Examining the level of individual ambidexterity : the role of business intelligence usage, knowledge sharing proactive personality and role ambiguity

(1)

Examining the level of individual

ambidexterity:

The role of Business Intelligence usage, Knowledge sharing Proactive

personality and Role ambiguity.

Name: Paul Robbert de Vries

Student number: 10684662

Date: 28-06-2016

Qualification: Executive Programme Business Studies, Strategy Track Institution: Amsterdam Business School, University of Amsterdam Thesis supervisor: B. Lima

(2)

(3)

iii

ABSTRACT

This research investigates the relationship between the usage of business intelligence tooling and an individual’s ability to balance and simultaneously execute explorative and exploitative activities, which is known as individual ambidexterity. Most research in the area of finding evidence for antecedents which drive ambidexterity focusses on the organizational level of analyses. Less is known of the antecedents of individual ambidexterity, making it difficult for organizations to build an ambidextrous organization, since it all starts with having

ambidextrous employees. This study addresses this gap in the literature by providing an in-depth understanding of the antecedents’ business intelligence usage and knowledge sharing in relation to the level of an individual’s ambidexterity.

Based on an empirical survey, data has been collected from 138 individual business intelligence users working in multiple organizations. Results indicate that both business intelligence usage and knowledge sharing directly positively affect the level of individual ambidexterity. In additional the effects of two moderators, role ambiguity and an individual’s level of pro-activeness have been tested, since prior research indicates and addressed

opportunities to further investigates if individual’s behavior and characteristics influence their level of ambidexterity. Contrary to the expectations, only role ambiguity affects the

relationship between business intelligence usage and individual ambidexterity.

Although no significant support is found in this study on the effects of the moderators on the willingness to share knowledge, to positively affect the level of individual

ambidexterity, this research indicates they seem promising constructs for future research.

Keywords: business intelligence, knowledge sharing, proactive personality, role ambiguity, exploration and exploitation, individual ambidexterity

(4)

iv

1. INTRODUCTION

As competition intensifies and the pace of change accelerates, firms need to renew themselves by exploiting existing competencies and exploring new ones (Floyd & Lane, 2000). This frequently leads to tension in how to balance simultaneously explorative and exploitative activities, which is known as ambidexterity. A recent review article from Raisch & Birkinshaw (2008) provides a clear and transparent overview of various literature streams and models which cover antecedents, moderators, and outcomes of organizational

ambidexterity. Ambidexterity has been described as “the ability to simultaneously pursue

both incremental and discontinuous innovation” (Tushman & O'Reilly, 1996).

For firms to survive over the long run, they need to become ambidextrous by

mastering both adaptability (exploration) and alignment (exploitation), by implementing both incremental and revolutionary change (Tushman & O'Reilly, 1996). Several researchers such as He & Wong (2004) and Lubatkin, et al. (2006) investigated the relationship between ambidexterity and firm performance. They found positive effects for firms which score high on ambidexterity in relation to their firm performance.

The relationship between ambidexterity and firm performance has been recognized in prior research, this also counts for research focused on studying on how firms can become ambidextrous by investigating drivers which enhance organizational ambidexterity, also known as antecedents. Several literature streams contributed in trying to find explanations on how firms can become ambidextrous from angles such as organizational learning,

technological innovation, organizational adaptation, strategic management, and

organizational design (Raisch & Birkinshaw, 2008). These researches address both structural and contextual antecedents on how to become ambidextrous. Structural elements can be identified as formal organizational mechanisms, whereas contextual elements are identified as supportive soft elements, such as personal relationships, autonomy and proactive behavior (Jansen, et al., 2005; Mom, et al., 2009; Gibson & Birkinshaw, 2004).

However, recent studies addressed the need to uncover the effects of individual’s level of ambidexterity, since individuals who act ambidextrous, promote ambidexterity across various levels throughout the organization (Taylor & Helfat, 2009). Furthermore, by

primarily focusing on the organizational level of analyses, individual homogeneity is in effect assumed, though the main characteristics of exploration and exploitation are said to originate at the individual level (Keller & Weibler, 2015).

(8)

2 working environment positively affect an individual’s level of ambidexterity (Rogan & Mors, 2014; Bonesso, et al., 2014). This indicates that also from an individual level of analyses both structural and contextual antecedents need to be taken into account.

Since the digital revolution evolved, the business world has changed tremendously and provided enormous amounts of new opportunities to support employees in executing their daily activities (Chen, et al., 2012). A relatively new phenomenon to support individuals to measure outputs and make fact based decisions is the usage of business intelligence (BI) tooling (Popovič, et al., 2012). Business intelligence tooling provides actionable information delivered at the right time, at the right location, and in the right form to assist decision-making (Nagash, 2004).

Business intelligence tools allow employees to extract, transform and load data for analyses, and then make those analyses available in reports, alerts, and scorecards to support decision making (Davenport, 2006; Gollapudi, et al., 2012). The use of BI tooling has gained a lot of attention within firms, since information and monitoring performance has become more important in the fast changing and dynamic global business environment (Gandomi & Haider, 2015). The fast technological developments, large data sets and user friendliness of BI tooling provided opportunities for individuals to use BI tooling in their daily work to run reports, status charts and detect trends (Nudurupati, et al., 2011).

