, as listening to others has shown to be an important aspect of better decision making (Kausel
8 et al., 2015). Moreover, such exploitative behaviours may be detrimental to follower voice (i.e. sharing information, and ideas on ways to potentially improve work processes), as the potential costs of speaking up (i.e. humiliation, loss of social status) outweigh the benefits (i.e. promotion, recognition, status) (Detert & Burris, 2007). Their lack of empathy can lead to the leader being perceived detached and cold, and thus hindering the sense of belongingness in the followers, as they may feel left on their own with their problems (Rosenthal & Pittinsky, 2006; Volmer et al., 2016). In addition, through their arrogance, and exploitative behaviours, narcissistic leaders tend to take credit for their follower achievements and blame them for failure, ignoring the needs of their followers, this in turn may lead to increased stress for the followers and lower their sense of control (Volmer et al., 2016). Furthermore, narcissistic leaders are known to exhibit risk taking behaviours due to their overinflated sense of self, and belief that their decisions are the best, making employees execute tasks that they might not necessarily agree with. Thus, potentially leading to even more job stress for their followers (W. K. Campbell et al., 2011; Rosenthal & Pittinsky, 2006). Narcissists are extremely sensitive, and defensive towards criticism, and tend to be aggressive (Baumeister, Bushman, & Campbell, 2000; W. K. Campbell et al., 2011; T. A. Judge et al., 2009; Maccoby, 2000; Rosenthal & Pittinsky, 2006). This might also be evident in the interactions with followers who might doubt their abilities. Due to their overinflated sense of self, and the willingness to do whatever it takes to push their agenda, and satisfy their own needs fosters behaviours which can be potentially unethical (Maccoby, 2000). Furthermore, research shows that leaders can influence their followers through modelling appropriate behaviours, and that they are the ones who with their actions communicate which behaviours are acceptable and which are not (Anderson, Baur, Griffith, & Buckley, 2016). As a result, narcissistic leaders may foster environments where followers perceive hostility and unethical behaviours as acceptable, creating a toxic work environment.
9 Despite the consistent findings regarding the positive link between narcissism and leader emergence (Grijalva et al., 2015), the outcomes of narcissistic leaders (e.g. follower well-being and leader effectiveness), remain inconsistent. The same inconsistencies are evident in assessing the perceived leader effectiveness of narcissistic leaders. Where some research finds positive (Deluga, 1997), and some no effects of narcissism on perceived leader effectiveness (Resick et al., 2009). This is often associated with the duality of the narcissistic leader personality. Furthermore, these inconsistencies indicate that there might be other variables influencing this relationship. While it would be expected that wellbeing of followers suffers, so far it has been found that narcissists have no negative effect on wellbeing. Volmer, Koch, and Goritz (2016) found that narcissistic leaders in fact do not have any adverse effects on follower well-being. They suggest that this may be associated with the narcissists’ need for continuous positive feedback. As they relate more with individuals who have a high self-esteem, narcissistic individuals might be motivated to preserve their positive self-image in the eyes of followers, by promoting them. As a result of the need for constant positive feedback narcissistic leaders might be inclined to increase the loyalty and job satisfaction of their followers (Volmer et al., 2016).
Given the number of toxic characteristics related to narcissistic leaders the author will examine the link between narcissistic leaders and abusive leadership, but also consider an important moderation of interaction frequency.
2.2
Abusive leadership and the moderating role of
interaction frequency
Abusive leadership is defined as the subjective perceptions of followers, on the extent to which their leaders show sustained hostile behaviours (this definition excludes physical aggression) (Tepper, 2007). There are few important points that need to be highlighted to fully understand the domain of abusive leadership. First, it is based on a subjective perception of the followers, and their observations on the behaviours exhibited by their supervisor.
10 Second, abusive leadership is an intentional behaviour, with a purpose to cause harm. Finally, the display of such behaviours has to be continuous (Tepper, 2007). Typical behaviours associated with abusive leaders include angry outbursts, blaming followers, to save themselves embarrassment, taking credit for follower successes, being rude to followers etc. (Keashly, Trott, & MacLean, 1994; Tepper, 2000, 2007).
Consequentially, abusive supervision has a negative effect on perceptions of justice, it promotes alienation and feelings of helplessness and frustration (Mitchell & Ambrose, 2007; Tepper, 2000). Furthermore, abusive leadership is directly linked with lower follower life and job satisfaction, psychological distress, and higher levels of turnover (Mitchell & Ambrose, 2007; Tepper, 2000, 2007; Wu & Hu, 2009; Zellars et al., 2002). Abusive leadership is also shown to have negative effects on organizational citizenship behaviour (OCB) (i.e. discretionary extra role behaviours, which foster organizational effectiveness) (Zellars et al., 2002). Abused followers tend to feel frustration and a loss of control over the situation. Thus, they try to restore this sense of control by “punishing” their abusive supervisors, and choosing not to engage in OCB, or even exhibiting counter-productive work behaviours (Zellars et al., 2002). In addition, data shows that about 13.6% of US followers are affected by abusive supervision, which through absenteeism, lost productivity, and health care costs is estimated to cost the corporations $23.8 billion annually (B. J. Tepper, Duffy, Henle, & Lambert, 2006). Thus, clearly indicating, the practical importance of detecting and controlling acts of abusive supervision.
