A dissertation entitled
Leaders’ Self-Efficacy and Follower Psychological Empowerment: A Field Study that incorporates the Role of Empowering Leadership Behaviour and Leader-Member
Exchange. by Meike Bunt 10003712 Organisational psychology University of Amsterdam 2nd of November 2014 Amount of words: 11675 Supervisor Ms. Dr. B. Nevicka Sr. Dr. M.R.W. Hamstra
Abstract
Follower empowerment has received increased attention from scholars and business life because of the beneficial effects on both company profits and follower wellbeing. Previous research has shown that leaders can play a crucial role in the psychological empowerment of followers. This study looked at leaders’ self-efficacy because it is the most central mechanism in human behaviour. Leaders who have job related confidence, a higher self-efficacy, were expected to engage more often in empowering leadership behaviour but only to followers with whom they have a trusting, reciprocal relationship; high leader-member exchange (LMX). That increased empowering behaviour would be related to higher psychological
empowerment. Data was collected by distributing questionnaires amongst leaders and followers in a retail company in the Netherlands. The results showed that the
relationship between leaders’ self-efficacy and follower psychological empowerment depended on the level of LMX, but was not explained by empowering leadership behaviour. Theoretical (i.e. organisational culture) and statistical (i.e. lower power) explanations were given to clarify the lack of certain significant results. The most important implication is that LMX should be included in future research and practitioners should try to enhance LMX as much as possible.
Keywords: psychological empowerment, empowering leadership behaviour, leaders’ self-efficacy, leader-member exchange.
Table of Contents
Abstract……….……p. 2 Introduction……….…..p. 4 Empowerment………....p. 8 Leaders’ Self-Efficacy and Empowering Behaviour……….p. 11 Leaders’ Self-Efficacy, Empowering Behaviour and
Leader-Member Exchange………..…...…p. 15 Method.………..…...p. 18 Participants………...p. 18 Procedure……….…..p. 20 Measures………....p. 20 Leader self-efficacy...………...p. 20 Empowering leadership behaviour………...p. 21 Psychological empowerment………...…..p. 21 Leader-member exchange………...…...p. 21 Control variables………..….p. 21 Results………..…p. 22 Empowering Leadership Behaviour………...p. 23 Psychological Empowerment……….…..p. 25 Exploratory………...………...…...p. 28 Discussion………...…...p. 33 References………...…...p. 48
“People are responsible adults at home. Why do we suddenly transform them into adolescents with no freedom when they reach the workplace?”
- Ricardo Semler, CEO of Semler SA (2003).
Ricardo Semler is the pioneer in creating organisations with empowered employees. His employees determine their own salaries, working hours and make decisions that affect them and the company (Semler, 2004). His manufacturing company grew from 90 to 3000 employees and the annual revenue increased from 4 to 212 million dollars when he implemented a strategy to empower his employees. The employee
satisfaction is high and so is the annual growth, with a 900% increase in the first decade (de Ruijter, 2014). It seems like the code to create a successful company has been cracked, striking the attention of both scholars and practitioners in these tough economic times.
Empowerment is related to important organisational outcomes such as customer satisfaction and organisational performance and is therefore a sought after effect (Maynard, Gilson & Mathieu, 2012). Followers feel psychologically
empowered when they receive autonomy and responsibility, get to participate in decision-making, feel like their job has meaning and feel confident in their job (Spreitzer, 1995). These are all aspects that could be affected by leaders’ behaviour and leaders have indeed been found to be important drivers to the successful
empowerment of followers (Druskat & Wheeler, 2003; Jha, 2004). They have shown to enhance follower empowerment by engaging in empowering leadership behaviour such as; delegating autonomy and responsibility, coaching and sharing information (Konczak, Stelly & Trusty, 2000). Without the leaders’ guidance, the follower will not experience psychological empowerment (Ahearne, Mathieu & Rapp, 2005). This
indicates the crucial role of leaders’ behaviour in follower empowerment.
Not all leaders have managed to empower their followers, despite the apparent positive effects on the business (Forrester, 2000; Argyris, 1998). To explain this failure in empowering, studies have focused on finding the best leadership style (Kark, Shamir & Chen, 2003; Dvir, Eden, Avolio & Shamir, 2002; Jung & Sosik, 2002; Avolio, Zhu, Koh & Bhatia, 2004). They found a positive relationship between empowerment and a person-oriented leadership style, which is characterised by its non-directive and trusting nature (Tuuli, Rowlinson, Fellows & Liu, 2012). However some studies have suggested the importance of leader characteristics such as self-efficacy (Laschinger & Shiaman, 1994).
Self-efficacy, which is the belief in one’s own ability to complete tasks and reach goals in specific situations, is of essential importance because it is the most central and persuasive mechanism in determining human behaviour (Bandura, 1977). The level of self-efficacy influences the degree to which people experience threat, what kind of activities they like to engage in (e.g. difficult or easy tasks) and the behaviour they show (Fast, Burris & Bartel, 2013; Bandura, 2006; Bandura, 1977). Finn, Mason and Bradley (2007) found that when the leaders’ self-efficacy increased they demonstrated more positive leadership behaviour, such as openness to new experience. Adding to this, lower self-efficacy leaders engage more often in
destructive behaviours - such as showing aggression - than leaders with higher self-efficacy (Fast & Chen, 2009). Both of these studies show that leaders’ self-self-efficacy could potentially influence the interaction with followers. It could influence for example the degree to which leaders show empowering leadership behaviour. Some studies have already linked the self-efficacy of the leader to constructs related to empowering leadership behaviour. Research has shown that leaders’
self-efficacy influences information sharing (Maner & Mead, 2010) and the
encouragement of ‘employee voice’ (Fast, Burris & Bartel, 2013). Yet none of them linked the self-efficacy of the leader to the entire construct of empowering leadership behaviour, which includes coaching, information sharing and delegation of autonomy and responsibility. This study will add to the body of research by looking at self-efficacy and how it is related to the engagement in empowering leadership behaviour, which ultimately would be associated with follower psychological empowerment. Proposed is a different relationship than previous research on employee voice and sharing information might suggest. Delegation increases the leader’s dependency on the followers (Hakimi et al., 2010), which is not the case with information sharing or encouraging employee voice. When followers speak up, the leaders can decide themselves what to do with their opinion. However the leader loses control when delegating, because the followers control the quality delivered on the delegated tasks. Therefore we suggest an influence of another variable on the construct of empowering leadership behaviour; namely leader member exchange (LMX).
When focusing on the leadership style or characteristics, one assumes that leaders treat all followers in the same way and followers interpret that behaviour the same way (Dunegan, Duchon, & Uhl-Bien, 1992). While previous research showed that followers describe their leader’s behaviour differently (Zalesny & Grean, 1987) possibly because the leader treats his/her followers differently (Danserau, Graen & Haga, 1975). This different treatment could especially apply to the delegating aspect of empowering behaviour, because of the increased risk leaders take when delegating autonomy and responsibility to their followers.