As suggested by Boe-Lillegraven (2014), further research could investigate the effects of using BI tooling on individual ambidexterity, since BI tooling provides the opportunity to dynamically track and measure outputs, as well as identify new opportunities to exploite in current businesses. BI tooling enables individuals to share and exchange their findings generated from BI usage, and communicate novel information through the digital highway (Popovič, et al., 2012; Nagash, 2004). This study will thefore focus on the somewhat unadressed effects of BI usage on an individual’s level of ambidexterity.

According Sharma & Djiaw (2011) the use of BI tooling can enhance knowledge sharing between employees in where it’s not the quantity of the knowledge, but the ability to apply existing knowledge to create new knowledge by exchanging. This can be characterized as a structural element. A contextual element has been addressed by Nudurupati, et al. (2011) which emphasize that an individual’s attitude and behavior towards BI tooling plays an equally important role to enhance an individual’s level of ambidexterity. Structural elements are defined and coordinated from a firm level perspective, which is covered in procedures, vision and strategy. Contextual elements are supportive and driven from a less formal perspective. A supportive context enhances and embraces personal interactions, shared

(9)

3 beliefs and roots which is facilitated and embedded in both firms and individuals behavior (Nudurupati, et al., 2011).

Individual’s willingness to exchange knowledge isn’t only enhanced in case firms support and technically facilitates, but individual’s knowledge capabilities and proactive personality play a role as well to explain ambidexterity (Wu, et al., 2014; Lubatkin, et al., 2006). As mentioned in previous paragraphs several researchers such as (Sharma & Djiaw, 2011; Nudurupati, et al., 2011; Bititci, et al., 2012) provided suggestions to further examine and exploit the possible causal relationship effects of BI usage on knowledge sharing to enhance an individual’s level of ambidexterity. In case BI tooling is available for a large number of individual’s within a single firm, this provides opportunities to share outcomes and therefore I expect that BI positively affect the level of knowledge sharing which enhances an individual level of ambidexterity.

Another antecedent to explain the effects on individual ambidexterity has been studied by Bateman & Crant (1993) by arguing that proactive personality behavior can stimulate and enhance knowledge sharing. Proactive behavior is defined as self-directed, future-oriented behavior to improve the current situation for oneself or the organization (Shin & Kim, 2015). This relates to Birkenshaw & Gibson (2004) characteristics of ambidextrous individuals, who take initiative and are furthermore motivated to act spontaneously upon opportunities beyond their basic job requirements. An individual’s cognitive abilities, personality traits, social interactions and network capabilities have been identified as antecedents, which influences individual’s enactment in knowledge sharing and proactive personality behavior (Rogan & Mors, 2014; Good & Michel, 2013; Bateman & Crant, 1993).

Proactive behavior is behavior that directly alters environments Bateman & Crant (1993), which contains both structural and contextual elements. Contextual elements are autonomy, trust and managerial support (Shin & Kim, 2015). Structural elements could be identified as code of conducts and strict working procedures (Miles, et al., 1978). An individual’s proactive personality orientation could stimulate the effort individual’s put in networking and building ties with other’s to strengthen their willingness to exchange knowledge. Proactive personality could stimulate individual’s to enhance their social interactions with others and cope with multitasking (Shin & Kim, 2015).

Therefore, this research will explore if proactive personality moderates the effect of BI usage on individual ambidexterity and knowledge sharing, since an individual’s level of pro-activeness could determine if an individual is willing and open to exchange knowledge.

(10)

4 Proactive individuals are open for new ideas, improvements and initiate changes actively and therefore will become more ambidextrous over time then someone who doesn’t possess these traits (Pintrich, 2000).

Additionally role ambiguity as a moderator will be tested, since individual’s who experience role ambiguity might be willing to share knowledge to both gain and obtain information from others (Floyd & Lane, 2000). Individuals who fulfill multiple roles within a firm are willing to participate in activities that involve adaptation to new opportunities, but are clearly aligned with the overall strategy of the firm (Birkenshaw & Gibson, 2004). In case individuals are encouraged and supported by the firm to make their own choices between alignment and adaptation oriented activities, they will act ambidextrous.

Within a more general perspective, individual behavior context refers to the systems, processes, and beliefs that individuals enables and encourages to judge for themselves how to best divide their time between the conflicting demands for exploitation and exploration (Ghoshal & Bartlett, 1994). Based on above mentioned arguments this paper’s objective is to deepen and extend our understanding if BI usage affects individual ambidexterity. Current literature address a research gap on how, and to what extent BI usage can affect the level of individual ambidexterity (Jourdan, et al., 2008; Sharma & Djiaw, 2011). This research will contribute to the literature and practice, by illuminating the current state of research, if underlying personal behavioral aspects affect an individual’s level of ambidexterity, but

above all, if BI usage, as a new phenomenon, is a driver to explain individual ambidexterity. Therefore, this research will search for the answers on following research questions:

“ How does business intelligence usage affects the level of individual ambidexterity?” “ To what extent does knowledge sharing mediate the relationship between business intelligence usage and individual ambidexterity?”

(11)

5 As such, I will test following conceptual model as shown in figure 1.Fur

Individual Ambidexterity Sharing Knowledge Business Intelligence Usage Proactive Personality Role Amguity

Figure 1: Conceptual Model

The aim of this research is to test hypothesis, which contributes to the research era of

individual ambidexterity by testing if BI usage and knowledge sharing are antecedents, which positively affect individual ambidexterity. Furthermore, the conditional effect of proactive behavior and role ambiguity on these antecedents will be tested.

