Faculty of Economics and Business MSc Business Administration Master Thesis Leadership and Management
Why do leaders treat team members differently?
The effect of leader personality traits on
leader-member exchange differentiation
Anne Diederen
Supervisor: dr. C.K. Buengeler
Statement of Originality
This document is written by Student Anne Diederen 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.
Abstract
Background: The aim of this study was to examine the relationship between (Big Five) leader personality traits and leader-member exchange (LMX) differentiation. A distinction was made between the measurement of the actual level of LMX differentiation in teams, measurement of the level of differentiation as perceived by the leaders, and measurement of the level of differentiation as perceived by the followers.
Type of analysis: In order to calculate the actual level of LMX differentiation, the disparity measure of LMX was used. Scales were developed to directly measure leaders' and followers' perceptions of LMX differentiation. Hierarchical regression analysis was conducted on data on teams that were obtained using questionnaires in order to examine the effect of leader personality traits on each measure of LMX differentiation. Additional moderation analysis was conducted to test whether team interdependence or constraints moderate the relationship between leader personality traits and LMX differentiation.
Results: The actual level of LMX differentiation, the level of LMX differentiation that leaders perceive and the level of LMX differentiation that team members perceive are very dissimilar from one another. Hierarchical regression analysis showed that a negative relationship between leader agreeableness and leader-rated LMX differentiation exists. Additional moderation analysis revealed two significant interaction effects, namely between conscientiousness and team interdependence in the relationship with follower-rated LMX differentiation; and between extraversion and team interdependence in the relationship with leader-rated LMX differentiation.
Table of contents
Introduction ... 1
Literature review and hypotheses ... 4
Development of LMX Theory ... 5
LMX Differentiation Defined ... 7
The Big Five Personality Traits and LMX Differentiation ... 8
Neuroticism ... 9 Extraversion ... 10 Openness to experience ... 11 Agreeableness ... 12 Conscientiousness ... 14 Research Method ... 15 Method ... 15 Sample ... 16 Measures ... 17 Actual LMX differentiation ... 17 Perceptions of LMX differentiation ... 20 Personality ... 21 Control variables ... 22 Moderation variables ... 22 Results ... 23 Additional Analysis ... 27 Discussion ... 29 Theoretical Implications ... 30
Limitations and Directions for Future Research ... 37
Conclusion ... 39
References ... 40
Annex 1: Leader Survey ... 49
Introduction
Research shows that many team leaders tend to be friendlier, more inclusive, and more communicative with some team members than they are with others (Bauer & Erdogan, 2016). The degree to which the relationships between a team leader and each of the team members differ from one another such that the leader forms high-quality relationships with some team members and low-quality relationships with others, is referred to as leader-member exchange (LMX) differentiation (Ma & Qu, 2010; Boies & Howell, 2006). Some leaders tend to differentiate a lot, while other team leaders tend to treat team members fairly similarly.
Research into LMX differentiation is grounded on LMX theory. LMX theory focuses on the dyadic relationships between a leader and his or her followers. Each LMX relationship is unique and can be described in terms of either low-quality or high-quality (Bauer & Erdogan, 2016). By researching the relationships between leaders and followers, LMX theory attempts to understand what effects these relationships have on the individual, team, and organizational level (Gerstner & Day, 1997). LMX differentiation focuses on how these relationships develop and what effects they have when people work in teams. LMX differentiation takes into account that the dyadic relationships between the leader and each of the team members coexist and influence each other because in teams, multiple followers work together interdependently.
Social comparison theory is essential to understand LMX differentiation. When people work in teams, they pay attention to the way their colleagues are, what their colleagues can or cannot do and what their colleagues have achieved or have failed to achieve (Corcoran, Crusius, & Mussweiler, 2011). The main reason why people do this is to form beliefs and opinions about their own capabilities (Festinger, 1954). People especially tend to compare themselves to others when there are no objective standards against which they can measure
provides team members with information about the question what a 'normal' relationship with a team leader would be like, team members compare their own relationships with the team leader with the relationships that their colleagues have with the team leader. Team members notice when a leader is more communicative and more inclusive with some of their colleagues, so they are aware of differential treatment (Henderson, Liden, Glibkowski, & Chaudhry, 2009).
There has been a shift from work organized around individual jobs to work organized around team-based structures (Lawler, Mohrman, & Ledford, 1995). This means that properly functioning teams have become increasingly important for organizational success. Amongst others, research has shown that LMX differentiation affects outcomes such as followers' commitment, turnover rates, and individual or team performance (Henderson et al., 2009), so it is important to research LMX differentiation.
Given that some team leaders differentiate more than other team leaders do and that these differences can have important effects, the question arises why leaders differentiate. So far, antecedents of LMX differentiation have received little attention (Chen, He, & Weng, 2015), as research that has been conducted on LMX differentiation has mainly focused on its possible outcomes. This is a missed opportunity because researching the question why some leaders differentiate more than other leaders do, could lead to interesting insights.
With regard to LMX quality, research has established that both the leader and the follower influence the development of an LMX relationship (Graen & Uhl-Bien, 1995). The question, however, is which characteristics of leaders and followers influence the degree to which leaders differentiate between team members. Most of the attention in the already limited amount of research on antecedents of LMX differentiation on the individual level has gone to the contribution of followers. Although the contribution of followers is interesting and should not be neglected, there is good reason to consider the contribution of leaders. Inherent
to their tasks, leaders are typically more in control over the relationships that are built between him- or herself and the followers. Therefore, it has been argued that leaders have a greater influence on the development of LMX relationships than followers do and that their characteristics are important antecedents of LMX quality (Dienesch & Liden, 1986). The same argument can be used to support the claim that leader behaviour and characteristics are important antecedents of LMX differentiation, because the fact that leaders are the ones who are most in control in the development of the relationships with each of the team members is also important in relation to LMX differentiation. This means that leader characteristics might explain considerable variance in LMX differentiation and should, therefore, be investigated.
Research is needed on the antecedents of LMX differentiation in order to advance our understanding of the topic and to create theoretical convergence. Although research states that personal characteristics of supervisors can be important antecedents of LMX differentiation, only limited attention has been given to characteristics other than demographics (Bernerth, Armenakis, Feild, Giles, & Walker, 2007).
