Master’s Thesis
Master Business Administration – Strategy Track
Personal Characteristics and Individual Ambidexterity: The Mediating Role of
Relational Capital
Student: Jouke Willem Aize Reitsma Student number: 10894365
University of Amsterdam, Faculty of Economics and Business Supervisor: Dr. Bernardo Silveira Barbosa Correia Lima
Statement of Originality
This document is written by Student Jouke Reitsma 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.
TABLE OF CONTENT
TABLE OF CONTENT ... 3
INTRODUCTION ... 5
LITERATURE REVIEW ... 7
Organizational Ambidexterity ... 7
Different Ambidexterity Concepts ... 10
Employee Characteristics ... 13
Relational Capital ... 14
Theoretical and Managerial Implications ... 15
The Relationship between Characteristics and Relational Capital ... 16
The Relationship between Relational Capital and Individual Ambidexterity ... 18
The Research Model ... 20
RESEARCH METHOD ... 21
Sample ... 21
Data Collection Method ... 25
MEASURES ... 26
Agreeableness and Extraversion ... 26
Relational Capital ... 27
Individual Ambidexterity ... 27
RESULTS ... 29
Reliability Analysis ... 29
Factor Analysis ... 31
Common Method Bias Analysis ... 33
Correlation Analysis ... 36
Regression Analysis ... 39
DISCUSSION ... 44
Theoretical implications ... 47
Practical implications ... 48
LIMITATIONS AND FUTURE RESEARCH ... 49
Limitations ... 49
Future Research ... 50
REFERENCE LIST ... 51
APPENDIXES ... 56
APPENDIX A: Survey Cover Letter ... 56
APPENDIX B: Survey Questions and Answers ... 57
Appendix C: Factor analysis ... 61
INTRODUCTION
The world market is becoming more competitive and more dynamic with each day passing. Organizations are now more than ever struggling to survive the fierce worldwide competition. For an organization to survive on the long term it has to ensure that it makes a profit now and that it will make a profit in the future (March, 1991). The concept of focussing at both innovating your current core business to make it more efficient and at the same time investing resources and time to discover ‘the next big thing’ is called ambidexterity. Ambidexterity is a combination of exploitation, improving the efficiency on the short term, and exploration, looking for the next big innovation that will secure profits in the future.
In the field of strategy much research has been completed on ambidexterity but there is still much to learn about this relatively new concept. Research has concluded that an ambidexterity concept where employees themselves make the shift between exploration and exploitation is the most viable. This form of individual ambidexterity makes an employee the main source of ambidextrous behaviour. Because the individual is the key to success we need to know which attributes of individuals make a difference in their level of individual ambidexterity in order to be able to create an environment that stimulates the attributes that have a positive effect and the opportunity to make selections based upon these attributes with a positive influence.
This is the first research that combines the characteristics of the employee with his relational capital and level of individual ambidexterity. This research uses a survey to question employees to rate their personality, their relational capital and how ambidextrous they have behaved over the last year. Through this survey data is gathered to analyse the relationships between the characteristics of employees, their relational capital and their level of individual ambidexterity. With the data of the survey I try to find evidence that the character of an
employee has influence on the relational capital of that employee which on its turn has a positive influence on the level of individual ambidexterity of the employee.
This study contributes to the existing literature and the field of strategic practice in at least three ways. First the results of this study confirm the expectations that were based on previous literature but had not been empirically tested that the characteristics extraversion and agreeableness directly influence the level of relational capital of the employee. This result provides managerial practice with the evidence that the character of the employee is very important regarding their ability to gain a big relational capital at work which can improve the performance of the employee.
The second finding is that the relational capital positively influences the level of individual ambidexterity of the employee. This results gives organizations the opportunity to increase the level ambidexterity among its employees by facilitating relational capital at work through several activities and their organizational culture.
The last finding is a combination of the first two that these two characteristics extraversion and agreeableness mediated by relational capital have a positive influence on individual ambidexterity. This study proved that more characteristics than solely characteristics that sustain self-development can improve individual ambidexterity and extended our knowledge between a person’s character and the ability to perform ambidextrously.
In the first chapter the concepts that are used in this study; ambidexterity, relational capital, extraversion and agreeableness will be explained. An extensive literature review follows which provides the theory behind the hypotheses that are proposed for this study. The third and fourth chapters provide the details how the research has been conducted. The results based upon 183 respondent to the survey are analysed subsequently and conclusions are drawn. This thesis ends with a discussion and evaluation of the research.
LITERATURE REVIEW
In the field of business strategy one of the biggest dilemmas that scholars have been trying to solve for decades is finding a strategy that will be the answer for organizations in their search for a long-term successful existence. One of the concepts developed that could be the answer is ambidexterity. Much has been researched and conceptualized about ambidexterity but most answers raised new questions, there is still a lot we can learn about this promising concept. The first section will shortly explain the concept of ambidexterity and highlight its importance for organizations. The second section will introduce a second concept of ambidexterity, contextual ambidexterity, which was designed to dissolve the flaws of the first concept, structural ambidexterity. After the second section the theoretical and practical implications of this research will be mentioned. This chapters finishes with the hypotheses that are proposed at the start of this research.
Organizational Ambidexterity
One of the more recent theories that elucidates how organizations can survive on the long-term is for organizations to become ambidextrous. James March was in 1991 one of the first authors who wrote an article about organizational ambidexterity (March, 1991), which in his eyes is the key to success for organizations to keep an organization competitive in a dynamic environment on both the short- and the long-term. Organizational ambidexterity is a concept in which organizations simultaneously improve their efficiency on the short term, called exploitation, and look for innovations to secure future profits for the organization on the long term, called exploration.
Exploration and exploitation are two different concepts that complement each other but also compete for the organizations resources. As Levinthal and March mention in their article;
“The basic problem confronting an organization is to engage in sufficient exploitation to ensure its current viability and, at the same time, to devote enough energy to exploration to ensure its future viability” (Levinthal & March, 1993, p. 105). The concept of ambidexterity is vital, but hard to achieve, for the long-term survival of the organization. Balancing exploration and exploitation is as mentioned by Levinthal and March difficult to implement and sustain because both concepts require different environments. Exploration requires that organizations focus on the long term and that organizations have a flexible organizational structure that allows for new ideas to flourish and for experimentation to take place. Exploitation on the other hand needs a tightly controlled organizational structure and needs short term thinking and static routines which improve the current production methods (Levinthal & March, 1993; He & Wong, 2004). Due to the difference in time horizon the financial returns associated with exploration and exploitation are complementary but differ substantially. Exploitation provides returns on the short run due to higher cost efficiency, which is very attractive for managers who often need to deliver short-term results, while exploration is linked with in general higher returns on the long run, due to new ideas and the innovations that are a result of them. The fact that engaging in exploration often does not pay off in the short term inhibits managers that are under pressure to provide short term results to think about the future (March, 1991). The variance in returns also differs between an ambidextrous organizational strategy and an explorative organizational strategy. The returns from engaging solely in exploration shows a much greater variance than the returns on engaging in an ambidextrous organizational strategy due to the lack of exploitation in the solely explorative approach (March, 1991; He & Wong, 2004).