Quantitative data has been collected via an internet-mediated survey distributed among individuals, which make use of BI tooling. The hypotheses of the possible

relationships of observable variables are based or related to existing research and therefore this research follows a deductive approach.

This research is structured as follows: In the next chapter, a literature review will be presented to further explore the current state of research in the era of individual

ambidexterity, business intelligence, knowledge sharing mechanisms, proactive behavior and role ambiguity. The literature review will provide more in-depth findings, detailed theoretical background and link arguments for testing the hypothesis. Based on the findings within the literature, hypotheses are identified to measure direct, mediating and moderating effects on individual ambidexterity.

Chapter three describes detailed information about the context and set up of the survey and data collection. To test the hypotheses an internet survey has been distributed to individuals within my network and known contacts in four companies where participants were asked to participate by filling in a survey for academic research purposes. By collecting data from such a broad range of participants, bias is reduced and strengthens the external validity. The survey is based on questions used in prior research to increase validity and reliability. All constructs have been measured according a seven point Likert scale as defined by (Bateman & Crant, 1993).

(12)

6 describes detailed information about the tests which are used to analyze the data, reports descriptive statistics and contains validation analyses on the hypothesis.

In chapter five I will discuss the results in relation the literature review and report managerial implications. Chapter six indicates limitations of this research and provides suggestions for further research. In the last chapter an overall conclusion is presented.

2. LITERATURE REVIEW

A company’s ability to simultaneously execute today’s strategy while developing tomorrow’s arises from the context within its employees operate (Birkenshaw & Gibson, 2004). This is in line with the findings of Mom, et al. (2009) who emphasize that complementing formal structural coordination mechanisms with personal mechanisms increases the level of individual ambidexterity. Birkenshaw & Gibson (2004) further build on the organization literature in particular on Ghoshal & Bartlett’s (1994) framework for organizational effectiveness. This framework suggests that ambidexterity emerges from a supportive organization context.

Based on these arguments the question raises of whether BI usage can be one of these supportive structural drivers to enhance individual ambidexterity. Sharma & Djiaw (2011) as well as Jourdan, et al. (2008) have indicated this as a gap in the current research literature, since BI is a relatively new phenomenon, and therefore will be further investigated in this research.

2.1 Individual Ambidexterity

The term ambidexterity has been introduced in the academic research literature by Duncan (1976), in which he refers that ambidexterity is derived from the Latin literature, where it refers to the ability of an individual to use the right and left hand with equal skill. More recently the definition of ambidexterity given by O'Reilly & Tushman (2013) is frequently cited: “ Organizational ambidexterity refers to the ability of an organization to both explore and exploit – to compete in mature technologies and markets where efficiency, control, and incremental improvement are prized and to also compete in new technologies and markets where flexibility, autonomy, and experimentation are needed.

However March’s (1991) article frequently is cited as the catalyst for the current interest in the concept of ambidexterity (Raisch & Birkinshaw, 2008). According to March (1991) exploration is associated with search, discovery, experimentation, risk taking and innovations as whereas exploitation is associated with refinement, efficiency, production and

(13)

7 selection. There is clearly a contradiction between simultaneously acting and executing exploitative and explorative activities. Therefore, several researchers such as Benner & Tushman (2002) and Tushman & O'Reilly (1996) discussed the need for firms to establish a balance between exploitation and exploration activities.

Birkenshaw & Gibson (2004) have identified the concept of simultaneously balancing between exploitation and exploration by explaining the terms alignment and adaptability. In which alignment is associated to incorporate innovative ideas and lessons learned in the daily business routines and processes. Adaptability on the other hand is the ability of a firm to identify new ideas and opportunities. Tushman & O'Reilly (1996) focused on the innovative aspect of how to find the proper balance be defining ambidexterity as “The ability to

simultaneously pursue both incremental and discontinuous innovation”.

Birkenshaw & Gibson (2004) state that most researchers struggle to find comprehensive and significant evidence how to explain the right balance between simultaneously execute exploitative and explorative activities without separating these activities within a firm. They developed and explored the concept of contextual

ambidexterity, in which individuals make choices between alignment-oriented and adaptation-oriented activities in the context of their daily work (p. 49). Contextual

ambidexterity therefore bridges individuals and organizational level of analyses. Instead of trying to explain ambidexterity solely from a firm level perspective, they focus on both. This is line with the literature gaps in the era of individual ambidexterity as defined by (Raisch & Birkinshaw, 2008; Gupta, et al., 2006; Mom, et al., 2009). They address future opportunities for research on the drivers of individual ambidexterity, the cognitive processes that shape individual’s ambidexterity and possible effects of intellectual capital that drives the pursuit of becoming ambidextrous.

Prior studies on individual ambidexterity don’t provide a clear definition, but have in common that individual’s which can cope with and balance executing explorative and exploitative tasks are ambidextrous. Mom, et al. (2009) provides following definition of individual ambidexterity, but solely focus on managers as: ‘A mangers’ behavioral

orientation toward combining exploration and exploitation related activities within a certain period of time’ (Mom et al., 2009: p. 812). This study will follow Mom’s definition, whereas this research refers to individual ambidexterity.

Mom, et al. (2009) emphasize that both organizational integration mechanisms as well as interpersonal interaction enhance ambidexterity. However, their findings concerning

(14)

8 organizational integration mechanisms are less significant than for interpersonal interaction. BI tooling has been identified by Jourdan et al. (2008) as a formal integration

mechanism, which can be used by multiple hierarchical levels within an organization, whereby BI tooling can be used to support both cognitive differentiation and integration processes (Sharma & Djiaw, 2011).