It is interesting to pay attention to personality traits because they determine the way in which people behave (Lievens, Chasteen, Day, & Christiansen 2006). This means that measurement of personality traits can be used to predict a variety of outcomes (Hogan & Holland, 2003; Bernerth et al., 2007) and that it is worthwhile to investigate if LMX differentiation is among these outcomes. In the present study, the relationship between leader personality traits, in particular the NEO-PI or 'Big Five', and LMX differentiation shall be researched. To the knowledge of the author, this relationship has not been studied in prior research.
The aim of this thesis is to contribute to the so far limited body of knowledge on antecedents of LMX differentiation by asking: Do leader personality traits affect
leader-Figure 1. Research framework
The main purpose of this study is to extend the current knowledge of antecedents of LMX differentiation by exploring the influence of leader personality traits on LMX differentiation. A more practical contribution that this study aims to make, is to provide a newly created instrument to measure leaders' and followers' perceptions of LMX differentiation.
In order to be able to address the research question, in the next chapter an overview will be given of the current state of the literature on LMX differentiation and personality traits, and hypotheses will be proposed that are meant to answer the research question. In the ensuing chapter, the research methodology will be described. Subsequently, the results of the data analysis will be presented. The theoretical and practical implications of this study are described in the discussion chapter, as well as strengths and limitations of this study. Final conclusions in relation to the research question will be further translated into recommendations for future research.
Literature review and hypotheses
When one speaks of leadership, at least two actors are involved: a leader and a follower. Leadership theories mainly focus on characteristics or behaviours of a leader that make him or her successful and often have little consideration for the position of the follower. In this respect LMX theory is different from other theories. It departs from the premise that leaders form unique relationships with each of their followers. This means that LMX theory does not
Leader personality traits:
- Neuroticism - Extraversion - Openness - Agreeableness - Conscientiousness
attention for the role of followers in this theory, it is not surprising that many researchers have taken an interest particularly in the influence of followers on LMX relationships. Interesting to note, however, is that according to Dulebohn, Bommer, Liden, Brouer, & Ferris (2012), leader variables, such as leader personality, explain the most variance in LMX quality. This raises the question whether leader personality also explain variance in LMX differentiation.
In order to be able to properly address LMX differentiation, one should understand LMX theory. For this reason, an overview of the development of LMX theory will be given first. Subsequently, LMX differentiation will be defined. Lastly, the possible relationships between personality traits and LMX differentiation will be considered and hypotheses will be proposed.
Development of LMX Theory
In 1975, Dansereau, Grean, and Haga introduced the vertical-dyad linkage, which is considered the precursor of LMX theory. The vertical-dyad linkage was introduced as part of role theory, so role theory is at the roots of LMX theory (Boies & Howell, 2006). Another fundamental theory that has been important for the development of LMX theory is social exchange theory (Dulebohn et al., 2012). Role theory and social exchange theory both help to understand why people behave in certain ways and this understanding can help explain why people develop certain types of LMX relationships. Both theories are shortly described below. Role theory emphasizes that people are social actors who learn behaviours that are appropriate to the positions that they occupy in society (Solomon, Surprenant, Czepiel, & Gutman, 1985). This implies that people's behaviour can partly be predicted based on the roles that they have at the workplace or in daily life (e.g. being a parent or a friend). The roles of interest in LMX are the one of the leader and the one of the follower. According to role
development of an LMX relationship. The degree to which followers meet the expectations of the leader dictates the type of LMX relationship that will follow (Liden & Maslow, 1998).
Social exchange theory offers a frame of reference to help understand human behaviour in a different way than role theory does. According to social exchange theory, human behaviour is a function of payoffs (Blau, 1964). People value behaviours as costly or rewarding and base their behaviour on this analysis. When people expect that a certain activity will yield valued returns (e.g. the activity of the leader results in the follower to work harder or the activity of the follower will allow him or her to work flexible hours), people are motivated to undertake the activity (Liden, Sparrowe, & Wayne, 1997; Cook, Cheshire, Rice, & Nakagawa, 2013).
An LMX relationship tends to develop quickly based on initial interactions between a leader and his or her group members and stabilizes into a relationship that can range from low-quality (also referred to as out-group exchange) to high-quality (also referred to as in-group exchange) (Bauer & Green, 1996). Trust is one of the key components that distinguishes high-quality from low-quality LMX relationships (Nahrgang & Seo, 2016). From the leader's perspective, important factors for the formation of LMX are follower competence, dependability, and achievement (Duchon, Green, & Taber, 1986). The type of LMX relationship that a leader has with a follower affects the interaction between them. In low-quality LMX, the relationship between the leader and his or her followers is characterized by unidirectional downward influence, role-defined relationships, and a sense of loosely coupled fates. This means that low-quality LMX relationships are characterized by economic exchange based on formally agreed on, immediate, and balanced reciprocation of tangible assets, such as employment contracts (Dulebohn et al., 2012). When there is a low-quality LMX relationships the leader functions as an overseer. In contrast, when there is a high-quality LMX, the relationship between the leader and the follower can be described as a
partnership. High-quality LMX relationships are characterized by reciprocal influence, extra-contractual behaviour exchange, mutual trust, respect and liking, and a sense of common fate (Dulebohn et al., 2012).
LMX Differentiation Defined
LMX theory examines the relationship between a leader and a follower. This relationship is embedded in a broader context such as society, an organization, a work unit or a work team (Anand et al., 2016). LMX theory does not take into account that in teams, the leader is linked to multiple followers and that these linkages might influence each other. LMX differentiation, in contrast, does take these multiple linkages into account and can be defined as a process in which a leader forms different quality exchange relationships with team members through engaging in differing types of exchange patterns with each of them (Henderson et al., 2009).
The fact that LMX differentiation occurs at the team level means that it creates a context in which there is within-group variability in the nature and quality of the relationships of the leader and each of the team members (Henderson et al., 2009). Different team members have different levels of LMX-quality and they will evaluate their own relationship with the leader in comparison to the relationships of fellow team members with the leader (Schriesheim, Neider, Scandura, & Tepper, 2001).
Given that LMX differentiation occurs at the team level, it is important to consider what constitutes a work team. A basic definition of work teams is: "interdependent collections of individuals who work together towards a common goal and who share responsibility for specific outcomes of their organizations" (Sundstrum, De Meuse, & Futrell, 1990, p.120). Interdependence is crucial to label a group of individuals who work together as a team. This means that team members depend on each other in order to reach their shared goals.
