Effects of leader behavioural complexity on individual innovative ambidexterity : the mediating roles of organizational learning and LMX

(1)

Effects of Leader Behavioural Complexity on

Individual Innovative Ambidexterity:

The mediating roles of Organizational Learning and LMX

Master Thesis

Author : S.A.F. (Jelle) Arts Student number : 10311408

Date : 24-06-2015

Version : v1.0

(2)

2 Statement of Originality

This document is written by S.A.F. Arts 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.

S.A.F. (Jelle) Arts 10311408

(3)

3

Abstract

In this thesis, the influence of leader behavioral complexity (LBC) on individual innovative ambidexterity is analyzed using data from a single organization in the public sector of The

Netherlands. The data of the underlying study was retrieved by an online questionnaire taken by 234 respondents. This thesis rises to the call of Rosing, Frese, & Bausch (2011) to explore the construct of ambidextrous leadership to further enhance our comprehansion of the drivers behind innovative ambidexterity. A model, which includes the two mediator variables LMX and organizational learning, is constructed in order to grasp both the relational and the behavioral effect that leadership

establishes when striving for ambidexterity at an individual level. By performing multiple regression analyses and through a mediation test via bootstrapping method, the formulated hypotheses are put to the test. The results validate the proposed hypotheses and show that LBC shows great promise as an effective influencer of individual innovative ambidexterity. The results further indicate that LMX and organizational learning fully mediate this relationship.

(4)

4

1. Introduction

1.1. Research problem and research question

Within today’s highly dynamic business environments, many firms are struggling to gain and maintain a competitive advantage among competitors. Having a sustained competitive advantage implies that a firm possesses the ability to “implement a value creating strategy that is not simultaneously being implemented by any competitors and that these competitors are unable to duplicate the benefits of this strategy” (Barney J. , 1991). In light of the current speed of environmental changes, recent scholars have shown great interest within the theoretical potential of organizational ambidexterity: a firm’s ability to pursue the exploitation of current business processes while exploring new

opportunities for future demand at the same time (Birkinshaw & Gibson, 2004; Raisch, Birkinshaw, Probst, & Tuchman, 2009). While there is a consensus among scholars that ambidexterity plays a crucial role within today’s business environment (Gibson & Birkinshaw, 2004; Benner & Tushman, 2003), the path towards achieving it is still rather ambiguous (Gupta, Smith, & Shalley, 2006; Kortmann, 2012). Several scholars have investigated the effects of leadership upon firm

ambidexterity, since there seems to be an obvious theoretical connection between the two concepts (Jansen, Vera, & Crossan, 2009; Yukl, 2009). There are however two main limitations in current research surrounding this proposed connection. Firstly, ambidexterity is mostly explored at the firm-level (Benner & Tushman, 2002; He & Wong, 2004) or at the business unit-firm-level (Birkinshaw & Gibson, 2004; Jansen, van den Bosch, & Volberda, 2006), but there is a lack of research being conducted regarding ambidexterity at the individual level (Good & Michel, 2013). This seems rather strange since it has been previously indicated that the contrebution of individual actions upon achieving ambidexterity throughout an organization is quite significant (Gibson & Birkinshaw, 2004; Mom, Van den Bosch, & Volberda, 2007). Secondly, a meta-analysis conducted by Rosing et al. (2011) has shown that the research surrounding the effects of leadership upon ambidexterity has found to produce hetrogenetic results over time. Their main finding for the occurance of this phenomenon was the fact that prior scholars chose to seek connections between ambidexterity, which has a highly dynamic character, and a pre-defined leadership style which is, as a concept, statically defined by nature. Next to providing a meta-analysis, Rosing et al. (2011) also encourage the exploration of their formulated concept of ‘ambidextrous leadership’. In contrast to previously defined leadership styles, ambidextrous leadership does not emphasise on pre-defined approaches or behavior, but rather on the ability of a leader to be flexible in matching his/hers behavior towards the requirements of a sitiation and to the individual prospects of the innovation process. A similar phenomenon has been denoted by earlier scholars as ‘leader behavioral complexity’ (Denision, Hooijberg, & Quinn, 1995; Hart & Quinn, 1993; Hooijberg & Hunt, 1997), consisting of both behavioral repetoire and behavioral

(7)

7 differentiation. Although it has been generally accepted that leadership has a substantial direct influence upon follower behavior, in case of this behavior being innovative ambidexterity, it is likely this is established through creating a supportive environment for ambidexterity to flourish. The findings of Bass & Avolio (1993), that leaders who are striving for organizational renewal foster learning-supportive cultures in order to stimulate creativity, problem solving and experimentation, were later validated by Nemanich & Vera (2009). Hult, Ketchen, & Nichols (2003) also found that where leaders established organizational learning as an integrated aspect of the organizational culture, innovative behavior is likely to occur. Although organizational learning is likely to act as a catalist, it is unlikely that embedding this aspect within an organizational culture fully mediates the relationship between leader behavioural complexity and individual ambidexterity by itself. After all, there is a difference between being presented an opportunity and actually taking that opportunity. According to Rosing, Frese, & Bausch (2011) the occurance of Leader-member exchange (LMX) may act as a stimulus towards actually taking this opportunity, since the establishment of a high quality individual relationship between a leader and a follower may incline a follower to risk

experimentation. By adding an individual oriented derivative to the equation next to a more environmental oriented one, this study aims to make a contribution towards getting a better understanding of the relationship between leadership and ambidexterity on the individual level, by exploring the effects of leader behavioral complexity upon individual innovative ambidexterity, through the establishment of organizational learning and LMX. By doing so, this study rises to the call of Rosing, Frese, & Bausch (2011) to explore the construct of ambidextrous leadership in order to further enhance our comprehansion of the drivers behind individual ambidexterity. The main research question formulated for this study therefore is:

What are the effects of leader behavioral complexity upon individual innovative ambidexterity by means of establishing organizational learning and LMX?

1.2. Outline of this study

The next section will provide a review of the most relevant literature which also supports the hypotheses of this study (chapter 2). This is followed by a layout of the employed research method (chapter 3), and an overview of the results (chapter 4). Within chapter 5 the results will be discussed, and finally the limitations of this research are presented in chapter 6, followed by the conclusions of this research which will be drawn in chapter 7.

(8)

8

2. Theoretical foundation & hypotheses

Within this chapter, a theoretical foundation is presented regarding the proposed connection between leader behavioral complexity and innovative ambidexterity at the individual level. After an introduction to the topic of ambidexterity, the proposed relationship between the independent variable (leader behavioral complexity) and the dependent variable (individual innovative ambidexterity) will be discussed. The suggested mediating roles of organizational learning and leader-member exchange will be presented next. Matching hypotheses will be introduced throughout the literature review. This chapter finalizes by giving an overview of the hypotheses within a structural model.