Displaced aggression (i.e. “aggression towards an innocent victim uninvolved in the initial provocation”) (Martinez, Zeichner, Reidy, & Miller, 2008), due to a psychological contract breach (i.e. mutual expectations of inputs and outcomes between an employer and follower) is often mentioned as one of the key antecedents of abusive leadership (Keashly et al., 1994). Tepper (2007), found that when the leader perceived that there has been a
11 psychological contract breach (i.e. the leader did not receive what she/he expected, or that given promises were not met accordingly) she/he was more likely to be abusive (Tepper, 2007). This behaviour, links back to narcissistic individuals, who believe they are entitled to respect, admiration, and resources because they are superior to others, and often over evaluate their capabilities, expecting others to praise them (Rosenthal & Pittinsky, 2006). Furthermore, narcissistic leaders often attribute success entirely to themselves, and fail to acknowledge the contribution of other followers, thus expecting higher rewards from their employer (Skakon et al., 2010). Thus, they are more likely to perceive a breach of psychological contract and act abusively. Moreover, research shows that when receiving negative feedback narcissists use aggression as a means of self defence mechanism to maintain their positive self-image (Baumeister, Bushman, & Campbell, 2000; Martinez et al., 2008). Furthermore, Twenge and Campbell (2003), found that displaced aggression is highly associated with narcissism. When narcissists face ego threat, their inflated sense of self feels endangered, and their need for social dominance and a lack of concern for others, makes them to take out their aggression, not only on those directly involved in the rejection/ negative feedback, but also third parties (Twenge & Campbell, 2003). In addition, behaviours associated with abusive leaders, such as, blaming followers, to save themselves embarrassment, and taking credit for follower successes are typically associated with the dark side characteristics of narcissistic leaders (W. K. Campbell et al., 2011; Chatterjee & Hambrick, 2007; Rosenthal & Pittinsky, 2006). This shows a clear link between the dark side characteristics of narcissists and abusive leadership.
Nonetheless, as previously mentioned, narcissists also possess a set of “bright” side traits, such as charisma, charm, and assertiveness. Most of the evidence suggests that the bright side characteristics are often present at the emerging stage of relationships, and initially mask the darker and negative characteristics (Biesanz et al., 2007; W. K. Campbell &
12 Campbell, 2009; Leckelt et al., 2015; Ong et al., 2016). Campbell (2011) compared relationships with narcissistic leaders to eating a chocolate cake. Initially you are very excited and satisfied. However, if you have it every day over a longer period of time, the satisfaction wears off and the negative side effects become more evident (Biesanz et al., 2007; W. K. Campbell & Campbell, 2009; Leckelt et al., 2015; Ong et al., 2016). A similar idea is also expressed in the contextual reinforcement model, where the authors argue that narcissistic leaders shine at the emerging zone of relationships (i.e. short- term contexts, and early stages of relationships), while their negative characteristics become very evident in the later stages of interpersonal relationships (W. K. Campbell & Campbell, 2009). Campbell and Campbell (2009) later argue that the costs of narcissism mostly become evident between longer term acquaintances, and in continuing relationships. They define it as the enduring zone of relationships (W. K. Campbell & Campbell, 2009).
This idea is also further supported by recent research on the decrease of popularity of narcissists over time (Leckelt et al., 2015), and the Energy Clash Model (Sedikides, C., & Campbell, W. K, 2017). The findings show that popularity of narcissists decreases over time, through decrease in admiration of narcissistic individuals, and increase in narcissistic rivalry. This is due to the two pathways of narcissistic behaviour, where admiration has a positive effect in the early acquaintance situations, by the display of agentic behaviours (dominance, and assertiveness), while rivalry is related to negative evaluations, through the display of antagonistic behaviours (arrogance, aggression, and untrustworthiness) (Leckelt et al., 2015). With an increasing level of acquaintance, it becomes easier to observe such antagonistic behaviours, where they are more likely to be displayed as the interpersonal contexts become more intimate. Furthermore, the same behaviours may be perceived differently as the level of acquaintance increases. For example, narcissists’ dominance, and confidence may lead them to be perceived as leaders initially. However, after continuous displays of trying to control
13 and dominate every situation, it may potentially frustrate their interaction partners (Leckelt et al., 2015; Ong et al., 2016). Thus, indicating the importance of interaction frequency in the evaluations of narcissistic leaders.
