Mirror Mirror on the Wall, Who Is the best of them all? : leader’s narcissism and subordinate behavior

(1)

MSc. in Business Studies

– Leadership and Management Track

Master Thesis

Mirror Mirror on the Wall, Who Is the Best of Them All?

Leader’s Narcissism and Subordinate Behavior

University of Amsterdam June 30th, 2014 Final version

Supervisor:

Prof. Deanne Den Hartog

Author:

Liang Wang (10605460) Word count: 14777 Page number: 53

(2)

1

Abstract

Nowadays, narcissistic leadership is a hot and still debatable topic in the leadership field. Existing literature has two contrasting perspectives on narcissistic leadership. On one side, it is argued that narcissistic leaders are constructive for organizations, since they are visionary and confident, and they are more likely to be effective in uncertain and dynamic contexts. On the other side, it is believed that narcissistic leaders are detrimental for organizations, since they are arrogant, self-centered, exploitative, hypersensitive and paranoid. This research aims to find empirical evidence to demonstrate whether narcissistic leaders are constructive or destructive for their organizations, by investigating how narcissistic leaders influence their subordinates’ organizational commitment and job performance. A survey is used to collect dyadic data. The sample contains 161 dyads that match between leaders and employees. However, the results of this research do not support the first hypothesis that narcissistic leadership leads to lower levels of employees’ job performance, by reducing the levels of employees’ motivation. Similarity, the results do not support the second hypothesis that narcissistic leadership leads to lower levels of organizational commitment of employees, by creating poorer interpersonal relationships with their subordinates. Moreover, the third hypothesis is also rejected, the results show that tenure does not moderate the relationship between leader’s narcissism and the interpersonal relationships between leaders and employees. The research contributes to the existing literature by showing that narcissistic leadership does not absolutely lead to lower levels of employee performance and organizational commitment than non-narcissistic leadership, and the downside of narcissistic leadership is not found.

(4)

3

Introduction

When talking about narcissism, the very first conceptions come to mind include arrogance, egotism, hubris, grandiosity, paranoia, hypersensitivity, self-importance, self-inflation, and so on and so forth. It comes as no surprise that most of the perceptions associated with narcissism are negative. But undeniably, there is an upside of narcissism as well as the downside. Narcissists are often confident, dominant, creative and innovative, and these are some of the vital aspects for leadership. Hence, both in the past and nowadays, narcissists often rise to positions of power and leadership. Famous examples of successful narcissistic leaders include Bill Gates, Steve Jobs and Jeff Bezos, who are perceived as having narcissistic personality traits, since they often actively promote themselves and their personal philosophies, such as writing their own autobiography, and at the same time, they also dominate business successfully (Maccoby, 2000). Therefore, two scholars named Rosenthal and Pittinsky, respectively, together have created a term called narcissistic leadership to describe this phenomenon (Rosenthal & Pittinsky, 2006).

Narcissistic leadership has become a relatively hot topic in leadership field, since a contradiction in narcissism as leaders exists. On one side, narcissistic leaders are toxic for organizations, due to their negative narcissistic traits such as sensitive to criticism, lack of empathy, intensely competitive, and they are often poor listeners and dislike mentoring. However, on the other side, there are virtues of narcissistic leaders as well, since they are visionary and inspirational, and they attract followers (Maccoby, 2000). Thus, even the pros and cons of narcissistic leaders are widely studied and recognized, the paradox of whether narcissistic leaders are good or bad leaders still remains, no consensus is reached yet, and the two conflicting views lead to different conclusions by different researchers. Moreover, beside narcissistic leaders per se, one related question has started to catch more and more attention,

(5)

4 which is about the influence of narcissistic leaders on their subordinates. Therefore, the research question of this research paper is:

How is leader’s narcissism related to job performance and organizational commitment of their employees?

This paper focuses on this specific relationship because it is unclear in literature and is not fully studied by empirical research. Also, this relationship between leader’s narcissism and their subordinates’ performance has significant implications for practitioners as well, since it is important for them to know whether narcissistic leaders have additional positive or negative indirect impact on organizations via their influences on their subordinates.

To date, the paradox of whether narcissistic leaders are constructive or destructive still remains. However, the paradox might be more complicated than expected. I believe narcissistic leaders not only have direct impact on organizational performance, but also have indirect influences, which suggests that narcissistic leaders’ personality traits can have positive or negative impact on their subordinates, and therefore, these personality traits influence subordinates’ performance and behavior. Narcissistic leaders can have both positive direct and indirect impact on their subordinates and organizations. For example, in uncertain contexts, narcissistic individuals are preferred as the leaders, since they are often confident and visionary, even though other group members are aware of their negative characteristics such as arrogance and exploitativeness (Nevicka, De Hoogh, Van Vianen & Ten Velden, 2013). Moreover, narcissistic leaders are more effective in uncertain and dynamic environments (Nevicka, Ten Velden, De Hoogh & Van Vianen, 2011). However, narcissistic leaders can also have both negative direct and indirect impact on their subordinates as well as organizations. Narcissistic leaders reduce their subordinates’ performance by decreasing the levels of their motivation, engagement and commitment. Also, narcissistic leaders’ direct destructive impacts on organizations from their own actions include gambling attitudes and

(6)

5 risky behaviors, which suggests that sometimes narcissistic leaders might pursue their own personal interests such as prestige, at the cost of their organizations’ interests (Brunell, Gentry, Campbell, Hoffman, Kuhnert & DeMarree, 2008).

Once scholars and practitioners fully understand the role of narcissistic leaders, as well as both the direct and indirect impact of narcissistic leaders on their subordinates and organizations, they can better recommend whether it is better for organizations to hire narcissistic individuals, and also advise them whether and when to promote narcissistic individuals to positions of power and leadership. Moreover, if individuals with high narcissism scores are ultimately detrimental and toxic for organizations, then organizations might want to avoid hiring narcissists in selection and recruitment processes as much as possible by developing more advanced human resource practices, and also to prevent narcissists rising to positions of power and leadership.

In the following section, I first outline the debate about positive and negative sides of narcissistic leaders. Second, I develop my own hypotheses based on literature. Third, I explain the method that is used to collect dyadic data that match between supervisors and subordinates. Two questionnaires are used to collect data from supervisors and subordinates, respectively. The sample contains 161 dyads that match between supervisors and subordinates, demographics such as nationality, type of industry, type of job, age and gender are not restricted at all. Fourth, I conduct a simple mediation regression analysis to test the first hypothesis that suggests subordinate motivation mediates between leader’s narcissism and subordinate performance. Similarly, I conduct a simple mediation regression analysis to test the second hypothesis that suggests the interpersonal relationships between leaders and followers mediate between leader’s narcissism and subordinate commitment. Moreover, tenure, as the time period that leaders and their subordinates interact with each other, is introduced as a moderator and added into the third hypothesis. I conduct a moderated

(7)

6 mediation regression analysis to test the third hypothesis, which suggests that tenure moderates between leader’s narcissism and the interpersonal relationships between leaders and followers. Fifth, I present and discuss my results. In the last section, I conclude, point out limitations as well as future recommendations, and state my contributions.