2.1. Ambidexterity as a dynamic capability

Influential variables of organizational performance and effectiveness have always been a topic of interest in the literature on management and organizational theory (Yukl, 2008). An observable trend within literature on these topics, is that organizations operate in increasingly dynamic environments as time progresses. The phenomenon of an increasing change-rate which leads to a higher degree of environmental instability is also known as environmental dynamism (Yang & Li, 2011), which is typified by shortened product life cycles, continuous and rapid changing technology, and increasing competition (Volberda, 1996). The dynamic capabilities view, which is based upon Barney’s resource based view (Barney, 1991), suggests that in order to cope with these environmental changes, and in order to gain competitive advantage, organizations must be able to integrate, build and redesign internal and external competences at the same rapid pace (Teece, 2007, Eisenhardt & Martin, 2000). As organizations strive towards adaptation within their environment it is crucial to explore new ideas and business processes for future use, while at the same time exploiting current producs and services needed to leverage current business practices (Birkinshaw & Gibson, 2004; Raisch, Birkinshaw, Probst, & Tuchman, 2009). This challenge of continuously being aligned and efficient in managing today’s business demands while also being adaptive enough to make changes towards the

environment of tomorrow, is known as organizational ambidexterity (Gibson & Birkinshaw, 2004). 2.2. Types of organizational ambidexterity

Organizational ambidexterity can emerge in different ways and at different levels throughout an organization. While most ambidexterity research focuses at the organizational level, it remains a multilevel phenomenon which is essentially founded by individuals within the organization as they manage the dynamic context (Good & Michel, 2013). The concept of ambidexterity implies there is a constant trade-off to be made between exploration and exploitation or adaptability and alignment (Gibson & Birkinshaw, 2004, Simsek, 2009, Kortmann, 2012). Since there are several ways of managing this trade-off, several types of ambidexterity can emerge throughout the layers of an

(9)

9 organization. Kortmann (2012), acknowledges four types of ambidexterity which I’ll now briefly describe in order to set clear definitions throughout this study. The first type is structural

ambidexterity, which handles the trade-off between exploration and exploitation by separating these activities among two separate organizational units (Jansen, Tempelaar, van den Bosch, & Volberda, 2009). While technically this ‘solution’ meets the criteria of an organization persuing both paradoxal activities, some scholars argue that true organizational ambidexterity can only be achieved when both activities are combined within a single organizational unit (Raisch S. , Birkinshaw, Probst, & Tuchman, 2009). A second type of ambidexterity which also uses the concept of separation to ensure both activities are conducted is sequential ambidexterity. Unlike structural ambidexterity,

exploration and exploitation are not carried out at the same time, but in a sequential order (Brown & Eisenhardt, 1997). Where it is theoretically possible that sequential ambidexterity occurs at an individual level, it is more likely to occur at the bussiness-unit or firm level. The third ambidexterity type, is contextual ambidexterity, which refers to the capability of an organization to simultaniously align and adapt business processes within an organizational unit (Gibson & Birkinshaw, 2004). The difference between contextual ambidexterity and structural or sequential ambidexterity, lays within its nature to pursue both paradoxical agendas at the same time and within a single organizational unit (Kortmann, 2012). The occurrence of alignment implies that more than one entity of the organizational unit configures its activities so they are ‘aligned’ with each other. It can therefore be concluded that this type of ambidexterity also manifests itself at the organizational level. It likely occurs when senior managers of an organization implement a firm-wide supportive context towards applying aspects of alignment and adaptability (hence the contextual nature) which in turn enables individual ambidextrous behavior among business units and individuals.

2.2.1. Innovative ambidexterity

The fourth type of ambidexterity, which will also have the main focus of this study, is innovative ambidexterity. According to Simsek (2009), innovative ambidexterity Is defined as an organization’s actual exploration and exploitation performance, and Tuchman & O'Reilly (1996) describe innovative ambidexterity as “the ability to simultaneously pursue both incremental [exploitative] and

discontinuous [exploratory] innovation”. This type of ambidexterity does not depend on a separation mechanism perse, and it can occur throughout all levels of an organization. At first glance, innovative ambidexterity has quite some simularities compared to contextual ambidexterity. The main

difference between these two types lies within the level where a trade-off is being conducted: Where alignment & adaptability trade-offs are made at an organizational unit level, exploitative &

exploratory trade-offs can occur on all levels of an organization (including directly on the individual level). As a higher order capability, contextual ambidexterity functions as a facilitating capability for

(10)

10 other subordinate capabilities to emerge, such as innovative ambidexterity (Kortmann, 2012). In other words: senior managment of an organization can set a strategy in motion which creates opportunity (context) for innovative ambidexterity to emerge. Where the research of Kortmann (2012) puts focus on the relationship between contextual factors which can be implemented within an organizion in order to stimulate innovative ambidexterity, this study focusses on how supervisor behavior directly influences individual innovative ambidexterity (regardless of a managerial context).

2.3. Individual innovative ambidexterity & leadership

Jansen et al.(2009) argue that a basic need for organizational ambidexterity to occur, is that every individual needs the ability to constantly judge how to properly divide time between paradoxal demands. This ability is not likely to be solely evoked by creating proper contextual guidance at a strategic level. Based on earlier research, direct leadership also plays an important role when managing professional employees (Lee, 2008). The link between leadership and innovation also isn’t new as a concept and keeps gaining attention in current literature (Mumford et al., 2002, Rosing, et al, 2011). Some scholars even argue it’s the most important predictor of innovation and

organizational development in general (Mumford et al, 2002, Nemanich & Vera, 2009, Yukl, 2009). Leaders have a significant impact on how followers go about achieving their goals by influencing them on a daily basis (Witt & Boerkrem, 1989, Redmond et al., 1993), and according to Mumford et al. (2002) motivation and intellectual stimulation by leaders have a positive effect upon the

innovativeness of a follower. Although Rosing et al. (2011) state that the occurance of innovation implies by default that ambidexterity has to be present due to the cyclic nature of eploration and exploitation within innovation, most scholars do not explicitly point out if by ‘innovation’ they also refer to the occurrence of ambidexterity (Lee, 2008, Gumusluoglu & Ilsev, 2009, Oke et al., 2009) and if so, which type of organizational ambidexterity is their topic of interest. The majority of studies focusses on innovative context being created by leadership activities at a strategic level (Rosing et al., 2011), which implies that predictors of contextual ambidexterity are being researched. In order for true organizational ambidexterity to emerge throughout an organization however, the presence of innovative ambidexterity on the individual level is very important, if not the most important aspect according to Kortmann (2012). When interpreting prior findings and assumptions, it seems likely that there is a connection between leadership and the origin of innovative ambidexterity at an individual level. The main challenge which has presented itself regarding this proposed linkage, is that the connection seems to significantly vary in strength and even in direction throughout the course of time (Rosing et al., 2011).

