The bright and dark side of narcissism: What is the role of leader
narcissism on the relationship of employees and leaders?
The relationship between adaptive and maladaptive narcissism on LMX and the moderating effect of machiavellianism and proself orientation.
Ivira Bos
University of Amsterdam
Faculty of Economics and Business Master:
Track:
MSc Business Studies Leadership and Management Student number: 10003311
Student e-mail: ivirabos@hotmail.com Supervisors:
Second reader:
Dr, A. de Hoogh Prof. D. Den Hartog, Date:
Total amount of words
29-06-2015 15.890
ii
Statement of Originality
This document is written by Student Ivira Bos 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.
iii
ABSTRACT
Despite the growing amount of literature on narcissism, no consensus has been reached regarding narcissism’s impact on leadership. Part of the confusion has its roots in the multiple forms of narcissism, also referred as the dark and bright side of narcissism (Campbell et al., 2011; Ackerman et al., 2011). The present study focuses on both dimensions of narcissism, the adaptive (bright side) and maladaptive (dark side) in relationship with LMX. Further the moderating role of machiavellianism and proself orientation in the relationship between narcissism and LMX will be examined.Data were collected from two distinct samples, employees and leaders, from diverse organizations and occupational groups through internet based surveys. In total 159 dyads were used in this study. The results showed, narcissism was unrelated to LMX, however a moderation effect of machiavellianism for leaders high on machiavellianism showed a negative relationship between the different aspects of narcissism and LMX. A positive trend was found between aspects of narcissism and LMX when
machiavellianism of leaders was low. Further, no support was found for the moderation effect of proself orientation.The results of an explorative study about the multidimensionality construct of narcissism revealed that even if a third dimension was indicated, narcissism could be better treated as a construct consisted of 2 dimensions. Recommendations are given.
iv
Table of Content
1 INTRODUCTION...1
2 LITERATURE REVIEW...4
2.1 NARCISSISM. ... ...4
2.1.1 Multiple ‘flavors’ of narcissism...6
2.1.2 Maladaptive and adaptive narcissism...7
2.2 LMX ...7
2.3 NARCISSIM AND LMX...8
2.3.1 Maladaptive facet and LMX...9
2.3.2 Adaptive facet and LMX...10
2.4 MACHIAVELLIANISM...11
2.5 PROSELF ORIENTATION...15
2.6 CONCEPTUAL MODEL... ...19
3 METHOD... ...19
3.1 SAMPLE AND PROCEDURE...19
3.2 MEASURES...21
4 RESULTS... ...23
4.1 DATA PREPERATION ...23
4.2 TESTING THE MAIN EFFECTS...24
4.3 MODERATORE EFFECTS ...25
4.3.1 Machiavellianism...25
4.3.2 Proself orientation...32
4.4 EXPLORATIVE ANALYSIS...38
5 DISCUSSION... ...39
5.1 FINDINGS AND THEORETIAL IMPLICATION...39
5.2 PRACTICAL IMPLICATIONS...47
5.3 LIMITATIONS AND FUTURE RESEARCH...49
5.4 CONCLUSION...51
5.5 ACKNOWLEDGEMENTS...51
v
List of Tables
4 RESULTS... ...23
4.1 MEANS, STANDARD DEVIATIONS AND INTERCORRELATIONS OF MEASURES...24
4.2 RESULTS OF REGRESSION; TESTING THE MAINEFFECTS...25
4.3 RESULTS OF REGRESSION; NARCISSISM OVERALL, MACHIAVELLIANISM AND LMX...26
4.4 RESULTS OF REGRESSION; MALADAPTIVE NARCISSISM, MACHIAVELLIANISM AND LMX...28
4.5 RESULTS OF REGRESSION; ADAPTIVE NARCISSISM, MACHIAVELLIANISM AND LMX...30
4.6 RESULTS OF REGRESSION; NARCISSISM OVERALL, PROSELF ORIENTATION AND LMX...33
4.7 RESULTS OF REGRESSION; MALADAPTIVE NARCISSISM, PROSELF ORIENTATION AND LMX...35
vi
List of Figures
1 INTRODUCTION... ...1
1.1 CONCEPTUAL MODEL ...19
4 RESULTS... ...23
4.2 .SIMPLE SLOPES; NARCISSISM OVERALL, MACHIAVELLIANISM AND LMX...27
4.3 SIMPLE SLOPES; MALADAPTIVE NARCISSISM, MACHIAVELLIANISM AND LMX...29
4.4 SIMPLE SLOPES; ADAPTIVE NARCISSISM, MACHIAVELLIANISM AND LMX...31
4.5 SIMPLE SLOPES; NARCISSISM OVERALL, PROSELF ORIENTATION AND LMX...34
4.6 SIMPLE SLOPES; MALADAPTIVE NARCISSISM, PROSELF ORIENTATION AND LMX...35
4.7 SIMPLE SLOPES; ADAPTIVE NARCISSISM, PROSELF ORIENTATION AND LMX ...37
1
Introduction
A great part of leadership roles within modern organizations seems to be prevalent with narcissists (Campbell, Hoffman, Campbell & Marchisio, 2011).Rosenthal and Pittinsky (2006) argue that, leadership roles in all its forms, from the great historical figures to today’s business leaders, are often held by narcissist. The study concluded that most leadership positions are correlated with narcissism. Those findings can be partly explained by linking narcissism to leadership emergence. To further support this, a study of military school cades showed that narcissism positively predicted leadership development (Harms, Spain, & Hannah, 2011). Another study indicated that in leaderless group discussions (LGDs) narcissists tend to emerge as leader (Brunell, Gentry, Campbell, Hoffman, Kuhnert &
Demarree, 2008). All these findings are consistent with Campbell, Goodie and Foster’s (2009) conclusion that narcissists seek out leadership positions in organizations. It is not only
narcissists who are seeking for leadership roles; people tend to have a preference in having a narcissistic person as leader as well. For example, research of Schnure (2010) demonstrated that experienced interviewers in a selection procedure evaluated applications with narcissistic characteristics for a managerial job, as more favourably. Similarly, individuals with high scores of narcissism are more likely to be selected into leadership roles then individuals with low scores of narcissism (Grijalva, Harms, Newman, Gaddis & Fraley, 2014).
As described, leadership roles and narcissism are two related concepts. Because of the great influence leaders have on all kinds of outcomes, it is important to investigate whether being a narcissist and leader at the same time benefits or hinders the organization (O’reilly, Doer, Caldwell & Chatman, 2013; Grijalva et al., 2014). Existing research highlights both the positives and negatives of a narcissistic leader. On the negative side, Nevicka, DeHoogh, VanVianen, Beersma and McIlwain (2011) found a relationship between narcissism and a reduced group-level information. Having a group with a narcissistic leader is correlated with
2 lower levels of information exchange and therefore performance disadvantages. Other
findings suggest that narcissistic leaders are more likely to violate integrity standards and are more likely to create a destructive workplace (Blickle, Schlegel, Fassbender & Klein, 2006). Furthermore, narcissistic traits of a leader are generally viewed negatively by others, which translates into lower task and contextual performance, more counterproductive performance among subordinates (Judge, LePine, & Rich, 2006) and less happier employees (Blair, Hoffman, & Helland, 2008).