Various empirical studies have investigated the relationship between ambidexterity and organizational performance, but different studies found different results. He & Wong found out that ambidexterity has two effects on the performance of an organization, the interaction between exploitation and exploration has a positive effect on organizational performance, but
the relative imbalance between exploration and exploitation has a negative effect on organizational performance. According to He and Wong an organization is ambidextrous even if it engages only a little in exploration and a little in exploitation (He & Wong, 2004). Tushman and O’Reilly concluded on the other hand that one organizational culture that facilitates both exploration and exploitation at the same time is the key to success when supporting projects (Tushman & O'Reilly, 1996). O’Reilly and Tushman researched whether the ambidextrous organization design outperformed other organizational designs on supporting innovative projects. The results from this research show that the ambidextrous organizational design resulted in a higher percentage of achieved innovation goals compared to the functional, cross-functional and unsupported organizational designs (O'Reilly & Tushman, 2004). These studies indicate that ambidextrous organizational designs, when correctly implemented, have a positive effect on organizational performance. These empirical papers agree on the part that an ambidextrous strategy is good for the performance of the organization but come up with different causes.
Since these the two papers from March and Levinthal & March were published in the early 1990’s the research on organizational ambidexterity has significantly increased over time. The last couple of decades multiple subjects related to the concept; which antecedents cause organizations to transform into ambidextrous organizations, whether ambidexterity has a positive effect on an organization’s performance, how ambidexterity can be balanced, which internal and external organizational and environmental characteristics influence the success of ambidexterity and many more other topics which are related to organizational ambidexterity (He & Wong, 2004; Raisch, et al., 2009; Gupta, et al., 2006; Simsek, et al., 2009; Lavie, et al., 2010).
The importance of organizational ambidexterity for an organization’s quest of survival seems clear but there are still many questions left unanswered in this area of research. Even
after all this research in the field of ambidexterity there is much ambiguity about a concrete definition of organizational ambidexterity, under which circumstances it has a positive effect on organizational performance, which type of ambidexterity should be chosen and how to manage and implement this concept (Simsek, et al., 2009) .
Different Ambidexterity Concepts
In the previous section the concept and the importance of organizational ambidexterity for organizations has been explained. At the end of that section was concluded that there is a lot of ambiguity about how ambidexterity should be implemented. This section will highlight the different concepts of ambidexterity implementations and their advantages and disadvantages. The first concept, structural ambidexterity, is the subject of most of the early research on organizational ambidexterity and encompasses that some business units in an organization focus only on exploration and some focus only on exploitation. In this concept of ambidexterity both exploration and exploitation are implemented at the same time but they are separated among the business units of the organization. This means that a business focusses on either exploration or exploration but not at both. The business units each have their own organizational structure supporting the needs of the concept that this business unit uses. However according to several scholars, does structural ambidexterity bring as many problems with it as it solves for organizations. Structural ambidexterity leads because of the separation between the business units to isolation. This isolation between business units in an organization will make it difficult to link the new exploratory initiatives to the core business (Raisch, et al., 2009; Birkinshaw & Gibson, 2004).
In 2004 Birkinshaw and Gibson propose a new form of ambidexterity, contextual ambidexterity (Birkinshaw & Gibson, 2004). Contextual ambidexterity is the solution to the problem of isolation that arises when structural ambidexterity is pursued. In the concept of
contextual ambidexterity the split between exploitation and exploration is not made between an organization’s business units but in the time of each employee. All employees in a contextual ambidexterity organizational structure are ambidextrous and divide their time between exploration and exploitation (Birkinshaw & Gibson, 2004). Contextual ambidexterity requires a flexible organizational structure which allows employees to decide for themselves the appropriate amount of time that is needed to invest in exploitation of the current processes and routines and the time needed to explore future opportunities and innovations (Birkinshaw & Gibson, 2004). One of the differences between contextual ambidexterity and structural ambidexterity is that only one organizational structure and organizational culture is needed to support this ambidexterity concept instead of two. One organizational structure that facilitates the needs of both exploration and exploitation is tough to achieve but makes it easier to integrate different business units in the organization which encourages a free flow of knowledge, ideas and resources (Birkinshaw & Gibson, 2004).
In the contextual ambidexterity, according to Birkinshaw and Gibson, the split between exploration and exploitation is made by the individuals in the organization. Contextual ambidexterity where the individual is the main source of ambidextrous behaviour of the organization is called individual ambidexterity. There is not only one main definition for individual ambidexterity. Two very different definitions of individual ambidexterity previously used by scholars are; “the individual-level cognitive ability to flexibly adapt within a dynamic context by appropriately shifting between exploration and exploitation” (Good & Michel, 2013, p. 437) and the ability of an employee to both explore and exploit (Mom, et al., 2007; Mom, et al., 2009; Keller & Weibler, 2014). Good and Michel (2013) focus on the intelligence that is needed for employees to make an accurate distinction of when they should pursue exploitation and when to pursue exploration. In contrast to Good and Michel’s definition most other research focuses on the actual time that an employee spends engaging in exploration and exploitation.
This study will use the definition used in Mom, et al. 2007 and 2009 since the priority for an organization is that the individual can perform ambidextrously.
Within individual ambidexterity some scholars focus mainly on the role of the managers in the ambidextrous organizational process, this research concept of ambidexterity is often called managerial ambidexterity (Turner & Lee-Kelly, 2012; Ambos, et al., 2008; Good & Michel, 2013; Keller & Weibler, 2015; Mom, et al., 2015). At the level of managerial ambidexterity most research has focussed on the effect a manager’s characteristics, leadership style, the knowledge flows in the organization, and the influence that the experience of managers has on the level of ambidexterity achieved in the organization (Avolio, et al., 1999; Mom, et al., 2007; Keller & Weibler, 2014; Smith & Tushman, 2005). Till this day most research that has fixated on individuals in the organization has been focused on the managers of an organization (Good & Michel, 2013). However to implement individual ambidexterity the low-level employees of an organization need to be ambidextrous as well, or at least show ambidextrous behaviour (Birkinshaw & Gibson, 2004). By only focussing on the managers, the ambidexterity of low-level employees is neglected.
In the article of Birkinshaw and Gibson (2004) the characteristics of employees that are needed to implement a contextual ambidexterity innovation strategy in an organization are briefly discussed. The authors mention that the employees need to be comfortable focussing both at the short term and the long term. Employees also need to be able to work in an integrated organizational structure to increase the flow of knowledge and ideas through the organization (Birkinshaw & Gibson, 2004). The research on individual ambidexterity however lacks research into which types of persons are more ambidextrous or more suitable to work ambidextrously than others (Simsek, et al., 2009).