(11)

11 2.3.1. The need for dynamic leadership

The meta-analysis conducted by Rosing et al. (2011) shows that several different types of leadership are related to innovation, but the results show inconsistencies and contradictions over time.

According to Ancona et al. (2001), leadership needs to “match the complexity and the pace of innovation” in order to be effective. This implies that a single ‘staticaly’ defined leadership style is rarely fully synchronized with the innovation process, which would explain the hetrogenic results produced over time. Rosing et al. (2011) therefore propose that the style of a leader should

“specifically match the requierments that teams and individuals face within the innovation process”. Since innovation requires ambidexterious behavior from individuals, a leader should be able to flexibily stimulate both exploration and exploitation at the the opportune moment. The concept of a similar flexible leadership approach was presented earlier by Hooijberg & Quinn (1992), which is known as leader behavioral complexity. A behaviorally complex leader is someone who has the ability to “act within multiple roles and is able to apply behavior that circumscribe the requisite variety implied by an organizational or environmental context” (Denision, Hooijberg, & Quinn, 1995). Because leader behavioral complexity is a currently established phenomenon, and because it has many similar traits to the proposed ‘ambidextrous leaderhip concept’ of Rosing et al., this study will utilize leader behavioral complexity as a means to indicatively test their theory of flexible leadership being better suited to induce individual ambidexterity.

2.3.2. Leader behavioral complexity

The concept of behavioral complexity originates from the idea that a leader has to manage a network of varying relationships, which includes superiors, peers, subbordinates and external partners

(Hooijberg, 1992). As the size and differntiation of the network become more complex and the relationships within the network tend to shift, the ability to match his/hers behavioral repertoire to the needs of a specific situation and individual become more important to a manager (Hooijberg, 1996). There are two main components of behavioral complexity which can be distinguished when interpreting this description. Behavioral complexity calls for the presence of a varyity of leadership functions in order to adequatly engage the organizational or environmental context. Hooijberg (1996) has therefore defined the first component of behavioral complexity as behavioral repertoire. Behavioral repertoire finds its origins within the Competing Value Framework (CVF) which was presented by Quinn & Rohrbaugh in 1981. This framework consists of two competing dimensions; flexible versus stable structure, and internal versus external focus, and counts four separate quadrants. Each quadrant is distinct from the others but retains a specific spatial relationship with the quadrants sharing the same dimensions (Denision, Hooijberg, & Quinn, 1995). Cameron, Quinn, DeGraff, & Thakor (2006) have recently adjusted the labels of the four quadrants within the

(12)

12 framework so that they are well suited and applicable from the organizational level to the individual behavior level (compete, control, collaborate, and create). A clear overview of the relation between dimensions and quadrants is presented within figure1. Because the CVF is a framework that is specifically defined in terms of opposing roles, and asserts that leader effectiveness requires meeting and integrating the competing roles (Zaccaro, 2001), it has been applied within earlier studies as a means to measure behavioral repertoire (Denision, Hooijberg, & Quinn, 1995; Hooijberg, 1996). Within this study, behavioral repertoire will be measured according to the most recently adjusted CVF model by Cameron, Quinn, DeGraff & Thakor (2006).

Figure 1: The dimensions and quadrants of the CVF

According to Lawrence, Lenk, & Quinn (2009) “Good judgment is an important aspect of effective leadership”. It is therefore only logical that merely possessing a substantial behavioral repertoire by itself is not going to be effective when a leader does not know how and when to apply the elements of his/hers repertoire to a specific situation (Hooijberg, 1996). Behavioral differentiation is therefore defined as a second component of behavioral complexity, and refers to “the extend to which a manager varies the performance of leadership functions depending on the demands of the organizational situation” (Hooijberg, 1996). Only a combination of both behavioral repertoire and behavioral differentiation can account for effective behavioral complexity within a leader. Now that I’ve established a clear theoretical connection between individual innovative ambidexterity and leader behavioral complexity as a dynamic leadership concept, I would hypothesize that:

H1: Leader behavioral complexity has a significant positive effect on individual innovative ambidexterity.

(13)

13 2.4. Leader behavioral complexity: creating optimal context for ambidexterity

Leadership has found to have an effect upon followers on both a behavioral level, as well as on a relational level (Tierney, Farmer, & Graen, 1999; Lee, 2008). The effect of leadership on the

behavioral level, involves manipulating traits and behavior of followers that are closely linked to the individual, group or organizational outcome. Charismatic and transformational leaders for example, tend to manipulate their followers’ behavior by the way in which they generally manage their employees and by creating supportive context for this desireable behavior to occur (Bass, 1985; House & Shamir, 1993). The effect of leadership on the relational level is found within the dyad relationship between leader and follower (Lee, 2008). Graen & Cashman (1975) were the first to explore this relationship in depth, and subsequently founded their theory of leader-member exchange (LMX). High quality leader-member exchange has proven to have a positive effect on innovative behavior (Basu and Green, 1997; Rosing et al., 2011; Scott & Bruce, 1994).

Given the dual nature of the effects that leadership has according to Tierney, Farmer, & Graen (1999), it would be logical that both a positive behavioral and a relational derivative are being envoked when trying to influence innovative behavior. The following paragraphs of this section will look further into the suggested behavioral and relational fundaments, which are suggested to be established by leader behavioral complexity when striving towards individual innovative

ambidexterity among followers.

2.4.1. The behavioral establishment of leadership: organizational learning

Opposing the statement of March (1991) that explorative and exploitative learning are competing with eachother for scarce resources, I share the view of Tushman & O’Reilly (1996) which states that simoultaniously persuing both incremental and discontinuous learning can coexist and reinforce one another. March (1991) was however, one of the first scholars who pointed out the importance of persuing both explorative and exploitative learning when striving towards strategic renewal. Based upon this dichotomy, Crossan et al. (1999) developed a framework of arganizational learning. Their framework associates exploration with a learning process supporting novation: through intuition and interpretation, individuals develop new ideas and integrate these ideas into their teams. The next step would be to incorporate this knowledge throughout the firm. Exploitation on the other hand, supports continuity and is associated with a process based upon feedback: the firm captures best practices which are then transferred to individuals and teams (Crossan et al., 1999). The contexts in which people share existing knowledge and create new knowledge, is established by interpersonal relationships among team members when striving towards a common goal (Dutton & Heaphy, 2003). As also pointed out by Gherardi, Nicolini, & Odella (1998), we can conclude that learning therefore is