As mentioned above, research suggests narcissism can engender negative attributes. However, the large number of narcissists in important leadership positions indicates that narcissism is likely to also have positive aspects (Rosenthal and Pittinsky, 2006). The following authors are claiming leaders who are narcissistic can be positive for the
organization as well. Galvin, Waldman and Balthazard (2010) demonstrated that narcissism was positively related to charismatic leadership. This relation was both explained through the visionary boldness component of charisma and also through the component representing the tendency to take risks and be inspirational and exciting. In addition, Maccoby (2003) claims that the benefits to the organization of narcissistic leaders are enormous because such leaders tend to be self-aware of their own behavioral tendencies and consciously work to control them. Other authors argue that leaders who are narcissistic tend to be creative, self-aware, can often laugh at their own tendencies (Doyle and Lynch, 2008) and are extravert (Grijalva et at., 2014)
Despite the growing body of literature focusing on narcissism, no consensus has been reached regarding narcissism’s impact on leadership. Part of the confusion in findings stems from the existence of multiple ‘flavors’ or forms of narcissis (Campbell, Bonacci, Shelton, Exline & Bushman, 2011). Most studies do not make a distinction between multiple forms of narcissism and are only reporting effect sizes for global narcissism. Campbell et al. (2011)
3 refer to these seemingly contradictory forms of narcissism as the ‘bright’ and ‘dark’ side of narcissism. This terminology has been approved by a couple of authors and is sometimes attributed to both the adaptive and maladaptive narcissism (Ackerman, Witt, Donnellan, Trzesniewski, Robins & Kashy, 2011). To find out whether these contradictory findings in previous research are because of the existence of more dimensions, the present study focusses on both the adaptive and maladaptive dimensions of narcissism.
In addition, this study will focus on the relationship leaders have with their followers. The theory about leadership indicates that leadership is a multilevel phenomenon and
therefore leaders should always be seen in relationship to followers (Epitropaki & Martin, 2005). The relationship with followers, also known as the Leader–Member Exchange (LMX) quality, leads to important employee outcomes such as higher job satisfaction (Gerstner & Day, 1997) wellbeing, leader satisfaction (Epitropaki & Martin, 2005) organizational
commitment and citizenship behaviors (Townsend, Phillips & Elkins, 2000). These outcomes raise the need of investigating the effects of narcissism on LMX. In particular the effects of both, adaptive and maladaptive narcissism on LMX are investigated in this study. Further, machiavellianism and proself orientation are added as moderators to this model to see whether these constructs affect the relationship between narcissism and LMX. Machiavellianism is used as a moderator because most studies analysed the negative consequences of
machiavellianism leader without taking other variables, such as narcissism, into account (Paulhus & Williams, 2002). Further, studies looking at the comparison of both constructs found differences between narcissism and machiavallianism. Those differences involved: extraversion (Helgeson & Fritz, 1990), psychopathy (Lee & Ashton) and honesty (Hare, 1991). These differences between constructs could possibly contribute to adding more negative attributes to leaders and raises the question on how machiavellian influences the relationship between narcissism and LMX. Social value orientation and specifically the
4 proself orientation will be investigated within this model because of similar reasons. Most research on this construct was done in an experimental way with proself orientation being an independent variable investigating the direct effects. Not even one study provided information about comparisons between the narcissism and proself orientation constructs. Analysing the constructs separately showed that both the narcissism and proself orientation constructs have an overlap in the low quality relationships with followers, lack of empathy (Declercq & Bogeart, 2008) and the lack of fairness distribution (Campbell et al., 2011). The only difference between both is the lack in trust self oriented individuals have (Kanagaretnam, Mestelman, Nainar & Shehata, 2009). This difference again, introduces the question on how proself oriented leaders influence the relationship between narcissism and LMX.
The current research addresses the need to provide more insights in the role narcissism has within leadership in order to determine the type of advice offered to organization
regarding to the selection of narcissists for leadership roles. By using a multidimensionality construct of narcissism, the different components of narcissism will be analysed separately and linked to LMX. Another purpose is to provide insights in the moderating role of machiavellianism and proself orientation in the relationship between narcissism and LMX. Finally, an explorative study will be conducted to investigate whether narcissism exist, as proposed by Corry, Merritt, Mrug and Pamp (2008), of two dimensions.
Literature review and Hypothesis
Narcissism
Narcissism has been used with increasing frequency to describe behavior in
organizations and especially in explaining the destructive behaviors of leaders (Campbell et al., 2011). The confusion about whether narcissism in leaders hinders or benefits the
organization partly stems from how best to define narcissism and seems to be an obstacle in this increased interest in research. (Miller and Campbell, 2008).
5 The social-personality literature, the psychiatric literature and the clinical psychology are all using two different concepts of narcissism (Campbell et al., 2011). Both the psychiatric and clinical literature defines narcissism as a personality disorder (NPD). The NPD refers to inflexible and enduring characters associated with a desire for admiration, a lack of empathy and grandiosity (American Psychiatric Association, 2000). The social-personality literature describes narcissism as a trait that is distributed normally in population and for which there is no qualitative cut-off for (Foster & Campbell, 2007). There is a debate whether research should focus on trait narcissism or on NPD (Campbell et al, 2011). The problem with NPD is that the prevalence of someone with NPD is relatively low, while the prevalence of those with narcissistic symptoms, without the sufficient distress to cross the line into the clinical
disorder, is much bigger. Using the narcissism trait reduces the chance of missing the more common occurrences of narcissism (Stinson, Dawson, Goldstein, Chou & Huang, 2008). Having this in consideration, it is preferable to focus on trait narcissism rather than NPD (Campbell et al, 2011).
Following the social personality literature, narcissism is a personality trait that is characterized by a sense of personal superiority (Campbell, Bonaci & Shelton 2004; John & Robins, 1994), grandiosity (Morf & Rhodewalt, 2001), dominance, a desire for power
(Emmons, 1987) and the need for attention and confirmation of superiority (Bogart, Benotsch, & Pavolic, 2004). Moreover, narcissists are related to arrogance, entitlement, self-absorption and hostility (Rosenthal & Pittinsky, 2006). Additionally, narcissists show an unusual high level of self-love, they feel entitle to praise and admiration and believe that they are uniquely special (Morf & Rhodewait, 2001).
The most widely used measurement method of narcissism in social and personality psychology, which is also increasingly used by clinical psychologists, is the NPI (Miller & Campbell, 2008). Recent estimates indicate that within social and personality psychology
6 75% of researchers uses this instrument to measure narcissism (Cain, Pincus & Ansell, 2008). Such estimates suggest that the NPI will continue to play a great role in future research of narcissism. Because the NPI interpretation is clouded by conflicting studies assessing its factor structure it is important to develop a better understanding of the NPI (Ackerman et al., 2011).