Literature review

Narcissistic leadership

The term narcissism comes from a Greek myth about a man called Narcissus, who fell in love with his own reflection in water (Nevicka, Ten Velden, De Hoogh & Van Vianen, 2011). Nowadays, narcissism is a personality trait consisting arrogance, lack of empathy, hypersensitivity and anger, insatiable need for recognition, superiority, power and admiration (Rosenthal & Pittinsky, 2006). These negative personality traits are also the downsides of narcissistic leaders.

Narcissistic leaders can be defined as the leaders who make their decisions and judgments mainly based on their arrogance, egotism and self-importance, no matter these decisions and judgments are good or bad (Rosenthal & Pittinsky, 2006). Moreover, narcissistic leaders are mainly motivated and driven by their underlying personal egocentric needs for admiration and power, instead of their genuine concerns for the organizations they lead (Kets de Vries & Miller, 1997). Similarly, it is argued that the main reason why narcissists seek leadership is because they want to garner admiration and power, in order to fulfill their egotistical needs (Glad, 2002). All these personality traits and behaviors of narcissistic leaders are harmful for organizations, and it seems like organizations should avoid hiring narcissists due to the downsides of their personality traits. However, the reality

(8)

7 tells a different story. Many famous and successful leaders, including both political and business leaders, are actually narcissists. Well-known examples include Alexander Hamilton (Chernow, 2004), Steve Jobs (Robins & Paulhus, 2001), Benjamin Netanyahu (Kimhi, 2001), and George W. Bush (Krugman, 2005; Suskind, 2004). Even though narcissists have their shortcomings, they also have many positive attributes that are so essential for leadership, for instance charisma and grand vision (Deluga, 1997), and these positive attributes are one of the reasons why narcissists emerge and are perceived as effective and successful leaders.

While there are many legendary examples of successful narcissistic leaders, it cannot be concluded that narcissists are good leaders, or narcissistic leaders are always effective and valuable for organizations. The existing literature on narcissism and leadership outlines both the pros and cons of narcissistic leaders, therefore, a debate on whether narcissistic leaders are good or bad has started to catch more and more attention, however, no consensus is reached yet.

Narcissistic leader’s emergence and effectiveness

In order to emerge as leaders, people need to be perceived as leaders in their followers’ eyes. Abundant research shows that individuals with high narcissism scores tend to emerge and be perceived as leaders in group-settings, due to their personality traits such as charming, confident and attractive (Brunell, Gentry, Campbell, Hoffman, Kuhnert & DeMarree, 2008; Nevicka, Ten Velden, De Hoogh & Van Vianen, 2011). Likewise, narcissists also emerge and flourish in dynamic, novel, chaotic and crisis situations (Rosenthal & Pittinsky, 2006). This is because narcissists are often visionary, creative and imaginative, hence, they emerge and are perceived as leaders in these kinds of situations since they often come up with innovative ideas, they are able to provide a future direction and guide others (Campbell, Goodie & Foster, 2004; Robins & Beer, 2001). Also, narcissists are supremely confident, they are more

(9)

8 likely to make bold decisions, both good and bad ones, and these bold decisions are usually required in dynamic, chaotic and crisis situations (Galvin, Waldman & Balthazard, 2010). Furthermore, since narcissists are perceived as confident, powerful, decisive and strong-minded, and these attributes are essential for leaders, especially in uncertain contexts. Therefore, narcissists are also preferred as leaders in uncertain contexts, even when individuals are aware of the negative aspects of narcissistic leaders, such as egotism and exploitativeness (Nevicka, De Hoogh, Van Vianen & Ten Velden, 2013).

With regard to the effectiveness of narcissistic leadership, there are also two opposing views. On one hand, because there are many famous and successful narcissists who dominate in politics and business, they attract numerous followers, and hence, it is widely believed that narcissists are born as leaders, they excel at being leaders, and they are generally effective and successful leaders (Brunell, Gentry, Campbell, Hoffman, Kuhnert & DeMarree, 2008).

On the other hand, narcissistic leaders are considered as ineffective in general. Because narcissists are arrogant and egoistical, they are more likely to promote and self-inflate, in order to garner more attention and admiration. Empirical research shows that narcissists are more like to engage in counterproductive behaviors such as sabotage and deception at workplace, with the aim to overrate themselves and underrate others (Judge, LePine & Rich, 2006). Furthermore, because narcissists are self-centered, and they are eager for admiration and recognition from others, hence they often behave in egoistic ways, and they are perceived as coldly indifferent to others as well as lack of empathy. These egocentric behaviors are not viewed positively by their co-workers, and hence, narcissists are not perceived as effective in the eyes of their peers and subordinates (Penney & Spector, 2002). In addition, empirical research also demonstrates that even though narcissistic leaders are perceived as effective leaders, they actually hinder group performance (Nevicka, Ten Velden, De Hoogh & Van Vianen, 2011).

(10)

9 The effectiveness of narcissistic leaders is still debatable, since constructive (helpful) and reactive (unhelpful) narcissistic characteristics are not exclusive, they may inhere in one leader but emerge in different situations, and narcissistic leaders could be helpful and effective for organizations in short to medium term, but harmful and problematic in long term (Stein, 2013). Also, there are many other factors that are able to influence the effectiveness of narcissistic leadership, such as contextual factors. For example, one research shows that narcissistic leaders perform better at decision-making in uncertain situations with ambiguous and misleading information, than non-narcissistic leaders (Byrne & Worthy, 2013). Therefore, even though narcissistic individuals often emerge and are perceived as leaders in team settings, the effectiveness of narcissistic leadership cannot be concluded since contextual factors play an important role in determining it.

Leader’s narcissism and employee performance

Even though it is believed that in general narcissistic leaders are effective leaders due to displayed authority, this does not imply that narcissistic leaders also improve employee performance. In other words, effective narcissistic leaders do not necessary improve the levels of employee performance. For instance, one empirical research suggests that narcissistic leaders are generally perceived as effective leaders, nevertheless, they hamper group performance by inhibiting information exchange among group members (Nevicka, Ten Velden, De Hoogh & Van Vianen, 2011). This example demonstrates that the relationship between leader’s narcissism and employee performance is negative, since narcissistic leaders negatively affect group performance. Information exchange is used as a mediator that influences this negative relationship between leader’s narcissism and employee performance. Moreover, there are many other factors that are also capable of influencing this negative relationship between leader’s narcissism and employee performance. Therefore, this research

(11)

10 paper focuses on subordinate motivation as one potential mediator that plays a similar role as information exchange, and it is expected to mediate the effect of leader’s narcissism on employee performance.