(14)

14 a social process of interaction and participation which occurs within groups of people. Multiple scholars have established that the role of a leader also plays an important part within the social process when striving towards organizational learning (Shin & Zhou, 2003; Vera & Crossan, 2004). In addition, Vera & Crossan (2004) specifically point out that it takes multiple leadership roles to succesfully stimulate the arisal of organizational learning. Although the CVF quadrants which form the basic outline for leader behavioral repertoire oppose one another in terms of focus and structure (Denision, Hooijberg, & Quinn, 1995), all the underlying constructs of each of the quadrants contain elements of organizational learning (Lawrence, Lenk, & Quinn, 2009). A clear bifurcation of

explorative and exploitative learning amongst the quadrants cannot (and must not) be made, since the CVF is a circumplex model and each of the quadrants has related traits to its adjectent. In particular, the quadrants that are present in the flexible dimension of the CVF (creation &

collaboration) stimulate a work-environment, which contributes towards development of (future) business practices. The collaboration quadrant encourages participation of employees within the innovation process, and the creation quadrant focuses on the overall establishment a psychologically safe environment where employees are appreciated for their initiative and ideas. The stable

dimension of the CVF provides a balance by focusing on current operations. The control quadrant is meant to establish ground rules and clarity for employees, and the competition quadrant puts focus on gaining a competitive edge within today’s business environment (Lawrence, Lenk, & Quinn, 2009). The presence of both explorative and exploitative learning-stimulants is clearly present, and I would therefore hypothesize that:

H2: Leader behavioral complexity has a significant positive effect on organizational learning.

The presence of an environment where employees are stimulated to learn new and improved practices, which subsequently are accepted and implemented, has been found to be a major catalyst for innovation (Bass & Avolio, 1993; Schein, 1997). Within this study, I build upon the findings of Nemanich & Vera (2009) by suggesting that this ‘enabling environment’ is founded through the concept of organizational learning. Rather than solely by a dyadic relationship between leader and follower, I would suggest that organizational learning complements the relation between leader behavioral complexity and individual innovative ambidexterity by underlining the cultural aspects of learning as a process of participation and interaction among employees. I would therefore

hypothesize that:

(15)

15 2.4.2. The relational establishment of leadership: leader-member exchange

As I have stated earlier, creating an environmental context is not the only important aspect when it comes to fostering ambidexterity. The dyadic relationship between leader and follower has also proven to play an important role in the fostering of innovative behavior among employees (Basu & Green, 1997; Scott & Bruce, 1994). Graen and Scandura (1987) state the reason for high quality dyadic relationships to debouch into innovative behavior among employees being that, in these cases, leaders are in inclined to give their followers a higher level of discretion and autonomy. It is also interesting that Mumford et al. (2002) found that in the opposite case of a low quality dyadic relationship being present, innovation was not likely to occur due to followers feeling too pressured within their work-environment. Individual employee consideration increases employees’ self-esteem, and by encouraging employees to express their individual needs they tend to show more initiative (Nemanich & Vera, 2009). The dyadic relionship between leader and follower has been a prominent topic within leadership research ever since Graen and Cashman (1975) presented their theory of leader-member exchange (LMX). In general, Leader-Member Exchange theory suggests that a leader develops a unique relationship or exchange with each follower, rather than applying the same ‘style’ to all their followers (Dansereau et al., 1975; Graen and Scandura, 1987).

Denison et al. (1995) established that when leaders are able to enact a broad scala of roles, they are better at handling (conflict) situations. This implies that a broad behavioral repertoire envokes empathy within a leader, which in turn leads to the establishment of a higher quality relationship with the individual in question (Wu, Steward, & Hartley, 2010). According to Hooijberg (1996) however, merely posessing a behavioral repertoire is not sufficient. In order to establish an

appropriate relationship, one must be able to effectively switch between different roles in order to handle paradoxes and contradictions. The updated defenition of behavioral complexity (LBC) by Hooijberg (1996), containing both behavioral repertoire and behavioral differentiation, therefore shows great potential towards positively stimulating leader-member exchange. I would therefore hypothesize that:

H4: Leader behavioral complexity has a significant positive effect on leader-member exchange.

In their meta-analysis, Rosing et al. (2011) confirm that LMX has been found to be consequently related to aspects of innovation in a positive way. Nemanich & Vera (2009) suggest that this connection is present due to the high quality dyadic relationships between leader and employees, which creates an environment where the employee feels safe to engage in innovative behavior. A learning effect is enhanced by this sort of environment, because employees tend to be more open

(16)

16 and likely to communicate with their colleagues (and their leader) due to the established

psychological safety (Schein, 1997). As an increase in communication leads to better integration among teammembers, this wil stimulate both explorative and exploitative learning. Individuals are likely to experience that their contribution is valued by their leader (and by other teammembers), which increases perceived support for exploration. The fact that communication between leader and follower (or amongst followers) is enhanced by LMX increases recombination of knowledge, and therefore exploitation. Since LMX tends to influence both explorative and exploitative innovation, I hypothesize that:

H5: Leader-member exchange has a significant positive effect on individual innovative ambidexterity.

2.5. The mediating role of organizational learning and LMX

Althoug many scolars have contributed to scientific literature surrounding the topics of leadership, innovation and the indirect variables which are included in such cases (Rosing et al., 2011), few have specifically taken the dual effect that leadership envokes into account (Tierney et al., 1999; Lee, 2008). By doing so, it seems only logical that a substantial part of the relationship between the analyzed variables is not accounted for. By taking both relational and behavioral effects of leadership into account, this study aims to give a more holistic view on this matter.

As proposed in the previous paragraph, leader behavioral complexity will influence behavioral traits within followers that lead to the establisment of organizational learning. On the relational spectrum of the leadership effect, the appopriate application of the leaders’ behavioral repertoire, known as behavioral differentiation, is suggested to positively influence the dyad relationship between leader and follower (LMX). It must be noted that both of these proposed effects are assumed to occur when a leader actually intends to let his followers engage individual innovative ambidexterity, since the concept of LBC is founded on the principle that a leader adjusts his/hers behavior according to the required organizational outcome. If the assumption of Tierney et al. (1999) is correct, and the combination of the behavioral and the relational effect of leadership accounts for its total effect, it would be likely that the combination of both these effects funtion als a mediator between LBC and individual innovative ambidexterity. I would therefore hypothesize that:

H6: The combination of organizational learning and leader-member exchange fully mediates the effect of leader behavior complexity on individual innovative ambidexterity.