When leaders delegate they become more dependent on their followers (Hakimi et al., 2010) because they cannot control followers’ actions on the delegated
tasks (Wagner, Cummings, Smith, Olson, Anderson, & Warren, 2010) and therefore do not know what quality work will be delivered. It has been suggested that leaders make a cost-benefit analysis when they delegate (Milewski & Lewis, 1997), in line with the theory of equity. This theory proposes that people look at the input-output ratio when engaging in activities (Walster, Walster & Berscheid, 1978). When delegating, input can be seen as the leaders’ costs and output as the benefits. The leaders’ cost-benefit analysis for delegation could consist of: the risk of the increased dependency and cognitive efforts involved with monitoring on the cost side and time saving (Moore, 1982) and follower development on the benefit side (Milewski & Lewis, 1997).
Leaders make sure that the risk of the increased dependence that accompanies delegation is compensated by the benefits. Therefore leaders might be selective to whom they show empowering behaviour. Several scholars have proposed leaders would only delegate to competent followers with whom they have a relationship of mutual trust, reciprocity, and emotional support: a high leader-member exchange (LMX) (Liden, Wayne & Sparrowe, 2000; Chen, Kirkman, Kanfer, Allen & Rosen, 2007; Leana, 1986; Bauer & Green, 1996; Yuki & Ping Fu, 1999) to ensure the benefits would be equal compared to the costs. This situational factor of leadership was previously linked to psychological empowerment and indeed a positive
relationship between the leader-follower relationship and psychological empowerment has been found (Gomez & Rosen, 2001).
Most research on leadership and psychological empowerment did not take the situational factor of LMX into consideration. While it is important to take situational factors into consideration, according to the contingency theory. This theory states that there is no best leadership style but that this depends on the leader, follower and
situation (Gerstner & Day, 1997). Therefore this study goes beyond looking at just leadership style or leaders’ characteristic and takes into account quality of the leader-follower relationship and how this relates to the influence of leaders’ self-efficacy on empowering leadership behaviour and thereby follower psychological empowerment. First, background information about the empowerment literature will be provided, after which the relationship between psychological empowerment and empowering leadership behaviour will be explained. Then, the relationship between leaders’ self-efficacy and empowering leadership behaviour will be described, while taking into account the quality of the relationship between the follower and leader. Last, the relationship between leaders’ self-efficacy and psychological empowerment will be explained including the plausible effects of empowering leadership behaviour and the relationship between the leader and follower.
Empowerment
Empowerment goes hand in hand with independence and autonomy (Kark, Shamir & Chen, 2003) and is, despite being a relatively new construct, studied frequently. It is positively associated with many organisational outcomes, such as work satisfaction (Seibert et al., 2011; Greasley, Bryman, Dainty, Price, Soetanto & King, 2005) and due to a positive spill over effect from work to family (Gryzwacz & Marks, 2000), also related to positive outcomes such as greater perceived meaning in life (Krishnan, 2012).
There are two different forms of empowerment: structural and psychological. Structural empowerment refers to the extent that organisations provide: support, access to information and opportunities to learn (Kanter, 1993). Structural empowerment is strongly intertwined with job design and organisational climate
(Tuuli, Rowlinson, Fellows & Liu, 2012) and can be seen as the organisational measures taken to facilitate empowerment (Hales, 2001). This paper focuses on the experienced cognitions and feelings of followers, which is the psychological view of empowerment (Spreitzer & Quinn, 2001). The focus lies on followers’ reactions because an organisation can implement a lot of measures to facilitate empowerment, however the beneficial effects will not arise until the followers feel empowered (Laschinger, Finegan, Shamian & Wilk, 2004; Psoinos & Smithson, 2002). Spreitzer (1995) developed a measure of the construct of psychological empowerment based on the dimensions from Conger and Kanungo (1988). The construct includes four different dimensions: competence, meaning, impact and self-determination (Spreitzer, 1995). Competence is the belief people have in themselves that they can fulfil a task successfully, also referred to as self-efficacy of the follower (Bandura, 1977). Meaning is the value their tasks hold for the follower. Impact is the extent to which the follower perceives his or her behaviour to make a difference on operating outcomes (Thomas & Velthouse, 1990). Lastly self-determination is defined as the autonomy in the initiation and continuation of work behaviours and processes (Greasley et al., 2005).
A great number of studies have shown that leaders influence followers’ experienced empowerment (Kark et al., 2003; Dvir et al., 2002; Jung & Sosik, 2002; Avolio et al., 2004; Gomez & Rosen, 2001). Leaders fulfil an important role because they determine if the follower receives the opportunity to feel empowered (Jha, 2004), for example by showing empowering behaviour or not. For a feeling of psychological empowerment to occur, the leader can show certain behaviour such as delegating autonomy and responsibility, which is considered of essential importance for follower empowerment (Konczak et al., 2000). Receiving more autonomy and responsibility
will increase the followers’ experienced control in their work; self-determination. However psychological empowerment cannot arise without the support of the leader (Ahearn et al., 2005; Psoinos & Smithson, 2002; Forrester, 2000).
Followers make a clear distinction between empowerment and abandonment (Greasley et al., 2005). Followers felt abandoned when they received autonomy and were ‘left alone’ afterwards. Thus having control over the work process and tasks is insufficient for creating empowered followers. To create empowered followers it is important that the leader stays involved and gives guidance to the follower when needed (Greasley et al., 2005). Solely delegation might therefore not be enough to empower followers. It is essential for leaders to include behaviour such as coaching and information sharing (Arnold, Arad, Rhoades & Drasgow, 2000). Therefore this paper defines empowering leadership behaviour as: the delegation of autonomy and responsibility, information sharing and coaching (Konczak, et al., 2000).
When leaders share information they can increase the meaning of the followers’ job by positioning their tasks in the bigger picture and emphasizing the importance of them. Also, the coaching nature of empowering leadership behaviour – through the encouragement to persevere and face difficult challenges- is found to enhance follower competence, part of psychological empowerment (Ahearne et al., 2005). Follower competence will also be positively influenced through the delegation of responsibility and autonomy, which is often perceived as a reward from the
management for good performance (Zhang, Tremaine, Milewski, Fjermestad & O’Sullivan, 2012). This recognition for their efforts will lead to a higher experienced competence. Lastly followers will experience an enhanced feeling of control, self-determination, through the distribution of autonomy and responsibility. Consequently, it has been found that when leaders demonstrate more empowering behaviour, the
psychological empowerment of the followers increases as well (Seibert et al., 2011; Zhang & Bartol, 2010).
Signals from practitioners have however suggested that some managers fail to empower their followers (Argyris, 1998). Scholars have confirmed this; leaders seem to differ in the degree to which they engage in empowering leadership behaviour (Zhang & Bartol, 2010). This implies that it is likely that another variable influences empowering behaviour. An explanation might be found in the extent to which the leader feels capable to finish tasks successfully: self-efficacy.