There are several motivation theories that suggest motivation is positively related to effort and hence performance. One famous motivation theory is expectancy theory, it has three components, which are expectancy, instrumentality and valence, and they imply that effort leads to performance, performance leads to outcomes, and outcomes are important or valued, respectively (Grant & Shin, 2011). The expectancy theory indicates that employees are motivated if they believe their outcomes are important and valued, and motivated employees will exert more effort, and hence their performance will improve. In short, subordinate motivation is positively related to employee performance, and employee performance will drop if their levels of motivation decrease.

The expectancy theory suggests that employees engage in certain types of behaviors when they are motivated to do so, and employees are motivated if their outcomes are important and valued. In addition, it is also essential that employees believe their performance will be rewarded, in forms of increased pay, recognition or promotion. Since narcissistic leaders are arrogant and self-important, they tend to overestimate themselves and devalue others (Judge, LePine & Rich, 2006), therefore, subordinates may assume that their narcissistic leaders will not view their outcomes as important or valuable, and they will also be devalued by their narcissistic superiors, hence, subordinate motivation will decrease. In other words, with respect to the expectancy theory, it is reasonable to assume that there is a negative relationship between leader’s narcissism and subordinate motivation.

Similarly, equity theory suggests that employees need to feel fairness in order to be motivated, and they expect their outcomes such as pay and recognition are matched with their inputs (Grant & Shin, 2011). However, narcissistic leaders have insatiable needs for

(12)

11 recognition and power, they think they deserve more because they are superior and others are inferior (Glad, 2002), hence, subordinates might believe that they will not get fair treatments from their narcissistic leaders, and as a result, subordinate motivation will decrease. According to the equity theory, it is also logical to assume that leader’s narcissism is negatively related to subordinate motivation.

According to identification-based motivational and social learning processes, ethical leadership affects followers’ self-concept and beliefs by role modeling, and thus positively influences followers’ motivation and performance (Den Hartog & Belschak, 2012). In other words, subordinate motivation will increase if they perceive their leaders as ethical. However, one empirical research reveals that narcissistic leaders are perceived as ineffective and unethical by followers in highly ethical contexts. In addition, unethical narcissistic leaders’ actions and behaviors also convey negative messages to their followers, such as lack of integrity and fairness (Hoffman, Strang, Kuhnert, Campbell, Kennedy & LoPilato, 2013). Moreover, narcissistic leaders are known as manipulative and deceptive, and they are generally viewed as unethical by their followers (Glad, 2002). Because narcissistic leaders are unethical in the eyes of their followers, subordinate motivation will decrease. This reasoning also implies that leader’s narcissism is negatively related to subordinate motivation.

Narcissistic leaders often deceive, manipulate and intimidate so as to secure their power, authority and leadership positions, and these skills employed by them are not favorably valued by their subordinates (Glad, 2002). Moreover, due to their insatiable need for recognition, narcissistic leaders are more likely to self-promote and self-nominate, and they also tend to devalue others, especially their competitors (Hogan, Raskin & Fazzini, 1990), and this will increase the tendency for narcissistic leaders to employ these deceptive, manipulative and intimidating skills. Narcissistic leaders demotivate subordinates by

(13)

12 engaging in these types of behaviors, since their subordinates might believe their efforts and contributions are not recognized or fairly valued by their superiors, and hence, subordinate motivation will decrease.

In a nutshell, narcissistic leaders are exploitative, dishonest and manipulative, they take excessive credit for success but blame others for failure, they even blame others for their own mistakes and shortcomings (Campbell & Sedikides, 1999), and they also overvalue themselves and undervalue others (Morf & Rhodewalt, 2001). Therefore, I expect that narcissistic leaders will reduce subordinate motivation, and thereby hinder employee performance, since motivation is positively related to employee performance (Figure 1 - Mediating Effect of Subordinate Motivation).

Hypothesis 1: The negative relationship between leader’s narcissism and employee

performance is mediated by subordinate motivation.

Figure 1 - Mediating Effect of Subordinate Motivation

Leader’s narcissism and organizational commitment of employees

Narcissism is a personality trait, and personality traits play an essential role in interpersonal performance. Personality traits, such as the Big Five personality traits, which include Agreeableness, Extraversion, Openness, Neuroticism and Conscientiousness can be linked to interpersonal performance, and one empirical research shows that individuals with different personality profiles are likely to experience different types of interpersonal problems

Leader's Narcissism Subordinate Motivation

Employee Performance

(14)

13 (Nysæter, Langvik, Berthelsen & Nordvik, 2009). There are several ways to classify interpersonal performance, such as dominance versus submission, or hostility versus friendliness (Alden, Wiggins & Pincus, 1990; Gurtman, 1996). Narcissists are often classified as dominant and hostile, which are capable of deteriorating interpersonal relationships and decreasing interpersonal performance. Hence, it is argued that narcissism is negatively related to interpersonal effectiveness and interpersonal performance (Blair, Hoffman & Helland, 2008).

Narcissists view themselves as superior and others as inferior, they are insensitive to others’ needs and concerns. Due to narcissists’ self-enhancement and devaluation of others, narcissistic leaders tend to see themselves as extraordinary performers, overestimate their own performance, and derogate those who outperform them or whom they believe threaten or rate them less favorably (Judge, LePine & Rich, 2006). These hostile and aggressive behaviors of narcissistic leaders can jeopardize the interpersonal relationships between them and their subordinates. Moreover, narcissistic leaders are grandiose, they always seek feedback for admiration and recognition, and they also engage in hostile behaviors such as verbal aggression that turn people away (Morf & Rhodewalt, 2001). Besides, they devalue those who provide negative feedback to them (Kernis & Sun, 1994). All these behaviors have negative consequences on the interpersonal relationships between narcissistic leaders and their subordinates. Therefore, it is reasonable to assume that leader’s narcissism is negatively related to the interpersonal relationships between leaders and employees.

Amorality is one of the personality traits of narcissists, and it suggests that narcissists lack senses of morality and ethic, and they do not care about right and wrong, either. Moreover, narcissists might cross moral and ethical boundaries if necessary, with the aim to fulfill their personal desires (Rosenthal & Pittinsky, 2006). Research confirms this viewpoint by demonstrating that narcissistic leaders are perceived as ineffective and unethical by

(15)

14 followers in highly ethical contexts (Hoffman, Strang, Kuhnert, Campbell, Kennedy & LoPilato, 2013). Unethical narcissistic leaders damage the interpersonal relationships between them and their subordinates, since unethical leaders are not able to build trusting relationships. In addition, subordinates see unethical leaders as deceptive, lack of integrity and fairness, which results in damaged interpersonal relationships between them. Thus, it is expected that leader’s narcissism will be negatively related to the interpersonal relationships between leaders and followers.

Furthermore, narcissism is also called the dark side of charisma (Den Hartog & Koopman, 2001), which emphasizes on the unattractive personality traits of narcissists, and it suggests a negative relationship between narcissism and interpersonal skills (Campbell, Hoffman, Campbell & Marchisio, 2011). Also, narcissistic leaders often engage in haughty behaviors that are called “oral rage”, which indicates that narcissistic leaders tend to express their anger orally in aggressive ways towards their co-workers at work (Rosenthal & Pittinsky, 2006). Therefore, I expect that narcissistic leaders will have poor interpersonal relationships with their subordinates. In other words, leader’s narcissism is negatively related to the interpersonal relationships between leaders and employees.