Employee Characteristics
To find out which types of persons are more individually ambidextrous than others information about the classification of the personality or characteristics of persons is needed. In the field of psychology and human resource management multiple researches have been devoted to the characteristics of employees and their effect on their performance in various situations (Barrick & Mount, 1991; Digman, 1990). Multiple scholars have tried to identify a combination of characteristics which could be used for classifying personality attributes. Most scholars defined similar characteristics which can be combined to create ‘the big five’ characteristics which encompass most of an individual’s behaviour (Barrick & Mount, 1991). The big five characteristics are; extraversion, emotional stability (or neuroticism), agreeableness, conscientiousness and openness to experience (Barrick & Mount, 1991; Digman, 1990). The first characteristic, extraversion, measures how open the personality of a person is. Behaviours associated with extraversion are, being sociable, assertive and active. The second characteristic emotional stability or neuroticism is linked with being anxious, depressed, emotional or worried. The agreeableness of a person is related to the dimensions, cooperative, good natured and tolerant. The characteristic conscientiousness is associated with being careful, organized and responsible. The last characteristic openness to experience, sometimes called intelligence, is linked with being broad minded, original and imaginative (Barrick & Mount, 1991).
Keller and Weibler did empirical research about the connection between two of these big five characteristics and their link with exploration and exploitation. The results of their research show that conscientiousness is positively linked with exploitation and that openness to experience is positively linked with exploration (Keller & Weibler, 2014). The relationship between intellect and ambidexterity has also been explored in multiple studies and turned out to be positive (Good & Michel, 2013; Raisch & Birkinshaw, 2008; Kang, et al., 2012). The characteristics conscientiousness and openness to experience are linked with the capability of
an individual to learn and develop himself (Keller & Weibler, 2014). Ambidexterity also has been linked with learning and developing individuals (March, 1991). This makes the link between these characteristics and ambidexterity seem logical. This research sheds light on the self-developing characteristics of an individual but completely neglects the characteristics that are used when individuals are working with other people, something that is very important for every organization and occurs frequently (Alchian & Demsetz, 1972). These characteristics are agreeableness and extraversion (Barrick & Mount, 1991).
The overlooked characteristics extraversion and agreeableness are not linked with self-development which is directly connected with individual ambidexterity. These two characteristics are used when individuals have to work together with each other (Barrick & Mount, 1991). To research a link between these two characteristics and individual ambidexterity the interaction of the individual with his colleagues needs to be measured as a mediator. Relational capital is a concept that measures this interaction of the individuals at work.
Relational Capital
Relational capital is, alongside human capital and structural capital, one of the three components of intellectual capital (Carson, et al., 2004). “Relational capital refers to the quality of a person’s relationships with other organizational members in terms of the degree to which she perceives those relationships to be close and trustful” (Mom, et al., 2015, p. 810) .
Different scholars have different ideas about which dimensions relational capital encompasses. Moran splits relational capital in two dimensions, structural embeddedness and relational embeddedness. Structural embeddedness encompasses the structural configuration of a person’s social network. Relational embeddedness is directed at the quality of a person’s network and how the network affects performance (Moran, 2005). According to Kale et al.
relational capital is about the level of mutual trust and respect that exists between people in a relationship (Kale, et al., 2000). When looking at relational capital in alliances Sambasivan et al. built relational capital on the basis of three dimensions; communication, trust, and commitment (Sambasivan, et al., 2011). Trust and commitment are seen as dimensions in the before mentioned researches as well, but by adding communication the authors created a dimension to measure the medium of the flow of information. For this research the configuration of relational capital that consists of the dimensions relational embeddedness and structural embeddedness is taken due to the agreement of previous research about the inclusion of these dimensions in relational capital. The relational embeddedness encompasses the level of trust between the individuals, the structural embeddedness includes the level of closeness between the individuals (Moran, 2005).
To measure if the characteristics of an individual have influence on the level of individual ambidexterity of that individual through their relational capital, the effect of these characteristics on relational capital will be measured and in turn the relationship between relational capital and individual ambidexterity will be measured. This leads to the following research question that is developed for this thesis:
What is the effect that the level of agreeableness and level of extraversion of an employee has on the level of individual ambidexterity through the relational capital of the employee?
Theoretical and Managerial Implications
Multiple review articles mention the need for future research which uses contextual ambidexterity, or individual ambidexterity, as the dependent variable and research that explores which characteristics employees need to possess to be individually ambidextrous (Simsek, et al., 2009; Raisch, et al., 2009). Some research on the effect employees’ characteristics have on
the success of ambidexterity has already been explored but not all of this research has been tested empirically (Birkinshaw & Gibson, 2004). The conclusion is that the link between the employees’ characteristics and individual ambidexterity has not yet fully been explored.
This study will add to the existing literature by testing the effect personal characteristics have on individual ambidexterity through relational capital. Some studies have already shown the relationship between some characteristics of the ‘big five’ but the relationship between the other characteristics and individual ambidexterity has been neglected (Keller & Weibler, 2014). This study will extend the knowledge we will have about the connection between a person’s character and individual ambidexterity.
This research also has practical implications. Intra-organizational social capital of employees have a positive effect on their performance (Maurer, et al., 2011). Because of the positive effect a better explored relationship between employees’ characteristics and social, or relational, capital and ambidexterity could benefit managers who try to achieve individual ambidexterity in their team or organization. Another contribution of this research is for organizations that use a contextual ambidexterity innovation strategy and need to know which characteristics to look for in new employees. This study has the goal to find out if the characteristics are related with individual ambidexterity in order to guide managers in organizations in their process of selecting the right person for the job.
The Relationship between Characteristics and Relational Capital
For the characteristics agreeableness and extraversion to have an influence on individual ambidexterity through the mediator relational capital there first needs to be a significant relationship between the characteristics and relational capital. Previous research found out that the characteristics agreeableness and extraversion are of influence of a person’s behaviour towards others by stimulating interaction (Barrick & Mount, 1991). An empirical research
among project group members in a software and consulting firm shows that agreeable people have the tendency to trust both their peers and their managers (Mooradian, et al., 2006). Trust is an important component of relational capital which could lead to argue that agreeable people could have a high level of relational capital (Moran, 2005).
Agreeable employees are seen as approachable whereas extraverted employees have a tendency to be sociable (Dulebohn, et al., 2012). Being approachable and having the tendency to be sociable are two characteristics that have a positive influence if a person wants to be part of a network of colleagues at work. This resulted from a research done by Asendorpf and Wilpers in 1998. This study showed that agreeable individuals had less conflicts with peers and that extraverted individuals felt that they could rely on their peers as did it have a positive influence on the amount of contacts and the time spend with contacts (Asendorpf & Wilpers, 1998). Relying on peers means a high level of trust among peers. This means that extraverted people would trust their peers more than others leading to a higher level of relational capital. If extraversion and agreeableness have a positive association with the amount and the quality of interactions and on trust among colleagues it is likely that these characteristics have a positive influence on relational capital since this measures the interactions and their perceived quality and trust. This leads me to argue to the following hypotheses:
Hypothesis 1: The level of agreeableness of an employee is positively related with the relational capital of the employee.