It is important to be aware and to consider whose perception of LMX differentiation is being assessed. In theory, the quality of a dyadic LMX relationship and the level of LMX differentiation in a team are treated as real and measurable. In research, however, one has to work with the perceptions of the team leader and the team members on the quality of their LMX relationships and on the level of differentiation. Because the 'true' quality of an LMX relationship and the 'true' level of differentiation cannot directly be measured but can only be approached, it is likely that different people reflect differently on the same relationship. Research consistently finds that the perspective of a leader and those of the followers concerning their mutual relationship are generally not in agreement (Gerstner & Day, 1997; Sin, Nahrgang, & Morgeson, 2009; Schyns, 2016). It could happen that a leader perceives that he or she treats all the team members fairly similarly, while team members perceive that this leader does tend to differentiate between team members. This implies that it is likely that the perception of the leader and the perceptions of the team members of the level of LMX differentiation in a team will oftentimes also not be in agreement. This possibility shall be taken into account in the current study.
The Big Five Personality Traits and LMX Differentiation
Personality traits are powerful individual differences that persist over time, are real, causal and can be studied. They connect emotions, learning, thinking, attention, the patterning of behavioural outputs, and clinical problems (Harkness, 2009). The Big-Five personality traits are neuroticism, extraversion, openness to experience, agreeableness, and conscientiousness. They are relatively stable among working age adults (Cobb-Clark & Schurer, 2012), and the dimensions apply across different cultures (Judge et al., 1999).
Neuroticism
Neuroticism is the opposite of emotional stability, hence could also be referred to as emotional instability. Neuroticism refers to a lack of positive psychological adjustment and emotional stability (Judge et al., 1999). It has been conceptualized as a heightened sensitivity to negative stimuli (Tellegen, 1985). This means that neurotic people respond negatively to stress and uncertainty (Judge et al., 2002). Costa and McCrae (1992) measured anxiety, hostility, depression, self-consciousness, vulnerability, and impulsiveness as parts of neuroticism. People who score high on neuroticism experience - and thus have to deal with - negative emotions relatively often. Neurotic individuals tend to make negative evaluations of themselves, others and their experiences. Gunthert, Cohen, and Armeli (1999) studied how people who score high on neuroticism respond to minor stressors that occur on a daily basis. The authors found that neurotic people use more self-blame, wishful thinking and hostile reactions to cope with their daily stressors. Furthermore, neurotic people are severely limited in their social skills (Judge et al., 1999) and are therefore less likely to establish long-term relationships (as these require social skills, along with commitment and trust) (Raja, Johns, & Ntalianis, 2004; Bernerth et al., 2007).
One could reason that neuroticism leads to a more negative view of relationships and will, therefore, be negatively related to LMX quality (Schyns, 2016). Bernerth et al. (2007) tested this assumption but did not find evidence to support this claim. The question is if, and how, neuroticism is related to LMX differentiation. Because neurotic people respond negatively to stress and uncertainty (Gunthert, Cohen, & Armeli, 1999), it is likely that they try to avoid situations that cause these emotions. As neurotic individuals are limited in their social skills (Raja et al., 2004), they might have difficulty to get a clear sense of what are appropriate social interactions and what not. Therefore, they might feel inclined to create
team members. This would mean that they treat everybody according to the same guidelines, hence that they treat everybody fairly similarly. It is therefore expected that neurotic leader will not differentiate to a large extent. This leads to the following hypothesis:
Hypothesis 1: Leader neuroticism relates negatively to LMX differentiation.
Extraversion
Extraversion is the opposite of introversion, which describes individuals that are generally more reserved, aloof, sober, and task-oriented (Barrick & Mount, 1991). Extraversion is often associated with sociability, but it includes more. People who score high on extraversion are not only socially oriented (outgoing, affectionate, friendly, and talkative), but also active (adventurous and assertive). Furthermore, extraverts tend to be dominant and ambitious (Costa & McCrae, 1992). They often have many close friends and are likely to take on leadership roles (Judge et al., 1999). Of the big five personality traits, extraversion is the strongest correlate of leadership emergence, what suggests that extraversion is the most important trait to have to become a leader (Judge et al., 2002).
One could argue that the fact that extravert leaders are socially oriented and tend to communicate more than there introvert counterparts will lead them to have more high-quality LMX relationships with followers than introvert leaders do. This would affect LMX differentiation as well because the level of differentiation is lower when leaders have high-quality relationships with many followers. The findings of different researchers about the relationship between leader extraversion and LMX quality, however, are not consistent. Some researchers claim that there is indeed a positive relationship, while others state that evidence does not support this claim or that extraversion only relates to some dimensions of LMX (Schyns, Maslyn, & Van Veldhoven, 2012; Nahrgang, Morgeson, & Ilies, 2009). This means that the relationship between extraversion and LMX quality does not provide a solid starting
point based on which one can build a case for a relationship between extraversion and LMX differentiation.
According to Hills and Argyle (2001), introverts are more selective and have only a few close friends with whom they have special relationships. This means that it is likely that introverts will create a small in-group, which is associated with a high level of LMX differentiation, hence that introvert leaders differentiate more than extravert leaders do. Given that introversion is opposite to extraversion, this leads to the following hypothesis:
Hypothesis 2: Leader extraversion relates negatively to LMX differentiation.
Openness to experience
"The personality dimension that most centrally influences social and interpersonal phenomena is Openness" (McCrae 1996, p.323). People who score high on openness are original, imaginative, have broad interests and are daring (Costa & McCrae, 1992). Common traits associated with openness to experience include being imaginative, cultured, curious, original, broad-minded, intelligent, and artistically sensitive (Barrick & Mount, 1991).
Open individuals are more likely to emerge as leaders and to be effective leaders than closed individuals are (Judge et al., 2002). Leaders who score high on openness tend to think outside the box and challenge others (Bernerth et al., 2007). Empirical research so far has not supported hypotheses that relate leader openness to experience to LMX quality (Schyns, 2016). So how does leader openness to experience relate to LMX differentiation?