(17)

2.6. Structural Model

Figure 2: Structural model with hypotheses

Team Orientation Systems Orientation Learning Orientation Memory Orientation H3 H2 H5 H4 H1 / H6 Leader Behavioral Complexity Behavioral Repertoire Behavioral Differentiation Individual Innovative Ambidexterity Exploration Exploitation Leader-Member Exchange Organizational Learning

(18)

3. Research methodology

This chapter gives an overview of the applied research methodology of this study. It provides an overview of the sample, the data collection process, and which individual measures have been applied to validate the hypotheses.

3.1. Sample

This study is based on a highly diverse sample of respondents within a single organization of the public sector in The Netherlands. The diversity of respondents is a desirable factor within this research, because this is best suited to reflect the diversity of human relations, culture and

conditions (Allmark, 2004). Since the unit of analysis is at the individual level, this will help secure the validity of the results and conclusions. A questionnaire was created which covered all the constructs that are displayed in the structural model. The actual content which was used shall be presented in full at the individual measurement paragraph of this chapter. From the last week of February 2015 until the third of April 2015, the questionnaire was send throughout the organization by senior managers, and reached approximately 2.000 respondents. I was unable to determine the exact number of individual invitations, because the questionnaire was also send out and promoted by my colleagues through their personal networks (all within the same organization). The senior managers broadcasted a reminder after three weeks in order to enhance the response rate. This finally resulted in 395 participants, giving a response rate of 19.75%. Unfortunately, 161 questionnaires turned out to be incomplete. Since the setup of the study did not allow participants to leave out any questions, these results were discarded from the sample. The final sample consists of 234 respondents, leading to a response rate of 59.24% based on 395 participants.

3.2. Non-response bias

In order to determine if the collected sample is relevant to the study, it is important to interpret the ‘quality’ of the respondents. According to Kortmann (2012), this is achieved by checking for a non-response bias and by performing a key informant check. According to Dillman (2000), a non-non-response bias occurs due to the fact that respodents typically differ from individuals who do not respond to a questionaire. Since the latter group is impossible to define, Armstrong & Overton (1977) determined that late respodents are assumed to be similar to respondents, which makes testing for non-response bias possible. I’ve addopted the approach by Li & Calantone (1998) who labeled the first 75% of respondents as ‘early’ and the remaning 25% as ‘late’ respondents. In order to check for a non-response bias, a Mann-Whitney U test was performed. Table 1 presents the results of the conducted Mann-Whitney U test for all constructs used in the conceptual model of this research.

(19)

19 Table 1: Results of the Mann-Whitney U test

The results show that there is no significant difference between the early and late respondents of the sample. I can therefore conclude that non-response bias is not influencing this research.

3.3. Key informant check

As mentioned earlier, a second check to see if the sample is appropriate and usable for this study is the key informant check. Normally, the term ‘key informant’ refers to an individual who is

particularly qualified to answer questions regarding the topic of investigation (Kumar, Stern, & Anderson, 1993). Since the unit of analysis of this study is at the individual level, and because the study is being conducted within a single organization, in order for the results and conclusions to be generalizable it would be prefereable in this case if the sample consists of a diverse population. In contrast to performing a regular key informant check, I’ve therefore conducted a ‘generalization check’ of the sample in order to ensure the data is usable. In order to do so, the management level and the age of the respondents were reviewed. The respondents’ age were grouped in periods of ten years, and the management level of the individual employees was determined by categorization of their pay grade level. Table 2 contains an overview of the respondents’ management level and age within the sample. Based on the results, I can conclude that there is no reason to assume that the population is homogenitic, and therefore the sample is appropriate for this study.

(20)

20 Table 2: Age and Management Level of the respondents

3.4. Descriptive statistics

In order to complete the general view of the dataset, table 3 was conducted, containing the

descriptive statistics. As concluded in the previous paragraph, the distribution of management level seems to be showing no abnormal anomalies within the dataset. With an average of 3.98 and a std. deviation of 3.94, the years a respondent has been on his current job stands out within this overview, especially compared to the maximum of 31 years. The explanation behind these numbers lies within the human resource policy of the organization in question. Depending on individual employment agreements, an employee either chooses for a career path containing job switches every three to five years, or he/she chooses for a more stable career which does not compel the employee to switch jobs. The majority of the employees within this organization falls under the first category, which explains the shown statistics. Team size was measured categorically, and ranges from teams consisting of 1-5 members to 25-30 members, with an average of 11-15.

(21)

21 3.5. Common method bias

A potential source of common method variance lies within the social desirability of respondents (Kortmann, 2012). This particular study might be susceptible to this variance, because respondents were asked to answer questions regarding the skills of their superiors. In order to decrease the chance of this variance to occur, respondents were assured anonymity in the questionnaire’s invitation and once more in the introduction. Second, the items of the dependent and independent variables were positioned at random throughout the questionnaire in order to make it more difficult for respondents to link items. According to Kortmann (2012) this will help prevent respondents from giving socially desirable answers. In order to ensure no common method variance (CMV) occurred, the Harman’s single-factor test was conducted after collection of the data. According to Chang (2010), the Harman’s single-factor test is the most commonly used test by scholars to assess wether or not CMV has occurred. “This method loads all items from each of the constructs into an

exploratory factor analysis to see whether one single factor does emerge or whether one general factor does account for a majority of the covariance between the measures; if not, the claim is that CMV is not a pervasive issue”. The Harman’s single-factor test resulted in one factor accounting for 35% of the total variance. Since the results indicate that this factor does not account for the majority of covariance between measures, I conclude that there is no common method bias present.

3.6. Individual measures

In this paragraph the constructs and items of the used variables in this study are introduced. All used items were derived from validated instruments, and were slightly adapted to fit the circumstances of the research setting. Since all respondents are from The Netherlands, the items were translated into Dutch via a double verified translation. All English items are presented in appendix 1.

3.6.1. Leader behavioral complexity

As explained in the literature review, behavioral complexity is a second order construct formed out of behavioral repertoire and behavioral differentiation. Behavioral repertoire was measured by a total of 36 items which were adopted from Lawrence, Lenk, & Quinn (2009). Each of the four repertoire quadrants was measured by 9 items. The score for leader behavioral repertoire was measured by calculating the mean of the separate quadrant scores. Behavioral differentiation was measured using a 5 item scale introduced by Wu, Steward, & Hartley (2010). Finally, the score for leader behavioral complexity was computed by calculating the mean score of behavioral repertoire and behavioral differentiation, since both weigh equally. All scales were found reliable according to their Chronbach’s Alpha (see table 4).