The multiple ‘flavors’ of narcissism. Several researchers have examined the
structure of the NPI by applying factor analytic techniques. One of the first researchers who tried to find different components of narcissism was Emmons (1984), who conducted a principal components factor analysis. The well-known Emmons’ factors were:
Leadership/Authority, Self-Absorption/Self-Admiration, Superiority/ Arrogance, and
Exploitiveness/Entitlement. Raskin and Terry (1988) were interested in different components as well and replicated the way Emmons (1984) analysed the NPI. They concluded that Emmons’s (1984) selection criterion was too conservative in choosing only four NPI factors. Raskin and Terry introduced their seven components: Authority, Vanity, Entitlement, Self-Sufficiency, Superiority, Exhibitionism and Exploitiveness. Using confirmatory factor
analysis, Corry et al. (2008) showed narcissism to have two dimensions: Leadership/Authority and Exhibitionism/Entitlement. According to Ackerman et al. (2011), even when the internal consistency of both dimensions found in the study of Corry et al. is high, there was an important dimension missing. Ackerman et al. (2011) identified three dimensions by proposing that the dimension Exhibitionism/Entitlement should be splitted up in two new dimensions namely: the Grandiose Exhibitionism and the Entitlement/Exploitativeness. Ackerman’s three dimensions consisted of: Leadership/Authority, Grandiose Exhibitionism, and Entitlement/Exploitativeness. Ackerman et al. argue that such a distinction is consistent with Brown, Budzek and Tamborski (2009) theoretical claims about the importance of separating entitlement and grandiosity. According to Ackerman, Donnellan and Robins
7 (2012), the two-factor model of Corry et al. and the three-factor model of Ackerman et al. are effective measurements on both empirical and theoretical grounds. They both provided a good range of precision. However, Ackerman stated that a combination of both models, by using the model of Corry’s et al. (2008) as a basis, is best in practice. In their proposed model the leadership/authority dimension of Corry et al. is shown to have the best precision and
coverage and is, except for one item, similar with the leadership/authority scale of Ackerman et al (2011). Further, Ackerman et al. suggested to delete items 14, 24 and 25 from the Exhibitionism/entitlement scale of Corry et al. (2008) as they provided negligible levels of information and are better predictors of the subdimension entitlement. Eliminating those items support the Ackerman’s idea that entitlement does not belong on a scale that appears to reflect grandiosity (Brown et al., 2009). Even if, the Raskin and Terry model is the most currently used model, Maxwell, Donnellan, Hopwood and Ackerman (2012) stated the proposed model of Ackerman (2012) is better due to its simplicity.
Maladaptive and adaptive narcissism. Based on the four factor model of Emmons
(1987), the components: entitlement, exploitativeness and exhibitionism have been considered maladaptive because of their poor social adjustment, whereas the label authority/leadership is considered relatively adaptive based on their shared relations with self-confidence and
assertiveness. Ackerman et al. (2011) also stated that narcissism turned out to be both maladaptive (negative) and adaptive (positive). Their study showed that, because of their association with desirable mental health, the authority/leadership factor reflects the adaptive side of narcissism whereas the exploitative and entitlement factors refers to maladaptive factors (Watson & Biderman, 1993; Watson & Morris, 1991).
LMX
Graen and Uhl-Bien (1995) developed the leader-member exchange theory, focusing on the two-way relationship between leaders (supervisors) and subordinates (employees). The
8 theory states that leaders do not treat all their followers equally; therefore, they develop different quality of relationships. The quality of these leader-member exchange (LMX) relationships influences subordinates' responsibility, decision influence, access to resources and performance (Deluga, 1994).The relationship between leaders and employees is built on social exchange and are characterized by support, commitment, loyalty and trust (Uhl-Bien & Maslyn, 2003 Cropanzano & Mitchell, 2005). This mutual reciprocation that embodies high LMX relationships results in an increased attachment between both leaders and followers (Ferris, Liden, Munyon, Summers, Basik & Buckley, 2009). High quality LMX relationships further influences job outcomes such as an increase in task performance, organizational commitment and satisfaction and a decrease in turnover rate (Gerstner & Day, 1997). Ilies, Nahrgang and Morgeson’s (2007) meta-analysis showed that employees involved in high quality LMX relationships ‘pay back’ their leaders by engaging in citizenship. There was found a correlation of .37, which supported previous meta-analytically relationships between LMX and citizenship. Specifically, the meta-analysis of Ilies et al. (2007) showed that high quality LMX relationships are positive related to organizational commitment, positive attitudes towards the organization, procedural and distributive justice, greater job
responsibility and perceptions of impact and self-determination. Ilies et al. are ending their study suggesting that a leader, with a high quality relationship with employees, therefore improves several organizational outcomes.
Narcissism and LMX
The research about the main effects of narcissism on LMX suggest a negative relationship. For instance, narcissistic leaders seem to be more likely creating destructive workplaces (Blair, Hoffman, & Helland, 2008), to be good in creating a ‘blame’ and ‘toxic’ culture (Hogan et al., 1994), and tend to use their power for personal aggrandisement (Post, 1993). In addition to those findings, Judge et al (2006) demonstrate that narcissists are lacking
9 empathy, arrogantly fantasizing grandiose dreams and manipulate outcomes towards their own interests which all translates into awkward interpersonal interactions. They add that others generally view narcissistic leaders negatively, which results in lower job performance and fewer examples of organizational citizenship among subordinates (Judge, LePine, & Rich, 2006). Blickle et al. (2006) suggest that because narcissistic leaders are more likely to violate integrity standards their employees are less satisfied with their job. All of the above mentioned statements indicate a negative relationship between narcissism and LMX, therefor the first hypothesis is as following:
Hypothese 1A: Narcissism is negatively related to LMX
Maladaptive facet and LMX. The next section focuses on the relationship between
the maladaptive side of narcissism and LMX. Individuals who score high on the maladaptive facet (Exhibitionism/Entitlement) are more likely to show uncooperative and unsocial dominating behavior (Barry et al., 2003; Meurs & Perrewe, 2011). Furthermore, these individuals view themselves as more important than others and easily manipulate people (Emmons, 1984; Raskin & Terry, 1988). One of the dimensions of the maladaptive facet is entitlement. Entitlement is related to taking more resources than a system can support and to demanding more salary (Campbell, Bonacci, et al., 2004; Campbell, Goodie, et al., 2004). A sense of entitlement of narcissists is also related to the belief that they deserve special treatment. These findings, from a leader point of view, all suggest a negative relationship between maladaptive narcissism and LMX. From the employee point of view, a lack of empathy from leaders will result in frequent feelings of inequity (Huseman, Hatfield, & Miles, 1987). When employees feel that their contributions are not ‘properly’ acknowledged, they will suffer from decreased perceptions of distributive justice, with a bigger change of being unsatisfied (Campbell, Goodie et al., 2004). Therefore:
10 Hypothese 1B: The ‘maladaptive’ facet is negatively related to LMX
The adaptive facet and LMX. In contrast with maladaptive narcissism, adaptive
narcissism (Leadership/Authority facet) has been related to variables considered socially desirable including self-confidence, assertiveness and independence (Raskin & Terry, 1988). In addition, Watson, Taylor and Morris (1987) found that adaptive narcissism appears to be related with extraversion that correlated positive with dominance and self-esteem.
Furthermore, adaptive narcissism was also found to be associated with positive personal adjustment and optimism (Hickman, Watson & Morris, 1996) and proves an advantage to individuals in social situations that require authority, leadership and a confident and social presentation. Holland (1985) argued, that adaptive narcissism appears to overlap with his ‘social’ personality’ which is defined as ascendant, persuasive, friendly, sociable and warm personality. There was also an indication for an overlap with the ‘enterprising’ personality, which is conceptualized as possessing leadership, agreeable, speaking ability, talkative, self-confident and optimistic. Lastly, Hill and Yousey (1998) determined that adaptive narcissism is associated with extraversion, warmth, dominance and social boldness. All of these positive outcomes of adaptive narcissism lead to the following hypothesis:
Hypothese 1c: The ‘adaptive’ facet is positively related to LMX.