While there is a common assumption in the current literature that over time narcissistic leaders are being perceived more negatively, it is important to distinguish between the level of acquaintance (i.e. captured by interaction frequency and relationship intensity) and simply the length of acquaintance. Only knowing someone for a longer period does not necessarily mean that you have had the chance to observe their behaviours and establish relationship with this person. Research shows that increasing distance between the leader and followers is associated with leaders being perceived more positively (Shamir, 1995). This is because the followers are unable to observe their behaviours and weaknesses in a day to day setting, which in turns leads them to view leaders in a more superficial way, fostering an illusion of leaders being more charismatic (Shamir, 1995). Furthermore, the notion of interaction frequency as a key variable in establishing close relationships is also supported by the social psychology literature (Biesanz et al., 2007). Biesanz et al. (2007) argue that over time, as people become more exposed to behaviours of others in social situations, they form their perceptions about specific someone, rather than relying simply on the first impression. Therefore, the current research proposes that with an increasing interaction frequency, antagonistic and abusive behaviours (i.e. the socially toxic aspects of narcissism) will become more and more evident, undermining the initial impressions of their positive traits. So only those followers who interact with narcissistic leaders frequently would be likely to observe their abusive leadership. Followers with less contact, on the other hand, would be less likely to perceive narcissistic leaders as abusive because they would be less likely to experience the negative side of narcissistic leaders.
14 Thus, in line with research about declining popularity over time, this paper proposes that:
H1: The relationship between narcissistic leadership and abusive leadership will be positive only when the interaction frequency is high and not when the interaction frequency is low.
2.3
Abusive leadership and follower well-being
Up to 60% of all lost working days in European Union are caused by work-related stress, the cost of which reaches up to EUR 20,000 million each year (Skakon et al., 2010). This, indicates not only the theoretical but also practical importance of understanding the factors influencing follower wellbeing. The general definition of well-being is “the condition of being contented, healthy, or successful” (Well-being, 2017). Which can be related to both – physical and mental wellbeing. However, the organizational literature focusing on leader effects on follower well-being considers mainly psychological well-being as the physical wellbeing is less likely to be linked with leader behaviours and practices (Kuoppala et al., 2008; Volmer et al., 2016; Wu & Hu, 2009). The most common constructs linked with work related wellbeing are job satisfaction, stress, and follower turnover intentions (Horn, Taris, Schaufeli, & Schreurs, 2004; Kuoppala et al., 2008; Skakon et al., 2010; Volmer et al., 2016).
There are a multitude of studies showing that leader behaviours, and particular leadership styles (e.g. transformational, transactional, authentic) have an effect on follower well-being (Kuoppala et al., 2008; Skakon et al., 2010; Tepper, 2007; Wu & Hu, 2009; Zellars et al., 2002). In addition, leader-follower relationships are one of the most common stressors in organizations (Tepper, 2000). The existing theory mentions two key reasons for such findings. First, leaders are known to be directly responsible for evaluating follower performance. Thus, during interactions with their supervisor followers might feel increased
15 anxiety and pressure. Second, research shows that autonomy is one of the essential needs of humans. However, the presence of a supervisor may diminish the feeling of autonomy, as followers might feel that they are being controlled and monitored. Such feelings can lead to irritation, and followers constraining their emotions (Bono, Foldes, Vinson, & Muros, 2007). Furthermore, Skakon et al. (2010) found that a specific type of leadership style can directly impact follower well-being either positively (e.g. transformational leadership) or negatively (e.g. abusive leadership), and that leaders have an essential role in creating an environment where followers can experience positive well-being and thrive. This outlines the potentially devastating effects of abusive leadership on follower wellbeing.
Indeed, there is a large body of research linking abusive leadership with negative follower outcomes, follower wellbeing being one of the most prevalently affected outcomes (Bono et al., 2007; Keashly et al., 1994; Kuoppala et al., 2008; Mitchell & Ambrose, 2007; Schyns & Schilling, 2013; Wu & Hu, 2009; Zellars et al., 2002). Reactance theory argues that people have an innate need for having a sense of personal control, whereas abusive leaders tend to express anger towards their followers, and blame them for things that the followers have not done (Mitchell & Ambrose, 2007). As such, this creates feelings of a loss of control which leads to anxiety, helplessness, and frustration (Mitchell & Ambrose, 2007). Moreover, abusive leadership is linked to sustainable detrimental behaviours (e.g. ridiculing followers, not giving followers credit for their work, putting followers down in front of their colleagues etc. (Tepper, 2007), thus facilitating a continuously stressful environment. In line with Hypothesis 1, we expect that the relationship between narcissistic leadership and employee wellbeing will be negative only when the interaction frequency is high and when the followers are more exposed to the abusive behaviour of the narcissistic leader, and not when the interaction frequency is low. Therefore, the current research proposes the following:
16
Hypothesis 2:
H2a: Abusive leadership has a negative relationship with follower well-being.