Self-efficacy is the most central and persuasive mechanism of human agency, which is the idea that people are more than just onlookers of their behaviour
(Bandura, 1997). People can influence and control their behaviour and self-efficacy plays a crucial role in determining whether they engage in certain behaviour. People who score higher on self-efficacy tend to think in self-enhancing manners; their actions would lead to certain desired effects. While people who score lower on self-efficacy tend to think in self-debilitating ways; their actions would not lead to a desired effect (Bandura & Locke, 2003). People do not have an incentive to act if they do not believe that their actions lead to a desired effect (Bandura, 2006). Therefore self-efficacy might play a crucial role in determining whether leaders show
empowering leadership behaviour.
Leaders’ Self-Efficacy and Empowering Behaviour
There are two different kinds of self-efficacy: leaders’ self-efficacy and leadership efficacy (Hannah, Avolio, Luthans & Harms, 2008). Leadership efficacy is a multi-level perspective that includes collective and follower efficacy. It relates to leaders’ confidence in the organisation, in their teams and followers, but not necessarily the
confidence in their own leadership skills (Hannah et al., 2008). This study will focus on the leaders’ self-efficacy, which is focused on the leaders’ cognitions about themselves. By doing so this paper will isolate –as much as possible– the effect that leaders’ cognitions about their specific behaviours and capabilities have on their engagement in the construct of empowering leadership behaviour, which has not been studied before. Adding to this, by isolating the effects of leaders’ cognitions,
outcomes of this study can be used to improve leaders’ selection process. Campbell (1956), Kenan (1948), and Stogdill and Shartle (1955) already mentioned a few decades ago that higher-level, more experienced, managers were more likely to report the use of delegation than lower-level, less experienced, managers. One might argue that experience therefore plays an important role in determining the level of delegation. However the self-concept theory states that people’s self-concept, self-efficacy, is built upon self-perceptions and external experiences (Mischel & Shoda, 1995). Thus job related experience contributes to the level of self-efficacy (Ormrod, 1999), which was confirmed by previous research (McCormick, Tanguma & Lopex-Forment, 2002).
As the scholars already mentioned these higher-level managers had more experience, which thus would have resulted in a higher self-efficacy. It is possible to speculate that these higher-level managers had encountered task related success more often than lower-level managers, which also could have enhanced their self-efficacy positively. It could therefore have been that those higher-level experienced managers had higher self-efficacy than lower less experienced managers. Possibly this could have been the underlying factor in determining whether leaders reported the use of delegation.
empowering leadership behaviour. The first one originates from the tasks leaders like to occupy themselves with. Leaders are in general reluctant to delegate difficult and important tasks to their followers (Leana, 1986), because those tasks are an essential part of their work, which distinguishes them from their subordinates. When leaders delegate, initially the easier tasks are distributed to the followers. Leaders with lower self-efficacy like to engage in these easier activities because they know they will succeed in these (Bandura, 1977), and will therefore be less likely to delegate these tasks. They will also be disinclined to delegate more difficult tasks, because followers might come to the leader for guidance. When followers ask for advice on these
difficult tasks and leaders are not able to give it, they can ‘lose face’ because leaders are expected to be more competent and know better than their followers (Magee & Galinsky, 2008; Tyler & Chen, 2001). Lower self-efficacy leaders do not necessarily feel confident in this advice giving aspect of their job and will therefore avoid it. While, on the contrary, leaders with higher self-efficacy like to engage in challenging activities (Bandura, 1977). Delegating easier tasks, which they like less, creates more time to spend on challenging tasks. Therefore leaders with higher self-efficacy will be more likely to delegate the easier tasks to some of their followers.
The last reason why leaders’ self-efficacy might play an important role is the perceived threat that lower efficacy leaders experience. Leaders with lower self-efficacy experience, more often than leaders with higher self-self-efficacy, a threat to their ego (Fast et al., 2013; Anderson, 1992). This ego threat, caused by the feeling of being easily replaced (Gist, 1987), has shown to have negative effects on the leader’s behaviour, such as increased engagement in aggressive behaviour towards their followers (Fast & Chen, 2009; Scheepers & Ellemers, 2005). People will always try to defend their power and therefore avoid situations in which their ego can be
threatened (Kermer, Driver-Linn, Wilson & Gilbert, 2006; Quist & Resendez, 2002). Previous research has indeed confirmed that leaders with lower self-efficacy try to prevent potential threats by not sharing information with skilled followers (Maner & Mead, 2010), because skilled followers could form a larger threat to the leader once they receive more information. Another threat that a leader can
experience is related to ‘employee voice’. Leaders who score higher on self-efficacy have shown to encourage followers to speak up more than leaders who score lower on self-efficacy, because lower self-efficacy leaders experience a threat to their ego when their followers speak up (Fast et al., 2013), which they want to avoid. Leaders with higher self-efficacy might encourage their followers to speak up because they think more often in opportunities than in threats. This thinking in opportunities has been argued to exist because higher self-efficacy people perceive situations as controllable; their actions can influence the final result (Krueger & Dickson, 1993). Higher self-efficacy leaders would focus on the potential benefits of the well-informed skilled follower, such as the additional help and time saving, instead of the threat those followers might become.
Taking into consideration the preferences in task engagement and prior research regarding employee voice and information sharing it would seem logical to suggest that leaders’ self-efficacy positively influences empowering leadership behaviour. However empowering leadership behaviour also consists of delegating autonomy and responsibility (Konczak et al., 2000). When leaders delegate autonomy and responsibility, they lose control and become more dependent on their
subordinates (Hakimi, van Knippenberg & Giessner, 2010). Which creates a
vulnerable position because followers might act in self-interest or deliver poor quality work. Leaders cannot directly control their followers’ behaviour (Wagner et al.,
2010). They can hold their followers accountable, but in the end leaders are still held responsible for the result of the delegated tasks by the top management (Hakimi et al., 2010). Therefore it seems unlikely that leaders would take the risk by showing
empowering behaviour to all their followers, especially since people are generally inclined to reduce risk (O’Toole & Lawler, 2006).
Risk is not the only negative aspect in delegation; the loss of control and the cognitive efforts involved with monitoring the followers can also be considered as downsides of delegating. Despite downsides there are also advantages when delegating such as: follower development and the time the leader saves (Moore, 1982). Leaders are suggested to ensure that the downsides of delegation are compensated with the advantages (Milewski & Lewis, 1997). This cognitive evaluation is also referred to as the cost-benefit analysis. Leaders might only show empowering behaviour to those they feel are reliable, competent and trustworthy to maximise the possibility that the benefits are equal to the costs.
Leader’s Self-Efficacy, Empowering Behaviour and Leader-Member Exchange Followers have reported to receive different degrees of responsibility and autonomy from one leader (Zhang & Bartol, 2010), which indicates that leaders indeed
differentiate between followers (Danserau et al., 1975). Differentiation can originate from different levels of trust (Pellegrini & Scandura, 2006) and competence (Gomez & Rosen, 2001; Yuki & Ping Fu, 1999). Trust and follower competence are related to the degree to which leaders engage in delegating and thus possibly the leaders’ empowering behaviour (Liden et al., 2000; Bauer & Green, 1996).