Machiavellianism
Previous literature about machiavellianism consistently incorporates citations by Christie and Geis (1970, p39). As pioneers, they define machiavellianism as “a process by which the manipulator gets more of some kind of reward than he would have gotten without manipulating, while someone else gets less, at least within the immediate context”. Christie and Geis (1970) developed statements and implemented them into the measurement of the normal personality construct and showed reliable differences in respondents. The respondents
11 who agreed with these items were labelled as being more machiavellian and were more likely to behave in a cold an manipulative way. This measurement created by Christie and Geis is today’s most used measurement method of the machiavellianism construct.
According to McHoskey, Worzel and Szyarto (1998), respondents who are scoring high on machiavellianism could be described best as: domineering, impersonal, cold, cynical, deceitful, exploitative, impervious, practical, suspicious, superficial charm, glibness and duplicity. In addition to this explanation, Deluga (2001) added the traits manipulative, deceitful and exploitative and stated that someone scoring high on machiavellian is not distracted by social influences, emotional issues, interpersonal concerns. Besides these traits regarding machiavellianism, behavior patterns can also be attributed to a person scoring high on the machiavellianism construct. Just like charismatic leaders, machiavellians are skilled in creating a desired image (Gardner and Avolio, 1995). In addition Christie and Geis (1970), suggested a person scoring high on Machiavellianism is capable of presenting an image of depersonalized coolness under pressure.
Machiavellianism, narcissism and LMX
A first glance at the literature of both the machiavellian and narcissism construct indicates some similarities. These similarities did not stayed unnoticed; a lot of research is devoted only to compare narcissism with machiavellianism. Paulhus and Williams (2002) demonstrated that the correlation between narcissism and machiavellianism ranged around .25. According to their study this correlation could be explained with one of the traits of the Big Five, agreeableness. Both variables are negatively correlated to agreeableness. The research of Jakobwitz and Egan (2006) supported these findings and found a correlation with agreeableness of both variables as well. Narcissism was correlated to agreeableness with scores from r=-.34 tot r=-.43 and machiavellianism was correlated with scores varying between r=-.47 and r=-.41. They concluded that both narcissist and machiavallian people are
12 not agreeable. McHoskey (1995) supported these findings partly. He agreed on the statements about a correlation between both constructs but following his findings this correlation could be better explained with interpersonal manipulation. The findings of his study showed that both, narcissism and machiavellianism, indicated an overlap in the propensity for
interpersonal manipulation, just as Geis (1978) suggested. Another explanation for the founded correlation comes from Watson and Morris (1991). Their research, about the similarities between both narcissism and Machiavellianism, demonstrated similarities in the inability to take perspective of others and moreover the lack of empathy. They even
demonstrated that machiavellianism was positively associated with the exploitativeness and entitlement aspects of narcissism, which stands for maladaptive narcissism. Furthermore, there is an explanation which comes from the overlap both constructs have with dominance. For example Gurtman (1991, 1992) showed that both constructs fall into the upper-left quadrant of the interpersonal circumplex and are both associated with: arrogance, dominance, and a lack of interpersonal warmth. Paulhuls and Williams (2002) added to this that the commonality between Machiavellianism and Narcissism can be found in socially malevolent character and is expressed in behavior like self-promotion, emotional coldness, duplicity and aggressiveness (Paulhus & Williams, 2002).
Nonetheless the overlap in agreeableness, interpersonal manipulation, lack of empathy and dominance, most studies conclude that both constructs are measuring different traits and are therefore not equivalent (Glad, 2002). Conger (1997) also compared both constructs and concluded, based on factor analysis, that both narcissism and machiavellianism differ a lot from each other and that they should exist apart. Helgeson and Fritz (1999) demonstrated that a difference between both narcissism and machiavellianism is associated with extraversion and openness. Their research showed that narcissism is positively correlated with extraversion and openness whereas machiavellianism is not. This finding was also supported by Lee and
13 Ashton (2005). According to their findings, narcissism is correlated with extraversion by both the big five model (r=.46) and a more recent created structural model of personality developed by Lee and Ashton (2004), the HEXACO model (r=.49). Again, machiavellianism did not showed any correlation with extraversion. These findings may suggest that when a leader is both narcissistic and machiavellianism, the characteristics of machiavellianism rule some of the positive characteristics of narcissism, namely extraversion and openness, out. This may indicate that the link between narcissism and LMX will be stronger when leaders are both machiavellian and narcissistic. Another finding of Lee and Ashton (2005) suggest that the dark trait psychopathy has a much higher correlation with machiavellianism (r=.66) than it has with narcissism (r=.37). Psychopathy, which refers to a pattern of remorseless
manipulation, callous and exploitation of others was investigated as a psychological cause of antisocial and criminal behaviors (Hare, 1991). Having a leader that also tends to have machiavellian characteristics seems to add more psychopathy characteristics which possible results in a higher chance of negative outcomes. Another difference between narcissism and machiavellianism is the association with the honesty-humility trait. Narcissism shows to be highly correlated with the modesty facet (r=.62) whereas machiavellianism is not. Further, narcissism is correlated less with both fairness and sincerity facets then these facets are correlated with machiavellianism. Overall these findings suggest that machiavellians are less modest, fair and sincere. It can be imagined that when a leader is besides being narcissistic also machiavellian this, because of lower scores of honesty, results in a greater impact on LMX. In conclusion, in comparison with narcissists machiavallian are not extravert, they are associated with psychopathy and are less honest (Helgeson & Fritz, 1990; Lee & Ahston, 2005; Hare, 1991).
Finally the direct link between machiavellianism and LMX will be discussed. While high machiavellians may be successful at the workplace, there is substantial literature
14 supporting that high machiavellianism leaders are not having a good relationship with their followers. According to Christie and Geis (1970) people who score high on machiavellianism manipulate more and are persuaded less than people scoring low on machiavellianism. Other research shows that leaders with high scores of machiavellianism tend to have higher levels of mistrust, cynicism, egocentricity, and propensity for interpersonal manipulation (McHoskey, 1998) and show more interpersonal coldness (Wiggins & Broughton, 1985). Additionally, leaders scoring high on machiavellianism care little about the values of others (Yurtsever, 2003). Further research suggests that, high machiavellian leaders exhibited a highly directive leadership style with little consideration for interpersonal concerns, such as: managing tension between followers or showing consideration for followers’ feelings (Deluga, 2001). These findings indicates that high machiavellianism leaders have a lot of negative impacts on followers. Moreover, high machiavellianism would lead to less satisfaction with superiors (Walter, Cole and Humphrey, 2011)
Overall, based on the above-mentioned literature, it is suggested that leaders being narcissistic and machiavallian contribute to more negative characters and therefore are more negative related to LMX. These more negative characters as a result of being
machiavellainism can be described to machiavallians as being less extravert, having a link with psychopathy and being dishonest. Further because of the negative relationship between machiavellianism and LMX, it is suspected that machiavellianism would moderate the relationship between narcissism and LMX. The following hypothesis could be introduced.
Hypothese 2A: The negative relationship between leader narcissism and LMX will be moderated by leader machiavellianism, such that the relationship will be stronger for leaders higher on machiavellianism.
15 narcissism and LMX will be moderated by leader machiavellianism, such that the relationship will be stronger for leaders higher on machiavellianism.