A handful of empirical research shows that there is a negative relationship between leader’s narcissism and organizational commitment of employees, which suggests that the higher the scores of leaders’ narcissism, the lower the levels of employees’ organizational commitment. The reason is because organizational commitment of employees is related to the interpersonal relationships between leaders and employees. Employees are more committed to their leaders and organizations if they have better interpersonal relationships with their leaders. However, narcissistic leaders are manipulative and exploitative, these attributes are negatively related to relationships-oriented behaviors, hence, employees are less committed due to poorer interpersonal relationships between narcissistic leaders and employees

(16)

15 (Campbell, Hoffman, Campbell & Marchisio, 2011). Therefore, I expect that narcissistic leaders will have poor interpersonal relationships with their subordinates, and hence reduce employees’ organizational commitment, since the level of organizational commitment of employees is positively related to the interpersonal relationships between leaders and employees (Figure 2 - Mediating Effect of Interpersonal Relationship).

Hypothesis 2: The negative relationship between leader’s narcissism and organizational

commitment of employees is mediated by the interpersonal relationships between leaders and employees.

Figure 2 - Mediating Effect of Interpersonal Relationship

The presence of narcissistic leaders is oftentimes charming, charismatic and attractive in the beginning (Deluga, 1997), since they are extraverted, humorous, entertaining and creative (Miller & Campbell, 2008). However, the pleasant presence is used to cover the inner dark side of narcissistic leaders. Deep down they are often cold, indifferent, ruthless, abusive, exploitive and aggressive (Rosenthal & Pittinsky, 2006). Also, narcissistic leaders are often perceived as abusive and manipulative (Hogan, Curphy & Hogan, 1994). From this viewpoint it is reasonable to assume that narcissistic leaders are appealing, engaging and glamorous in the short term, due to their strong social skills and charisma. However, their personal glory fades with time, and their negative personality traits shine through in the long term, which can potentially hamper the interpersonal relationships between narcissistic leaders and their co-workers as well as subordinates. Moreover, empirical research shows that

Leader's Narcissism Interpersonal Relationship

Organizational Commitment

(17)

16 narcissists often receive lower peer likability than non-narcissists, due to their negative personality traits such as arrogance, egotism and hubris, which suggests that narcissists tend to have poorer interpersonal relationships with their co-workers than non-narcissists, especially in the long term (Harms, Wood & Roberts, 2006). Therefore, tenure, defined as the time period that the narcissistic leaders work together with their subordinates, is believed to moderate the relationship between leader’s narcissism and the interpersonal relationships between narcissistic leaders and their subordinates. In other words, I expect that the interpersonal relationships between narcissistic leaders and their subordinates will worsen over time (Figure 3 - Moderating Effect of Tenure).

Hypothesis 3: The negative relationship between leader’s narcissism and organizational

commitment of employees is mediated by the interpersonal relationships between leaders and employees, and the negative relationship between leader’s narcissism and the interpersonal relationships is moderated by tenure, such that this relationship is weaker/stronger for lower/higher values of tenure.

Figure 3 - Moderating Effect of Tenure

In a nutshell, Figure 4 – Conceptual Model shows that the combination of the three hypotheses formulated above in the literature review section. Moreover, this conceptual

Leader's Narcissism Tenure Interpersonal Relationship Organizational Commitment

(18)

17 model also captures the main purpose of this research paper, since it demonstrates the way the research question ‘How is leader’s narcissism related to job performance and organizational commitment of their employees?’ will be answered. This research paper tries to find empirical evidence to show that leader’s narcissism is related to employee performance through subordinate motivation, and leader’s narcissism is related to organizational commitment of employees through the interpersonal relationships between leaders and employees.

Figure 4 – Conceptual Model

Data and method

The first part of this section explains the research method of this research. Next, the sample is described. Last, a detailed description of all the variables used in this research is presented, together with their measurement instruments.

Research method

A survey was used to collect cross-sectional data, and the survey consisted two questionnaires, one for supervisors and the other for subordinates, and they needed to match.

Leader’s Narcissism Subordinate Motivation Employee Performance Interpersonal Relationship Organizational Commitment Tenure

(19)

18 The purpose of using two separate questionnaires is to collect dyadic data, which match between subordinates and supervisors. The format of both questionnaires was closed ended, participants needed to choose the extent to which they agree or disagree with each statement in the questionnaires. In addition, in order to collect demographics such as age, gender, level of education and tenure, open-ended questions were used at the end of the questionnaires, and categories were determined after data collection phase without overlap or gap. Two versions were used for both questionnaires, one was English version, and the other was Dutch version. Participants could choose which language they prefer. Double translation was used to ensure that the contents in both versions were identical. Both questionnaires were self-administered, online- and paper-based questionnaires were used together. An invitation letter explaining the purpose of the survey, instruction and participants’ right was included, and it was also used as the cover page of both questionnaires (see Appendix A - Questionnaire). All participants were anonymous, they participated voluntarily, and there was no return for participating in this research survey. All information was kept confidential, and all data collected were used for research purposes only. Individual participants could not be traced back by using the information they provided.

Snowball sampling was used as the sampling technique for the survey. I mainly used my personal contacts to collect data, and I also asked my participants to send the questionnaires to their acquaintances, with the purpose to expand the sample size. Therefore, one advantage of using snowball sampling is a high response rate because of personal contacts. I sent out 232 dyadic questionnaires and received 161 completed dyads, thus, the response rate was 69%. However, one limitation of using snowball sampling is that the sample might not be very representative, since the participants were not randomly drawn from the whole population.

(20)

19

Sample description

The sample size is 161 dyads, it included people (both leaders and subordinates) who were employed at that time worldwide. There was no restriction on age, gender, level of education, type of industry, position or nationality. These demographics were measured at the end of the questionnaires for both supervisors and subordinates.

Both employees and supervisors worked in a wide range of jobs, including education, government, health, sport and wellness, accountancy, business service, medical, ICT, transport, real estate, sales, banking, internet, import and export, IT, trade, tourism, e-commerce, aviation, automotive, retail, consultancy, logistics, fashion, software development, and engineering. Of the employees, 37.9% were male; and of the leaders, 64% were male. The average age of the employees was 30.86 years (SE = 10.19); and the average age of the leaders was 40.78 years (SE = 10.39). The employees have worked for their supervisors for 2.86 years (SE = 3.11) on average. Regarding to the level of education, 77.7% of the employees have obtained an undergraduate level or above; and 90.6% of the supervisors have obtained an undergraduate level or above. With respect to nationality, 83.9% of the employees were Dutch; 13.7% of the employees were Chinese; 2 (1.24%) employees were Japanese. Also, there was 1 (.62%) Korean employee and 1 (.62%) Ghanaian employee. Moreover, 82.6% of the leaders were Dutch; 13% of the leaders were Chinese; 3 (1.86%) leaders were British. Also, there was 1 (.62%) Korean leader; 1 (.62%) Japanese leader; 1 (.62%) Pakistani leader; and 1 (.62%) Singaporean leader. Concerning the type of job, 34.8% of the employees had part-time jobs, and 65.2% of them had full-time jobs.