Openness to Experience is positively related to universalism (Roccas, Sagiv, Schwartz, & Knafo, 2002). Universalistic values (i.e. values that are guided by the idea that rules and policies apply equally to all in all situations), in turn, are negatively related to LMX differentiation (Ma & Qu, 2010). This implies that openness will also be negatively related to
Leaders who score high on openness to experiences will usually feel comfortable among all sorts of people (McCrae, 1996). Leaders who score high on openness are more likely to see and value differences (Ekehammar & Akrami, 2003; Homan et al., 2008), so they will probably also feel comfortable in diverse groups. Leaders, on the other hand, who score low on openness will find it hard to understand and adapt to other people's perspectives (Gurtman, 1995). People who score low on openness will feel most comfortable among people who share their ideas and values (McCrae, 1996). Closed leaders are likely to form small in-groups with followers who have similar ideas and values. This is associated with a high level of LMX differentiation. Given that open leaders, on the other hand, feel comfortable among all sorts of individuals, they are less likely to feel inclined to create a small in-group. This leads to the following hypothesis:
Hypothesis 3: Leader openness relates negatively to LMX differentiation.
Agreeableness
People who score high on agreeableness are cooperative and likeable. They have a tendency to trust other people and care for them, and they are cheerful and gentle (Judge Higgins, Thoresen, & Barrick, 1999). Agreeable people are harmonious in relationships with others. Common traits associated with agreeableness include being courteous, flexible, trusting, good-natured, cooperative, forgiving, soft-hearted, and tolerant (Barrick & Mount, 1991).
The relationship between agreeableness and leadership is ambiguous. On the one hand, there are aspects of agreeableness that are important for leaders, such as altruism, tact, and sensitivity (Judge, Bono, Ilies, & Gerhardt, 2002). Agreeable leaders are more motivated to achieve interpersonal intimacy and they are more able to form social alliances than leaders who score low on agreeableness (Buss, 1991). On the other hand, agreeableness is related to modesty, which is not a trait that leaders tend to have (Judge et al., 2002). LMX is a social
relationship, for the quality of which agreeableness is the most relevant personality trait (Bernerth et al., 2007). Schyns, Maslyn, and Van Veldhoven (2012) found that leader agreeableness is related to two dimensions of LMX, namely the dimensions affect and professional respect. Leaders who are agreeable are most likely to establish positive relationships with each of their followers and they are more likely to be able to sustain a larger number of relationships at the same time (Schyns, 2016). This should affect LMX differentiation as well. When the level of LMX differentiation is high, a few of the team members are privileged over the others. When a leader has high-quality relationships with many followers, he or she differentiates less.
Agreeable people are social. They tend to include others more than less agreeable individuals, which means that they broaden their empathic circles and create a larger community (Graziano & Eisenberg, 1997). When work is organized in teams, agreeable leaders are more likely to encourage team members to work together and to share information, regardless of status and power differences (Heller, Judge, & Watson, 2002). Agreeable leaders are likely to encourage team members to participate by creating an inclusive environment. In order to ensure that all team members equally feel part of the team and no one feels left out, it is likely that agreeable leaders will try to build similar relationships with each of the team members, hence differentiate less than leaders who score low on agreeableness. Another argument to expect a negative relationship can be based on the work by Bernerth et al. (2007), who found that agreeable leaders are more likely to have high-quality LMX with each followers, which is also associated with a low level of LMX differentiation. This leads to the following hypothesis:
Conscientiousness
Conscientiousness is related to self-control. Conscientiousness is positively related to overall job performance (Barrick & Mount, 1991) and to success at work (Judge et al., 1999). This could be explained by the traits of conscientious people. Common traits associated with conscientiousness include being careful, thorough, responsible, organized, planful, hardworking, achievement-oriented, and persevering (Barrick & Mount, 1991). More negative traits that are associated with conscientiousness are rigidity, inflexibility, intolerance of ambiguity, ethnocentrism, and compulsive tendencies (Peterson, Smith, Martorana, & Owens, 2003).
Conscientiousness is more strongly related to leader emergence than to leader effectiveness (Judge et al., 2002). Conscientious people tend to have socially prescribed impulse control, meaning that they think before they act, follow rules, plan, and prioritize (Witt, 2002). So far, the relationship between leader conscientiousness and LMX quality is not yet fully clear (Schyns, 2016). Whereas Bernerth et al. (2007) found a positive relationship between leader conscientiousness and high-quality LMX, Schyns, Maslyn, and Van Veldhoven (2012) found a negative relationship between conscientiousness and two dimensions of LMX, namely loyalty and contribution. These findings contradict each other.
Conscientious leaders will probably show more concern with follower performance and will feel more inclined to help followers than leaders who are less conscientious do (Bernerth et al., 2007). When conscientiousness is viewed in the light of social exchange theory (Blau, 1964), one would expect conscientious leaders to be aware more often of the occasions that they need to repay followers for their acts. Moreover, conscientious leaders will also more often actually repay followers for their acts than non-conscientious leaders would (Bernerth et al., 2007). This implies that conscientious leaders are focused on making sure that every team member gets what he or she deserves. As a result, the leader will treat
each team member on an individual basis, which means that it is likely that conscientiousness will increase differentiation. Therefore, the following hypothesis will be tested:
Hypothesis 5: Leader conscientiousness relates positively to LMX differentiation.
Research Method
In this chapter, the research method that has been used to conduct this study is presented. The procedure, the sample and the measures are described. As LMX differentiation is of central importance to this study and there are several possibilities to conceptualize and examine differentiation, the choices that have been made concerning these calculations are extensively explained.
Method
This thesis is part of a research project called “Antecedents and Consequences of Differentiated Leadership” that is being conducted at UvA and led by Dr. Claudia Buengeler (Assistant Professor, Amsterdam Business School, UvA). Two separate questionnaires were developed and distributed online (using the online survey tool Qualtrics) to leaders and the members of work teams in all sorts of organizations in the Netherlands in order to gather the data that were used for this thesis and for another thesis.
Data collection started in April 2016 and ended in June 2016. The personal networks of the researchers involved were used to recruit as many teams as possible within the given period of time. The surveys were returned anonymously. In order to be able to link the surveys of leaders to the surveys of their team members, matching codes were used.
The targeted respondents were Dutch, so the questionnaires were also drawn up in Dutch. The scales used were mostly adopted from English-language articles. For some scales,
Dutch translations were available and thus used. For others, the items were translated from English to Dutch using the back-translating technique (Brislin, 1970).
In order to collect the appropriate data for this research, teams in their entirety were asked to fill in the questionnaires. The questionnaires contained an introduction in which the purpose of the study, the length of the questionnaire, and the procedural aspects were outlined. On top of that, ethical aspects such as the guarantee of the protection of anonymity were stressed.