Hypothese 2C: The positive relationship between the adaptive dimension of leader narcissism and LMX will be moderated by leader machiavellianism, such that the relationship will be weaker for leaders higher on machiavellianism.
Proself orientation
According to the social psychological theory, social value orientation is a relatively stable personality trait. To some extent, individuals differ in their social value orientation, which is defined by people’s preferences for outcome distributions between the self and the interdependent other (Messick & McClintock, 1968). A number of social values have been recognized but most authors distinguish individuals from having a prosocial orientation or a proself orientation. The proself orientation is sometimes divided into an individualistic and competitive orientation (Smeesters, Warlop, Van Avermaet, Corneille & Yzerbyt, 2003), but according to van Lange en van Kleef (2008) in many situations these two orientations are functionally equivalent. Individuals with a proself orientation are more individualistic and competitive. These individuals, have a high concern for their own outcomes, and either tend to maximize the relative advantage over other's outcomes or tend to ignore the outcomes of interdependent others. In contrast, an individual with a prosocial orientation values equality, cooperation and good outcomes for themselves as well as for the others. Prosocial individuals do not like receiving good outcomes at the expense of others. They prefer 'win-win solutions', in which themselves and the opponents receive high outcomes (Smeesters et al. 2003). Current research about social value orientation, specifically proself orientation, so far has only investigated the direct consequences of prosef orientation. In order to make
16 analysed and after, the link between narcissism and proself orientation will be discussed. Studies about proself orientation are mostly done in an experimental setting trying to imitate real-life situations about negotiation (Beggan, Messick, & Allison, 1988). The findings suggest that when people prepare for a negotiation, individuals with a proself orientation choose more competitive heuristics such ‘your gain is my loss’ whereas prosocial individuals choose more cooperative heuristics such as ‘equal split is fair’ (Dreu & Boles, 1998). The explanation for this difference can be found in the way proself oriented individuals perceive cooperative heuristics, such as finding cooperating with others as ineffective. In addition, results found by van Van Lange et al. (1997) demonstrate that prosocial oriented individuals are more attached to their opponents then proself oriented individuals are and, therefore choose cooperative heuristics over competitive heuristics. Furthermore, individuals with a proself focus care less about fairness distribution (De Cremer, Tyler, & den Ouden, 2005) and have less sense of social responsibility (De Cremer & Van Lange, 2001). In their study, Dreu and Boles (2004) found another reason for prosocial-oriented persons to act in a more cooperative manner. According to their study, prosocial negotiators behave in more trusting ways than proself oriented negotiators, which results in better bonding with their opponents. Results of Gillespie, Brett and Weingart (2000) showed indeed that prosocial oriented individuals have more trust in their opponents and make more concessions then prosocial individuals do. According to the study of Kanagaretnam et al. (2009) this difference in trust can be explained using the game theory. People with a proself orientation expect others to be proself orientated as well; in conclusion they may expect their opponents not to cooperate. For this reason, they are less likely to invest and therefore identified as less trustful. For
individuals with a prosocial behavior it is the other way around, as they think opponents will react in the same way as they would do, they are likely to invest and are perceived as more trusting by their opponents. Another explanation for the findings suggesting that proself
17 oriented individuals are less able to bond with their opponents than prosocial oriented
individuals comes from their lack of empathy, which allows people to better predict others’ behaviors and actions and make more educated choices (Singer & Fehr, 2005). In another study conducted by Declercq and Bogeart (2008), it was revealed that a prosocial value orientation was positively related to empathy whereas a proself value orientation was
negatively related to empathy. In addition, individuals with a proself orientation are worse in putting themselves in another’s person’s place and in interpreting others people’s feelings and expectations. In conclusion, opponents perceive proself oriented individuals as less trustful, caring less about fairness distribution, being less attached with their opponents and lacking empathy. Most of these characteristics are similar to characteristics of the narcissism
construct, as Campbell et al. (2011) stated; narcissists care less about the fairness distribution, are less attached to their opponents and that are lacking empathy. The great overlap between proself orientation and narcissism is therefore predicted to have a reinforcing effect on the already predicted negative relationship between narcissism and LMX. A difference between both constructs can be found in the lack in trust proself oriented individuals have. This lack in trust indicate that a leader, being narcissistic and proself orientated at the same time, is likely to act in a more negative way towards opponents then it would have done without being distrustful.
The above-mentioned literature shows that leaders with a proself orientation are less likely to have high quality relationship with followers. Moreover, a proself orientation reinforces the attributes related to the attachment with opponents, lack of empathy and the fairness distribution. The lack of trust is not present in the construct of narcissism and therefore is expected to make the consequences of being both narcissistic and machiavellian more negative. According to those conclusions, the following hypotheses are proposed.
18 Hypothese 3A: The negative relationship between leader narcissism and LMX is moderated by leader proself orientation, such that this relationship becomes stronger for leaders higher on proself orientation.
Hypothese 3B: The negative relationship between the maladaptive dimension of leader narcissism and LMX is moderated by leader proself orientation, such that this
relationship becomes stronger for leaders higher on proself orientation.
Hypothese 3C: The positive relationship between the adaptive dimension of leader narcissism and LMX is moderated by leader proself orientation, such that this relationship becomes weaker for leaders higher on proself orientation.
Figure 1, clarifies the propositions made in the present study. A conceptual research model has been created that links narcissism overall, maladaptive narcissism and adaptive narcissism to LMX with both the moderating effects of machiavellianism and proself
19 orientation.
Figure 1. Conceptual model
METHODS
The research design used in this study is cross-sectional in nature and will be assessed using a self-report internet mediated survey.
Sample and Procedure
The data, in this present study, were collected as part of a more general survey about the effects of leadership behavior on employees within organizations. Because of the broader
Narcissism Adaptive Narcissism Maladaptive Narcissism Machiavellianism LMX Proself orientation
+
+
+
+
+
-
--
-20 study, there is also data collected that will not be used in this study. As mentioned before, this study is specific about how adaptive versus maladaptive narcissism in leaders affects the perception of an employee about LMX. In order to say something about the effects of
leadership behavior, dyads are examined. The employees were asked to complete a
questionnaire regarding their immediate leaders (subordinate survey). Their leaders were then asked to complete a similar questionnaire about their subordinates (leader survey). The respondents were approached during a 4-week period mainly by using own contacts which is similar to the convenience sampling method. As a result, the sample of this study will consist supervisor/employee dyads working in different organisations. In a few cases, owners of companies or employees working in the HRM department were asked if they, the company or a part of the company, wanted to participate in this study of leadership and behavior. Using a convenience sampling method was necessary because of the limited resources available to conduct respondents and because of the fact that potential subjects were easily available. Respondents who were willing to participate in the study received an email explaining the intent of the survey and pointing them to the web-link for the online survey and an unique code to make their response anonymous. The surveys started with an informative introduction letter, stressing the voluntary character and ensuring the confidentiality of their responses. Participation was voluntary and respondents were also offered access to the results of this study by filling in mail-adresses at the end of this survey.