Description of measures

(21)

20 disagree with each statement, where 1 means strongly disagree and 7 means strongly agree. All measures are at the individual level.

Independent variable

Leader narcissism (40 items; α = .934). Leader’s narcissism was assessed by means of the Narcissistic Personality Inventory (NPI) (Raskin & Terry, 1988). Raskin and Terry define leader’s narcissism as a combination of authority, exhibitionism, superiority, vanity, exploitativeness, entitlement, and self-sufficiency. An example item of the NPI scale is “I am a born leader” (+). Items were rated on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).

Dependent variables

Organizational commitment (6 items; α = .895). Organizational commitment was assessed by means of the Organizational Commitment Scale (Meyer, Allen & Smith, 1993). Meyer, Allen and Smith define organizational commitment as a combination of affective, continuance, and normative commitment. An example item of the Organizational Commitment Scale is “I really feel as if this organization's problems are my own” (+). Items were rated on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).

Leader perception of employee in-role behavior (7 items; α = .854). Employee performance was assessed by means of In-Role Behavior (IRB) Scales (Williams & Anderson, 1991). Williams and Anderson define IRB as employees’ required activities. An example item of the IRB Scales is “Adequately completes assigned duties” (+). Items were rated on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).

(22)

21

Mediating variables

Leader-Member Exchange (LMX) (11 items; α = .895). Interpersonal relationship was assessed by means of the LMX Scale (Liden & Maslyn, 1998). Liden and Maslyn define LMX as the exchange relationship between a leader and an employee. An example item of the LMX Scale is “My supervisor is a lot of fun to work with” (+). Items were rated on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).

Motivation (6 items; α = .721). Motivation was assessed by means of the Motivation Scale (Hackman & Lawler, 1971). Hackman and Lawler define motivation as a combination of variety, autonomy, task identity and feedback dimensions. An example item of the Motivation Scale is “I do high quality work” (+). Items were rated on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).

Moderating variable

Tenure, as a moderator, is measured in terms of years at the end of the questionnaires in the demographics. Tenure is defined as the number of years the leaders supervising their subordinates. An open-ended question was used to measure tenure. The purpose of defining tenure this way is to measure and emphasize on the time period in which leaders and their subordinates interact with each other.

Control variables

Demographics including gender, age, level of education, tenure, frequency of contact, type of job, and nationality for both leaders and subordinates were included as control variables. Moreover, survey source and survey language were also included in the control variables. These personal characteristics and background information were collected at the end of the questionnaires in the demographics.

(23)

22 Gender is included as a control variable, since male and female might have different approaches towards the same topic (Eckel & Grossman, 1998). In particular, male and female might differ in their interpretation of work, leadership and performance. Empirical evidence shows that female leaders are perceived as less effective than their male counterparts (Eagly & Karau, 2002), and it is also more difficult for females to become leaders (Eagly & Carli, 2007). Moreover, male and female leaders often engage in different types of behavior, for example, female leaders are more transformational whereas male leaders are more transactional, and this suggests that males and females have different leadership styles (Eagly, Johannesen-Schmidt & Van Engen, 2003). Thus, it is important that the gender of leaders is controlled for in the research. Furthermore, employees’ gender should also be controlled for. Empirical research suggests that females are more likely to choose and engage in less difficult tasks than males, and engagement in challenging tasks is positively related their performance evaluation (De Pater, Van Vianen, Fischer & Van Ginkel, 2009). Therefore, the gender of both leaders and employees should be controlled for.

Age of both supervisors and employees should be controlled for as well. Empirical evidence shows that leaders with different ages behave differently, in particular, younger leaders are more likely to engage in relationship-building activities than older leaders (Gilbert, Collins, & Brenner, 1990). Regarding to employee age, empirical research suggests that age

and job performance are not related in general (McEvoy & Cascio, 1989). However, one

recent study shows that younger and older employees are perceived differently in terms of

key job performance dimensions (Bertolino, Truxillo & Fraccaroli, 2013). Therefore, it is

better to control employee age in this research.

Research has showed that the level of education is somehow positively related to

individuals’ knowledge, skills and abilities (KSAs), and KSAs are also positively related to

(24)

23

education might be perceived as more effective, and employees with higher levels of

education might be perceived as more capable and productive. Thus, the levels of education

of both leaders and employees should be controlled for.

Two contrasting theories are used to explain the relationship between tenure and job

performance. One is human capital theory, which argues that employees gain more

knowledge and skills with longer tenure, and thus their performance will improve. The other

is based on job design, which argues that the longer the tenure, the more bored employees get,

and employee performance decreases since they are less motivated (Ng & Feldman, 2013).

The effect of tenure on job performance is still debatable, however, it is better to include

tenure of both leaders and employees in the control variables.

Other demographics including type of job and frequency of contact are also controlled

for. Type of job is controlled for, since people who do full-time jobs might perform

differently than people who do part-time jobs. Frequency of contact is controlled for, since it

might influence the interpersonal relationships between leaders and employees.

Nationality of both leaders and employees is controlled for, since people come from

different countries might differ in many dimensions (Drogendijk & Slangen, 2006). Hofstede

has created four cultural dimensions to show that people with different nationalities differ,

which are ‘power distance’, ‘uncertainty avoidance’, ‘individualism’ and ‘masculinity’

(Hofstede, 1983), later on, Schwartz has extended the four cultural dimensions to seven

dimensions, which are ‘conservatism’, ‘intellectual autonomy’, ‘affective autonomy’,

‘hierarchy’, ‘egalitarian commitment’, ‘mastery’ and ‘harmony’ (Schwartz, 1994). These

cultural dimensions can be used to explain the success of multinational enterprises as well as

the behaviors of employees. Because people come from different countries have different

perceptions, behaviors and performance, therefore, nationality is controlled for. Similarly, the

(25)

24 In this research, paper survey and online survey were used together. The main reason to use both paper survey and online survey is because it was easier and more convenient for researchers to collect data. The survey source is also controlled for, just in case if these two survey methods lead to differences in responses and information collected. Overall, all the mentioned control variables are outlined in Table 1 - Control Variables.