Sample
In this study, the population of interest consists of leaders and their corresponding team members in the Dutch workforce. This is a large population and the sampling frame is unknown. It was not possible to use random sampling techniques, so we used convenience sampling instead (Field, 2009). We tried to reach as many teams as possible using our personal networks. All teams that were willing to participate, were included in the study. As the snowball method was part of our strategy, the exact response rate is impossible to determine. 71 fully completed leader questionnaires and 202 fully completed team member questionnaires were returned. The team leaders were aged between 21 and 70, with an average age of 43.8 years. 64.8% of the leaders were male, 35.2% of the leaders were female and most of the leaders (84.5%) worked fulltime. The team members who responded were aged between 19 and 66 years, with an average age of 36.6 years. 54% of the team members were male and 46% of the team members were female. Compared to the leader sample, the share of team members who worked part-time was much larger: 57% of the team members worked fulltime and 43% of the team members worked part-time. Responses were used when at least the leader and two of the members of the same team had completed the questionnaires. In total, data of 68 leaders (and thus 68 teams) combined with 188 team members were used.
3 fully completed leader questionnaires and 14 fully completed team member questionnaires were eliminated.
Measures
The data collected with these questionnaires are not only used for this thesis, but also for other research papers. For this reason, the questionnaires contained variables that were not used in this thesis. The main measures for this study are leader's personality traits and LMX differentiation. More measures were added to be able to control for their effects or to run moderation analysis. The leader, as well as the team members, were asked to answer questions on relevant demographics, including gender (m/f, nominal variable), age (ratio) and educational level. Items were measured using validated 5-point or 7-point (completely disagree - completely agree) Likert scales (interval).
Actual LMX differentiation
The variable that is of central importance to this study, is LMX differentiation. In this study, LMX differentiation is measured in multiple ways, allowing to calculate the actual level of LMX differentiation and to examine the perceived level of differentiation both by leaders as well as by followers.
Measuring differences is challenging. Often in research, the precise meaning of differences, oftentimes termed diversity, is not clear (Harrison & Klein, 2007). Harrison and Klein (2007) define diversity as "the distribution of differences among the members of a unit with respect to a common attribute" (p. 1). The common attribute of which the distribution of differences is of interest in this study is LMX quality. The use of the word 'unit' in this definition implies that one number describes an entire group, which in this study is a work
The actual differentiation scores were calculated based on the direct measure of the LMX-7 by Graen and Uhl-Bien (1995) in the follower survey. This scale consists of seven items that ask each follower to describe his or her relationship with the leader. This scale is one of the measures that is most frequently used in LMX research (Dulebohn et al., 2012). An example of an item is: "How well does your leader understand your job problems and needs?" (Not a bit / a little / a fair amount / quite a bit / a great deal). We used the Dutch translation of these items from Van Breukelen, Konst, and Van der Vlist (2002). The responses were collected at the individual (team member) level and analyzed at the team-level. We used intraclass
correlation coefficients (ICC[1]) and ICC[2]) and the rwg to test whether it was reasonable to
aggregate these individual measures to the team level based on Biemann, Cole, and Voelpel (2012). The results showed agreement across followers' ratings and demonstrated adequate interrater reliability (ICC[1] = .26, ICC[2] = .50, mean Rwg = .97).
The question that needs to be addressed is what would be the most appropriate way to calculate LMX differentiation based on the aggregated scores of LMX. LMX theory is based on the premise that relationship quality tends to vary considerably within a team; within-group agreement is thus not expected. For this reason, using the mean or the average is not an appropriate measure of diversity (Bliese, 2000).
Harrison and Klein (2007) distinguish between three types of diversity, namely separation, variety, and disparity. It is important to be clear on which of these three types of diversity is being used in the current research. In order to justify the diversity type of choice in this study, one must understand the implications of all three types of diversity first. Separation refers to differences in position and opinion among unit members and reflects disagreement or opposition within the group. Variety refers to differences in kind or category, primarily of information, knowledge, or experience among unit members. Disparity refers to differences in concentration of valued social assets or resources such as pay and status among
unit members. When disparity is high, it means that a few of the unit members are privileged over many others. As it is clear that the aim of this study is not to measure differences in kind or category, variety is not the appropriate choice of measuring diversity. The use of separation and disparity in this study are both defendable, the question is which one is preferable.
When measuring separation, one measures differences in position. Examples of theories for which separation measures are used are similarity attraction, selection and attrition, and social categorization theory (Harrison & Klein, 2007). An example of research that measured diversity as separation can be found in research conducted by Williams and O'Reilly (1998), who researched the relation between diversity of attributes such as age, values and personality and within-unit behavioural and social integration. When measuring disparity, one measures differences in portion or possession of a valued asset or resource. Examples of theories for which disparity measures are used are distributive justice theory, tournament theory, stratification, status hierarchy or status characteristics theory. Disparity highlights the relative comparisons among team members with regard to a highly valued asset or resource. It suggests that diversity (i.e. disparity) leads to negative outcomes such as within-unit competition, suppression of voice, or reduced communication.
Applied to LMX differentiation, what would one be measuring when one uses separation measures and what would one be measuring when one uses disparity measures? The standard deviation is used to measure separation. This means that when a score is low, there is perfect agreement among the team members on the quality of LMX that they have with their leader, so everybody in the team values his or her LMX relationship with the leader in the same way. When the score is high, this means that the team members are equally split at the opposing endpoints, which can be visualized as there being two 'camps'. A moderate score means that team members show some spread or disagreement.
The Coefficient of Variation (CoV) is used to measure disparity, which is the standard deviation divided by the mean. The CoV reflects both the distances between team members and the dominance of (the concentration of the resources in) those who have higher amounts of the attribute. When a score is low, it means that all members of the team have the same LMX quality. When the score is high, it means that some team members outrank all others so that some team members have high-quality LMX whilst all others have low-quality LMX. When the score is moderate, it means that some team members outrank others, but that the differences among team members are compressed.
The most important consideration when opting for one of these two measures is whether one wants to measure attitudes within the team and differences in the position that team members hold regarding LMX or that one wants to measure the amount to which team members perceive to possess high-quality LMX. When LMX diversity is approached as disparity, high-quality LMX is considered a socially valued and desired resource, which is in line with the way LMX is generally approached (Law-Penrose, Schwind Wilson, & Taylor, 2016). Moreover, measuring diversity as disparity expresses the idea that diversity on a given attribute matters less when the resource is typically high than when it is typically low. This means that when all team members have relatively high-quality LMX relations with the leader, diversity matters less than when all team members have relatively low-quality LMX relations with the leader, which makes sense, especially when LMX is considered a resource (Law-Penrose et al., 2016). For these reasons, the disparity measurement is used in this study to calculate differentiation on the basis of the LMX-7 scores.