Data were collected from two distinct samples, employees and leaders, from diverse organizations and occupational groups. From the 287 employees how received an email with unique link and code a total of 234 employees completed the surveys, representing a 81.1% employee response rate. An total of 289 emails were send to leader resulting in 215 completed leader-surveys, representing a 74.4%leader response rate. Some employees and leaders did not complete the survey in an appropriate manner, resulting in 56 unusable leader and 75
21 unusable employee surveys. These unusable surveys could not be used in the matched-sample analyses, which resulted in 159 completed matched leader-employee dyads. Of the 159 leaders, 61% were man with ages ranged from 20 to 62 years, averaging 42.573 years (s.d. = 10,434). Over 80.5 percent was holding a bachelor degree or higher (high school: 5.0%, MBO: 11.9%, HBO: 37%, University: 43.4%, different: 1.9%). The mean time working with the organization and dyadic tenure with employee was rated at 12.26 (SD=9,808) and 6.01 (SD=5,600). Respectively 6.9 % of leaders was working with their employees on daily basis, 28.9 % on a weekly basis 1.9% on a monthly basis.
From the employees side, 47.8 % was men with ages ranged from 18 to 64 years, averaging 33.780 years (SD = 11.448). Of all employees 40.9 percent was working part-time and 59.1 percent fulltime. Over 72.3 % was holding a bachelor degree or higher (high school: 5.7%, MBO: 18.2 %, HBO: 37.1 %, University: 35.2 %, different: 3.1%). The mean time working for this organization, in years and months, was rated 5.723 (SD= 7,026) and 4.1720 (SD =3,690) respectively.
Measurements
The specific measures are described below, along with the results of calculation the Cronbach alpha coefficients for the various measures. The leaders completed the surveys for the measures of narcissism and machiavellianism whereas the employees completed the measures of LMX and proself orientation.
Narcissism. Raskin and Hall (1979) base the questionnaire of the narcissism construct
on the Narcissistic Personality Inventory (NPI), which is developed. The NPI is a self-report, 40-item forced choice measure in which respondents must choose between two divergent statements. Because this study is only interested in maladaptive and adaptive narcissism some items are dropped. As described in the literature review, the model of Corry et al. (2008) is used as a base. In accordance to suggestions made by Ackerman et al. (2012) some items
22 were deleted. The adaptive narcissism represent the nine items namely: 1, 8, 10, 11, 12, 27, 32, 33, 36. Maladaptive narcissism is covered by eleven items 3, 4, 7, 15, 19, 20, 28, 29, 30, 38, 39 . Examples of items are ‘I think I am a special person,’ and ‘I like to look at myself in the mirror’. The present study uses a 7 point Likert scale for the 20 narcissistic expressed items in order to provide consistency in the formats and reduce confusion among respondents. Overall Cronbachs alpha was measured, giving 0.89. Cronbachs alpha, measured for adaptive narcissism was is .83 and for maladaptive narcissism is .83
Machiavellianism. To assess Machiavellianism, a short version of the the Mach-IV
scale developed by Christie and Geis (1970) was used. This shorter version was developed by Den Hartog and Belschak (2012) who found some items to be aggressive. This questionnaire consists of 8 items with a self-report Likert type sale and scores ranging from 1 to 7, where 1= totally disagree; 4 = neutral and 7= totally agree. The survey composed of 20 items each consisting of a statement. The statements can be classified into three categories describing duplicity tactics, opinions about human nature and concerning themes of abstract morality. Sample items include: ‘‘it is wise to flatter important people’’ and ‘‘most people are basically good and kind’’. Cronbachs alpha is .82
Leader Member Exchange. Leader member exchange was measured from the
employee perspective using a 11-item scale constructed by Liden and Maslyn (1998). This in literature often used (Wang, 2005) survey includes four dimensions; mutual affection,
contribution, loyalty and professional respect. The scale exist of a 5-point scale ranging from 1 (strongly disagree ) to 5 (strongly agree). Examples of sample items are; ‘My supervisor would come to my defense if I were 'attacked' by others’ (loyalty), ‘My supervisor is the kind of person one would like to have as a friend’ (Affect), ‘I admire my supervisor's professional skills’ (Professional respect). and ‘I do work for my supervisor that goes beyond what is specified in my job description’ (contribution). Cronbachs alpha was measured at: .88.
23 Proself orientation. The proself orientation scale was developed by Carter and Greer
(2013) based on the value orientation theory of de Dreu and McCusker (1997). Social value orientation is a relatively stable personality trait, which is distinct from peoples' specific and variable preferences for outcome distributions in specific situations of conflicting interests (Dreu & McCusker, 1997). A 12 item questionnaire was created involving the three social value orientation, namely: the social value orientation, proself orientation and the competitive orientation (Carter & Greer, 2013). Because present study only focuses on proself orientation, only these four items were used, included: ‘Found her/his own goals more important than others’. Again this questionnaire is a self-report Likert type sale, with scores ranging from 1 to 7, where 1= totally disagree; 4 = neutral and 7= totally agree. Cronbach alpha was: .82.
Results
Data preparation
Before running the analyses; the dataset was examined to determine whether variables were normally distributed. In order to do so, skewness and kurtosis were calculated. With all variables a normal distribution could be assumed, except for LMX. The skewness scores of LMX, -1.55 (SD=0.195), which means most frequent scores are grouped towards right. Scores of kurtosis, 4.399 (SD=0.389), indicated a pointy and heavy tailed distribution. To be more accurate a kolmogov-smirnov test was carried out. Scores of the kolmogov-smirnov test showed three of the six variables were significant. Lmx:D(147) = .14, p<.001 Proself: df(147)=.12, p=.001 machtotal: df(147)=.11 , p <.001. This means that the distribution for these variables could be abnormal. However in large samples of data these test can be significant even when the scores are only slightly different form a normal distribution (Field & Gillet, 2010) Therefore, we also looked at the Q-Q plots, P-P plots and histograms. The plots and the histogram showed a normal distribution, except for LMX. Also prominent for
24 the LMX data were two outliers that were detected in the boxplot. Even though the outliers differentiated more then 2.5 SD from the mean we decided to include them in the analyses, as they do seem to represent data.
All items used in the questionnaire were indicative except for one item in the proself-scale, preliminary to conduct in the analysis, this item was recoded. After which the variables for: LMX, narcissism, maladaptive narcissism, adaptive narcissism, machiavellianism and proself orientation, were computed. Lastly these variables were centred to the mean to lessen multicollinearity problems (Field, 2009). Below you find Table 1, which depicts the means, standard deviations, and inter correlations among the variables.
Table 1
Descriptive Statistics and Correlation Matrix
Variable Mean SD 1 2 3 4 5 6 1. Narcissism 4.01 16.08 (.89) 2. Adaptive 4.49 7.74 .87** (.83) narcissism 3.Maladaptive 3.62 9.98 .93** .63** (.83) narcissism 4. Machiavellianism 2.67 7.82 .21* .14 -.20* (.82) 5. Proself orientation 3.06 4.68 .17* .12 .18* .18* (.82) 6.LMX 5.55 8.93 -.10 -.08 -.10 -.12 -.43** (.88)
a. Reliabilities for each scale are listed between parentheses on the diagonal. N = 159; *p < .05. ** p < 0.01.
Testing the main effects
Hypotheses were tested using the function ‘Pairwise deletion’. Only cases without missing data in each pair of variables are being analysed. There is a probability this function introduces bias into estimates and mathematically inconsistent results.
25 The first hypothesis predicted a negative relationship between narcissism overall and LMX. A regression analyses was conducted to see whether this hypothesis was supported by the data. The analysis showed a non-significant negative relationship between narcissism overall and LMX β= -.10, t(148)= -1.27, p = .207. This means, that despite that the results showed a trend in the predicted direction, hypothesis 1A was not supported.