H2b: The moderating effect of interaction frequency between a leader and a follower on the relationship between narcissism of the leader and follower well-being will be explained through abusive leadership
2.4
Abusive leadership and perceived leader effectiveness
There is a wide array of research focusing on leader effectiveness (Brunell et al., 2008; D. J. Campbell, Bommer, & Yeo, 1993; Deluga, 1997; Lord, Foti, & De Vader, 1984; Tepper, 2000; Yukl, 2012). Leadership effectiveness is related to leader’s ability and contribution to the organizational and individual level goals (Schyns & Schilling, 2013). The current literature outlines several ways of assessing leadership effectiveness, mainly distinguishing between objective (e.g. organizational performance, team performance), and subjective measures (e.g. follower perceptions, 360 feedback ratings) (D. J. Campbell et al., 1993; Yukl, 2012). However, measures like organizational performance, are usually influenced by a multitude of external variables (e.g. economy) out of the leader’s control. Thus, most research focuses on the perceived leader effectiveness, using follower perceptions in assessing effectiveness (Deluga, 1997; Owens, Johnson, & Mitchell, 2013; Resick et al., 2009; Rosenthal & Pittinsky, 2006; Tepper, 2007; Yukl, 2012). Perceived leader effectiveness is explained through the traits approach, which argues that specific behaviours and traits (i.e. neuroticism, extraversion, openness to experience, agreeableness) of the leader will be the main predictors of their effectiveness (D. J. Campbell et al., 1993; House, Spangler, & Woycke, 1991; Judge, Bono, Ilies, & Gerhardt, 2002; Resick et al., 2009; Yukl, 2012). This can be explained through the implicit leadership theory, which argues that individuals have internal attributions of preferred leader personality characteristics, values and leader attributes (Lord et al., 1984). When these attributions are in line with the actual
17 observed leader traits and behaviours, followers will perceive the leader as effective (Lord et al., 1984).
There is an extensive body of literature showing that abusive leadership is negatively associated with perceived leader effectiveness (Brunell et al., 2008; Judge et al., 2002; Mitchell & Ambrose, 2007; Tepper, 2000, 2007; Wu & Hu, 2009; Yukl, 2012; Zellars et al., 2002). Tepper (2007) argues that a leader with a low score on agreeableness would show relatively little concern about the consequences of their behaviour, thus exhibiting more argumentative and conflictive behaviours, and departing from the traits attributed to a prototypical leader, thus being associated with low leader effectiveness. In addition, as previously noted, abusive leaders are typically associated with continuous hostile behaviours towards their followers, thus lowering their perceptions of justice and sense of control (Zellars et al., 2002). A loss of control in turn may cause the followers to see the leader in a negative light, make them feel helpless and disconnected, and perceive the leader as unfair (Zellars et al., 2002). Therefore, as through frequent interaction the followers would be better able to observe the negative traits of narcissistic leaders and perceive them as abusive, the followers would consequently see them also as less effective through this exposure to abusive behaviours. Thus, this study further suggests that the relationship between narcissistic leadership and perceived leader effectiveness will be negative only when the interaction frequency is high and the leader is perceived as abusive, and not when the interaction frequency is low (see full conceptual model in figure 1).
H3a: Abusive leadership has a negative relationship with perceived leader effectiveness.
H3b: The moderating effect of interaction frequency on the relationship between leader narcissism and perceived leader effectiveness will be explained through abusive leadership
18
Figure 1. Research conceptual model.
Narcissistic leadership Abusive leadership Follower well-being Perceived leader effectiveness Interaction frequency H1 H2 H3
19
3. Method
3.1
Participants
The total sample consisted of 128 leader-follower dyads. The leaders' age ranged from 29 to 59 years (M = 39.91 SD = 11.00), they consisted mostly of males (male 70%, and female 30%), and slightly more than a half had a high level of education (51% HBO or university). The leaders had spent from 1 to 27 years in their current position (M = 5.60, SD = 6.65), and they supervised a maximum of 500 followers (M = 37.37, SD = 77.67). The employee sample consisted of respondents ranging from 18 to 63 years in age (M = 35.48, SD= 12.27), mostly female (60%) and the majority had completed higher education (65% Masters, PhD or other post-graduate education). The number of years spent in the current position ranged from only a few months up to 40 years (M = 5.80, SD = 7.69). Finally, the amount of time that the leader and follower had spent working together ranged from less than a year to over 19 years (M = 2.53, SD = 2.83).
3.2
Procedure
The data was gathered by administering 2 different types of online surveys, one for the leaders and one for the followers. The surveys could be completed either in English (74% of respondents) or in Dutch (26% of respondents). The surveys were first distributed to leaders, who each then had to nominate 1, 2 or 3 followers, to whom the follower survey would be forwarded later. Due to the limited time available for data collection, convenience sampling was used in approaching the leaders. The leaders were mainly selected from the network of the 5 students collecting the data. First, an initial request was sent to ask if they would be willing to participate. Second, the email containing all the necessary information and the survey link was sent to leaders, willing to participate. To increase the response rate, the leader/follower pairs were offered a small incentive – 3 pairs had the chance to win a voucher worth 40EU (20 EU each).
20 The survey was sent to 128 leaders, and 97 leaders completed the survey, thus resulting in a response rate of 75.78%. Each leader was provided with an individual code which they had to enter before starting the survey. This code was also sent to the followers, and was later used to match the leader and follower answers. In total the leaders nominated 203 followers, out of which 128 completed the survey, thus resulting in a follower response rate of 63.05%, which is slightly above an average response rate for online surveys, which is usually somewhere between 30% - 40%. This was achieved, by sending several reminders to both, the leaders and followers.