Hypothesis 2: The level of extraversion of an employee is positively related with the relational capital of the employee.
The Relationship between Relational Capital and Individual Ambidexterity
Relational capital describes how people organize and value their network. As previously established does this research use the concept where relational capital exists out of two parts, relational embeddedness and structural embeddedness. The first dimension perceives the quality of the network of an individual where the second comprehends the structure of the network (Moran, 2005).
The effect of relational capital on individual ambidexterity has not directly been researched before, to my knowledge, but links between relational capital and alliance ambidexterity and between relational capital and exploration and exploitation have been researched (Mom, et al., 2015; Tiwana, 2008; Tsai & Ghoshal, 1998; Schildt, et al., 2003). Ghoshal and Tsai studied the influence of social capital on value creation through product innovation. Product innovation is highly associated with exploration which is one part of ambidexterity (March, 1991). The authors found empirical evidence that trust and trustworthiness had a positive effect on value creation by product innovation through the mediating role of resource exchange and combination (Tsai & Ghoshal, 1998). Maurer found in a research a positive relationship between trust and product innovation through knowledge acquisition (Maurer, 2010). Trust is the basis of the relational embeddedness dimension of relational capital (Moran, 2005). This study suggests a positive link between relational capital and the explorative part of ambidexterity.
Quite recently a research was published about the effect that relational capital can have on the engagement in explorative activities by employees (Mom, et al., 2015). This study shows that relational capital has both a negative and a positive effect on the engagement in exploratory behaviour of employees. The negative effects is due to the alignment of goals amongst organizational members and the positive effect was due to the knowledge an employee acquires from other organizational members (Mom, et al., 2015).
Tiwana did research on the effect the relational capital of an alliance has on alliance ambidexterity. Relational capital was divided in strong and bridging ties in an alliance. Ties between organizations in an alliance could be comparable with ties between employees in an organization, because both work together to achieve common goals. This study shows that strong ties, fully mediated by knowledge integration, had a positive effect on alliance ambidexterity. Another finding in this research is the interaction between strong and bridging ties in the alliance, also fully mediated by knowledge integration, had a positive effect on the level of ambidexterity in the alliance (Tiwana, 2008).
Schildt, et al. explored the connection between the level of integration in an external corporate venture and its effect on exploration and exploitation. This research argues that the more integrated the external venture is, with acquisition as the most integrated external venture, the better it is for exploitative purposes compared to explorative purposes (Schildt, et al., 2003). If the integration in a network of individuals has the same effect than relational capital could have a positive effect on the exploitative part of ambidexterity through tight integration in the network of the individuals.
The first section of this chapter proposes that higher levels of agreeableness and extraversion positively influence the level of relational capital. Most researches show positive signs for a relationship between relational capital and exploration and exploitation (Tiwana, 2008; Schildt, et al., 2003; Tsai & Ghoshal, 1998), and there is only one study that shows both a positive and a negative effect on exploration (Mom, et al., 2015). This leads me to argue:
Hypothesis 3: The relational capital of an employee is positively related to the level of individual ambidexterity of the employee.
Hypothesis 3a: Relational capital positively mediates the relationship between agreeableness and individual ambidexterity.
Hypothesis 3b: Relational capital positively mediates the relationship between extraversion and individual ambidexterity.
The Research Model
In this study relational capital is a mediating variable because of the expected lack of a direct relationship between the characteristics and individual ambidexterity, but the propositioned positive effect of agreeableness and extraversion on relational capital and of relational capital on its turn on individual ambidexterity.
Figure 1: The research model where agreeableness and extraversion are the independent variables, relational capital is a mediator and individual ambidexterity the dependent variable.
Agreeableness Extraversion Individual ambidexterity Relational capital
RESEARCH METHOD
Sample
To be able to test the hypotheses data is needed about the characteristics, the relational capital and level of individual ambidexterity of employees. Because this research is conducted with an organizational perspective in mind the population for this research consists of all employees but because this study is conducted in the Netherlands the sample will primarily be filled with Dutch employees. To most accurately create a sample that approaches the population the focus of this research lays on employees of whom most have a full-time job (which is defined for this research as working more than 32 hours per week).
Because the population is very big and there is no sampling frame available the sample that is approached to gather data for this research consists out of several organizations positioned in several different industries in the Netherlands as well as individuals in the network of the researcher whom are working in various organizations in the Netherlands. This type of non-random sampling is called convenience sampling (Saunders & Lewis, 2015). The danger of convenience sampling is that the individuals in the sample are not chosen randomly, and therefore not all persons in the population have the same chance of being selected for the sample. This sampling method could lead to a sample which is skewed and does not accurately represent the population. However due to the lack of an accurate population frame, limited resources and a short timeframe, the convenience sampling method is an acceptable sampling technique to find a representative sample for this type of research (Lewis, et al., 2012). By approaching different types of organizations and individuals active in multiple industries a diversified sample is gathered making the results as lowly biased as possible.
Besides the independent, agreeableness & extraversion, mediating, relational capital, and dependent variables, individual ambidexterity, the survey contains questions that will
provide data for control variables. One of the functions of the control variables is to assess if the sample is representative of the population. The second function of the control variables is to assess whether a change in the dependent variable is actually due to a change in the independent variable or the mediating variable and not because of a change in a third variable (Lewis, et al., 2012).
In previous research the variables cognition has already been linked to ambidexterity. By adding the level of education as a control variable the influence of the level of education on ambidexterity should be controlled for (Good & Michel, 2013; Kang, et al., 2012). Being ambidextrous means an individual is able to focus on two different concepts. It seems logical that if an individual has a part-time job and only works one day a week there is a smaller chance that the person behaves ambidextrously due to the small amount of time he or she works. With the control variables working hours per week and the amount of jobs an individual holds this influence should be controlled for. The survey is sent to individuals in many different types of organizations. To assess whether the organization and the specifics of the organization that a respondent works for has any influence on the variables in the model the control variables industry type, organizational size and organizational age have been added. Previous research has found that the tenure of managers is negatively related to exploration due to the lack of fresh ideas (Fernandez-Mesa, et al., 2013). To control for this effect the control variables tenure and function level have been added. The last control variables employee age and gender have been added to the survey to assess if the sample is representative of the population. The questions and answer possibilities used to measure the control variables can be found in table B1 in appendix B.
Sending out the online survey to collect data resulted in 267 respondents. Of these 267 respondents 184 had completed the entire survey. After analysing the results one respondent
was dropped due to the sidedness of the answers that were given. With the data of the remaining 183 respondents the results are analysed.