Table 1 – Control Variables

Control variable Items

Gender 1 - Male

2 - Female

Age Open-ended question

Level of Education 1- High school or below

2 - College 3 - Bachelor 4 - Master

5 - Doctorate or above

Tenure Open-ended question

Type of Job 1 - Part time

2 - Full time

Frequency of Contact 1 - Daily

2 - Weekly 3 - Monthly

Nationality Open-ended question

Survey Language 1 - Dutch

2 - English

Survey Source 1 - Paper survey

2 - Online survey

Results

This section reports the results of this research. First, data preparation including cleaning data, handling miss data, recoding counter-indicative items and computing scale variables is demonstrated. Subsequently, reliability analysis is conducted, and the results are reported. The Cronbach’s alpha coefficients were obtained for the independent variable leader’s

(26)

25 narcissism, the two dependent variables organizational commitment and employee performance, as well as the two mediating variables LMX and subordinate motivation. Next, descriptive statistics are showed for the continuous and the categorical variables. Moreover, a correlation analysis is conducted, and the correlations are reported. Furthermore, a normality analysis is conducted, in order to check whether the two dependent variables organizational commitment and employee performance, the two mediating variables LMX and subordinate motivation, and the moderating variable tenure are approximately normally distributed or not. Finally, two simple mediation regression analyses and a moderated mediation regression analysis are conducted in order to test the hypotheses formulated in the theoretical section.

Data preparation

My sample consists 161 dyads, which match between subordinates and supervisors. After collecting data, the first step is data cleaning. Because I collected data together with my thesis group members, the raw data contained more items than I need. Therefore, I dropped all irrelevant items, and only kept the items that are needed for my analysis. Overall, my data include 7 categories. The first category is my independent variable, which is leader’s narcissism, and it contains 40 items. The second category is organizational commitment, which is a dependent variable, and it contains 6 items. The third category contains leader perception of employee in-role behavior, in other words, employee performance, which is the other dependent variable, and it contains 7 items. The fourth category is Leader-Member Exchange (LMX), which is a mediator and represents the interpersonal relationships between leaders and employees, and it contains 11 items. The fifth category is subordinate motivation, which is the other mediator, and it contains 6 items. The sixth category is tenure, which is the moderator, and it is measured by using the time period that employees have worked for their leaders. The last category is demographics, which include age, gender, level of education,

(27)

26 type of job, frequency of contact, industry, and nationality.

The second step is to deal with missing values. I used frequency in SPSS to check how many missing values are there for each item. 11 out of 40 items in leader’s narcissism scale have 1 missing value each, and the missing percentage for each item is 0.6%. Organizational commitment scale has no missing values. Only 1 item in employee performance scale has 1 missing value, and the missing percentage is 0.6%. Only 1 item in LMX scale has 2 missing values, and the missing percentage is 1.2%. Only 1 item in subordinate motivation scale has 1 missing value, and the missing percentage is 0.6%. All demographics for both employees and leaders have no missing values. The missing percentages for all items with missing values are less than 10%, therefore, I can use hotdeck to replace missing values. I ran correlation and found a deck variable that correlates the highest with each variable with missing values, and I also made sure that the correlation is significant at 0.01 level, then I used hotdeck to replace missing values. I repeated the step for each variable with missing values, since each variable with missing values correlates highly with different deck variables. I reran frequencies to check and ensure all missing values have been replaced. From now on, items with replaced values are used for further analysis.

The third step is to recode counter-indicative items. Only items number 6 and 7 in employee performance scale are counter-indicative (see Appendix A - Questionnaire). I used SPSS to transform and recode these two items into same variables. Because I use a 7-point Likert scale, the scale midpoint is 4, so old and new values will mirror the scores around 4. Therefore, (1=7), (2=6), (3=5), (4=4), (5=3), (6=2), and (7=1). Moreover, recoded items are used for further analysis.

The fourth step is to compute items into scale variables. All items under the same scale are computed into a scale mean. The same step is repeated to compute the scale mean for each scale. This means that 40 items under leader’s narcissism is computed into a leader’s

(28)

27 narcissism scale mean, 6 items under organizational commitment is computed into an organizational commitment scale mean, 7 items under employee performance is computed into an employee performance scale mean, 11 items under LMX is computed into a LMX scale mean, and 6 items under subordinate motivation is computed into a subordinate motivation scale mean. These 5 scale means are used as scale variables for further analysis.

Reliability analysis

The fifth step is to compute scale reliability. All items under the same scale are put together to compute scale reliability. The values of the Cronbach’s alpha coefficient for leader’s narcissism (α = 0.934), organizational commitment (α = 0.895), employee performance (α = 0.854), LMX (α = 0.895), and subordinate motivation (α = 0.721) are all larger than 0.70 (Table 2 – Reliability Statistics), and this suggests that all these variables are internally consistent (Pallant, 2010). Moreover, the values of the corrected item total correlations for each item under each scale are all larger than 0.30, and the differences between the values of the Cronbach’s alpha if item deleted and the values of the Cronbach’s alpha for each item under each scale are all smaller than 0.10. Therefore, if any one item under a scale variable is deleted, the reliability of the scale variable will not be significantly impacted. Overall, reliability for my scales is relatively high and acceptable, and all variables have a good internal consistency, which means the data collection techniques I used yield consistent findings.

(29)

28

Table 2 – Reliability Statistics

Variable Cronbach’s Alpha N

Narcissism .934 40 Commitment .895 6 Performance .854 7 LMX .895 11 Motivation .721 6

Descriptive statistics

There are ten continuous variables used in this research paper. The independent variable is leader’s narcissism, the two dependent variables are employee performance and organizational commitment, the two mediating variables are subordinate motivation and interpersonal relationship (LMX) between leaders and subordinates, and the moderating variable is tenure. These six variables are all continuous variables. Moreover, four control variables employee age, leader age, employee tenure and leader tenure are also continuous variables.

Table 2 – Descriptive Statistics for Continuous Variables, demonstrates the descriptive statistics for the ten continuous variables mentioned above. Number of observations, minimum, maximum, mean and standard deviation are showed. The independent variables leader’s narcissism has a range from 2.40 to 6.53, with a mean of 4.26, and the standard deviation is 0.73. The first dependent variable organizational commitment has a range from 1.83 to 7.00, with a mean of 4.99, and the standard deviation is 1.09. The second dependent variable employee performance has a range from 2.86 to 7.00, with a mean of 5.78, and the standard deviation is 0.75. The first mediating variable LMX has a range from 2.63 to 7.00, with a mean of 5.43, and the standard deviation is 0.88. The second mediating variable subordinate motivation has a range from 3.33 to 7.00, with a mean of 5.74, and the standard deviation is 0.66. A 7-point Likert scale is used for all these 5 variables mentioned above, and

(30)

29 the possible range is from 1 to 7.

Descriptive statistics for the demographics are also showed in Table 3 – Descriptive Statistics for Continuous Variables. The moderating variable tenure is defined as the number of years the leaders supervising their subordinates, and it has a range from .10 to 20.00, with a mean of 2.86, and the standard deviation is 3.11. The control variable employee age has a range from 20 to 61, with a mean of 30.86, and the standard deviation is 10.19. The control variable leader age has a range from 22 to 62, with a mean of 40.78, and the standard deviation is 10.39. The control variable employee tenure has a range from .20 to 30.00, with a mean of 4.21, and the standard deviation is 4.86. The control variable leader tenure has a range from .30 to 38.00, with a mean of 9.35, and the standard deviation is 7.86. Employee tenure is defined as the number of years the employees have been working for their organizations, and leader tenure is defined as the number of years the leaders have been working for their organizations, hence, there is a difference between the moderating variable tenure and these two control variables.