3.3
Measures
Leader narcissism. Leader narcissism was measured by using the NPI 16 item scale
(Ames, Rose, & Anderson, 2006), which is a shortened version of the NPI 40 item scale. This scale is one of the most commonly used tools in trying to capture subclinical narcissism and is based on a self-report measure, thus it was filled in by the leader him of herself. The NPI 16 is a forced choice scale, in which participants must choose which one of two options best applies to them. An example item is: “I think I am a special person” (narcissistic option) versus “I am no better or worse than most people” (non-narcissistic option). The initial scale showed a low reliability (Cronbach's a = .57). To improve the internal consistency of the measure, two items were removed from the initial 16 item scale. This resulted in an improved reliability (a = .60), which, however, was still lower than the acceptable standard value of .70. Therefore, additional tests were run to assure construct validity of leader narcissism. For example, narcissism is known to be correlated with self- esteem (Sedikides, C., & Campbell, W. K, 2017), which was also found to be true in the acquired data set, as narcissism (M = 0.39, SD = 0.18) was significantly correlated with self-esteem (M = 3.33, SD = .41) r = .25,
p < .001. This is similar to what has been found in previous research r = .22, p < .05
21 It appears that the construct validity is sufficient, thus the author will proceed with using the 14-item scale. The narcissism score was calculated by averaging the items, with 1 coded as the narcissistic response and 0 as the non-narcissistic response. Thus, the minimum possible score of the scale was 0, while the maximum score was 1. In the current sample, these were 0 and 1, respectively.
Abusive leadership. Abusive leadership was measured using a 5-item scale adapted
from Tepper (2000) (Mitchell, M, S., & Ambrose, M. L., 2007). It captures a follower’s perception of the leader’s abusive behaviours. It includes 5 statements (e.g. "My supervisor makes negative comments about me to others", "My supervisor tells me I am incompetent"), which the follower must assess based on a seven-point scale varying from 1 = strongly
disagree to 7 = strongly agree. The scale proved reliable, and thus answers were averaged
into one score (a = .91).
Interaction frequency. Interaction frequency was based on 2 items, answered by
followers: the number of direct/ face to face interaction with the leader per week, and the number of indirect interactions with the leader per week. As the type of interaction may affect the level of acquaintance, for example because indirect contact may not necessarily allow the follower to observe leader behaviours, it is important to separate between face-to-face and indirect contact. Both items will thus be used separately.
Perceived leader effectiveness. Perceived leader effectiveness was measured using 3
items (e.g. How capable is your supervisor as a leader?) from a validated scale by De Hoogh, Den Hartog and Koopman (2005). Answers could be given on a 7-point scale, ranging from 1 = “not at all”, to 7 = “very”. The scale proved reliable, so answered were averaged across the three items. (a = .90).
Follower well-being. Follower well-being was initially measured by using three key
22 was measured with 3 items (Todorova, Bear, & Weingart, 2014) (a = .76), turnover intention with 3 items (Adams & Beehr, 1998) (a = .92), and experienced job stress with 4 items (Keller, 2001). However, after preliminary data analysis, the job stress measure showed a very low reliability (a = .39). Therefore, to assure adequate validity, this measure was excluded when computing the well-being scale, which thus consisted of the items to measure job satisfaction, and turn-over intentions. Example items include “Most days I am enthusiastic about my work” (job satisfaction), and “I often think of quitting this job and finding another” (turn-over intentions). Answers could be given on 7-point scales, ranging from 1 = “strongly disagree” to 7 = “strongly agree”, and the reliability over the 6 items was sufficient. Thus, a follower well-being score was calculated after the scales were standardized, and turnover intentions were reverse coded (a = .72).
Controls. Potential control variables include gender, which is a common control used
in leadership research, as the gender stereotypes have shown to influence the perceptions of accepted behaviours of leaders (e.g. females are expected to behave in a more participative and supporting way, while it is more acceptable for male leaders to exhibit more dominant behaviours) (De Pater, Van Vianen, Fischer, & Van Ginkel, 2009). Also, narcissism is correlated with gender, males are typically more narcissistic (Grijalva et al., 2015). Second, the length of relationship between leader and follower will be used as a control variable to assure that it is the frequency of interaction rather than the length of relationship that influences how narcissistic leaders are perceived by their followers.
4. Results
This section will present the results of data analysis, based on the attained sample. First the author will discuss the correlations matrix. Second, the hypotheses will be tested using linear regression analysis, and if significant, PROCESS will be used to further probe
23 the interaction effects (Hayes, 2013). The proposed hypotheses will be tested in chronological order.