Hypotheses 1B proposed that maladaptive narcissism would be negatively related to LMX. Another regression analysis demonstrated a non-significant negative relationship between maladaptive and LMX β= -.10, t(151)= -1.26, p = .209. Despite being in the predicted direction, hypothesis 1B was not supported.
Lastly, hypothesis 1C predicted a positive relationship between adaptive narcissism and LMX. The regression analysis showed that the data did not support this hypothesis. The result showed a negative non-significant relationship between adaptive narcissism and LMX β= -.08, t(150)= -.96 p = .337. Hypothesis 1C was not supported. The results of the main effects can be found in Table 2.
Table 2
Regression analyses to test the main effects on LMX.
b (SE) β
Overall narcissism -.06 (.05) -.10
Maladaptive narcissism -.09 (.07) -.10
Adaptive narcissism -.09 (.09) -.08
a. N = 149; *p < .05. ** p < 0.01.
Testing moderator effects
Machiavellianism. Hypothesis 2A proposed that the negative relationship between
leader narcissism and LMX would be moderated by leader machiavellianism such that the relationship will be stronger for leaders higher on machiavellianism. Before conducting the analyses interaction term of narcissism total x Machiavellianism was computed. A
26 hierarchical multiple regression was carried out to test whether the data supported the
hypothesis. Results showed that here was no significant association between the main effect of narcissism and Machiavellianism on LMX. The total model explained a variance of 2 %, 𝑅2= .02, F(2,144)=1.45, p =.238. Adding the interaction term of narcissism x
machiavellianism significantly improved the prediction of the model. R2= .09, F (1,143)= 14.44, p < .001. Again, there were no significant main effects, however the interaction term proved to be significantly associated with LMX β= -.30, t(143)= -3.80, p < .001. The total model explained a variance of 11 %, see Table 3.
Table 3
Results of Hierarchical Moderated Regression Analysis for narcissism overall and machiavellianism on LMX
Steps and predictor variables LMX
β 𝑅2 ∆𝑅2 Step 1 Narcissism -.10 Machiavellianism -.08 .02 Step 2 Narcissism -.06 Machiavellianism -.11 Narcissism x Machiavellianism -.30**.11**.09** * Note. N=144; *p < .05. ** p < 0.01.
To have a closer look of the moderation effect, simple slopes were depicted in a graph (Figure 2). High machiavellianism was calculated by adding one standard deviation to the mean, low machiavellianism by subtracting one standard deviation. Again hierarchical
27 multiple regressions were conducted. The regression for high machiavellianism showed a significant negative relationship between narcissism overall and LMX, β= -.42, t(143)= -3.48, p = .001. For low machiavellianism the analysis indicated that there was a non-significant positive relationship between narcissism overall and LMX, β= .20, t(143)= 1.84, p = .067). Concluding that overall narcissism was more strongly, negatively correlated to LMX when machiavellianism is high, as predicted in hypothesis 2A.
Hypothese 2B suggested that the negative relationship between ‘maladaptive’ narcissism and LMX would be moderated by machiavellianism, such that the relationship will be stronger for leaders higher on machiavellianism. The interaction term of maladaptive narcissism x
machiavellianism was calculated. Hierarchical multiple regression analysis were run to test this hypothesis. Results indicated that there were no significant main effects of maladaptive narcissism and Machiavellianism on LMX. The model without the interaction term explained a variance of 2% 𝑅2= .02, F(2,147)=1.47, p =.234. By including the interaction term of
LM
X
Narcissism
low mach high mach
Figure 2. Simple slopes of the interaction effect of narcissism and machiavellianism on the variable LMX
28 maladaptive narcissism x machiavellianism the model significantly enhanced the prediction, R2
= .07, F (1,146)= 10.46, p = .002. The model including the interaction effect explained a total variance of 9% (see Table 4). The interaction term was significant, β= .26, t(146)= -3.23, p =.002.
Table 4
Results of Hierarchical Moderated Regression Analysis for maladaptive narcissism and machiavellianism on LMX
Steps and predictor variables LMX
β 𝑅2 ∆𝑅2 Step 1 Maladaptive narcissism -.07 Machiavellianism -.11 .02 Step 2 Maladaptive narcissism -.09 Machiavellianism -.084
Maladaptive narcissism x Machiavellianism -.26** .09** .07**
* Note. N=144; *p < .05. ** p < 0.01.
A differentiation was made for high and low machiavellianism to look at the possible different effects. Simple slopes were plotted (see Figure 3). The slope for low
machiavellianism showed a non-significant positive relationship between maladaptive and LMX, β=.19 t(146)= 1.65, p = 0.101. While the slope for high machiavellianism
demonstrated a significant negative relationship between maladaptive narcissism and LMX, β= -.36, t(146)= -2.99, p = .003. The hypothesis proposing that there would be a stronger negative relationship between maladaptive narcissism and LMX for higher values of machiavellianism is supported.
29 Figure 3. Simple slopes of the interaction effect of maladaptive
narcissism and machiavellianism on the variable LMX
Hypothesis 2C predicted that the positive relationship between ‘adaptive’ narcissism and LMX would be moderated by machiavellianism, as such that the relationship would be weaker for high machiavellianism leaders. First the interaction term was computed. The hierarchical multiple regression analysis showed that here was no significant association between the main effect of adaptive and Machiavellianism on LMX. The total model explained variance of 2% 𝑅2= .02, F(2,146)=1.28, p =.280. Adding the interaction term of adaptive x machiavellianism significantly improved the prediction of the model. R2= .09, F (1,145)= 14.76, p < .001. There were no significant main effects, however the interaction term proved negatively significant β= -.31, t(145)= -3.84, p < .001. The total model explained a variance of 11%, see Table 5.
LM
X
Maladaptive Narcissism
low mach high mach
30 Table 5
Results of Hierarchical Moderated Regression Analysis for adaptive narcissism and machiavellianism on LMX
Steps and predictor variables LMX
β 𝑅2 ∆𝑅2 Step 1 adaptive narcissism -.06 Machiavellianism -.11 .02 Step 2 adaptive narcissism -.10 Machiavellianism -.05
adaptive narcissism x Machiavellianism -.31** .11** .09**
* Note. N=144; *p < .05. ** p < 0.01.
To investigate how machiavellianism influences the relationship between adaptive narcissism and LMX, a hierarchical multiple regression analysis was conducted for high (+ 1 sd) and low machiavellianism (- 1 sd) (see Figure 4). Data demonstrated that there was a significant negative relationship between adaptive narcissism and LMX when
machiavellianism was high, β=- ,40, t(145)= -3.40, p < .001. A different trend was detected when machiavellianism was low. When Machiavellianism was low, there was a positive trend. The positive relationship between adaptive narcissism and LMX was however non-significant, β= ,21, t(145)= 1.94, p = .054. The hypothesis is partly supported, the prediction was that the relationship between adaptive narcissism and LMX would be less positive when machiavellianism was high. The results showed its not only weaker, it is negative, see Figure
31 3.