Table 3 – Descriptive Statistics for Continuous Variables

Variable N Min Max Mean SD

Leader’s narcissism 161 2.40 6.53 4.26 .73

Organizational commitment 161 1.83 7.00 4.99 1.09

Employee performance 161 2.86 7.00 5.78 .75

LMX 161 2.63 7.00 5.43 .88

Motivation 161 3.33 7.00 5.74 .66

Tenure (leader supervising employee) 161 .10 20.00 2.86 3.11

Employee age 161 20.00 61.00 30.86 10.19

Leader age 161 22.00 62.00 40.78 10.39

Employee tenure 161 .20 30.00 4.21 4.86

Leader tenure 161 .30 38.00 9.35 7.86

There are ten categorical variables used in this research paper, namely, employee gender, employee level of education, employee nationality, type of job, leader gender, leader level of education, leader nationality, frequency of contact, survey source and survey language. All

(31)

30 these ten categorical variables are control variables, and they are included in the demographics. Also, all these categorical variables are filled in by all respondents, which indicates that the total frequency for each categorical variable equals to the sample size 161, and the total cumulative percentage for each categorical variable is 100%.

Table 4 – Frequencies for Categorical Variables, demonstrates the frequencies and percentages for the ten categorical variables mentioned above. The frequency of employee gender shows that 61 (37.9%) respondents were male, and 100 (62.1%) respondents were female in the total sample. With respect to the employee level of education, only 9 (5.6%) employees had a high school level or below, 27 (16.8%) of them have obtained a College degree, 60 (37.3%) of them have obtained a Bachelor level, 64 (39.8%) of them have obtained a Master level, and only 1 (.6%) employee from the whole sample had a Doctorate level or above. Regarding to employee nationality, the majority was Dutch, 135 (83.9%) of them were Dutch, and only 26 (16.1%) respondents were non-Dutch. Concerning the type of job, just 56 (34.8%) employees had part-time job, whereas 105 (65.2%) of them worked for full-time.

Moving on to the demographics of leaders, in terms of leader gender, 103 (64.0%) of the leaders were male, and only 58 (36.0%) of them were female, which shows a contrast with employee gender, since the majority of the employees were female. With respect to leader level of education, only 2 (1.2%) leaders reported that they have obtained a high school level or below. 13 (8.1%) of them have obtained a College level, 68 (42.2%) of them have obtained a Bachelor level, 72 (44.7%) of them have obtained a Master level, and there were 6 (3.7%) leaders who have obtained a Doctorate level of above. Similar to employee level of education, the majority of both employees and leaders have obtained either a Bachelor or Master level. However, leaders tend to have higher levels of education than employees, since leaders have less people who obtained a high school level or below, and more people who have obtained a

(32)

31 Doctorate level or above. Concerning the leader nationality, 133 (82.6%) of the leaders were Dutch, and only 28 (17.4%) respondents were non-Dutch, which is quite the same as employee nationality. With regard to frequency of contact, 95 (59%) dyads met daily, 59 (36.6%) dyads met weekly, and only 7 (4.3%) dyads met monthly.

Two forms were used in the survey, one is paper survey and the other is online survey. Looking at survey source, it demonstrates that 115 (71.4%) of the dyads used online survey, and only 46 (28.6%) of the dyads used paper survey. Moreover, two languages were used in the survey as well, one is Dutch and the other is English. Concerning the survey language, 134 (83.2%) of the dyads used the Dutch version, and only 27 (16.8%) dyads used the English version.

Table 4 – Frequencies for Categorical Variables

Variable Level Frequency % Valid % Cumulative %

Gender (employee) 1 - Male 61 37.9 37.9 37.9 2 - Female 100 62.1 62.1 100.0 Level of Education (employee)

1 - High school or below 9 5.6 5.6 5.6

2 - College 27 16.8 16.8 22.4 3 - Bachelor 60 37.3 37.3 59.6 4 - Master 64 39.8 39.8 99.4 5 - Doctorate or above 1 .6 .6 100.0 Nationality (employee) 1 - Dutch 135 83.9 83.9 83.9 2 - Non-Dutch 26 16.1 16.1 100.0 Type of Job (employee) 1 - Part time 56 34.8 34.8 34.8 2 - Full time 105 65.2 65.2 100.0 Gender (leader) 1 - Male 103 64.0 64.0 64.0 2 - Female 58 36.0 36.0 100.0 Level of Education (leader)

1 - High school or below 2 1.2 1.2 1.2

2 - College 13 8.1 8.1 9.3 3 - Bachelor 68 42.2 42.2 51.6 4 - Master 72 44.7 44.7 96.3 5 - Doctorate or above 6 3.7 3.7 100.0 Nationality (leader) 1 - Dutch 133 82.6 82.6 82.6 2 - Non-Dutch 28 17.4 17.4 100.0 Frequency of Contact 1 - Daily 95 59.0 59.0 59.0 2 - Weekly 59 36.6 36.6 95.7 3 - Monthly 7 4.3 4.3 100.0

(33)

32 Survey Source 1 - Paper 46 28.6 28.6 28.6 2 - Online 115 71.4 71.4 100.0 Survey Language 1 - Dutch 134 83.2 83.2 83.2 2 - English 27 16.8 16.8 100.0 Total 161 100.0 100.0 100.0

Correlations analysis

The sixth step is to run correlation with all scale variables. Table 5 – Correlation Matrix, demonstrates the results from the bivariate correlation analysis. The purpose of the correlation analysis is to show the dependence of two variables, in other words, the extent to which two variables vary together.

The independent variable leader’s narcissism does not correlate with the two dependent variables organizational commitment (β = .082, p > .05) and employee performance (β = -.022, p > .05), or the two mediating variables LMX (β = -.051, p > .05) and subordinate motivation (β = .041, p > .05). Leader’s narcissism does not correlate with the moderating variable tenure (β = -.036, p > .05), either. With respect to the control variables, the independent variable leader’s narcissism is only positively correlated with the control variables employee nationality (β = .329, p < .01), leader nationality (β = .207, p < .01) and survey language (β = .288, p < .01) at the 0.01 significance level. Since leader’s narcissism has a weak and positive correlation with both the nationality of employees and leaders, as well as the language used in the survey, it suggests that nationality and culture might play a role in the interpretation of narcissistic leadership. Furthermore, the independent variable leader’s narcissism is negatively correlated with the control variable leader tenure at the 0.01 significance level (β = -.208, p < .01). Moreover, the independent variable leader’s narcissism also has a weak and negative correlation with the control variable leader age at the 0.05 significance level (β = -.201, p < .05). However, the independent variable leader’s narcissism does not correlate with any other control variables.