(22)

22 3.6.2. Individual innovative ambidexterity

Following Nemanich & Vera (2009), I adopted a two step approach to develop a measure for individual innovative ambidexterity. First, respectively three and four items of the questionnaire separately measured both exploration and exploitation. Although the separate scales for exploration and exploitation had proven to be reliable when used in the research of Nemanich & Vera (2009), the Chronbach’s Alpha scores turned out to be lower than expected in this study. The scale containing items for exploration scored .54 and the exploitation scale scored .25 (see table 4). Although these reliability scores are generally not acceptable, it can be argued that the Chronbach’s Alpha value was negatively biased by the small amount of items within the scales (de Vaus, 2002; Bacon, 2004). Combined with the fact that both scales have previously been positively validated, I decided to proceed with the second step of the development process for an ambidexterity measure and calculated the multiplicative interaction between exploration and exploitation (labeled AM01 – AM12). The computed individual innovative ambidexterity scale which was later used in the multiple regression analysis, turned out to be reliable with a Chronbach’s Alpha of 0.90 (see table 4).

3.6.3. Leader-member exchange

Leader-member exchange was measured by the 11-item scale developed by Liden & Maslyn (1998). Respondents were asked to indicate on a five point likert scale wether or not they believed the statements regarding the relationship with their leader were valid. The scale proved to be reliable with a Chronbach’s Alpha of .90 (see table 4).

3.6.4. Organizational learning

Organizational learning was measured by a 17-item scale which was developed by Hult, Ketchen, & Nichols (2003). This instrument was designed to measure the presence of the four aspects of organizational learning within a team environement. These aspects are: team orientation, systems orientation, learning orientation and memory orientation. Since organizational learning is a second order construct, reliablity checks were performed on each of the individual constructs before computing a mean score for organizational learning. All scales turned out to be reliable with Chronbach’s Alpha scores > 0.70 (see table 4). One item in the systems orientation construct was dropped in order to improve reliabillity.

(23)

23 Table 4: Descriptive statistics of the used constructs

3.7. Control variables

Three control variables were introduced in order to control for confounding effects. Firstly, team size was included because the span of control of a leader tends to have an effect upon leadership (Doran, et al., 2004). Doran et al. (2004) found that when the span of control of a manager (and therefore team size) grew bigger, a decrease in employee statisfaction occurred. This implies that the relational effect of leaderhsip may be affected as team size increases. Secondly, the time an employee has been at his/hers current job was introduced as a control variable. It seems only likely that in order for leadership to have an effect upon follower behavior, time is a crucial factor. Thirdly, the

management level of the employee was taken into account as a control variable when testing the hypotheses. It was included as a method to test for the effect of contextual ambidexterity. Kortmann (2012) established in his research that innovative ambidexterity is influenced significantly by

contextual ambidexterity. Since executive managers at the top of an organization generally put contextual ambidexterity in place, the effect upon the capability of a leader to utilize it will be strongest within the upper management layers, and is likely to decrease in the lower management levels. Secondly, it is likely that as the management level of an employee increases, this individual encounters less negative effects inflicted by corporate governance. Table 3 presents an overview of the descriptive statistics of the control variables.

(24)

24 3.8. Analysis strategy

In order to test the proposed hypotheses one up to five, the collected data was analyzed using SPSS. The first part of the analysis consisted of performing multiple regression analyses between the independent, dependent and proposed mediator variables. In order to test hypothesis six, a multiple mediation test via bootstrapping procedure by Preacher and Hayes (2008) was applied to the model. They state that their approach has advantages compared to other methods because the effect of each of the mediator variables on the relationship between leader behavioral complexity and individual innovative ambidexterity is determined as part of the complete model. Within SPSS I used a macro called INDIRECT, which was designed by Preacher and Hayes1_{, to perform this analysis. The} designed test by Preacher and Hayes contains several steps in order to check for mediation. First, the relationships between the independent variable, leader behavioral complexity, and the mediator variables, LMX & organizational learning, were measured (a paths). Second, the relationships between the mediator variables and the dependent variable, individual innovative ambidexterity, were calculated (b paths). Third, the total effect of leader behavioral complexity upon individual innovative ambidexterity was computed (c path). In order to finally test if mediation occurs within the model, the direct effect of leader behavioral complexity upon individual innovative ambidexterity was measured once more, but this time with the proposed mediator variables present in the

equation (c’ path). If the a, b and c path all show significant relationships, the delta which is presented by calculating the c’ path points out if and whether mediation has occurred within the model. If the c’ path presents a significant but decreased effect, partial mediation occurs by including the mediator variables into the model. If the c’ path presents an insignificant result, full mediation has occurred. Figure 3 gives an overview of the different paths which have just been described.

Figure 3: Paths mediation analysis

(25)

25 Although the output from the c’ path implies whether or not (partial) mediation occurs, within a multiple mediation model it does not indicate which of the proposed mediator variables causes the mediation effect. In order to assess this, the confidence intervals (CI) of the bootstrapping sequence were analyzed. If the CI of a variable contains the value 0 within its range, the variable in question is not accountable for the mediation effect.

4. Results

This chapter provides an overview of the results, which have been conducted in SPSS. Based on the performed analyses, the suggested hypotheses were either accepted or rejected. This section will first elaborate on the correlations which were found between the subjected variables. The second paragraph will cover the actual testing of the hypotheses and finally, the effects of the control variables are presented.

4.1. Correlations

The bivariate correlations between all subjected variables were computed in order to measure the strength of the relationship between the variables. Spearman rho was used because it is often better suited when ordinal measures have been applied (Pallant, 2010). Table 5 gives an overview of the found bivariate correlations. The dependent, independent and mediator variables all seem to be significantly and strongly correlated to eachother. The control variables do not significantly correlate with the employed constructs of this study, nor with eachother. The exeption is the slight but significant correllation between management level and the time on current job (r = -.132; p < .005).

(26)

26 4.2. Hypotheses results

Hypothesis 1 proposes that leader behavioral complexity has a significant positive effect on individual innovative ambidexterity. The multiple regression analysis between the independent variable leader behavioral complexity and the dependent variable individual innovative ambidexterity confirms this hypothesis (ß = 0.54; ρ < .001). This means that, without mediator variables present in the model, leader behavioral complexity explains 54% of the variance in individual innovative ambidexterity.

Hypothesis 2 proposes that leader behavioral complexity has a significant positive effect on organizational learning. After conducting a multiple regression analysis between the independent variable leader behavioral complexity and the dependent variable organizational learning, the results state that this hypothesis is to be accepted (ß = 0.66; ρ < .001).

Hypothesis 3 proposes that organizational learning has a significant positive effect on individual innovative ambidexterity. A multiple regression analysis between organizational learning as independent variable and individual innovative ambidexterity as dependent variable confirms this hypothesis (ß = 0.61; ρ < .001).