Trust is an important aspect of the leader-member exchange (LMX). LMX refers to the different quality in relationships developed between the followers and
leaders. There are high and low quality relationships, which are characterized by different cognitions (Zhang, Tremaine, Milewski, Fjermestad & O’Sullivan, 2012). The formation of low quality relationships (low LMX) means that the relationship between the leader and follower is based on fulfilling contractual obligations (Liden & Graen, 1980). Behaviours and expectations are limited to what is written in the job descriptions. High quality relationships (high LMX) are, on the contrary,
characterized by: mutual trust, reciprocity, and emotional support (Chen et al., 2007). LMX and follower psychological empowerment have been linked by scholars in previous studies. Scholars found that followers who had high LMX relationships with their leader also experienced more psychological empowerment (Gomez & Rosen, 2001; Chen et al., 2007). Explanation might lie in the increased empowering leadership behaviour those followers receive. Followers with high LMX relationships have reported to receive more work-related benefits (Kim & George, 2005), of which empowering behaviour can be considered one (Spreitzer, 2007). The increased benefits might derive from the reciprocal nature of the high LMX relationship. When the leader gives the follower something extra (such as increased delegation), the follower will have to also give something positive back to the leader (such as effort). Leaders also have shown to like followers with whom they have high LMX
relationships better, which might result in work-related benefits such as more
information sharing with those followers (Gerstner & Day, 1997). Followers in a high LMX relationship with their leader also receive greater access to resources (Gerstner & Day, 1997) and responsibility (Kozlowski & Doherty, 1989; Liden & Graen, 1980) than followers with low LMX relationships. Which might be because leaders perceive those followers to be more competent and trustworthy (Dienesch & Liden, 1986). The costs of engaging in empowering leadership behaviour to those
competent, trustworthy (high LMX) followers would be relatively lower than for less competent, trustworthy followers (low LMX). Because there is a larger chance that those high LMX followers deliver good quality results due their competence and lower chance of acting self-interest, due to the higher trustworthiness. The followers’ higher competence also implies that they require less guidance, which could increase time saving (benefits) and decrease the amount of coaching needed, monitoring (costs). Thus for these high LMX followers, there is a larger chance that the costs are compensated by the benefits in the analysis.
Therefore we propose that LMX influences the relationship between leaders’ efficacy and empowering leadership behaviour. Leaders who have higher self-efficacy would be more likely to engage in empowering behaviour -such as delegating to their followers- but only if they have a high quality relationship with those
followers. Conversely, the lower the self-efficacy of the leaders, the less likely it would be for those leaders to engage in empowering behaviour. This originates from the increased threat they experience and because they like to engage in easier tasks. Furthermore, when LMX is low, we expect no relationship between leaders’ self-efficacy and empowering leadership behaviour. This empowering behaviour is proposed to be associated with psychological empowerment.
Followers can experience empowering leadership behaviour (i.e. distributing autonomy and responsibility) as recognition from the management (Zhang et al., 2012), which could enhance their feeling of competence. Delegating autonomy and responsibility also gives followers an increased feeling of task control. They have the autonomy to decide how to do it, which influences the self-determination dimension of psychological empowerment. Leaders who show more empowering leadership behaviour encourage followers to speak up through the coaching aspect, which
enhances the feeling of participation in the decision-making process. To sum it up, empowering leadership behaviour is expected to, as explained above, lead to higher psychological empowerment (Seibert et al., 2011). Therefore the following
hypotheses and model (as depicted in Figure 1) are proposed:
Hypothesis 1: Leaders’ self-efficacy is positively associated to empowering leadership behaviour, but only if LMX is high.
Hypothesis 2: Leaders’ self-efficacy will be positively related to
psychological empowerment when LMX is high, explained by empowering leadership behaviour.
METHOD
Participants
The data for this cross-sectional study was collected from different stores from a retail company in the Netherlands. The design used partly avoided the common method variance bias (Podsakoff & Organ, 1986), because the data was gathered from two different sources; the leader and follower. In total 40 leaders were contacted of which
Leaders’ self-efficacy Leader-member exchange Psychological empowerment Empowering leadership behaviour
31 leaders replied, a response rate of 78%. From the 31 leaders, 53 followers
responded to the questionnaire, out of the 93 approached, which is a response rate of 56%. From the 31 leaders who replied, five leaders had to be excluded from the analysis, because none of their followers filled in the questionnaire. This means that the analysis included 26 leaders and 53 followers, on average 2.03 (SD = 0.82) followers per leader filled in the questionnaire.
Demographics of the leaders showed that 61% were male and 39% female, on average 38.00 (SD = 9.42) years old. Their average work experience as leader was 56.69 (SD = 53.54) months and the average span of control was 22 followers per leader. From the leaders, 23% finished high school, 26% finished higher vocational education and 50 % had a bachelor or master degree from a university or university of applied sciences. 34% of the followers were male and 66% female. The average age was 32.50 (SD = 10.96) years and on average, the followers worked 23.3 hours per week. The average length of the leader-follower relationship was 17.51 (SD = 16.28) months. Of the follower participants 24.6% finished high school, 45.3% higher vocational education and 30.2% had a bachelor or master degree from a university or university of applied sciences.
A post-hoc power analysis was conducted using software package G*power (Faul, Erdfelder, Buch & Lang, 2009). The sample size of 53 was used to calculate the statistical power, the alpha level of p = .05, and two predictor values were added. Cohen (1977) recommended using the following effect sizes in the calculation: small (f2 = .02), medium (f2 = .15) and large (f2 = .35). The post-hoc analysis revealed that the statistical power was .13 for detecting a small effect, .68 for a medium effect and .97 for a large effect. Meaning that this study had more than the .80 desired power to detect large effects (Cohen, 1992), however medium to small effects were more likely
to be missed in this study. Procedure
All participants were randomly selected by using an online random name-picker tool. First, all leaders were entered into the name-picker, after which forty were selected. The followers were selected by entering all followers from the selected leader into the name-picker tool. The tool picked three followers per leader at random. The leaders received an email asking them to fill in a special questionnaire. In the attachment of the email information about the research was included. The research was described as: ‘Leader characteristics and behaviours and followers’ needs’. To fill in the
questionnaire each leader received a personal link, further instructions were given on the screen. The leaders first filled in some demographics, after which they filled in the questionnaires. Most followers did not have their own work email address and could not be approached directly. Therefore the leaders were requested by mail to send the link of the questionnaire to three followers, selected by us through the name-picker previously. Followers could either finish the questionnaire during work hours or in their private time. Followers were clearly informed about the anonymity and voluntary participation of the research.
Measures
Independent variables.
Self-efficacy leader. The self-efficacy of the leader was measured through the leader perspective. The leaders filled in the validated eight-item New General Self-Efficacy Scale (NGSE) by Chen, Gully and Eden (2001). The leaders could respond on a 5-item Likert scale ranging from 1= strongly disagree to 5= strongly agree. An example of one of the items was: “Compared to other people, I can do most tasks very well”. The test-retest and internal reliability have been found in previously conducted
research to be high (Chen et al., 2001) and the scale in the current study had a Cronbach’s Alfa of .78.