Perceptions of LMX differentiation
The perceptions on LMX differentiation were measured both in the leader questionnaire and in the follower questionnaire on a 5-point Likert scale. The direct questions were adapted
from the LMX-7. For the leader questionnaire, the items of the LMX-7 were rephrased in such a way that they asked for the degree to which the leader perceived to differentiate him or herself. An example can be found in the transformation of the LMX-7 item "How well does your leader understand your job problems and needs?" into the Dutch equivalent of "I understand the job problems and needs of some team members better than those of other team members."
Similar direct questions were developed to ask the followers about their perception of the degree of differentiation in the team. The same LMX-7 item was transformed into the Dutch equivalent of "My leader understands the job problems and needs of some team members better than those of other team members." Given that these items were collected from individual followers and analyzed at the team-level, intraclass correlation coefficients were used (ICC[1] = .34, ICC[2] = .59, mean Rwg = .90) to justify the use of aggregated scores.
Personality
There are many different ways to measure personality, among which are observational measures, test studies, and the use of questionnaires. In this study, the latter was used. The measurement of choice, the NEO Five Factor Inventory (NEO-FFI) by Costa and MacCrae (1992), was used on a validated 5-point Likert scale. The NEO-FFI is an abbreviated version NEO Personality Inventory (NEO-PI, also known as 'Big Five' or 'Five Factor Model of Personality') that was developed by Costa and McCrae in 1985. The five dimensions of this inventory (neuroticism, extraversion, openness, agreeableness, and conscientiousness) are often used in academic research to measure people's personality. The NEO-FFI consists of 12 items for each dimension, so in total there are 60 items on personality. The NEO-FFI scales
of the validity of the longer NEO-PI scales (John & Srivastava, 1999). An example of an item measuring neuroticism is: "I am no worrier".
Control variables
Previous research has shown that demographic variables such as age and gender may influence the LMX relationship (McColl-Kennedy & Anderson, 2005; Turban & Jones, 1988; Bernerth et al., 2007). So far, research has not established whether these demographic variables also influence LMX differentiation. Therefore, the demographic variables age and gender will be taken into account in the analysis of this study as control variables.
Moderation variables
Constraints - In this study, constraints will be considered as a possible moderator in the relationship between leader personality traits and LMX differentiation. Research has shown that constraints are important in relation to LMX quality (Henderson et al., 2009). Building a high-quality LMX relationship requires time (Dansereau et al.,1975), effort (Maslyn & Uhl-Bien, 2001), and resources (Liden et al., 1997). Constraints in time, effort, and resources were measured in the leader survey. The scale used in this study consists of a combination of items from different authors, namely 4 items from Steffensmeier (2008), 3 items from Landy, Rastegary, Thayer, and Colvin (1991), and 4 items from Van Breukelen et al. (2002). An example of an item is: "In my job, I am often in a hurry."
Team interdependence - Team interdependence is part of the definition in the literature of what constitutes a team. To be a 'real team' requires a high level of interdependency, a low level of interdependency means that the group should be labelled a 'working group' (Katzenbach & Smith, 1993). In the present study, teams were invited to participate when they define themselves as a team. It was not possible to control if each of the participating
teams, in fact, would fall within the scope of the definition of teams as is used in the literature, which means that some teams should actually be referred to as 'working groups'. Therefore, it is interesting to test if the effects of personality are stronger in teams where the level of team interdependence is high (i.e. teams that would fall within the definition of 'real teams') than the effects of personality are in teams where team members are less interdependent (i.e. teams that are actually 'working groups'). Team interdependence was measured in the leader survey. This scale is based on an adapted version of Pearce and Gregersen (1991) and consists of 5 items. An example of an item is: "The team members need to work closely with each other to do their job."
Results
The data obtained from the questionnaires were analyzed using IBM SPSS Statistics version 23. Counter-indicative items were re-coded and scale reliabilities were reviewed on the basis of Cronbach's alpha. All the scales that were used in the current study had Cronbach's alpha ≥ .7 and all of the items had a strong correlation with the total score of the scale (above .30). None of the items would substantially affect reliability if they were deleted. This means that scale reliability for each of the scales can be assumed (Field, 2009). Means, standard deviations, correlations and Cronbach’s alpha coefficients are reported in Table 1.
The perceived level of LMX differentiation based on the measure that was reported by leaders is slightly higher than the level of LMX differentiation reported by followers, as the mean score of leader-rated LMX differentiation is 3.412 and the mean score of follower-rated LMX differentiation is 3.178. Personality traits were also measured at 5-point Likert scale. When one looks at the mean scores of the leader personality traits, one finds that the leaders score relatively high on conscientiousness and extraversion, followed by agreeableness. The
Table 1
Means, standard deviations, and intercorrelation matrix for study variables
Variables M SD 1 2 3 4 5 6 7 8 9 10 11 12 1. Age 43.81 12.229 (-) 2.Gender 1.34 .477 .114 (-) 3. Neuroticism 2.1569 .520 -.327** -.052 (.829) 4. Extraversion 3.8088 .453 .165 .252* -.561** (.816) 5.Openness 3.2341 .517 .299* .194 -.031 .289* (.744) 6.Agreeableness 3.6434 .513 .258* .383** -.383** .592** .300* (.824) 7.Conscientiousness 3.8346 .475 .115 .042 -.474** .414** .028 .192 (.839) 8. Team interdependence 3.8412 .715 .071 -.042 -.032 .090 .074 .024 .133 (.911) 9. Constraints 2.8636 .527 -.075 -.365** .131 -.193 -.089 -.393** -.097 .138 (.825) 10. Actual LMX differentiation .0683 .0514 .173 .054 -.116 .188 .126 .051 .227 -.065 .062 (.796) 11. Follower-rated LMX differentiation 3.1781 .540 .225 -.263* -.306* .141 -.031 -.092 .273* .013 .267* .082 (.862) 12. Leader-rated LMX differentiation 3.4118 .555 -.091 -.212 .045 -.049 .041 -.285* .092 -.014 .434** .160 .290* (.745) Statistical significance: *p<.05; ** p<.01
The disparity measure of LMX is not significantly related to either one of the direct measures of perceptions of LMX differentiation. The correlation between leader-rated LMX and follower-rated LMX is significant and can be described as a moderate-sized relationship (.290*). The disparity measure of LMX does not show any significant correlations with leader personality or any of the other measures that were used in this study.