In table 1, see next page, the frequency of the control variables are shown. As requested almost all of the respondents are in possession of a full time job, which we classified as working more than 32 hours per week (90.2%).We see that significantly more men (67.8%) than women (32.2%) completed the survey which is not surprising since there are more men with a full time job in the Netherlands than women. Most respondents are highly educated, having a WO/University degree (29.0%) or an HBO/University of applied sciences degree (44.8%). The variables tenure, industry type, organizational size and function level are quite well distributed. Most respondents are currently working for a more mature organization which is at least 20 years old (53.3%).
Table 1: Frequency table control variables
Control variable level Frequency Percent Valid Percent
Cumulative Percent Employee age 0-20 2 1,1 1,1 1,1 (years) 21-40 118 64,5 64,5 65,6 41-60 60 32,8 32,8 98,4 >60 3 1,6 1,6 100,0 Gender Male 124 67,8 67,8 67,8 Female 59 32,2 32,2 100,0 Education None 1 0,5 0,5 0,5 High school 20 10,9 10,9 11,5 MBO 21 11,5 11,5 23,0 HBO 82 44,8 44,8 67,8 WO 53 29,0 29,0 96,7 PhD 6 3,3 3,3 100,0 Working hours/week 0-24 10 5,5 5,5 5,5 25-32 8 4,4 4,4 9,8 >32 165 90,2 90,2 100,0 Number of jobs 1 168 91,8 91,8 91,8 2 or more 15 8,2 8,2 100,0 Tenure 0-3 88 48,1 48,4 48,4 (years) 4-10 56 30,6 30,8 79,1 >10 38 20,8 20,9 100,0
Function level Not managerial 83 45,4 45,6 45,6
Middle management 53 29,0 29,1 74,7 Upper management 46 25,1 25,3 100,0
Industry type Product 78 42,6 42,9 42,9
Service 104 56,8 57,1 100,0 Organizational size 0-20 42 23,0 23,0 23,0 (employees) 21-100 35 19,1 19,1 42,1 101-500 42 23,0 23,0 65,0 >500 64 35,0 35,0 100,0 Organizational age 0-5 16 8,7 8,8 8,8 (years) 6-20 69 37,7 37,9 46,7 >20 97 53,0 53,3 100,0
Data Collection Method
This study will empirically test the propositioned hypotheses to validate the proposed relationships between the variables in the research model. A survey is a good method to collect data from multiple employees in multiple organizations which can be easily quantified and used to test the proposed hypotheses. A survey is a cheap and fast tool to gain response from a large number of individuals which makes it suitable to stay within the financial and time limits this research has to deal with (Lewis, et al., 2012).
Although a survey looks like the best alternative a survey does have its disadvantages. Once the survey is sent to the organizations and individuals in the sample it can’t be adjusted and therefore the survey needs to be complete and provide all the data which is needed for the research at the first attempt. To prevent the possibility of concluding during the processing of the results that data is missing the survey was pilot tested. The pilot test came back positive, all the data needed was collected, but also had some feedback how the language in some of the questions and answers could be improved so that the questions and answer possibilities are clear for all possible respondents. Another disadvantage of the survey is that if it’s send electronically it is filled out without the researcher present and therefore the respondent can’t ask questions about ambiguities. To make the survey and its intentions as clear as possible a cover letter (see appendix A) was included explaining the purpose of the survey and of this research. Even though the survey did not ask for very delicate information, full anonymity of the respondents and cooperating organizations has been guaranteed.
MEASURES
All of the variables in the research model of this study have been tested before by other scholars. These studies which are published in peer reviewed journals have used and validated several answer scales that could be used in the survey for this study. The benefit of these scales is that they have been tested in the field and have provided the information that was necessary for other researches. The answers to the survey questions are generally based on a Likert scale consisting out of five to seven possible answers. The Likert scale gives researchers the opportunity to treat ordinal data as interval data making it suitable for empirical analyses that are measuring for correlation and regression (Lewis, et al., 2012).
Agreeableness and Extraversion
The independent variables in this research are two of the “big-five” characteristics; agreeableness and extraversion. These characteristics are used throughout multiple research areas, providing multiple questionnaires with validated questions to measure these characteristics. For this study the questions from the Five-Factor Personality Inventory (FFPI), adopted from Hendriks et al. (1999), are used to measure the agreeableness and the extraversion of the individuals in the sample. The FFPI consists of questions for all five of the “big-five” characteristics which are measured through six questions each. The Cronbach Alpha of the questions used to measure agreeableness and extraversion are in this research 0.84 and 0.86 respectively. The questions are answered on a Likert scale which consisted out of five answers ranging from “Not at all applicable” to “Totally applicable” (Hendriks, et al., 1999). All questions and answer possibilities that are used to test the agreeableness and extraversion of the respondents are displayed in table B3 in appendix B.
Relational Capital
A previous study measures social capital by the splitting it in three dimensions; weak ties, structural holes and network size. Weak ties and structural holes are the structure of the network and network size determines the amount of resources of an individual. The downside of their calculation of the structural holes is that the individual has to quantify the proportion of his relations invested in certain others, persons the individual has a relation with (Seibert, et al., 2001). It can be difficult to quantify differences between subjective measures like the quality difference in relationships.
Another research that measures relational capital divides it in four factors; networking ability, apparent sincerity, social astuteness and interpersonal influence. These four factors were compiled of a total of 18 items and have Cronbach Alpha scores in this research ranging from 0.78-0.87 (Ferris, et al., 2005). These items are entered by the individuals themselves making this measure congruent with the measure of the characteristics agreeableness and extraversion which are also answered by the individuals about themselves. The 18 items are answered on a 7-point Likert scale ranging from “Strongly disagree” to “Strongly agree”. All 18 items are mentioned in table B4 in appendix B.
Individual Ambidexterity
Most researches have measured the amount of time that individuals spend on exploration and exploration and determined on the absolute value of the time spend and the relative difference between the both how ambidextrous an individual is (Keller & Weibler, 2014; Keller & Weibler, 2015; Mom, et al., 2007). To measure the ambidexterity at the individual level the 11-item scale that has been used in Mom et al. (2007) is used in the survey. These 11-11-items are split in six exploitation items, which have a Cronbach’s alpha of 0.81, and five exploration items, which have a Cronbach’s alpha of 0.86. The individuals in the sample answered the
question: “To what extend did you, last year, engage in work related activities that can be characterized as follows….” which is followed by the 11-items (Mom, et al., 2007, p. 918). The answers are given on a 7-point Likert-scale which ranges from “To a very small extent” to “To a very large extent”. All questions and the answer possibilities that are used to measure individual ambidexterity are stated in table B2 in appendix B.
Several scholars calculated ambidexterity as a result of the combination of exploration and exploitation in different ways. He and Wong (2004) subtracted exploitation from exploration and worked with an absolute result whereas Birkinshaw and Gibson (2004) multiplied the exploration score with the exploitation score (Birkinshaw & Gibson, 2004; He & Wong, 2004). For this research the average score of exploitation and exploration is used to measure individual ambidexterity. The definition of individual ambidexterity used in this research is that individuals can both explore and exploit and therefor is the average of the two, making them equal, important instead of the absolute difference between the two.