Another driver to explain individual’s ambidexterity has been addressed by Mom, et al. (2009) and Floyd & Lane (2000) by arguing that individuals should both refine and renew their knowledge, skills and expertise. According to Burgelman (1991) a firm is viewed as ecology of strategic initiatives which emerge in patterned ways. Individuals search for opportunities to express their skills through the pursuit of different types of strategic initiatives. Strategic initiatives can be driven by administrative and cultural mechanisms experienced from an individual level. This is in line with the concept of strategic renewal as proposed by (Floyd & Lane, 2000). They follow Burgelman’s (1991) argument that

experiments with new skills or market opportunities diverge from official strategy. Strategic renewal is an evolutionary process associated with promoting, accommodating, and utilizing new knowledge in order to bring about change in an organization’s core competencies and/or a change in its product market domain (Floyd & Lane, 2000). These arguments are consistent and in line with the concept of both emphasizing structural organizational mechanisms as well as taken personal cognitive characteristics into account (Birkenshaw & Gibson, 2004). These different attributes of how to enhance individual ambidexterity, provide a more in-depth insight and is supported by Raisch & Birkinshaw (2008) which emphasize that both structural as well as contextual approaches are generally complementary. However, the link with the opportunities that BI usage can provide in respect of enhancing and supporting individual ambidexterity remains undiscovered. Gupta, et al. (2006) doubt the fact that individual ambidexterity is very difficult to proof due to scarce research which provided evidence and answers on research questions in relation to simultaneously executing

exploitative and explorative activities. Therefore, this research will investigate in more depth if BI usage is a driver, which positively relates to individual ambidexterity.

2.2 Business Intelligence

Business Intelligence and analytics and the related field of big data analytics have become increasingly important in both the academic and the business communities over the past two decades (Chen, et al., 2012). The further digitalization and availability of corresponding data

(15)

9 from multiple data sources opened the door and opportunities for firms to gain competitive advantage (Chesbrough, et al., 2006).

Business Intelligence (BI) is a grand, umbrella term, introduced by Howard Dresner of the Gartner Group in 1989, to describe a set of concepts and methods to improve business decision making by using fact-based, computerized support systems (Ghazanfari, et al., 2011). Business Intelligence tools allow employees to extract, transform and load data for analyses, and then make those analyses available in reports, alerts, and scorecards

(Davenport, 2006; Gollapudi, et al., 2012).

An advantage of BI tooling is the fact that it is able to cope with large amount of data, known as big data from multiple data sources (Davenport, 2006). This data is used as input for BI tooling, to control, monitor and adjust firms’ processes to enhance ambidexterity (Boe-Lillegraven, 2014). Furthermore, in case the data can be transferred into useful information, the information can be used by individual’s to support decision making, on whether to further explore or exploit activities (Lönnqvist & Pirttimäki, 2006). The characteristics of big data are described by Gandomi & Haider (2015) as data which has a high-volume, a high level of variety and a high velocity rate. BI tooling uses data which individuals can use to examine of how business opportunities are exploited and/or explored in real-time as well as

longitudinally (Boe-Lillegraven, 2014; Galbraith, 2014). On the one hand, (big) data is used for incremental improvement and optimization of current business practices and services. On the other hand, new products and business models can be innovated based on insights

gathered from the use of data (Hartmann, et al., 2014).

However, to be able to transform (big) data into useful information via BI tooling, measuring performance mechanisms are required (Nudurupati, et al., 2011). BI tooling enables measuring performances, which is known as the output on which decisions can be made (Popovič, et al., 2012). Based on these arguments, several researchers have made a link between BI tooling as a structural tool, which can be used as contextual element, by means of measuring performances and enhance decision making (Franco-Santos, et al., 2012).

The aim of using BI tooling in organizations, according Sharma & Djiaw (2011) is to create and transfer knowledge throughout the organization. The creation and transfer of knowledge will enhance better decision-making, improve productivity, exchange best practices and improves new product developments. This is in line with the findings by Popovič, et al. (2012), who states that BI enables an organization to innovate and learn in ways that increase organizational knowledge; support decision processes and enables firms to take effective actions. Table 1 provides an overview of similar and intertwined findings from

(16)

10 several prior researchers.

However, besides the organizational attributes, an individual’s experience and behavior towards using BI tooling equally affects enhancing ambidexterity (Nudurupati, et al., 2011). Bititci, et al. (2012) state that knowledge workers are more open to gain insights and

information from data analytics then general workers, since they are characterized as individuals with higher level of educational skills and have better capabilities to cope with complex activities. However, the intensity of using BI tooling by knowledge workers depends on the decision-making style and decision-making culture. An open, informal culture and perceived trust enables the intensity of using BI tooling (Popovič, et al., 2012). Individual’s which understand and trust the data of performance measurement systems are willing to promote, share and pro-actively encourage others to use this system as well (Nudurupati, et al., 2011). In case BI tooling is available for a large amount of individuals within a firm, they will experience a certain fairness in relation to each other, which enables learning and

problem solving (Franco-Santos, et al., 2012).

As mentioned by Boe-Lillegraven (2014) there is empirical evidence linking big data to firm productivity and profitability (e.g. McAfee & Brynjolfsson, 2012), but most of the research is anecdotal and case based. Another aspect mentioned by Franco-Santos, et al. (2012) and Gandomi & Haider (2015) are the possible downsides of using BI tooling. It can lead to wrong focus due to overload of information, change of misinterpreting the

information and lack of quality of the data. Taken these arguments in consideration, I expect that the causal relationship between BI usage and individual ambidexterity is positively related, since both BI and individual ambidexterity contain structural as well as contextual elements which might lead to simultaneously executing explorative and exploitative activities.