Leader-rated LMX differentiation and follower-rated LMX differentiation correlate significantly with different personality traits. Neither the disparity measure of LMX, the direct measure of leader-rated LMX differentiation nor the direct measure of follower-rated LMX differentiation correlate significantly with leader extraversion or leader openness. Leader-rated LMX differentiation correlates negatively with leader agreeableness. There is a significant negative correlation between follower-rated LMX differentiation and leader neuroticism and a significant positive correlation between follower-rated LMX differentiation and leader conscientiousness.
The hypotheses were tested using hierarchical regression analysis. The model was tested three times, each time with a different dependent variable, namely with the actual level of LMX differentiation (based on the CoV of LMX) as the dependent variable, with follower-rated LMX differentiation as the dependent variable, and with leader-follower-rated LMX differentiation as the dependent variable. The results are shown in Table 2 to 4. The control variables age and gender were entered in the first step. The variable age was standardized. In the second step, the standardized leader personality traits were added. These tests showed one result. The effect of leader agreeableness on leader-rated LMX differentiation (B = -.380, t = -2.196) has p < .05, which implies that the interaction effect is significant (Aiken & West, 1991). This relationship is negative, which means that leaders who score high on agreeableness perceive to differentiate less than leaders who score low on agreeableness.
Table 2
Results of hierarchical regression analysis for predicting actual LMX differentiation
Step variables R R2 R2 Change B SE beta t
Step 1 .177 .031 Age .001 .001 .169 1.376 Gender .004 .013 .035 .287 Step 2 .315 .099 .068 Age .001 .001 .162 1.169 Gender .003 .014 .025 .188 Neuroticism .008 .017 .086 .506 Extraversion .021 .021 .184 1.008 Openness .006 .014 .056 .406 Agreeableness -.013 .016 -.131 -.799 Conscientiousness .021 .015 .196 1.369 Statistical significance: *p<.05; ** p<.01 Table 3
Results of hierarchical regression analysis for predicting follower-rated LMX
Step variables R R2 R2 Change B SE beta t
Step 1 .367 .135** Age .011* .005 .258 2.221 Gender -.331* .132 -.292 -2.515 Step 2 .505 .255 .120 Age .010 .006 .221 1.752 Gender -.279* .139 -.246 -2.013 Neuroticism -.185 .160 -.178 -1.151 Extraversion .177 .198 .149 .896 Openness -.035 .132 -.033 -.262
Conscientiousness .179 .148 .157 1.210 Statistical significance: *p<.05; ** p<.01
Table 4
Results of hierarchical regression model for predicting leader-rated LMX differentiation
Step variables R R2 R2 Change B SE beta t
Step 1 .222 .049 Age -.003 .006 -.068 -.557 Gender -.238 .142 -.204 -1.678 Step 2 .383 .147 .098 Age -.003 .006 -.061 -.450 Gender -.155 .152 -.133 -1.019 Neuroticism .012 .176 .012 .071 Extraversion .137 .217 .112 .630 Openness .166 .145 .155 1.146 Agreeableness -.380* .173 -.351 -2.196 Conscientiousness .149 .163 .127 .913 Statistical significance: *p<.05; ** p<.01 Additional Analysis
The data gathered for this study allowed to explore if the variables team interdependence or constraints possibly moderate the relationship between leader personality traits and LMX differentiation. The effects of these variables are interesting to explore because team interdependence is part of the definition of what constitutes a team and because research has shown that constraints influence the development of LMX relationships (Henderson et al., 2009).
analysis again was used three times, each time with a different dependent variable (namely LMX differentiation based on the CoV of LMX, leader-rated LMX differentiation, and follower-rated LMX differentiation). In the first step, the standardized personality traits and standardized team interdependence were entered. In the second step, the interaction term of each of the standardized personality traits with the standardized team interdependence was entered.
In the model that was run with the CoV as the dependent variable, no significant relationships were found. With follower-rated LMX differentiation as the dependent variable, the interaction term between team interdependence and leader conscientiousness is significant (B = -.191, t = -2.433, p < .05), see figure 2.
Figure 2. The moderating effect of team interdependence on the relationship between leader conscientiousness and follower-rated LMX differentiation
Simple slope analysis can be conducted to answer the question whether the gradients significantly differ from zero (Dawson, 2014). The result of the simple slope analysis was not significant.
With regard to leader-rated LMX differentiation, the interaction term between extraversion and team interdependence is significant (B = -.190, t = -2.120, p < .05), see
gradients significantly differ from zero. The result of this simple slope analysis was also not significant.
Figure 3. The moderating effect of team interdependence on the relationship between leader extraversion and leader-rated LMX differentiation
In order to examine the moderating effect of constraints on the relationship between the five personality traits and LMX differentiation, the same procedure was followed as was done with team interdependence. In the first step, the mean centered personality traits and mean centered constraints were entered. In the second step, the interaction term of each of the personality traits with constraints was entered. The model was again tested with the three different LMX differentiation variables as the dependent variable. There were no significant interaction effects found in either one of the models.
Discussion
The central concept in this study is LMX differentiation, which departs from the premises that leaders develop unique relationships with each of their followers, that these relationships coexist when people work in teams and that the coexisting relationships between a leader and each of the team members influence each other. It is important to understand why leaders
become increasingly important for organizational success. Researchers acknowledge the influence of personality traits on the development of LMX relationships but have not considered the effect of leader personality traits on LMX differentiation. The primary purpose of this study was to investigate this association, as this apparent gap in our knowledge is important both to researchers and practitioners. This study was designed to answer the research question: Do leader personality traits affect leader-member exchange differentiation? Results revealed that leader agreeableness affects leader-rated LMX differentiation. This relationship and the findings from the additional moderation analysis will be discussed below. Potential limitations of this research will be acknowledged and suggestions for future research will be given.
Theoretical Implications
Statistical measures can assess either 'actual' or 'perceived' differentiation. In this study, the disparity measure of LMX was used to approximate the actual level of differentiation within a team. The leader-rated and the follower-rated measures were used to indicate the perceived level of differentiation within each team. Kristof-Brown, Zimmerman, and Johnson (2005) stated that perceptions are more powerful than reality. This can partly be explained by the fact that people can only base their behaviours on their perceptions, as they do not have all the information about what really is (Harrison & Klein, 2006). This raises the expectation that the measures of the perceived levels of differentiation will lead to more interesting insights than the measure of the actual level of differentiation, which is consistent with the findings in the current research.