Leader-member exchange. Leader-member exchange was measured by using the seven-item LMX-7 questionnaire (Scandura & Graen, 1984). This test measures the one-dimensional construct of LMX (Joseph, Newman, Sin, 2011). Items were rated on a 5-point scale ranging from 1 = strongly disagree to 5 = strongly agree. An example of one of the items: “My leader knows about my problems and talents”. The Cronbach’s Alfa was .93.
Dependent variables.
Empowering leadership behaviour. To measure the leader’s empowering behaviour nine items from the Empowering Leadership Scale (Ahearne et al., 2005) were included in the online questionnaire. Followers rated the items on a 5-point Likert scale ranging from 1=strongly disagree to 5= strongly agree. An example of the item: ´My manager believes that I can handle demanding tasks’. The Cronbach’s Alfa was .90.
Psychological empowerment. To measure psychological empowerment the followers had to fill in Spreitzer’s (1995) Psychological Empowerment Instrument (PEI). This is a 12-item questionnaire that measures the different constructs:
competence, impact, meaning, and self-determination. An example of an item: “The work I do is important to me”. Items were rated on a 5-point Likert scale ranging from 1= strongly disagree to 5 = strongly agree. The Cronbach’s Alpha for the PEI was .85.
Control variables. Based on the literature we included two different control variables. The first one is the length of time that a follower has worked together with a leader; this is known to affect their LMX relationship (Dienesch & Liden, 1986). To
measure this, the question: “My manager leads me for … months” was added. The average length of the leader-follower relationship was 17.51 (SD =16.28) months. The second control variable is leaders’ experience, since it is related to self-efficacy
(McCormick et al., 2002) and shown to influence the use of delegation (Campbell, 1956; Kenan, 1948; Stogdill & Shartle, 1955). To measure leaders’ experience they had to answer the following question: “How many months experience as a leader do you have?” (M = 56.69, SD = 53.54).
RESULTS
As explained above, five leaders were excluded from the analyses. None of the other participants were excluded based on the outliers test. We used the Shapiro-Wilks test to estimate the normality of the data, because this test has shown to have the greatest power (Razali & Wah, 2011). The Shapiro-Wilks test proved to be non-significant for all variables except leader-member exchange D(53)=.95, p < .05. Meaning that the data was normally distributed, except for the leader-member exchange data. The Kolmogorov-Smirnov test for LMX was however not significant D(53)=.10, p > .05. We assumed that the significant test for LMX was not an issue since the values for skewness were within the standards of < 2 and kurtosis < 7 as well (Finch, West & Mackinnon, 1997; West, Finch & Curran, 1995)
Table 1 shows the means, standard deviations, correlations and reliabilities from the variables measured at the follower level. Leaders’ self-efficacy, with a mean of 4.00 (SD = 0.40), was not significantly correlated with any of the variables
measured at the level of the followers and therefore excluded from Table 1. The control variable leader experience (M=56.69, SD=53.54), did not correlate
Table 1
Descriptive Statistics and Correlation Matrix
Variable Mean SD 1 2 3 4 1. Length of the 17.51 16.28 Leader-Follower Relationship (months) 2. Leader-member 4.06 0.65 .40** (.93) Exchange 3. Empowering 3.91 0.62 .44** .75** (.90) Leadership behaviour 4. Psychological 3.78 0.52 .33* .58** .44** (.85) Empowerment
a. Reliabilities for each scale are listed between parentheses on the diagonal. N = 53; *p < .05. ** p < 0.01.
Empowering leadership behaviour
The first hypothesis predicted a positive relationship between leaders’ self-efficacy and empowering leadership behaviour when LMX is high. Taking into account the random intercept – because of the nested data- a linear mixed model was conducted. The variance estimate parameter was redundant, meaning that the nested data did not affect the results. A hierarchical multiple regression analysis, in line with the
procedure from Cohen and Cohen (1983), was conducted to evaluate if the association between leaders’ self-efficacy and empowering leadership behaviour depended on the quality of the leader-member exchange. First leaders’ self-efficacy and LMX were centered to the mean to reduce multicollinearity problems (VIF<10). After which, the leaders’ self-efficacy-by-LMX interaction was computed (Aiken & West, 1991). The control variable was entered in the hierarchical regression model in step 1, the two predictors in step 2 and the interaction in step 3, for results see Table 2.
Table 2
Results of Hierarchical Moderated Regression Analysis for leader member exchange (LMX) and Leader Self-efficacy on Empowering leadership behaviour and
Psychological Empowerment.
Steps and predictor variables Empowering Psychological
Leadership Empowerment
Behaviour
β 𝑅𝑅2 ∆𝑅𝑅2 β 𝑅𝑅2 ∆𝑅𝑅2 Step 1
Length relationship leader and .44** .20** .33* .11* follower
Step 2
Length relationship leader and .19 .09
follower
Leaders’ self-efficacy .09 -.11
LMX .67** .60** .40** .56** .36** .26**
Step 3
Length relationship leader and follower .18 .07
Leaders’ self-efficacy -.35 -1.74*
LMX .02 -1.83
Leaders’ self-efficacy X LMX .82 . 60** .01 3.01* .44** .07* Note. N=53; *p < .05. ** p < 0.01.
Results indicated that the length of the relationship between the leader and follower was significantly positively associated with empowering leadership behaviour β= .44, t(51)= 3.53, p = .001, resulting in a 𝑅𝑅2= .20, F(1,51)=12.46, p =.001. Entering the main effects of LMX and leaders’ self-efficacy significantly improved the prediction of empowering leadership behaviour ∆R2= .40, ∆F (2,49)= 24.35, p < .001. A significant positive association between LMX and empowering leadership behaviour was found β= .67, t(49)= 6.69 , p < .001. There was, as expected, no significant association between leaders’ self-efficacy and empowering leadership behaviour β= .09, t(49)= 1.00 , p = .321. The interaction effect was also
not significant β= .82, t(48)= .81, p= .424. Meaning that the relationship between leaders’ self-efficacy and empowering leadership behaviour did not depend on the level of LMX. Therefore, the first hypothesis is not supported by the data.
Psychological empowerment
The second hypothesis proposed that leaders’ self-efficacy is positively associated with psychological empowerment explained by increased empowering leadership behaviour when LMX is high. Based on the findings of above we knew that the mediated moderation model, as depicted in Figure 1, would not apply to our data. The Structural Equation Modeling (SEM), with the R-package Lavaan version 0.5-16 (Rosseel, 2012) likewise confirmed that the model did not fit onto our data. We used the Chi-Square (χ2), Root Mean Square Error of Approximation (RMSEA),
Goodness-of-Fit Index (GFI), the Tucker-Lewis Index (TLI) and the Normed Fit Index (NFI) to evaluate the fit of the model. The fit measures for the model are compared to the ideal values, provided by Lomax and Schumacker (2012). They stated that RMSEA had to be less or equal to .08, GFI greater or equal to .90, TLI greater or equal to .90 and NFI greater or equal to .90. Table 3 shows the results and fit measures of our proposed model based on the data. Demonstrated is that GFI is the only fit measure that is considered acceptable. The RMSEA, TLI, NFI showed a less than acceptable fit.