4.1
Descriptive statistics
Table 1 shows the most important results of correlations matrix. The analysis showed that leader narcissism was significantly related to the number of hours’ leaders worked per week r = .18, p = .040. Contrary to previous research, the data sample did not show a significant relationship between leader narcissism and gender r = -.09, p = .314. In addition, it is important to look at the correlations with dependent variables, which include abusive leadership perceived leader effectiveness, and follower wellbeing. As expected, abusive leadership was negatively correlated with perceived leader effectiveness r = -.42, p < .001, and employee wellbeing r = -.23, p = .011, which indicates that the variables are mutually related. Furthermore, abusive leadership also showed a significant positive relationship with the tenure of the leader r = .19, p = .026. Finally, employee wellbeing was not only significantly positively related with employee age r = .21, p = .020, and gender r = .19, p = .027, but also negatively significantly related with both follower and leader mother tongue r
= -.27, p < .001. Females (M = 5.79, SD = 1.09) were found to have a higher wellbeing than
24 Table 1. Correlations matrix Variables M SD 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 1. Leader age 39.91 11.00 - 2. Leader gender 1.30 0.46 -.12 -
3. Leader mother tongue 1.13 0.33 -.16 .01 - 4. Leader tenure 5.59 6.65 .45* -.25** -.13 - 5. Leader span 37.37 77.67 .28** -.13 .19* .08 - 6. Follower age 35.48 12.28 .55** -.24** -.09 .21* .22* - 7. Follower gender 1.60 0.49 -.34 .21* -.17* -.04 -.19* -.07 - 8. Follower mother tongue 1.19 0.39 -.15 -.09 .61** .01 .09 -.15 -.06 - 9. Leader narcissism 0.39 0.18 -.07 -.09 -.04 -.04 .17 -.04 -.02 -.07 (.60) 10. Abusive leadership 1.43 0.75 -.37 .40 -.05 .19* -.12 -.09 -.08 -.01 .08 (.91) 11. Follower Wellbeing 5.60 1.16 .11 -.12 -.27** .04 -.02 .21* .19* -.27** .12 -.23* (.72) 12. Leader Effectiveness 5.77 0.95 .02 -.04 .02 -.19* .05 -.16 .03 -.03 .07 -.42** .49** (.90)
Note. N = 128. Gender (both for leader and follower) was coded as (1) = male, (2) = female; nationality (both for leader and follower) was coded
25 Additional analysis showed a significant difference in the wellbeing of followers, when the leaders were Dutch (M = 5.72, SD = 1.10) versus another nationality (M = 4.77, SD = 1.32), t (126) = 3.16, p = .002. The same pattern was also evident for the nationality of the followers, where the Dutch followers (M = 5.75, SD = 1.04) experienced significantly more wellbeing than followers of other nationalities (M = 4.94, SD = 1.45), t (126) = 3.18, p = .002. This could potentially indicate that the Dutch natives experience higher wellbeing overall than the employees from other countries, indicating some cultural influence.
The results of the correlations table were used to determine if extra control variables are needed when conducting the regression analyses. As leader tenure correlates with abusive leadership it will be used as a control in determining if interaction frequency moderates the relationship between leader narcissism and abusive leadership, to exclude it as an alternative explanation. Leader tenure will also be used as a control in analysing the relationship between abusive leadership, leader effectiveness, and testing the full model, as it significantly correlates with the perceived leader effectiveness. Finally, follower age and gender will be used as controls when analysing the relationship between abusive leadership and employee wellbeing.
4.2
Test of the Hypotheses
4.2.1 Testing Hypothesis 1
The next step in data analysis will include hypothesis testing. This will be done by means of hierarchical regression. The first hypothesis predicted that the relationship between narcissistic leadership and abusive leadership will be positive only when the interaction frequency is high and not when the interaction frequency is low. This hypothesis was analysed twice: once for direct interaction frequency, and once for indirect interaction frequency. The results of the analysis with direct interaction frequency are shown in Table 2. In the first step of hierarchical multiple regression leader tenure was entered as a control
26 variable. This model was statistically significant F (1, 126) = 5.05, p = .026 and explained 4% of variance in abusive leadership. In the second step leader narcissism and direct interaction frequency were entered, and the total variance explained by the model was 5%. The introduction of leader narcissism and direct interaction explained additional 1% of variance in abusive leadership, after controlling for leader gender and tenure, which was non-significant DR2= .01; F (2, 124) = .52, p = .596. In the final model the moderation was tested by adding the interaction term, this model showed that the moderating effect was non-significant F (1, 123) = .12, p = .725, which means that direct interaction frequency does not moderate the relationship between leader narcissism and abusive leadership. Thus, leader tenure was the only variable that significantly positively predicted abusive leadership b = .21, p = .023. Table 2.
Hierarchical regression model of narcissistic leadership and abusive leadership, and direct interaction frequency as a potential moderator
R2 DR2 B SE b t Step 1 .04* Leader tenure .02 .01 .20* 2.24 Step 2 .05 .01 Leader tenure .02 .01 .20* 2.29 Leader narcissism .34 .36 .08 0.95
Direct interaction frequency -.00 .08 -.03 -0.36
Step 3 .05 .00
Leader tenure .02 .01 .21* 2.30
Leader narcissism .34 .36 .09 .96
Direct interaction frequency -.00 .01 -.03 -.33
Direct_int_frequency*Leader_narc -.03 .08 -.03 -.35
Note. * * p < .05, ** p < .01.