Figure 4. Simple slopes of the interaction effect of maladaptive narcissism and machiavellianism on the variable LMX
Lastly to look at which of the narcissism aspects would be the stronger predictor, both were added in the hierarchical multiple regression. In step 1 the main effects for adaptive narcissism, maladaptive narcissism and machiavellianism were added to predict the LMX score. The model proved not significant, 𝑅2= .02, F(3,143)=.96, p = .413. Adding the interaction terms of adaptive narcissism x machiavellianism and maladaptive narcissism x machiavellianism significantly improved the model, the total explained variance resulted in 11%, R2= .09 F (2,141)= 7.42, p = .001. In the final model only one of the variables proved as a significant predictor, namely the interaction term between adaptive x machiavellianism, β= -.23, t(141)= -2.04, p = .044. This interaction term predicted the level of LMX best and there was a negative relationship between the variables. The relationship between maladaptive
LM
X
Adaptive narcissism
low mach high mach
32 narcissism x machiavellianism and LMX was also negative, however not significant, β= -.11, t(141)= -1.00, p = .321.
To conclude, the data showed that the strength of the relationship between overall narcissism, adaptive, maladaptive narcissism on LMX depend on the level of
machiavellianism. The relationship between the different aspects of narcissism and LMX was stronger for higher levels of machiavellianism. When machiavellianism was high, a
significant negative relationship was found. A positive, non significant, trend was discovered between aspects of narcissism and LMX when machiavellianism was low. The interaction between machiavellianism and adaptive narcissism proved to be the better predictor of LMX in comparison to maladaptive x machiavellianism.
Proself orientation. The first hypotheses proposed that the negative relationship
between overall narcissism and LMX would be moderated by leaders’ proself orientation, such that the relationship would be stronger for leaders high on proself orientation. Before conducting the analyses the interaction term of narcissism total x proself focus was computed. A hierarchical multiple regression was conducted to see whether the hypothesis would match with the data. The analysis showed that the first model with only the main effects for
narcissism and proself orientation was significant 𝑅2= .20, F(2,147)=18.35, p < .001. There was a significant negative association between proself and LMX, β= -.44, t(147)= -5.89, p < .001. The interaction term was added in the second step. Adding the interaction term did not significantly improve the model, R2= .02 F (1,146)= 3.53, p = .062. Again a main
33 Table 6
Results of Hierarchical Moderated Regression Analysis for narcissism overall and proself orientation on LMX
Steps and predictor variables LMX
β 𝑅2 ∆𝑅2 Step 1 Narcissism -.02 Proself orientation -.44**.20** Step 2 Narcissism -.01 Proself orientation -.42**
Narcissism x Proself orientation -.14 .22* .02
* Note. N=144; *p < .05. ** p < 0.01.
Despite that the interaction term was not significant simple slopes were depicted in a graph to illustrate the effect of proself orientation on the relationship between narcissism overall and LMX (see Figure 5). The hierarchical multiple regression analyses did not reveal any significant relationships. When leaders scored low on proself orientation there was a non-significant positive relationship between narcissism overall and LMX, β= .11, t(147)= 1.06, p = .291. A non-significant negative association was found between narcissism overall and LMX when leaders scored high on proself orientation, β= -.13, t(147)= -1.37, p = .172. The results show a more negative trend for leaders high on proself orientation however,
34 Figure 5. Simple slopes of the interaction effect adaptive
narcissism and proself orientation on the variable LMX
Hypothesis 3B proposed that the negative relationship between maladaptive
narcissism and LMX depended on the level of proself orientation, such that the relationship will be stronger for leaders higher on proself orientation. The interaction term was computed. The hierarchical multiple regression showed that the model with just the main effects was a significant predictor of the score on LMX, 𝑅2= .19, F(2,150)=17.69, p < .001. Again the relationship between proself orientation and LMX proved significant, β= -.432, t(150)= -5.78, p < .001. Including the interaction term to the model did not significantly improve the model R2
= .01, F (1,149)= 2.62, p = .108. There was however again a significant relationship between proself orientation and LMX, β= -.409, t(149)= -5.41, p < .001. Taking a closer look on the interaction term did not show a significant relationship between maladaptive
narcissism and proself orientation on LMX, β= -.121, t(149)= -1.62, p = .108.
LM
X
Narcissism
low self high self
35 Figure 6. Simple slopes of the interaction effect maladaptive
narcissism and proself orientation on the variable LMX Table 7
Results of Hierarchical Moderated Regression Analysis for maladaptive narcissism and proself orientation on LMX
Steps and predictor variables LMX
β 𝑅2 ∆𝑅2 Step 1 Maladaptive narcissism -.23 Proself orientation -.43**.19** Step 2 Maladaptive narcissism -.01 Proself orientation -.41**
Maladaptive narcissism x Proself orientation-.12 .20* .01
* Note. N=144; *p < .05. ** p < 0.01.
To further investigate the role proself orientation plays on the relationship between maladaptive narcissism and LMX, simple lines were plotted (see Figure 6). When leaders scored high on proself orientation there was a non-significant negative relationship between maladaptive narcissism and LMX, β= -.120, t(149)= -1.25, p = .212. While there was a non-significant positive relationship between maladaptive narcissism and LMX when proself orientation was low, β= .092, t(149)= .897, p = .371. Resulting in a non supported hypothesis. LM X maladaptive narcissism low self high self
36 Figure 6. Simple slopes of the interaction effect maladaptive
narcissism and proself orientation on the variable LMX
Hypothesis 3C predicted that the positive relationship between adaptive narcissism and LMX would be moderated by the proself orientation of the leader, such that there would be a weaker relationship for leaders high on proself orientation. First the interaction term was computed. The hierarchical multiple regression demonstrated that proself focus had a
significant positive relationship with LMX, β= -.44, t(149)= -5.94, p < .001. The total model explained a variance of 20%, 𝑅2= .20, F(2,149)=18.21, p < .001. Adding the interaction term did significantly improve the prediction of the model, R2= .02, F (1,148)= 4.08, p = .045. In the final model there were two significant variables. Firstly, there was a significant negative association between proself orientation and LMX, β= -.42, t(148)= -5.59, p < .001. Second, there was a significant negative relationship between the interaction term of proself orientation and adaptive narcissism on LMX, β= -.15, t(148)= -2.02, p = 0.045.
Table 8
Results of Hierarchical Moderated Regression Analysis for adaptive narcissism and proself orientation on LMX
Steps and predictor variables LMX
β 𝑅2 ∆𝑅2 Step 1 Adaptive narcissism -.03 Proself orientation -.44**.20** Step 2 Adaptive narcissism -.01 Proself orientation -.42**
Adaptive narcissism x Proself orientation -.15* .22* .02
37 * Note. N=144; *p < .05. ** p < 0.01.
To get more insight in the role proself focus plays in the relationship between adaptive narcissism and LMX a graph was depicted (see Figure 7). Both slopes proved to be non-significant. There was a non-signifcant negative relationship between adaptive narcissism and LMX, when leader scored high on the proself orientation scale β= -.14, t(148)= -1.48, p = .141. On the contrary, when leaders scored low on the proself orientation scale there was a non-significant positive relationship between adaptive narcissism and LMX, β= .12, t(148)= 1.18, p = .240. Conclusively, the hypothesis is not supported by the data.
Figure 7. Simple slopes of the interaction effect aladaptive narcissism and proself orientation on the variable LMX
The last hierarchical multiple regression was conducted to investigate whether the adaptive or maladaptive narcissism interaction with proself orientation would be stronger. In the first step the main effects for proself orientation, maladaptive narcissms and adaptive narcissism were added, the model predicted LMX significantly, 𝑅2= .20, F(3,146)=.96, p >
LM
X
aladaptive narcissism
low self