Table 3
Results and fit measures of the SEM conducted to analyse the entire model proposed.
Fit χ2 RMSEA GFI TLI NFI
measure
(50)= .312 .90* .47 .47
18.44
p < .001 N= 53
*= Fit measure with an acceptable or good fit.
To see whether the data partly supported hypothesis two, we conducted a couple additional analyses. Again, the nested data did not affect the results. First, in line with previous research and our predictions, we detected a significant positive correlation between empowering leadership behaviour, the proposed mediator, and psychological empowerment, the depended variable β= .44, t(51)=3.52, p = .001. After which, a hierarchical multiple regression was executed to test whether the relationship between leaders’ self-efficacy and psychological empowerment depended on the level of LMX. The length of the relationship between the leader and follower, the control variable, was significant positively associated with psychological
empowerment β= .33, t(51)= 2.48, p = .017. Adding the main effects of LMX and leaders’ self-efficacy significantly improved the predictability of the model ∆R2
= .26, ∆F (2,49)= 9.83, p < .001. Again, a positive, significant association between LMX and psychological empowerment was found β= .56, t(49)= 4.42, p < .001. Leaders’ self-efficacy was not significantly associated with psychological empowerment β= -.11, t(49)= -.964, p = .340. To test whether the interaction effect would result in a better prediction of psychological empowerment, the interaction was included in the last step. The interaction was significant β= 3.01, t(48)= .2.48, p = .017 and it significantly improved the prediction of ∆R2 = .07. ∆F (1,48)= 6.15, p = .017. The
final model explained 44% variance of psychological empowerment (see also Table 2).
Simple slopes for high (= 1 standard deviation above the mean) and low (=1 standard deviation below the mean) LMX were plotted to shed light on how LMX influenced the relationship between leaders’ self-efficacy and psychological empowerment, see Figure 2. The length of the relationship between the leader and follower was not significant when the interaction effect was added and therefore excluded from the simple slope regression analysis. The slope for low LMX revealed a significant negative association between leaders’ self-efficacy and psychological empowerment β= -.37, t(49)=-2.56, p = .014, which is considered a medium, almost large effect, d =0.73 (Cohen, 1988). This effect was found contrary to the prediction that there would be no significant association between leader self-efficacy and
psychological empowerment when LMX was low. The direction of the slope for high LMX was in line with our prediction, which was that there would be a positive relationship between leaders’ self-efficacy and psychological empowerment for followers with high LMX relationships. This positive association was however not
significant β= .18, t(49)=1.09, p = .281. The results revealed that leaders’ self-efficacy was more strongly (and negatively) related to psychological
empowerment for lower
P sy chol og ic al E m pow er m ent Leader Self-Efficacy High LMX Low LMX
Figure 2. Simple slopes of the interaction effect of LMX and leader self-efficacy on psychological empowerment.
levels, rather than higher levels of LMX. It is important to underline that the post-hoc analysis revealed a power of .68 for detecting medium effects; this is below the preferred .80. The positive relationship would have probably had a stronger level of significance, given more statistical power.
Conclusively, we found a positive significant relationship between
empowering leadership behaviour and psychological empowerment. The relationship between leaders’ self-efficacy and psychological empowerment depended on the level of LMX, but was not explained by empowering leadership behaviour; therefore hypothesis two was only partly supported. It might be that another variable influenced this relationship1.
Exploratory
This study was part of a larger study, therefore some additional exploratory analyses were conducted, based on questions that arose from previous research. In most empowerment research – as in this study- an excess in received autonomy and responsibility is not measured. When followers score high on psychological empowerment it can either mean that their needs are satisfied, or potentially exceeded. Exceeding followers’ autonomy needs can cause stress (Ryan & Deci, 2000), and is therefore something that should be avoided. Need fulfilment is defined
1
To see if there were other variables explaining the relationship between the interaction effect and psychological empowerment, some additional analyses were conducted. The question that could have measured the leaders’ trust in the followers was included: ‘I trust my followers’. However no correlation was found, contrary to previous research (Gomez & Rosen, 2001), with LMX β= .04, t(51)= .27, p = .787 or psychological empowerment β= -.12, t(51)=-.84, p = .406. The failure of finding an effect could have to do with the small sample size, poor measurement or ceiling effect (M = 4.19, SD =0.69).
as whether the distributed levels of autonomy and responsibility match the actual desired level of autonomy and responsibility (Deci & Ryan, 2000).
There are some contradictory results within the empowerment research that lead us think that need fulfilment is an important factor in this area. Ahearne et al. (2005) found that the followers with little work experience had positive reactions to their leaders’ empowering behaviour. The other study from Kark et al. (2003) found that those followers had negative reactions to the empowering behaviour. Our proposition is that this had to do with different needs of the followers; one group might have had a lower need for autonomy and therefore negative reaction to empowering leadership behaviour that exceeded their needs. While the other group might have had a higher need for autonomy and therefore positive reaction to the empowering behaviour that fulfilled their needs. The quality of the relationship between leader and follower might influence this need fulfilment as well.
Followers who have higher LMX relationships receive increased attention from the leader (Dunegan, Uhl-Bien & Duchon, 2002) and leaders communicate more often with high than low LMX followers (Chen et al., 2007). When the follower and leader communicate more and the relationship is based on trust, followers might be more open about their needs and wishes. This could increase the chance of need fulfilment. Therefore we would expect that the level of LMX will be positively
associated with need fulfilment, but we expect that this relationship will be influenced by another variable.
LMX might increase the verbal communication; the followers are more likely to express their preferences. However there might be situations in which followers do not feel comfortable expressing their feelings or needs regarding the distributed autonomy and responsibility (Cancian, 1987), for example in team meetings or other
public settings. Non-verbal communication will be important when followers do not feel able to express their needs.
When followers cannot express their needs, their facial expression might still show their feelings towards the distributed autonomy and responsibility. For example: when followers experience too little autonomy and responsibility they might show disappointment or frustration in their expression. While they might have an anxious or uncomfortable facial expression when they experience too much autonomy and
responsibility. The capability to recognise these emotions is also referred to as emotional intelligence (EI) (Jordan & Troth, 2010).
The emotional intelligence of the leader already proved to be related to team empowerment (Erkutlu & Chafra, 2012), but again an excess in empowerment was not included in this study. Higher EI leaders have been suggested to anticipate on followers’ needs more adequately, because they recognize these changes in their followers’ facial expression better (Erkutlu & Chafra, 2012). They would thus adapt the distributed levels of autonomy better in line with their followers’ needs, based on their non verbal communication; facial expression.
Therefore we expect an interactive effect between both verbal and non-verbal communication on need fulfilment, as depicted in Figure 3. Followers with higher LMX would be more likely to express their needs towards their leader, however when the situation restricts them from freely communicating, leaders with higher emotional intelligence will still recognise their needs. Thus we would expect that leaders’ EI will
moderate the relationship between LMX and empowerment need fulfilment. Such that Leader-member exchange Need fulfilment Emotional Intelligence
Figure 3. The proposed interaction effect of leader-member exchange and emotional intelligence on need fulfilment.
there will be a positive relationship between LMX and need fulfilment, which would be stronger when leader’s EI is high.