RESULTS
In this section the results of the data gathered by the online survey will be analysed. The first analysis of the results section is a reliability analysis to check whether the items measure the variable with high accuracy. The second analysis is a factor analysis, this analysis will group items together which measure the same element. After the factor analysis a common method bias analysis measured if the use of one method influenced the answers. To establish a connection between multiple variables a correlation analysis is completed. After establishing a correlation between multiple variables a regression analysis will provide information about the direction, positive or negative, of the effect and the magnitude of the effect that one variable has on another.
Reliability Analysis
To assess whether the questions, items, used in the survey measure the same element with high accuracy a reliability analysis is conducted. In the reliability analysis the Cronbach alpha coefficient of a variable needs to be above 0.7 to indicate a good reliability of the items used to measure the same variable (Field, 2013). In table 2 the reliability scores are shown of each variable.
Table 2: Reliability analysis of the discrete variables
(sub) Variable # Items Cronbach alpha Cronbach alpha if 1 item is deleted Agreeableness 6 0,662* 0,676* Extraversion 6 0,484* 0,509* Relational capital 18 0,907 0,909 Exploration 5 0,841 0,856 Exploitation 6 0,761 0,752
The reliability analysis shows that the Cronbach alpha coefficient of the independent variables agreeableness and extraversion are below 0.7. The reliability of the items used to assess these variables are therefore not up to the set standard. The reliability analysis also shows that if one of the items was deleted the Cronbach alpha score would not increase till above 0.7 therefor no items were deleted. Loewenthal discusses the Cronbach alpha coefficient for psychological tests and scales. His conclusion is that a Cronbach alpha coefficient between 0.6 and 0.7 should be acceptable for these scales (Loewenthal, 2001). The Cronbach alpha coefficient of agreeableness is within this margin, which could lead to the inclusion of the variable with the current items for further analysis. The Cronbach alpha coefficient of extraversion however is even with this lower limit still too low. Some items seem to measure a different element of the variable than others. To analyse if this is true a factor analysis of all the items of all the variables is executed to determine if there are multiple factors.
For the mediating variable, relational capital, and the sub-variables of the dependent variable, exploration and exploitation, the Cronbach alpha scores are above 0.7 (ranging from 0.761 to 0.907) which indicates a good reliability. The highest Cronbach alpha scores that occur for these variables if one item is deleted are slightly higher than the Cronbach alpha score if all items are included but the increase in the reliability is not significant enough (< 0.1) to exclude valuable data, therefore no items were deleted.
Factor Analysis
The factor analysis checks for the amount of factors that exist in a sample of questions or items. For all the items the Kaiser-Meyer-Olkin measure (table 3) verified the adequacy for the analysis, KMO = 0.796 (needs to be >0.6). Bartlett’s test of sphericity χ2 (820) 3615.747, p < 0.001 (needs to be below 0.05) indicates that correlations between items are significant and large enough for a factor analysis (Field, 2013).
Table 3:KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy. 0,796 Bartlett's Test of Sphericity Approx. Chi-Square 3615,747
df 820
Sig. 0,000
According with Kaiser’s criterion ten factors are found which have an eigenvalue above 1, see table C1 in appendix C. These ten factors in total explain 67.683% of the total variance, see table C1 in appendix C. Table 5 shows the values of each of the items corresponding with each of the factors. The questions that correspond with the items in table 5 can be found in the tables B1 to B4 in appendix B.
The factor analysis confirms the results of the reliability analysis for the variable extraversion. The reliability analysis showed low reliability which might have been the result of two different factors in one variable. The factor analysis groups the extraversion items in two factors. After the factor analysis a new reliability analysis of the two factors of extraversion will be conducted.
The other independent variable agreeableness also consists of 6 items which are placed into two different factors. Consistent with the extraversion variable are the reversely coded items (items 4 till 6) grouped together in one factor and the normally coded items are grouped together in one factor. The mediator relational capital was measured with 18 items in total.
These 18 items are split into 3 factors in the factor analysis. Two factors containing each 8 items and one factor containing only 2 items.
The dependent variable individual ambidexterity which is divided in the two sub-variables exploration and exploitation was measured with 11 items in total. The 6 items used to measure exploitation are all grouped into one factor where the 5 items used to measure exploration are separated into 2 factors. For the correlation and regression analyses the factor values will be calculated by taking the average scores of the items included in the factors. The average total value of the variable is then the average of each of the factor values that are included in the variable.
A reliability analysis of the 2 factors of the variable extraversion is conducted to check if the Cronbach alpha coefficient of each of the two factors is high enough to continue the analyses. Table 4 shows the results of the second reliability analysis. The first factors has a Cronbach coefficient of 0.690 which is slightly below the 0.7 threshold but above the 0.6 threshold mentioned by Loewenthal (Loewenthal, 2001). The reliability value of factor 1 would increase if one items would be dropped (0.736) but this increase is not large enough (< 0.1) which leads to an inclusion of all the items in the factor.
Factor 2 with all items included shows a high Cronbach alpha coefficient of 0.805. This is well above the limit and corresponding with factor one is the increase in the coefficient if one items is deleted too small to actually delete one item. For factor 2 all items are included for the rest of the analyses. For the other variables no second reliability analyses was run due to the fact that the reliability of all the items in the variable together proved to be sufficient.
Table 4: Reliability analysis extraversion factors
(sub) Variable # Items Cronbach alpha
Cronbach alpha if 1 item is deleted Extraversion F1 3 0.690 0.736*
Extraversion F2 3 0.805* 0.840*
Common Method Bias Analysis
Common method bias or common method variance is the difference in the results that are likely due to the measurement method instead of the concepts that the measurement method is supposed to measure (Podsakoff, et al., 2003). To analyse for this deviation in the results a Harman’s single factor analysis was run where only one factor was set by default and no rotation was used to calculate the factor loadings. The variance explained by the single factor was 20.543% which is significantly different from the total variance explained in the factor analysis 67.683% (table 5). The difference in variance explained is big enough to conclude that the bias does not threaten the validity of the results.