BI usage has been identified as an element which is used to support decision making, whereas individual ambidexterity is linked to alignment and adaptability. Furthermore, BI is used to measure performances, which might influence individual’s choices to either execute exploitative or explorative activities.

Therefore, this research will further extend and test the effects of BI tooling on individual ambidexterity. The following hypothesis will be tested:

Hypothesis 1: The use of Business Intelligence tooling positively affects the level of individual ambidexterity.

(17)

11 Table 1: Overview of prior research on business intelligence

Author / Year Theoretical Lens Unit of Analyses Key findings (Popovič, et al., 2012) Decision support systems

Firms BI usage is influenced by the quality of the information, access availability and decision making culture within firms.

BI definition: Qualitative information in well-designed data stores, coupled with business-friendly software tools that provide knowledge workers timely access, effective analyses and intuitive presentation of the right information, enabling them to take the right actions or make the right decisions.

(Ghazanfari, et al., 2011)

Evaluation of BI

Individual Regularly evaluating BI competences can advance decision support environments.

BI definition: BI systems refers to an important class of systems for data analysis and reporting that provide managers at various levels of the organization with timely, relevant, and easy to use

information, which enable them to make better decisions. (Chen, et al.,

2012)

BI and analytics Firms BI usage can be enhanced by training individuals as well as supporting a fact-based decision making culture.

BI definition: BI tools allow employees to extract, transform, and load data for analyses and then make those analyses available in reports, alerts, and scorecards.

(Sharma & Djiaw, 2011)

Knowledge management

Individual BI usage enhances knowledge sharing between individual’s, better decision making, sharing best practices, improve skills and encourage new product development.

BI definition: BI is a systematic process, by which knowledge needed for an organization to compete effectively, is created, captured, shared and leveraged

(Gollapudi, et al., 2012)

Architectural BI perspective

Firms An effective BI architecture should have

following attributes: Usability, common business view, scalability, reliability, leveraging existing infrastructure and be secured.

BI definition: BI is a computer based technique used in identifying, extracting and analyzing business data (Lönnqvist & Pirttimäki, 2006) Measuring BI usage

Individual Measuring BI usage provides useful information about the needs, information acquisition, analyses and utilization to improve the output of a

measuring performance system.

BI definition: BI is an organized process by which organizations acquire, analyze, and disseminate information from both internal and external information sources significant for their business activities and for decision making

2.3 Knowledge sharing

Within the current research era of learning literature, several researchers have found positive relations between knowledge sharing and the degree of learning from both organizational and individual level (Jansen, et al., 2009; Lichtenthaler & Lichtenthaler, 2009; Hirst, et al., 2009). Knowledge sharing can be defined as a social interaction culture, involving the exchange of

(18)

12 individual’s knowledge, experiences, and skills through the whole department or organization (Lin, 2007). Argote, et al. (2003) have analyzed multiple studies in the era of knowledge management. They argue that status, interpersonal interaction and incentives enhance individuals to share knowledge. Moreover, they identified gaps in current knowledge

management literature and suggest further research to explain how knowledge is embedded in an organization’s memory, whether knowledge persists through time or whether it depreciates and whether an individual’s experience has effects on learning.

(Mom, et al., 2009) argue that top down knowledge inflows are related to enhance exploitation as whereas bottom-up and horizontal knowledge transfer enhances exploration. This argument indicates there is a need for organizations to support and facilitate mechanisms which enhances knowledge sharing to achieve ambidexterity. BI tooling enables individuals to share their outputs dynamically and communicate novel information through the digital highway (Popovič, et al., 2012; Nagash, 2004). Furthermore BI tooling is fed by large amounts of data from multiple sources Nudurupati, et al. (2011). Therefore it can be argued that new knowledge can be generated by making use of the tooling. In addition, BI usage provides insights based on findings through measuring performances and supports decision making according Lönnqvist & Pirttimäki (2006) and Hartmann, et al. (2014), which might lead to transferring and sharing knowledge to strengthen and enrich their ideas for both explorative and exploitative purposes.

Gibson & Birkinshaw (2004) argue that firms should focus on professional development and knowledge transfer to create an atmosphere that enables individual ambidexterity. Furthermore, they state that communication throughout the firm is an important driver for individual ambidexterity. In line with the argument made by Floyd & Lane (2000), learning occurs as novel information is communicated from operating level individuals to middle managers. In addition Lubatkin, et al. (2006) state’s that top

management teams, level of communication directly influences how individuals deal with the contradiction between explorative and exploitative knowledge. A research performed by Mom, et al. (2009), underpins these findings by arguing that the acquisition of knowledge, from other individuals in the same firm, influences an individual’s explorative and

exploitative activities.

Learning can be enhanced in case individuals are willing and able to exchange and share knowledge with each other (Raisch & Birkinshaw, 2008). In current research, two main streams have been identified. The first group argues that exploitation is associated with reusing existing knowledge and therefore firms should only focus on learning from

(19)

13 exploration (Katila & Ahuja, 2002). The other group agrees and follows March’s approach, by arguing that learning occurs anyhow and therefore focus on the type and degree of

learning (Raisch & Birkinshaw, 2008). Following this line of reasoning, Im & Rai (2008) and Burgelman (1991) argue that inter organizational relationships between individuals positive enhance knowledge sharing for both exploitative and explorative activities.