Measures.
Emotional intelligence was measured by using the revised version from ‘Reading the Mind in the Eyes test´ from Vellante, Baron-Cohen, Melis, Marrone, Petretto, Masala, and Preti (2013). This test contained 33 items in which the leaders had to match one out of the four different emotions to the eyes demonstrated. For every question that was answered correctly, they received one point, thus the test scores could range from 0 to 33. The higher the score, the better the leader was able to recognize the emotions of others. The test retest reliability of the test was .83 (Vellante et al., 2013).
Need fulfilment was measured with one item: “The amount of responsibility and autonomy I receive is…”. The participants could respond by adjusting the point on the scale to the left and right. The scale was set in the middle (50), which meant perfect amount. The ultimate left (0) meant too little while the ultimate right (100) meant too much. Thus the less the score deviated from 50, the more the needs were fulfilled. The variable need fulfilment was recomputed for the analyses by deducting 50 from the score and multiplying the negative values by -1. By doing so, the closer the score was to 0, the better the need was fulfilled (M= 14.65, SD= 13.50). The higher the score on the variable need fulfilment, the less the need was satisfied. Results.
One participant was excluded from the analyses, because their leader did not fill in the EI test. The mean of the correct answers on the EI test was 21.84 (SD= 3.97). The data was not normally distributed, the Shapiro-Wilks test proved to be significant D (26)=0.83, p < .001, the skewness was -2.04, meaning that the data for EI was moderately skewed (Finch, West & Mackinnon, 1997).
To control for the lack in normality and nested data, a generalized linear mixed model was conducted to shed light on the relationship between LMX, EI and
empowerment need
fulfilment. Before analysing the data LMX and EI were standardized. The
relationship between LMX and the variable need fulfilment was not significant and the association between emotional intelligence and the variable need fulfilment also proved to be non-significant. However there was a significant association between the interaction term of EI and LMX and the variable need fulfilment, β=.28, F(1,48)= 14.15, p < .001. Simple slopes for high (1 standard deviation above the mean) and low (1 standard deviation below the mean) EI were plotted to shed light on how EI influenced the relationship between LMX and need fulfilment (see Figure 4). The slope for low EI revealed a significant negative
association between LMX and the variable need fulfilment β= -.24, F(1,48)=7.01, p = .011. LMX was thus associated with better need fulfilment when EI was low. The slope for high EI revealed a significant positive association between LMX and the need fulfilment variable β= .33, F(1,48)= 14.93, p < .001. LMX was thus negatively associated with need satisfaction when EI was high. Both relationships were against the predictions based on the literature.
V ar iab le n eed f u lf il m en t Low EI High EI LMX
Figure 4. Simple slopes of the interaction effect of LMX and EI on the variable need fulfilment.
Additional analyses were conducted to look at differences between followers who reported too much and too little autonomy and responsibility. The data was divided into two groups by splitting the file: above or below the average LMX score. After which we conducted an one sample t-test analysis to see if the values
significantly differed from 50 (perfect amount of autonomy and responsibility). The score on the need fulfilment significantly differed from the perfect amount when followers reported high LMX relationships t(26)= 3.17, p = .004. This meant that followers with higher LMX relationships experienced an excess in autonomy and responsibility.
DISCUSSION
Leaders experience threat and activities that they like differently due to their level of efficacy (Fast et al., 2013; Gist, 1987). Prior research regarding leaders’ self-efficacy demonstrated positive relationships with constructs related to empowering leadership behaviour such as coaching and information sharing (Fast et al., 2013; Maner & Mead, 2010). However the delegation construct of empowering behaviour involves risk because followers can deliver poor quality on delegated tasks. This increases the leaders’ dependency on the follower (Hakimi et al., 2010; Wagner et al., 2010). The equity theory suggests that leaders would want to ensure that the
downsides are compensated by the benefits of delegating (i.e. follower development and time saving, Moore, 1982). This cognitive evaluation of the downsides and advantages of delegation is also referred to as the cost-benefit analysis (Milewski & Lewis, 1997). Higher self-efficacy leaders, who feel comfortable with empowering, will selectively show more empowering behaviour to high LMX followers. These followers are perceived as trustworthy and competent (Gomez & Rosen, 2001).
Higher empowering behaviour would be associated with higher levels of followers’ psychological empowerment. According to the contingency theory including factors such as LMX is of essential importance because it allows for followers’ individual interpretation of leadership style and situation. This influences followers’ perspective of the leader (Gerstner & Day, 1997).
This study showed that the relationship between leaders’ self-efficacy and psychological empowerment depended on the level of LMX. Though we predicted that there would be no significant association between leaders’ self-efficacy and psychological empowerment for low LMX, we found a significant negative
association. The differential cost-benefit analyses from the leaders might explain the negative association (Milewski & Lewis, 1997).
One of the costs in the analysis is the loss of control; when leaders do tasks themselves they have more certainty about the quality of the results. Leaders perceive low LMX followers as less skilled; thus less likely to perform well on delegated tasks (Gomez & Rosen, 2001; Liden & Graen, 1980). Leaders with higher self-efficacy see themselves as very capable of delivering good results, while leaders with lower self-efficacy are less convinced about their own ability to perform well on tasks (Bandura, 1977). Lower self-efficacy leaders might therefore expect a smaller gap between the delivered quality of the follower and themselves. This quality gap is larger for higher self-efficacy leaders, because they see themselves as capable of delivering good quality results. This larger gap makes it less beneficial for them to empower low LMX followers. Consequently, higher self-efficacy leaders might prefer to do the task themselves, resulting in lower psychological empowerment of followers. While lower self-efficacy leaders would delegate more to low LMX followers, thus empowering
them more. This would explain the negative association between leaders’ self-efficacy and psychological empowerment for low LMX followers.
While we did not find the positive relationship between leaders’ self-efficacy and psychological empowerment to be significant for high LMX, it was in the expected direction. This lack in significance might have originated from a statistical limitation that this study has. The sample size of the study is relatively low, which is a problem because it increases the chance of committing of a type II error (Huck, 2004). That is, failing to detect an effect in the sample, even though one exists. The increase of type II error and with that, smaller power for detecting small to medium effects, could be an explanation why we failed to find a significant positive
relationship between leaders’ self-efficacy and psychological empowerment for higher LMX levels.
There might also be a theoretical explanation why the prediction for the relationship between leaders’ self-efficacy and psychological empowerment for high LMX was not supported. This study and the study from Fast et al. (2013) argued that lower self-efficacy leaders’ empower less because of an increased experienced threat. The organisational culture is a variable that could not influence the outcomes of the experimental studies, but could however play a role in this field study.
Organisational culture influences the way people consciously and
subconsciously think, perceive, feel and therefore consequently act (Schein, 1990). Pool (2000) found that organisational culture influences the leaders’ experienced stressors and job tension. This could also apply to the leaders’ experienced threat, since this is a stressor as well. Pool’s (2000) suggestion was that when looking at stressors, one should take environmental aspects – such as organisational culture - into consideration.