Research on LMX often finds that leader personality is not significantly related to leader-rated LMX quality, while it does find that leader personality is significantly related to follower-rated LMX quality (Goodwin, Bowler, & Whittington, 2009). In the current study, it
was explored if the same is true for measurements of LMX differentiation. The correlational relationships indicate that the answer to the question which leader personality traits relate to leader-rated LMX differentiation is distinct from the answer to the question which leader personality traits relate to follower-rated LMX differentiation.
The main theoretical contribution of the current research to the extant literature is that the use of different measurements of LMX differentiation (either the actual level of LMX differentiation, the LMX differentiation that leaders perceive, or the LMX differentiation that team members perceive) leads to results that are very dissimilar from one another. This is in line with research on LMX that states that the agreement between leader and follower perspectives on their mutual relationship is low (Schyns, 2016; Gerstner & Day, 1997; Sin et al., 2009).
A possible explanation for this discrepancy can be found in the argument of Yammarino and Dansereau (2002), who pose (with regard to LMX quality) that leaders and followers base their evaluations of their mutual relationship on different dimensions of LMX. LMX differentiation does not necessarily imply that a leader treats followers differently in all aspects of the relationship, it could be that leaders treat followers distinct in some aspects and similarly along other aspects (Cogliser & Schriesheim, 2000). If leaders and followers give more weight to different aspects of their relationship when they assess their mutual relationship, this could explain why their overall assessment of the same relationship is distinct. This would imply that the different perceptions of LMX differentiation have different antecedents (Schyns, 2016), which is in line with the findings in this study.
In order to test if the data support the hypotheses of this study, hierarchical regression analysis was performed with each of the three measures of LMX differentiation as the dependent variable. In this analysis, one significant relationship was found. The analysis
LMX differentiation. Leaders who score higher on agreeableness perceive to differentiate less than leaders who score low on agreeableness. This negative relationship is in line with the hypothesis on the relationship between agreeableness and LMX differentiation. The hierarchical regression analyses do not support the other hypotheses.
An interesting question is why leader agreeableness significantly relates to leader-rated LMX differentiation and not to the actual level of LMX differentiation or to follower-rated LMX differentiation. A possible explanation could be sought in the fact that the measurement of the leader personality traits and leader-rated LMX differentiation both were measured in the questionnaire of the leaders. According to Costa and McCrae (1992), there is substantial evidence that self-reports, in general, are trustworthy. However, research indicates that individuals have a tendency to present themselves favourably with respect to social norms and standards (Zerbe & Paulhus, 1987). This means that leader-rated LMX differentiation could indeed reflect the level to which leaders actually perceive to differentiate, but it could also indicate the level to which leaders think they should differentiate. Similarly, self-rated personality traits can be interpreted in three different ways. The personality scores can be viewed as a measure of what the person is really like, as a measure of the person's self-concept (i.e. the way the person thinks he or she is), or as the person's self-presentation (i.e. the way the person wishes to be viewed by others) (Costa & McCrae, 1992).
Assuming that the personality scores reflect what the respondent is really like, it is interesting to compare this finding to the findings in the literature that link agreeableness to LMX quality. Nahrgang et al. (2009) found a significant relation between agreeableness and leader-rated LMX when the relationship had somewhat matured and no significant relationship between agreeableness and follower-rated LMX. Sears and Hackett (2011), in contrast, did find a significant relationship between agreeableness and follower-rated LMX. According to Bernerth et al. (2007), agreeable leaders are more likely to have high-quality
LMX relationships with each follower. This implies that agreeableness should relate to low levels of actual level of differentiation. This relationship, however, was not found in the current study.
It is not possible to determine to what extent the responses in this study were influenced by leaders giving socially desirable responses. An interesting question is whether the tendency to give socially desirable responses responses is stronger with regard to certain personality traits, for example agreeableness, than it is with regard to other personality traits, or that the self-deceptive bias plays a role in all personality dimensions, as is suggested by John and Srivastava (1999).
If the leaders' scores on personality do not strictly reflect true personality, it is understandable that no significant relationships were found between agreeableness and the actual level of LMX differentiation or follower-rated LMX differentiation. Self-concept scores or self-representation scores are more random than actual personality scores. If, in addition, leader-rated LMX differentiation would indicate the level to which leaders think they should differentiate and not the level to which they perceive to actually differentiate, the significant relationship between agreeableness and leader-rated LMX differentiation would merely indicate that there is a relationship between how agreeable leaders think they are (or how agreeable they would like others to think they are) and the level to which leaders think they should differentiate. Perhaps it means that leaders who value agreeableness have a negative connotation with LMX differentiation as it implies that the leader has low-quality LMX relationships with some followers, which would be in line with research stating that people who score high on agreeableness have a high concern for relationships (Rocas et al., 2002).
LMX differentiation. Simple slope analyses were not significant for either one of the interaction effects that were found (i.e. the interaction effect between conscientiousness and team interdependence on follower-rated LMX differentiation, and the interaction effect between extraversion and team interdependence on leader-rated LMX differentiation). This does not necessarily mean that the results that were found should not receive attention, as the question whether or not the lines differ from the horizontal plane might not be that important and many researchers do not even conduct simple slope analysis (Aiken & West, 1991).
The analysis of the moderating effect of team interdependence on the relationship between leader personality traits and leader-rated LMX differentiation showed a significant interaction effect between extraversion and team interdependence. The interaction effect is negative, which means that when team interdependence is high (so when the team can be considered a 'real team'), extravert leaders perceive to differentiate less than when team interdependence is low. Nahrgang et al. (2009) found that leader extraversion relates significantly to leader-rated LMX and not to follower-rated LMX. The fact that the interaction effect between extraversion and team interdependence was only found with leader-rated LMX differentiation supports Schyns' (2016) observation that extraversion seems to be more related to leader-rated LMX than to follower-rated LMX and that this may mean that extravert leaders overestimate the quality that they have with their followers. It would be interesting if future research would test if this claim can be supported by empirical evidence.
The interaction term between team interdependence and conscientiousness is also negative and significant when the analysis is based on follower-rated LMX differentiation. This indicates that when team interdependence is high (i.e. a 'real team'), followers perceive that conscientious leaders differentiate less than when team interdependence is low. This finding contradicts the hypothesis in this study about the relationship between