(34)

33 The first dependent variable organizational commitment is positively correlated with the other dependent variable employee performance at the 0.01 significance level (β = .295, p < .01). Moreover, the dependent variable organizational commitment is also positively correlated with the two mediating variables LMX (β = .365, p < .01), and subordinate motivation (β = .397, p < .01) at the 0.01 significance level. Furthermore, the dependent variable organizational commitment also has a weak and positive correlation with the moderating variable tenure at the 0.05 significance level (β = .191, p < .05). However, the dependent variable organizational commitment does not correlate with any control variables.

The second dependent variable employee performance is positively correlated with the mediating variable LMX at the 0.01 significance level (β = .263, p < .01). Likewise, the dependent variable employee performance is also in a weak and positive correlation with the other mediating variable subordinate motivation at the 0.01 significance level (β = .215, p < .01). Furthermore, there is a weak and negative correlation between the dependent variable employee performance and the control variable employee nationality at the 0.01 significance level (β = -.208, p < .01). Also, the dependent variable employee performance is negatively correlated with the control variables type of job (β = .175, p < .05), leader nationality (β = -.171, p < .05) and survey language (β = -.187, p < .05) at the 0.05 significance level. Moreover, the dependent variable employee performance also has a weak and positive correlation with the control variable leader age at the 0.05 significance level (β = .181, p < .05). However, the dependent variable employee performance does not correlate with any other control variables.

The first mediating variable LMX is positively correlated with the other mediating variable subordinate motivation at the 0.01 significance level (β = .339, p < .01). However, the mediating variable LMX does not correlate with the moderating variable tenure, or any other control variables.

(35)

34 The second mediating variable subordinate motivation is only negatively correlated to the control variable frequency of contact at the 0.01 significance level (β = -.230, p < .01). However, the mediating variable subordinate motivation does not correlate with the moderating variable tenure, or any other control variables.

The moderating variable tenure is positively correlated with the control variables employee age (β = .291, p < .01), employee tenure (β = .373, p < .01), leader age (β = .355, p < .01) and leader tenure (β = .409, p < .01) at the 0.01 significance level. The moderating variable tenure also has a weak and negative correlation with the control variable leader level of education at the 0.01 significance level (β = -.271, p < .01). Moreover, the moderating variable tenure is also negatively correlated with the control variables employee nationality (β = -.168, p < .05), survey source (β = -.170, p < .05) and survey language (β = -.170, p < .05) at the 0.05 significance level. The moderating variable tenure is not correlated with any other control variables, since the Person correlation coefficients are insignificant.

The first control variable employee age is positively correlated to the control variables employee tenure (β = .613, p < .01), leader age (β = .342, p < .01) and leader tenure (β = .281, p < .01) at the 0.01 significance level, and it is also positively correlated to the control variable employee gender at the 0.05 significance level (β = .183, p < .05), The control variable employee age is negatively correlated to the control variables employee nationality (β = -.223, p < .01), survey source (β = -.425, p < .01) and survey language (β = -.225, p < .01) at the 0.01 significance level, and it is also negatively correlated to type of job (β = -.196, p < .05) and leader nationality (β = -.202, p < .05) at the 0.05 significance level.

The second control variable employee gender is positively correlated to the control variable leader gender (β = .319, p < .01) at the 0.01 significance level, and it is also positively correlated to the control variable leader age (β = .164, p < .05) at the 0.05 significance level. The control variable employee gender is negatively correlated to the

(36)

35 control variable type of job (β = -.275, p < .01) at the 0.01 significance level, and it is also negatively correlated to the control variables leader nationality (β = -.182, p < .05) and survey language (β = -.163, p < .05) at the 0.05 significance level.

The third control variable employee level of education is positively correlated to the control variables leader level of education (β = .386, p < .01) and survey source (β = .231, p < .01) at the 0.01 significance level, and it is also positively correlated to the control variable frequency of contact (β = .186, p < .05) at the 0.05 significance level. The control variable employee level of education is negatively correlated to the control variable employee tenure (β = -.218, p < .01) at the 0.01 significance level, and it is also negatively correlated to the control variable leader tenure (β = -.162, p < .05) at the 0.05 significance level.

The fourth control variable employee nationality is positively correlated to the control variables type of job (β = .214, p < .01), leader nationality (β = .867, p < .01) and survey language (β = .978, p < .01) at the 0.01 significance level, and it is also positively correlated to the control variable survey source (β = .165, p < .05) at the 0.05 significance level. The control variable employee level of education is negatively correlated to the control variable employee tenure (β = -.217, p < .01) at the 0.01 significance level, and it is also negatively correlated to the control variables leader age (β = -.162, p < .05) and leader tenure (β = -.179, p < .05) at the 0.05 significance level.

The fifth control variable type of job is positively correlated to the control variables leader nationality (β = .266, p < .01) and survey language (β = .223, p < .01) at the 0.01 significance level. The control variable employee level of education is negatively correlated to the control variable survey source (β = -.284, p < .01) at the 0.01 significance level, and it is also negatively correlated to the control variable leader gender (β = -.158, p < .05) at the 0.05 significance level.

(37)

36 variables leader age (β = .268, p < .01) and leader tenure (β = .331, p < .01) at the 0.01 significance level. The control variable employee level of education is negatively correlated to the control variables survey source (β = -.374, p < .01) and survey language (β = -.221, p < .01) at the 0.01 significance level, and it is also negatively correlated to the control variable leader nationality (β = -.186, p < .05) at the 0.05 significance level.

The seventh control variable leader age is positively correlated to the control variable leader tenure (β = .573, p < .01) at the 0.01 significance level. The control variable employee level of education is negatively correlated to the control variable survey source (β = -.162, p < .05) at the 0.05 significance level.

The eighth control variable leader gender only has a weak and negative correlation with the control variable survey source (β = -.155, p < .05) at the 0.05 significance level.

The ninth control variable leader level of education only has a weak and negative correlation with the control variable leader tenure (β = -.156, p < .05) at the 0.05 significance level.

The tenth control variable leader nationality is positively correlated to the control variable survey language (β = .891, p < .01) at the 0.01 significance level, and it is also positively correlated to the control variable survey source (β = .181, p < .05) at the 0.05 significance level.

The eleventh control variable leader tenure is negatively correlated to the control variable survey source (β = -.263, p < .01) at the 0.01 significance level, and it is also negatively correlated to the control variables survey language (β = -.171, p < .05) at the 0.05 significance level. However, the control variable leader tenure does not correlate with any other control variables. Lastly, the control variable survey source has a weak and positive correlation with the control variable survey language at the 0.05 significance level (β = .173, p < .05).

(38)

37 The descriptions presented above together with table 4 – Correlation Matrix show the extent to which two variables vary together, as well as the direction they vary. More importantly, they demonstrate that all correlation coefficients of the independent, dependent, mediating and moderating variables are below the 0.8 criterion of acceptability, therefore, there is no multicollinearity problem. However, some control variables are highly correlated.