All previous studies demonstrated that higher self-efficacy leaders engage more often in encouraging employee voice and information sharing than lower self-efficacy leaders (Fast et al., 2013; Maner & Mead, 2010). However, in all but one study, organisational culture could not influence the results because of the research design. The only study where organisational culture could have affected the results was the one conducted in a large organisation in oil and gas exploration and refinery by Fast et al. (2013). That culture is characterised by its reporting, technical, and highly competitive orientation. This study’s organisational culture is characterised by its protective and cohesive nature. This protective environment could have reduced the experienced threat for lower self-efficacy leaders, which removed the barrier to empower followers. The education level of the followers could also have contributed to reducing the threat. Fast et al. (2013) reported that the respondents were highly educated and trained in the hard sciences. The followers from our organisation were on average less educated. This could imply that they did not form a threat to the lower self-efficacy leaders. It stresses the importance for future studies to include
organisational culture and leaders’ experienced threat as variables as well as having a larger sample size. By doing so we can determine whether the non-significant positive relationship derived from the low power or because of other intervening variables. This study also revealed a significant positive relationship between
empowering leadership behaviour and psychological empowerment, confirming results found in previous research (Seibert et al., 2011; Zhang & Bartol, 2010). However, contrary to our predictions, the relationship between leaders’ self-efficacy and empowering leadership behaviour was not affected by the level of LMX.
Empowering leadership behaviour also did not explain the relationship found between leaders’ self-efficacy and LMX on psychological empowerment. This could be
explained from a statistical or theoretical perspective.
From a statistical point of view, a complicating methodological factor is that leaders’ self-efficacy was on average high with relatively little variance. This methodological issue was also previously specified in previous studies (Fast et al., 2013).2 It makes effects on empowering leadership behaviour difficult to detect. We still found a relationship with psychological empowerment despite the limitation. This might indicate that the effect of leaders’ self-efficacy and LMX on psychological empowerment would be larger than on empowering leadership behaviour. As mentioned previously, another explanation is the small sample size and with that, higher chance of missing small to medium effects.
Besides statistical, there could also be a theoretical explanation. It seems like the followers’ perceived empowering leadership behaviour did not explain the interactive effect of leaders’ self-efficacy and LMX on psychological empowerment. Explanation could be found in the way we measured LMX. LMX was measured from the followers’ perspective, while our argumentation stems from the leaders’. We argued that when LMX is high, leaders see followers as more trustworthy and competent.
Prior research showed that the correlations between the leaders’ and followers’ perspective of LMX were relatively low (Sin, Nahrgang & Morgeson, 2009). This could either mean that leaders and followers have different views about the quality of their relationship or reflect measurement problems (Yukl, 2010). This lower correlation has been argued to exist because of bad operationalization, biases and translation of the scale (Zhou & Schriesheim, 2011). However leaders and
2
Leaders’ average score on the self-efficacy questionnaire, as found by Fast et al., 2013, was 4.46 with a standard deviation 0.47.
followers could also have different views about the quality of their relationship (London, 1995). If the latter is correct, we cannot draw the conclusions we did. A study by Gomez and Rosen (2001) showed however that our line of argumentation could still be valid. They demonstrated a positive association (.47) between leaders’ trust and followers’ perceived LMX. They found that followers feel more empowered when the leaders trust them more. Trust could explain why we did not find the proposed relationships. Namely, leaders would be disinclined to engage in empowering leadership behaviour if they do not trust those followers, regardless of follower perceived high LMX. Conversely, leaders might demonstrate empowering leadership behaviour if they trust those followers, even if followers perceive the LMX as low. This would explain why there was no relationship of leaders’ self-efficacy and LMX on empowering leadership behaviour.
However there was a positive relationship between empowering leadership behaviour and psychological empowerment. When looking at the argumentation that leaders’ trust plays a role, we encounter an issue. It would imply that low LMX followers who have leaders who trust them -hence show empowering behaviour- would experience higher psychological empowerment. Our study did not find that relationship; low LMX followers experience less psychological empowerment. However, previous research found a variable that influences the relationship between empowering leadership behaviour and psychological empowerment.
People differ in the degree to which they view themselves as someone who wants to be empowered (Zhang & Bartol, 2010). This is the so-called empowerment role identity. The association between empowering leadership behaviour and
psychological empowerment is smaller when people have low role identity compared to high identity. Thus it could have been that low LMX followers, have leaders who
trust them and show empowering behaviour. If they do not see themselves as
someone who wants to be empowered (low role identity) the empowering behaviour would result in lower psychological empowerment.
Future research regarding leaders’ self-efficacy, LMX and psychological empowerment should include other variables to see if empowering leadership behaviour does explain the relationship. Above, a couple of suggestions are given such as leaders’ trust, LMX from leaders’ perspective and role identity.
Limitations and strengths
This study has some limitations and strengths. A problem with the measurement of LMX and empowering leadership behaviour arose in this study; the correlation was .75. Which could mean that they measured the same construct. To investigate whether this was the case, a confirmatory factor analysis was conducted. Indeed all the items of both scales were significant and all loaded strongest on one of the components. This makes it likely that they measured the same construct. This was in particular a problem when looking at the relationship between LMX and empowering leadership behaviour. However, it seems unlikely that the other relationships with psychological empowerment were found because of the high correlation between LMX and
empowering leadership behaviour. This is because empowering leadership behaviour was excluded from those analyses. Future research should include other measures for LMX to see if those still correlate highly with the measurement for empowering leadership behaviour.
The cross-sectional nature of this research was a limitation as well, this makes causal reasoning impossible. A longitudinal field study could resolve this issue. Another limitation of cross-sectional research is the possible vulnerability to mono
method variance. However, Spector (1987) argued that method variance is a problem with poorly designed scales and less of a problem with well-designed, multi-item, validated scales. All of the scales used in this study were validated by previous research and their Cronbach’s Alfa were very high: ranging from .78 to .93.
Additionally data was gathered from multiple sources, which also reduced the method variance. Therefore the risk of mono method variance was reduced as much as
possible.
The fact that this study was conducted in the field could be seen as a disadvantage as well. This is due to the decreased control over external factors. Leaders contacted followers to participate in the study. Followers were informed about the anonymity of this research, but social desirable answers from followers cannot be ruled out. However, the chance that the particular results of this study are based on the social desirable answers is very unlikely. We found sufficient
differentiation within variables such as LMX, empowering leadership behaviour and psychological empowerment. If social desirability explained the associations, then we probably would not have detected the variations we did. In this field study there was no control over which of the followers we selected, actually filled in the
questionnaire; the leaders could have asked others. However again, it seems unlikely that the leaders selected only satisfied followers themselves, since there was sufficient variation in follower experienced psychological empowerment. The amount of
followers per leader that filled in the questionnaire was also a limitation. These differed between leaders; in some cases all three followers participated, in others only one follower per leader filled in the questionnaire. More followers per leader will make the data more reliable, thus a distorted view could have derived from this study’s limitation. Low amount of followers per leader also makes generalising more