Another aspect to explain knowledge sharing among individuals is the level of internal connectedness. Internal connectedness encourages and enhances individuals to develop trust and cooperation, which facilities knowledge exchange (Jansen, et al., 2006; Mom, et al., 2009). These researchers also have mentioned the effect of cross-functional interfaces between individuals as driver in respect of knowledge sharing. Cross- functional interfaces increase corporation between individuals, since they think and act outside the narrow confines of their own jobs and take others interests, beliefs and perspectives into account (Floyd & Lane, 2000; Jansen, et al., 2009).

Based on these arguments, it can be argued that BI tooling could be a mechanism to support knowledge sharing. BI tooling is an interactive tool, which copes with both historical data and can simulate new scenario’s to generate new insights (Davenport, 2006). Therefore it foresees in capturing existing knowledge, create new knowledge, but above all it can be stored and reused at any time. BI tooling captures and stores the intellectual capital of a firm, which can be used at any time by individuals. Furthermore, another benefit of BI tooling is the fact that users are interconnected and can share their findings with each other (McAfee & Brynjolfsson, 2012). Given these various findings, it could be assumed that individual’s which are supported and provided with mechanisms to share and exchange knowledge become more ambidextrous, compared to individual’s which are less supported and given the right mechanisms to share knowledge. In sum, individual’s which score high in their

willingness to share and exchange knowledge due to supportive availability of BI tooling, might lead to higher abilities to cope with balancing exploitative and explorative activities. This leads to following hypothesis:

Hypothesis 2: There is a positive mediating effect of knowledge sharing between BI usage and individual ambidexterity

2.4 Proactive personality

An individual’s level of pro-activeness could contribute to their willingness to socially interact and share knowledge with others (Shin & Kim, 2015). Proactive behavior has been identified as one of the individual innovative behavioral constructs, which is positively

(20)

14 associated with higher expertise and problem solving skills (Campbell, 2000). Furthermore, individual’s innovative behavior has been extensive linked to explain the level of individual ambidexterity based on individual’s abilities, such as creating, introducing and applying new ideas once supported and facilitated by their firms (Gibson & Birkenshaw, 2004; Wu, et al., 2014) . Bateman & Crant (1993) describe a proactive person as someone who is not

constrained by situational forces, but rather looks for opportunities, shows initiative, and takes action to implement changes. Individuals who are highly proactive are likely to have a greater sense of work-role self-efficacy and ‘persevere until they bring about meaningful change’ (Seibert, et al., 1999: p. 417).

Since BI usage, provides individuals with new insighs, trends and opportunities, it can be argued that individual’s which are highly proactive are motivated to share and exchange their knowledge with others to find support for their initiatives to change and innovate. Within the current literature there is an absence of research of the association between BI usage and personal characteristics according to Jourdan et al.’s (2008) extensive review. This gap in current literature has also been acknowledged by Wang (2014) who suggests to further investigate these effects. To be able to investigate the above mentioned proposals following section will describe in more depth findings and results of current research in the era of proactive behavioral characteristics.

Individual innovation behavior is based on an individual’s engagement in generating and applying new ideas and approaches in the workplace (Wu, et al., 2014). Rogan & Mors (2014) argue when individual’s experience a high level of job autonomy and are able to divide their own time this positively strengthens their level of pro-activeness. These findings are in line with Gibson & Birkinshaw (2004), who address that a supportive organizational context of stretch, discipline and trust will encourage employees to take initiatives and are motivated to act spontaneously upon opportunities beyond their basic job requirements. Furthermore, proactive individual’s search for information, socially interaction and feedback in order to improve changing situations (Shin & Kim, 2015). Individuals benefit from the acquisition of different kinds of information (Floyd & Lane, 2000). This is in line with Mom, et al. (2009) statement that ambidextrous individuals refine and renew their skills, knowledge and expertise. In addition, it can be argued that individual’s which can be

characterized as highly proactive are motivated to support innovation for either explorative or exploitative purposes. This research therefore assumes that proactive personality strengthens both the willingness to share knowledge as well as become more easily ambidextrous. These assumptions lead to the following hypothesis:

(21)

15

Hypothesis 3a: Proactive behavior moderates the relation between BI usage and individual ambidexterity such that this relation is stronger for individual’s with proactive personality traits, but lower for individuals with reactive personality traits.

Hypothesis 3b: Proactive behavior moderates the relation between BI usage and knowledge sharing such that this relation is stronger for individual’s with proactive personality traits, but lower for individuals with reactive personality traits.

2.5 Role ambiguity

Individuals who fulfill multiple roles within a firm are willing to participate in activities that

involve adaptation to new opportunities, but are clearly aligned with the overall strategy of the firm (Birkenshaw & Gibson, 2004). Role ambiguity has been described by Kahn, Wolfe, Quinn, Snoek and Rosenthal (1964) as the single or multiple roles that confront the role incumbent, which may not be clearly articulated in terms of behaviors or performance levels. Breaugh & Colihan (1994) have further refined the definition of role ambiguity to job

ambiguity and indicate that job ambiguity possesses three distinct aspects: work methods, scheduling and performance criteria (Bauer & Simmon, 1994; p. 3).