As the moderation effect with direct interaction frequency is non-significant, Hypothesis 1 must be rejected. However, it is also important to look at the moderating effect of indirect interaction. This analysis is shown in Table 3. The first step in the hierarchical regression, just as in the previous analysis includes the control variable, leader tenure, and the
27 significance of step one is already reported in the previous section. Step 2 includes also the measures of leader narcissism and indirect interaction frequency, and is statistically significant F (2, 124) = 3.30, p = .04. The total variance explained by the second model is 9%, thus signifying an increase of 5%. In the second step two out of four variables were significant with leader tenure recording a stronger effect (b = .22, p = .010) than indirect interaction frequency, which was found to be negatively related to abusive leadership (b = -.21, p = .019). Step 3 includes the interaction term (b = -.09, p = .311), for testing the moderation effect, and is non-significant F (1, 123) = 1.03, p = .311. Thus, Hypothesis 1 cannot be confirmed, neither with direct, nor with indirect interaction frequency.
Table 3.
Hierarchical regression model of narcissistic and abusive leadership, and indirect interaction frequency as a potential moderator
R2 DR2 B SE b t Step 1 .04* Leader tenure .02 .01 .19* 2.24 Step 2 .09* .05 Leader tenure .26 .01 .22* 2.62 Leader narcissism .45 .35 .11 1.28
Indirect interaction frequency -.02 .01 -.20* -2.37
Step 3 .09 .01
Leader tenure .03 .01 .26* 2.81
Leader narcissism .44 .35 .11 1.25
Indirect interaction frequency -.01 .01 -.18* -1.98
Indirect_int_frequency*Leader_narc -.04 .04 -.09 -1.02
Note. * * p < .05, ** p < .01.
4.2.2 Testing Hypothesis 2
Hypothesis 2a proposes that abusive leadership has a negative relationship with follower well-being. Hierarchical linear regression analysis is conducted to test this hypothesis, and results are shown in Table 4. In the first step the control variables of follower age and gender were entered. The model significantly predicts follower wellbeing F (2, 125)
28 = 5.94, p < .001. The control variables explain in total 9% of the variance of follower wellbeing. In step 2 abusive leadership is added to the model, and the new model explains in total 12% of variance, which means that adding abusive leadership resulted in an increase of variance explained by 4%. The model shows significant results F (1, 124) = 5.15, p = .025: there is a significant negative relationship between abusive leadership and employee wellbeing, b = -.19, p = .025. This means that Hypothesis 2a can be confirmed, as abusive leadership significantly predicts employee wellbeing.
Table 4.
The relationship between abusive leadership and employee wellbeing
R R2 D R2 B SE b t Step 1 .29 .09* Follower age .02 .01 .22* 2.58 Follower gender .50 .20 .21* 2.47 Step 2 .35 .12* .04 Follower age .02 .01 .20* 2.39 Follower gender .43 .20 .19* 2.30 Abusive leadership -.29 .13 -.19* -2.27 Note. * * p < .05, ** p < .01.
Next, Hypothesis 2b was tested. This hypothesis predicted that the interaction frequency between a leader and a follower will have a moderating effect on the relationship between leader narcissism and follower well-being. First, as in the previous analysis, the control variables follower age and gender were entered (Table 5). In step two the variables of leader narcissism and direct interaction frequency were added. The acquired model was non-significant F (2, 123) = 1.34, p = .264, and the model explained only additional 2% of variance in employee wellbeing. The interaction did not have a significant relationship with employee wellbeing b = .06, p = .476.
29 Table 5.
The moderating effect of direct interaction frequency between the leader and follower on the relationship between leader narcissism and employee wellbeing
R2 DR2 B SE b t Step 1 .09* Follower age .02 .01 .22* 2.58 Follower gender .50 .20 .21* 2.47 Step 2 .11 .02 Follower age .02 .01 .23* 2.66 Follower gender .51 .20 .22* 2.51 Leader narcissism .86 .54 .14 1.61
Direct interaction frequency -.01 .02 -.03 -.38
Step 3 .11 .00
Follower age .21 .01 .22* 2.56
Follower gender .51 .20 .22* 2.53
Leader narcissism .87 .54 .14 1.62
Direct interaction frequency -.01 .02 -.04 -.44
Direct_int_frequency*Leader_narc .09 .12 .06 .72
Note. * * p < .05, ** p < .01.
Furthermore, also when testing the model with indirect interaction frequency the results remained non-significant F (2, 123) = 1.79, p = .170 (Table 6), and adding leader narcissism and indirect interaction frequency to the model only explained additional 3% of variance in employee wellbeing. Because Hypothesis 2b predicted a significant interaction between leader narcissism and interaction frequency, which was however not found, the mediating role of abusive leadership cannot be tested. Thus, Hypothesis 2b cannot be accepted.
30 Table 6.
The moderating effect of indirect interaction frequency between the leader and follower on the relationship between leader narcissism and employee wellbeing
R2 DR2 B SE b t Step 1 .08* Follower age .02 .01 .22* 2.58 Follower gender .50 .20 .21* 2.47 Step 2 .11 .02 Follower age .02 .01 .21* 2.54 Follower gender .53 .20 .22* 2.62 Leader narcissism .78 .54 .12 1.46
Indirect interaction frequency .01 .01 .08 1.01
Step 3 .11 .00
Follower age .20 .01 .20* 2.37
Follower gender .53 .20 .22* 2.62
Leader narcissism .79 .54 .12 1.47
Indirect interaction frequency .01 .01 .07 .81
Indirect_int_frequency*Leader_narc .03 .06 .04 .52
Note. * * p < .05, ** p < .01.