Pattern Matrix Factor Item Relational capital F1 Extraversion F1 Exploitation F1 Exploration F1 Agreeableness F1 Agreeableness F2 Relational capital F3 Extraversion F2 Relational capital F2 Exploration F2 Exploration 1 0,126 0,009 -0,102 -0,834 -0,018 -0,067 0,027 -0,034 -0,026 -0,017 Exploration 2 0,149 -0,034 -0,001 -0,868 -0,037 -0,025 -0,018 -0,002 0,110 -0,032 Exploration 3 0,022 -0,074 -0,005 -0,802 0,051 -0,081 -0,090 -0,020 0,003 0,064 Exploration 4 -0,131 -0,046 0,161 -0,474 -0,047 0,137 0,067 0,084 -0,161 -0,474 Exploration 5 0,004 0,100 0,080 -0,283 0,044 0,323 -0,007 -0,052 -0,114 -0,407 Exploitation 1 0,046 0,171 0,352 -0,233 -0,135 0,042 -0,097 -0,062 -0,217 -0,047 Exploitation 2 -0,048 0,027 0,492 -0,201 0,091 -0,109 -0,110 -0,059 -0,074 0,148 Exploitation 3 0,175 0,044 0,756 0,124 -0,116 0,010 0,030 0,007 -0,072 0,073 Exploitation 4 0,008 -0,032 0,565 -0,041 0,090 -0,175 0,085 -0,011 0,083 -0,009 Exploitation 5 -0,022 -0,013 0,752 0,092 -0,011 0,237 -0,021 -0,008 0,015 -0,034 Exploitation 6 -0,090 0,014 0,623 0,041 -0,039 0,142 0,009 0,096 -0,027 -0,273 Relational capital 1 0,798 0,016 0,106 -0,078 0,075 0,033 0,080 0,040 0,082 0,044 Relational capital 2 0,666 -0,069 -0,065 -0,065 -0,055 0,064 0,037 0,095 -0,167 0,031 Relational capital 3 0,685 -0,098 0,030 -0,070 -0,081 -0,071 -0,037 0,034 -0,122 0,018 Relational capital 4 0,740 -0,023 -0,007 -0,009 -0,147 -0,007 -0,104 0,019 -0,103 0,004 Relational capital 5 0,773 -0,017 0,052 -0,063 0,068 0,040 -0,115 -0,027 0,063 0,028 Relational capital 6 0,634 -0,032 -0,151 -0,082 -0,006 0,037 -0,176 -0,042 -0,061 -0,143 Relational capital 7 0,127 0,054 0,022 0,011 0,059 -0,014 -0,164 -0,039 -0,580 0,007 Relational capital 8 0,012 0,024 -0,032 0,032 0,087 -0,025 0,113 -0,010 -0,885 -0,018 Relational capital 9 0,340 0,236 -0,058 0,018 0,196 0,128 -0,100 0,080 -0,056 -0,253 Relational capital 10 0,238 -0,196 0,024 -0,008 0,053 -0,150 -0,358 -0,056 0,014 -0,229 Relational capital 11 0,194 -0,060 0,079 0,044 -0,066 -0,036 -0,592 -0,073 0,067 -0,216 Relational capital 12 0,095 -0,315 0,071 0,004 -0,157 -0,213 -0,237 0,207 -0,047 -0,277 Relational capital 13 0,263 -0,183 0,057 0,115 0,065 -0,234 -0,106 0,105 -0,189 -0,287 Relational capital 14 0,117 -0,096 0,111 0,082 0,056 -0,290 -0,558 -0,109 -0,046 -0,261 Relational capital 15 0,152 0,102 0,002 -0,072 0,067 0,030 -0,567 0,154 -0,162 0,190 Relational capital 16 -0,082 0,069 -0,017 -0,054 0,123 0,049 -0,747 0,095 -0,060 -0,022 Relational capital 17 0,059 0,062 -0,041 -0,058 -0,122 0,185 -0,649 0,097 -0,011 0,086 Relational capital 18 -0,014 -0,033 -0,051 -0,027 0,078 0,080 -0,754 0,057 0,005 0,084
Agreeableness 1 0,108 0,152 -0,084 0,032 0,768 0,142 -0,013 0,004 0,009 -0,104 Agreeableness 2 0,071 -0,036 0,083 -0,018 0,729 -0,022 -0,013 0,020 -0,133 0,068 Agreeableness 3 -0,150 -0,084 0,001 0,000 0,584 -0,035 -0,024 -0,002 -0,001 0,019 Agreeableness 4 0,111 -0,098 0,012 0,025 0,124 0,590 -0,040 -0,054 0,141 -0,081 Agreeableness 5 0,037 -0,223 0,050 0,193 -0,013 0,728 -0,177 -0,074 -0,065 -0,012 Agreeableness 6 0,031 -0,490 0,005 -0,091 -0,005 0,313 -0,044 0,062 -0,070 0,060 Extraversion 1 -0,090 0,490 0,094 0,067 -0,055 -0,081 -0,019 0,178 -0,226 0,189 Extraversion 2 -0,129 0,635 0,102 -0,038 -0,018 -0,090 -0,050 0,205 -0,144 0,105 Extraversion 3 -0,059 0,622 -0,004 0,040 0,014 -0,018 -0,034 0,153 -0,150 -0,113 Extraversion 4 0,050 0,267 -0,054 0,060 -0,011 -0,075 -0,183 0,448 0,040 -0,061 Extraversion 5 -0,026 -0,171 0,011 0,010 0,006 -0,077 0,044 1,011 -0,015 0,010 Extraversion 6 0,108 0,169 0,002 -0,002 0,031 0,076 -0,056 0,713 0,061 0,017 Eigen value 8,423 4,095 3,343 2,522 2,368 1,770 1,580 1,317 1,248 1,085 % of total variance explained 20,543 9,987 8,154 6,151 5,775 4,317 3,854 3,213 3,044 2,645
Extraction Method: Principal Axis Factoring.
Rotation Method: Oblimin with Kaiser Normalization. a. Rotation converged in 16 iterations.
Correlation is a relationship between the results of two variables. Table 6 shows the results of the bivariate correlation analysis. In the table all the significant correlations are marked with an asterix (*). For this analysis the nominal variables gender and industry type are transformed into discrete variables. In this section the most important correlations will be discussed.
The first independent variable, agreeableness, is positively significantly correlated with three of the control variables. Employee age (Pearson correlation coefficient = 0.182; Sig. =
0.014), tenure (Pearson correlation coefficient = 0.163; Sig. = 0.028), and organizational size
(Pearson correlation coefficient = 0.192; Sig. = 0.009) all are positively correlated with agreeableness. The control variable that is negatively significantly correlated with agreeableness is industry type (Pearson correlation coefficient = -0.169; Sig. = 0.022). Industry type is a nominal variable where the product industry received the value 1 and the service industry received the value 2. Higher values of agreeableness are more common in the service industry than the product industry. The correlation coefficient between agreeableness and the mediator relational capital is positive and significant as well (Pearson correlation coefficient =
0.163; Sig. = 0.028). This significant correlation makes it possible for agreeableness and
relational capital to have a positive regression coefficient as well. Between agreeableness and individual ambidexterity no significant correlation relationship was found.
The correlation relationships between the other independent variable, extraversion, and the control variables differ from the correlation relationships of agreeableness. None of the control variables is significantly correlated with extraversion. However similar with the relationship between agreeableness and relational capital is the correlation relationship between extraversion and relational positive, and significant (Pearson correlation coefficient = 0.324;
Sig. = 0.000). The correlation between extraversion and individual ambidexterity was not
The mediating variable relational capital is found to be positively correlated with the control variables employee age (Pearson correlation coefficient = 0.173; Sig. = 0.019) and function level (Pearson correlation coefficient = 0.255; Sig. = 0.001). Very important for the hypotheses is that relational capital is significantly positively correlated with individual ambidexterity (Pearson correlation coefficient = 0.374; Sig. = 0.000) which means there is a possibility of a positive regression relationship.