In case individuals are encouraged and supported by the firm to make their own choices between alignment and adaptation oriented activities, they will act ambidextrous (Birkenshaw & Gibson, 2004). This provides an opportunity to investigate when individual’s, which are supported in the decision-making process of which the outcome affects them are supported to cope with role ambiguity, they are also likely to act more ambidextrous. One of the main features of BI usage is to support individuals in decision making. Therefore it can be argued that individuals scoring low on role ambiguity are likely to act more ambidextrous. This argument has been addressed by Sawyer (1992) as an opportunity for further

investigation, since the level of available information provides major opportunities to contribute to current research.

However, according Floyd & Lane (2000), fulfilling multiple roles can also lead to role conflict. Role conflict can occur due to differences in goal achievement, between individuals who wear two heads and due to conflicting behavior norms and priorities. Another downside has been identified by Bonesso, et al. (2014) which argue that in case individuals are facing unclear information about role perceptions or incongruity among different role perceptions they are less motivated and stimulated to act ambidextrous. Jansen et al. (2009) and Floyd & Lane (2000) argue that role conflicts can be minimized through informal social integration, through organizational control mechanisms

(22)

16 such as market control, bureaucratic mechanisms or clan controls. Bonesso et al. (2014) introduced the term “full personal ambidexterity”, meaning that individuals which have a broad prior work experience and have balanced emotional and social competencies can be classified as “full personal ambidextrous” and are therefore better capable of dealing with role conflicts. These arguments indicate that role ambiguity as a construct influences the relationship between BI usage and knowledge sharing such that individual’s scoring high on role ambiguity are searching for social interaction, by means of exchanging knowledge. This leads to following hypothesis:

Hypothesis 4a: Role ambiguity moderates the relation between BI usage and

individual ambidexterity such that this relation is stronger for individuals with lower role ambiguity, but weaker for individuals with higher role ambiguity.

Hypothesis 4b: Role ambiguity moderates the relation between BI usage and

knowledge sharing such that this relation is stronger for individuals with higher role ambiguity, but weaker for individuals with lower role ambiguity.

3. METHODOLOGY

This research further builds on current empirical research, to explain conflicting findings in respect to individual ambidexterity. The aim of this research is to find explanations for causal relationships between the variables as mentioned in the research model. This research has a deductive approach and data has been collected via an internet-mediated survey. A mono method research design is used, since this research is based on a single data collection technique (Saunders & Lewis, 2012).

Researchers commonly use surveys as a research strategy, since it provides the opportunity to reach many possible participants, it is efficient, cost effective and data can be collected in a relatively short period of time. Due to resource and time constraints, a cross-sectional approach has been applied. To collect the data, an online questionnaire has been designed and distributed via Qualtrics.

The questionnaire has been distributed towards possible participants working in companies, known in my network, which make use of BI tooling. The convenience sampling technique with a snowball effect has been applied to be able to increase the sample rate. The participants have been contacted via an email containing an invitation to participate.

(23)

17 knows the researcher, which limits the usability of the respondents sample to a larger

population (Saunders & Lewis, 2012).

The invitation contained information about the purpose of the research, why their participation is valued, instructions and what types of questions could be expected. To increase validity and check for possible bias, some generic control variables has been included in the survey, such as hierarchical level, educational level and frequency of usage. The survey consists of eight categories. One general category containing the control variables, and the other eight variables as mentioned in the research model, BI usage, proactive personality behavior, role ambiguity, individual ambidexterity, exploration activities, exploitation activities and knowledge sharing. The last seven variables all have been measured according a seven point Likert scale; from (1) strongly agree to (7) strongly disagree, as defined by (Bateman & Crant, 1993).

3.1 Data and Sample

The online questionnaire has been sent through a direct invitation to 97 persons within my personal network. The other questionnaires have been forwarded in 4 companies through relatives and known members from my personal network. The number of forwarded

invitations is therefore unknown. The original sample size of this research was N=161. After validating the completeness of the questionnaire through frequency distribution, performed in the statistical software tooling “Statistical software Package for Social Sciences (SPSS) 23 (16%) respondents were dropped out of the sample, since they responded on less than 20% of the questions. Therefore the sample size for performing analyses is N = 138. The respondents average completion time to complete the 67 questions was 6.43 minutes, which is an average duration for completing an online questionnaire.

The online survey has been held between 30th of April and 1st of June 2016. After this date, late participants haven’t been taken into account for the data analyses.

3.2 Measures

All measurements for this research were based on prior or related performed quantitative research to ensure validity and reliability of the scales. However, since the unit of analyses of some of the used researches differed, the questions in the questionnaire have been adjusted to fit for individual measuring purposes. Furthermore, the questionnaire has been distributed among both English native speakers as well as Dutch participants. Therefore the

(24)

18 participant.

The questionnaire has been sent to four participants to pre-test the content of the questionnaire. The goal of this pilot is threefold: (1) get an indication of the expected duration of full filling the complete questionnaire, (2) to check if the questionnaire is complete, and (3) test the internal validity of the translated measures. The questionnaire has been sent to two English native speakers and two Dutch participants. Based on the results and feedback of these four participants, no changes have been made to the questionnaire. The invitation letter and a complete list of the scale items, including translations, can be found in Appendix A: Invitation letter & Survey items.