Hypothesis 4 proposes that leader behavioral complexity has a significant positive effect on leader-member exchange. The results of the multiple regression analysis between these two variables confirm this hypothesis (ß = 0.80; ρ < .001). A firm 80% of total variance in LMX seems to be explained by leader behavioral complexity.

Hypothesis 5 proposes that leader-member exchange has a significant positive effect on individual innovative ambidexterity. Multiple regression analysis between LMX and individual innovative ambidexterity shows that this hypothesis is also to be accepted (ß = 0.57; ρ < .001).

Up until this point, the conducted analyses and the produced results have shown that there are significant relationships between all the individual variables of this study. In order to ‘grasp the coherence’ of all these individual relationships and to put my theoretical foundation and the structural model (figure 2) to the test, the final analysis which was conducted aims to combine all variables within one model. The results were produced using the INDIRECT macro from Preacher & Hayes, as described in paragraph 3.8 of this thesis. In order to ensure standardized coefficients can be reported, all variables which were included in this model were first standardized (the INDIRECT macro does not provide standardized coefficients). Leader behavioral complexity was entered into

(27)

27 the model as the independent variable, and individual innovative ambidexterity was stated as the dependent variable. Both organizational learning and LMX were included simultaneously as mediator variables in the model, in order to test for a combined mediation effect. The SPSS output of the INDIRECT macro has been added to this thesis and can be found in appendix 2.

First, in compliance with hypothesis 1, the results indicate that leader behavioral complexity

positively influences individual innovative ambidexterity (ß = 0.54; t (232) = 9.86; ρ < .001). Second, in compliance with the results from hypotheses 2 and 4, it was found that leader behavioral complexity positively influences LMX (ß = 0.80; t (232) = 20.60; ρ < .001) and organizational learning (ß = 0.66; t (232) = 13.44; ρ < .001). Finally, also in compliance with the results of hypotheses 3 and 5, it was found that individual innovative ambidexterity is positively influenced by both LMX (ß = 0.31; t (232) = 3.72; ρ < .001) and organizational learning (ß = 0.43; t (232) = 6.46; ρ < .001). Because the a, b and c path were all significant, a mediation test by means of the bootstrapping method with bias corrected confidence estimates was analyzed. A 95% confidence interval of the direct effects was obtained with 1000 bootstrap resamples (Preacher & Hayes, 2008). The results indicate that both LMX (ß = 0.25; CI = .12 to .38) and organizational learning (ß = 0.28; CI = .20 to .40) have a mediating role in the relationship between leader behavioral complexity and individual innovative ambidexterity. In addition, the c’ path shows that the direct effect of leader behavioral complexity on individual innovative ambidexterity became insignificant when controlling for LMX and organizational learning (ß = 0.01; t (232) = 0.14; ρ = .89), which indicates that full mediation has occurred. Hypothesis 6, which proposed that the combination of organizational learning and leader-member exchange fully mediates the effect of leader behavior complexity on individual innovative ambidexterity, is

therefore to be accepted.

Figure 4: Results mediation analysis *P <. 001

(28)

28 4.3. Effects of the control variables

Although the computed correlation effects (table 5) indicate that none of the control variables within this study is likely to significantly explain part of the variance induced by the predictor variables onto the outcome variable, I’ve conducted hierarchical multiple regression analyses to verify this

assumption. Since full mediation occurred when testing the structural model as a whole (figure 4), I controlled for the influence of team size, management level and time on current job along the a path and the b path of the model.

The results (table 6) show that although management level does seem to significantly explain part of the variance between LMX and individual innovative ambidexterity, and the years an employee has been at his/hers job partially explains the variance between organizational learning and individual innovative ambidexterity, these influences are very modest.

The total variance explained by the connection between LMX and individual innovative ambidexterity was 35%; F (4,229) = 30.62; ρ = < .001, of which 3% was accounted for by the combination of all control variables. Management level was the only control variable which proved significant in this model (ß = -0.17; ρ < = .01). This implies that management level has a negative influence upon ambidexterity.

The total variance explained by the connection between organizational learning and individual innovative ambidexterity was 39%; F (4,229) = 36.48; ρ = < .001, of which 3% was accounted for by the combination of all control variables. Time on current job was the only control variable which proved significant in this model (ß = -0.11; ρ < = .05). This implies that as an employee gets more experienced in his/hers job, less ambidextrous behavior occurs.

(29)

29

5. Discussion & implications

This study attempts to explore the way in which innovative ambidexterity at an individual level is achieved through the presence of leadership aspects. It does so by evaluating the role of leader behavioral complexity (LBC) upon followers ambidextrous behavior through the establishment of LMX and organizational learning. In order to test these influences, six hypotheses were employed representing the positive influence of leader behavioral complexity on individual innovative

ambidexterity, the dualistic effect leader behavioral complexity has while tempting to do so, and the mediating effect of LMX and organizational learning that occurs in the process. The data and the results of this study are in line with the developed theory and confirm all hypotheses. The findings are further discussed in this chapter, followed by their theoretical and practical implications.

5.1. Discussion

First of all, the results of this study confirm the significance of the role which leader behavioral complexity plays in positively influencing the level of individual innovative ambidexterity. Although multiple scholars previously established that leadership in general tends to positively influence ambidexterity (Mumford et al., 2002, Rosing, et al, 2011), this was never explicitly confirmed in the case of LBC. By doing so, this research has also taken a first step into verifying the concept of

‘ambidextrous leadership’, which was theoretically founded by Rosing et al. (2011). It must be noted that the concept of ambidextrous leadership has similar traits to leader behavioral complexity, but the two do not necessarily imply exactly the same. Both concepts establish leadership as a flexible competence which is able to change its form depending on relational and environmental situations. In case of ambidextrous leadership these circumstantial influencers are all related to the innovation process, whereas leader behavioral complexity does not apply this specific focus perse. In regards to this research, this implies that the measurements of the LBC construct did not have a specific focus towards the innovation process as well. An element which must be taken into account when conducting future research. However, as a first attempt to test the theory of Rosing et al. (2011) leader behavioral complexity has shown great promice.