Apart from relational capital the only other variable that is significantly correlated with individual ambidexterity is function level (Pearson correlation coefficient = 0.179; Sig. =
0.016). This positive correlation coefficient dictates that managers in high management
positions are on average more individually ambidextrous than managers in middle management positions and non-managers.
The correlation analysis shows some very insightful significant correlations between some control variables. Function level is found to be negatively correlated with gender (Pearson correlation coefficient = -0.201; Sig. = 0.006). The nominal variable gender was quantified by giving men the value 1 and women the value 2. This means that women have on average a lower function level than men. Another interesting correlation is between the level of education and the size of the organization a respondent is working for (Pearson correlation
coefficient = 0.246; Sig. = 0.000). This correlation states that bigger organizations (> 500
employees) have on average higher educated employees than smaller organizations (< 500 employees).
Table 6: Mean, Standard deviation and Correlation of the variables Variables M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1. Agreeableness 3,74 0,58 2. Extraversion 3,60 0,52 0,012 3. Relational capital 5,61 0,64 ,163* ,324** 4. Individual ambidexterity 4,66 0,77 0,074 0,066 ,374** 5. Employee age 2,35 0,53 ,182* -0,126 ,173* ,188* 6. Gender 1,32 0,47 0,141 -0,034 -0,075 -0,141 -,146* 7. Education 4,01 1,01 0,139 0,072 0,012 0,103 -0,024 0,066 8. Working hours/week 2,85 0,49 0,012 -0,096 0,083 0,050 -0,068 -0,095 -0,054 9. Number of jobs 1,08 0,28 0,056 0,030 -0,046 0,059 -0,009 ,177* 0,058 -0,069 10. Tenure 1,73 0,79 ,163* -0,096 0,143 0,006 ,532** -0,131 -0,019 0,090 0,029 11. Function level 1,80 0,82 -0,058 -0,125 ,255** ,179* ,403** -,201** -0,025 0,045 -0,023 ,401** 12. Industry type 1,57 0,50 -,169* 0,047 -0,063 -0,116 0,113 -0,017 -0,016 -,181* 0,058 0,126 -0,081 13. Organizational size 2,70 1,17 ,192** 0,099 0,083 0,125 -0,024 0,057 ,246** 0,053 -,162* -0,011 -,229** -0,125 14. Organizational age 2,45 0,65 0,138 0,081 0,059 -0,050 ,175* -0,041 0,030 0,025 -0,051 ,315** 0,026 0,008 ,354**
**. Correlation is significant at the 0.01 level (2-tailed).
Regression Analysis
The last analysis that is completed is the regression analysis. The regression analysis tests the direction and the strength of the relationship between the variables in research model to confirm or reject the five hypotheses. To reject or support the hypotheses regression analyses with three models have been run (Jansen, et al., 2009). The first analysis with model 1 was a linear regression analysis containing the control variables and the dependent variable individual ambidexterity. The second regression analysis was also a linear regression analysis but the independent variables extraversion and agreeableness have been added. To test the hypotheses 3a and 3b a mediating regression analysis, model 3, including the control variables, the independent variables and the mediating variable was performed by process from Hayes in SPSS (model 4) was run to find out if relational capital indeed has a mediating role (Hayes, 2013).
To test whether the changes in the results from the mediating and dependent variables actually are a result from a change in the independent variables all of the ten control variables have been integrated in the regression analyses. Table 7 shows a summary of all the regression coefficients that are significant. The regression coefficients of all variables, including the not significant relationships, are in table D1 in appendix D.
To test for mediation, hypotheses 3a and 3b, this study used the method explained by Baron and Kenny in 1986. These scholars name three conditions for mediation to exist. The first condition is that the independent variables have a significant relationship with the mediating variable, hypotheses 1 and 2. The second condition is that the mediator has a significant relationship with the dependent variable, hypothesis 3. The last condition is that when conditions one and two are controlled for there is no direct significant relationship between de independent and the dependent variables (Baron & Kenny, 1986).
Table 7: Linear and mediating regression analysis results
Individual Ambidexterity
Variables Model 1 Model 2 Model 3
Control variables Function level 0.160* 0.169* 0.080 Age 0.327* 0.334* 0.282* Organizational size 0.139* 0.134* 0.109* Independent variables Agreeableness 0.140 -0.029 Extraversion 0.041 -0.049 Mediating variable Relational capital 0.404**
* Regression analysis is significant for P < 0.05 ** Regression analysis is significant for P < 0.001
Relationships between control variables and variables from the research model
First the significant relationships between three of the control variables and individual ambidexterity are highlighted. The nominal and ordinal control variables were made quantifiable by assigning numbers to the results. The first answer receives the value 1, the second answer the value 2 and so forth, see appendix B table B1 for all the answer possibilities of the control variables that were given in the survey.
The results from the regression analysis suggest a significant positive relationship between function level and individual ambidexterity (B= 0.160; P < 0.05). Function level was measured as an ordinal variable with the options; not managerial, middle management & top management. The correlation coefficient (B) proposes that an increase in position level from not managerial to middle management or from middle management to top management will result in an average increase in the level of individual ambidexterity of the employee of 0.160. The second significant regression relationship is between the age of an employee and individual ambidexterity (B= 0.327; P < 0.05). The age of an employee is also measured as an
ordinal variable where the possible answers were; 0-20 years, 21-40 years, 41-60 years and >60 years old. These findings suggest that people increase their level of individual ambidexterity over time. The older employees get the more ambidextrous they become.
The last significant relationship involving a control variable is the relationship between the size of the organization an employee is working for and the level of individual ambidexterity of that employee is positive and significant (B= 0.139; P < 0.05). The variable organizational size was tested with the following possible answers; 0-20, 21-100, 101-500, >500 employees currently working for the organization.
Relationships between agreeableness, extraversion and relational capital
To test the first hypothesis a linear regression analysis is completed with the independent variable agreeableness and the mediating variable relational capital. All the control variables were added to control for their influence on the variables. The regression analysis shows that there is a positive and significant relationship between agreeableness and relational capital (B=
0.197; P < 0.05). According to the regression analysis higher levels of agreeableness result in
higher levels of relational capital. Being more tolerant and agreeable can give you a bigger network or more interaction in that network. This not significant relationship leads to supporting hypothesis 1.
To test the second hypothesis, which proposes a positive relationship between extraversion and relational capital, a linear regression analysis with the variables extraversion, agreeableness, relational capital and all control variables is completed. The regression analysis supports this hypothesis by showing a significant positive relationship between the independent variable extraversion and the mediating variable relational capital (B= 0.473; P < 0.001). Extraverted employees approach other organizational members easier than introverted people. By getting in touch with more colleagues the more extraverted employees could build a bigger