4.2.3 Testing Hypothesis 3
Hypothesis 3a argues that abusive leadership has a negative relationship with leader effectiveness. Hierarchical linear regression analysis was conducted to test this hypothesis, and results are shown in Table 7. In the first step the control variable of leader tenure was entered. The model significantly predicts leader effectiveness F (1, 126) = 4.26, p = .041. The control variable explains in total 3% of the variance in leader effectiveness. In step 2 abusive leadership is added to the model, and the new model explains in total 15.7 % of variance, which means that adding abusive leadership resulted in an increase of variance explained by 12.7%. The model shows significant results F (1, 125) = 24.22, p < .001 there is a significant negative relationship between abusive leadership and leader effectiveness, b = -.40, p < .001. This means that Hypothesis 3a can be confirmed, as abusive leadership significantly predicts leadership effectiveness.
31 Table 7.
The relationship between abusive leadership and perceived leader effectiveness
R2 DR2 B SE b t Step 1 .03* Leader tenure -.03 .01 -.18* -2.06 Step 2 .19* .16 Leader tenure -.02 .01 -.10 -1.24 Abusive -.51 .10 -.40* -4.92 Note. * * p < .05, ** p < .01.
The next step in analysis was to test hypothesis 3b and see whether direct interaction frequency moderates the relationship between leader narcissism and leader effectiveness. The results are presented in table 8. In the first step of hierarchical multiple regression leader tenure was entered as a control variable. This model was statistically significant F (1, 126) = 4.26, p = .041 and explained 3.3% of variance in leadership effectiveness. In the second step leader narcissism and direct interaction frequency were entered, and the total variance explained by the model increased to 5%. The introduction of leader narcissism and direct interaction explained additional 1.8% of variance in abusive leadership, after controlling for leader tenure. The model was non-significant R2 change = .01; F (2, 124) = 1.16, p = .317. In the final model the moderation was tested by adding the interaction term, this model showed that the moderating effect was non-significant F (1, 123) = .16, p = .686, which means that direct interaction frequency does not moderate the relationship between leader narcissism and leadership effectiveness b = -.04, p = .686. Thus, leader tenure was the only variable that significantly negatively predicted leadership effectiveness b = -.19, p = .038.
32 Table 8.
The moderating effect of direct interaction frequency between the leader and follower on the relationship between leader narcissism and perceived leader effectiveness
R2 DR2 B SE b t Step 1 .03* Leader tenure -.03 .013 .18* -2.06 Step 2 .05 .02 Leader tenure -.03 .01 -.20* -2.26 Leader narcissism .29 .45 .06 .65
Direct interaction frequency .02 .02 .12 1.35
Step 3 .05 .00
Leader tenure -.03 .01 -.19* -2.10
Leader narcissism .29 .45 .06 .65
Direct interaction frequency .02 .02 .12 1.37
Direct_int_frequency*Leader_narc -.04 .10 -.04 -.41
Note. * * p < .05, ** p < .01.
Furthermore, as presented in table 9, also when testing the model with indirect interaction frequency the results remained non-significant F (2, 124) = 1.35, p = .261, and adding leader narcissism and indirect interaction frequency to the model only explained additional 2% of variance in employee wellbeing. Because Hypothesis 3b predicted a significant interaction between leader narcissism and interaction frequency, which was not found, the mediating role of abusive leadership cannot be tested. Thus, Hypothesis 3b cannot be accepted.
33 Table 9.
The moderating effect of indirect interaction frequency between the leader and follower on the relationship between leader narcissism and perceived leader effectiveness
R2 DR2 B SE b t Step 1 .03* Leader tenure -.03 .013 .18* -2.06 Step 2 .05 .02 Leader tenure -.03 .03 -.19* -2.23 Leader narcissism .23 .45 .04 .50
Indirect interaction frequency .02 .01 .13 1.49
Step 3 .05 .00
Leader tenure -.27 .01 .19* -1.99
Leader narcissism .23 .45 .04 .49
Indirect interaction frequency .02 .01 .14 1.48
Indirect_int_frequency*Leader_narc -.01 .05 -.02 -.21
Note. * * p < .05, ** p < .01.
4.3
Exploratory Analyses
The proposed moderation of interaction frequency was not confirmed. However, the literature on abusive leadership and leader narcissism often discusses follower self-esteem as a buffer for the negative characteristics of narcissistic leaders, and consequences of abusive leadership (Tepper, 2007). Thus, the influence of follower self-esteem as a moderator will be examined, instead of interaction frequency. As this research was a part of a larger study where other variables were measured, including follower self-esteem, the data was readily available to test follower self-esteem as a potential moderator. Follower self-esteem was measured by the 10-item Rosenberg Scale (Rosenberg, 1965). E.g. “I feel that I have a number of good qualities” (α =.82). Answers were given on a 7-point Likert (1=totally disagree; 7=totally