A second interesting phenomenon originating from the results, is the confirmed presence of the dualistic effect of leadership. Both the defined behavioral component (organizational learning) and the relational component (LMX) were significantly and quite strongly influenced by leader behavioral complexity, confirming the theory of Tierney, Farmer, & Graen (1999). Particularly, LMX appears to be strongly influenced by leader behavioral complexity (ß = 0.80; ρ < .001). This was expected due to the fact that the entire concept of behavioral complexity is based around the ability of a leader to engage the environment and his/hers relations in the most appropriate manner (Hooijberg & Quinn,

(30)

30 1992). This is highly compliant with general LMX principles, stating that leaders develop a unique relationship or exchange with each follower, rather than applying the same ‘style’ to all their followers (Dansereau et al., 1975; Graen and Scandura, 1987). Although not quite as strong, the effects of behavioral complexity on organizational learning are also clearly visible in the results (ß = 0.66; ρ < .001). This substantiates the statement made by Lawrence, Lenk, & Quinn (2009), saying the individual quadrants of behavioral repertoire each support aspects of learning. A possible

explanation for LBC being less of an influence upon organizational learning as compared to LMX lies within the fact that organizational learning is likely to also be influenced by other factors. Where LMX soley relies on the leader-follower dyad relationship, the formation of organizational learning can be influenced by company policy and by the top-management vision. Since the data which was used in this study originates from a large organization within the public sector, the presence of a HR policy is likely to be of influence.

The results of this study also show that both LMX and organizational learning have a significant positive influence on individual innovative ambidexterity, which is in line with the proposed

hypotheses. It is interesting to notice that organizational learning influences individual ambidexterity stronger (ß = 0.43; ρ < .001) than LMX (ß = 0.31; ρ < .001). Taken together with the earlier discussed findings, the results show that although LMX as a relational component is stronger influenced by LBC, the behavioral component organizational learning has a bigger impact on the dependent variable of this study: individual innovative ambidexterity. My suggestion is that this phenomenon occurs because, although both organizational learning and LMX establish an enabling environment for individual ambidexterity to flourish (Bass & Avolio, 1993; Schein, 1997; Nemanich & Vera, 2009), organizational learning is perceived more tangible as opposed to LMX in this regard. According to Nemanich & Vera (2009) the establishment of a learning culture does not depend solely on a leader-follower dyad relationship, but is often part of an organization’s management foundation. This implies that the effect of organizational learning upon ambidextrous behavior may not solely originate from a leader-member dyad relationship, but can also be influenced by stimulating behavioral traits of peer-followers since organizational learning is partially founded as a generic concept within the organization. It would therefore be an interesting topic for future research to study the reinforcing effects of peer behavior on individual ambidexterity within an environment that stimulates learning.

The most promising element of this study’s results is the occurrence of a full mediation effect between LBC and individual innovative ambidexterity through LMX and organizational learning. By being able to significantly and completely reduce the direct effect LBC has upon individual

(31)

31 ambidexterity, LMX and organizational learning prove to be appropriate variables within the

structural model. Since there were also significant direct effects between all the independent, dependent and mediator variables, the occurrence of full mediation reflects the presence of the proposed interdependencies between the variables within the model. The latter can be seen as an indicator that the structural model as a whole supports the underlying theoretical basis as presented in chapter two.

The effects which the control variables accounted for within this study, have shown results which were not predicted but can be explained. I suggested that the time an employee has been acting on his job would positively influence the relationship between LBC and LMX, due to the fact that a dyad relationship is more likely to establish over time. This proved not to be significant nor of relevance within the dataset. There was however a slight but significant negative effect caused on the relationship between organizational learning and ambidexterity as the years of experience of an employee on his job increase (ß = -0.11; ρ < = .05). This might be explained through the

establishment of working routines that an employee forms over time, which makes this individual less perceptive of innovative possibilities. Although many scholars have studied the effects of team tenure (Tyran & Gibson, 2008; van Knippenberg & Schippers, 2007) on innovation, it would be interesting for future research to further study the effects of individual job experience on the individual’s ability to innovate.

A second unexpected effect was found in the results, where the management level of an employee shows to have a slight but significant negative effect on the relation between LMX and individual innovative ambidexterity (ß = -0.17; ρ < = .01). Initially, I suggested that as the management level of an individual increases, this individual would profit more from the presence of contextual

ambidexterity due to less attenuation, and he/she would be less inconvenienced by corporate governance, which both enhance the ability to engage in ambidextrous behavior. A possible

explanation for the opposite effect occurring in the results lies in the daily pursuits of the individuals acting in higher management levels. Although it is likely that senior managers are indeed less inconvenienced by their own organization’s corporate governance, their daily pursuits are also more likely to include a higher amount of dialogue with the organization’s external partners. This in turn implies that the organization’s internal processes (including the dyad relationship between leader and follower) have less of an absolute impact on the individual’s ambidextrous capacity in his/hers work environment.

(32)

32 Team size, the final control variable which was included in this study, was suggested to have a

negative impact on a leader’s ability to form a dyad relationship with his/hers followers. The results show that there is no significant indication that suggests this assumption is valid. A possible

explanation may be that fellow team members also influence an individual’s image of a leader, which compensates for the diminished direct effect. Second, there is also the possibility that the decrease in employee job satisfaction, which occurs as a leader’s span of control enlarges (Doran et al., 2004), is just not related to the leader-member dyad relationship.

5.1. Implications for theory

A lot research being conducted surrounding the topic of ambidexterity highlights eather the contextual aspects that are of importance for ambidexterity to emerge within an organization

(Birkinshaw & Gibson, 2004; Jansen, van den Bosch & Volberda, 2006; Kortmann, 2012) or the effects that leadership has upon its establishment (Jung, Chow, & Wu, 2003; Gumusluoglu & Ilsev, 2009; Rosing et al, 2011). Both perspectives contribute important and complemental insights surrounding the topic, and it can be argued that when combined these theories provide a holistic view on the matter. This study contributes to the leadership perspective surrounding innovative ambidexterity by researching the effect that a leader has upon his/hers direct followers (at the individual level), which is an angle that thas not been addressed often in the current literature. Most studies surrounding ambidexterity and leadership tend to address the influence of senior management upon establishing contextual ambidexterity. At the moment, research surrounding individual ambidexterity has an arrear as opposed to knowledge about the influencors of ambidexterity at the firm or the business-unit level (Good & Michel, 2013). It can therefore be argued that this study, together with prior conducted research at other levels of analysis and perspectives, provides tangible insight at the way ambidexterity is fostered throughout the organization.

One of the main contributions this research makes, is rising to the call of Rosing et al. (2011) by exploring the concept of ambidextrous leadership. Although numerous scholars have conducted research in an attempt to establish a connection between a leadership style and innovation or ambidexterity, the results have shown large inconsistencies and even contradictions over time (Rosing et al, 2011). By using leader behavioral complexity to simulate the traits of ambidextrous leadership, this study has confirmed that when leadership is percieved as a flexible competence that individual ambidexterity is positively influenced. By doing so, an important step within the field of leadership theory and organizational innovation theory has been taken. Future research, in which ambidextrous leadership can be further shaped, has to determine wether or not this view will

Effects of leader behavioural complexity on individual innovative ambidexterity : the mediating roles of organizational learning and LMX