Dependent variable: Individual ambidexterity. Individual ambidexterity has been measured based on two dimensions as developed by Mom, et al. (2009) to measure for an individual’s engagement in exploration and exploitation activities. Both exploration and exploitation were measured through seven items. They validated their construct through in-depth interviews, and confirmatory factor analyses. Items to measure exploration contain elements such as renewal aspects, policies and procedures to follow and searching for new business opportunities. Exploitation items contain aspects of routines and acting upon individual experiences. Although the separate items have been proven to be reliable in the research of Mom, et al. (2009), I performed a Cronbach’s alpha (α) analyses to check for reliability and bias in this particular research context. The items for exploration scored (α = .648) and for exploitation (α = .739) as reported in table 4. According Hair, et al. (1998) a minimum of (α = .70) of each item can be considered as a reliable measure. This threshold isn’t achieved for exploration activities, however the corrected item-total correlations indicate that all the items have a good correlation with the total score of the scale >.30. In line with Mom, et al. (2009) the constructed dependent variable individual ambidexterity has been calculated by

multiplying the constructs for exploration and exploitation which results in a Cronbach’s alpha of (α = .707). Furthermore prior studies such as Gibson & Birkinshaw (2004); He & Wong (2004) and Lubatkin, et al. (2006) also combined exploration and exploitation measures to assess ambidexterity and reported higher levels of Cronbach’s alpha, which indicates the scale as reliable.

Independent variable: BI usage. To measure an individual’s attitude and experience with BI

usage a selection of 10 items from two scales have been selected. Since this research takes both structural and contextual elements into account, BI usage has been measured on two

(25)

19 scales, supportive decision making and measuring performances. These factors have also been acknowledged by Lönnqvist & Pirttimäki (2006) as suitable for measuring BI usage as driver for supporting decision-making and measuring performances. These two scales are used in this research since it captures the two core features in which BI usage foresees. Since the items for both aspects primarily focus on the experience of usage, they’re measured in a single construct.

For supportive decision making this study developed a four-item scale based on inputs and suggestions as mentioned in current decision support systems research (Ghazanfari, et al., 2011; Neely, et al., 2000; Braz, et al., 2011). This has been done, since no prior empirical research measured for this construct in this particular context, however the four items are in line with the directions and suggestions as made by above mentioned researchers. An example of an item states: ‘I make decisions, based on the outputs retrieved from BI analyses’.

To measure for the effects of BI usage on measurement performance output, a six-item scale adopted from Berman & Wang (2000) has been used. These six-items have been selected from their section containing “outcomes of performance measurement”. Their research isn’t however linked to BI, but captures the aspect of measuring performances as best alternative for the context of this research. Therefore, the items have been rephrased slightly to make them suitable and linked with BI for this research. An example of an item, as included, is; ‘Based on the output of BI usage, I adjust my working routines’.

The computed mean scale proved to be reliable with a Cronbach’s alpha of (α = .788).

According Field (2013) a Cronbach’s alpha value higher than (α > .70) indicates a strong and reliable scale.

Mediating variable: Knowledge sharing. To measure for the effects of knowledge sharing,

two lower order constructs have been chosen, (1) tacit knowledge sharing and (2) Trust in co-workers. The measure for tacit knowledge sharing captures the extent to which individuals have the capabilities and willingness to communicate (Katila & Ahuja, 2002; Floyd & Lane, 2000). According Lin (2007) tacit knowledge sharing is affected by distributive justice and instrumental ties, which BI tooling both provides. Therefore the four-item scale measuring tacit knowledge sharing of (Lin, 2007) is used for this research.

The measure for trust in co-workers captures the extent to which individuals engage in the support for social interactions and openness towards connectedness (Floyd & Lane, 2000; Argote, et al., 2003). According Lin (2007), trust exists when individuals perceive that

(26)

20 their co-workers possess such qualities of trustworthiness and believe that the co-workers would repay them by doing the same thing when they share knowledge with others (p. 415). Since BI tooling provides the opportunity to share outputs and is available for large amount of individuals, which enhances fairness, measuring for trust in co-workers seems appropriate for this research. Therefore, the 5-item scale to measure for trust in co-workers has been adopted from (Lin, 2007).

Since knowledge sharing, is a second order construct, reliability checks were performed on each of the individual items, before computing a mean score for knowledge sharing. All items turned out to be reliable since Cronbach’s alpha scores (α > 0.70).

Moderating variable: Proactive personality. Within the research era of measuring

individual’s proactive behavior many constructs have been used, such as differences in proactive personality, personal initiative, openness towards change and self-efficacy (Crant, 2000). According Wang (2014), an individual’s proactive personality affects the impact, towards presenting and sharing outcomes of BI usage, with others. This is line with the findings of Wu, et al. (2014), which used proactive personality as a control variable to measure for the willingness to share and exchange knowledge with others. Although both, Wang (2014) and Wu, et al. (2014) haven’t used proactive personality as a direct moderator between BI usage and knowledge sharing, there are elements and arguments to incorporate proactive personality as a moderator in the line of reasoning of this research.

According Shin & Kim (2015) individual’s pro-activeness could contribute to their willingness to socially interact and share knowledge with others. Furthermore, these

individuals actively seek for opportunities to cooperate with others, and search for interaction to share efforts (Birkenshaw & Gibson, 2004). To identify if proactive personality positively moderates the effects between BI usage and knowledge sharing a 9-item scale was used (Seibert, et al., 1999). This scale is a shorten version of Bateman & Crant (1993) measure for proactive personality and captures individual’s dispositions towards proactive behavior. This shorten version has been chosen to secure an adequate response rate and have been measured on a 7-point Likert scale. ‘No matter what the odds, if I believe in something I will make it happen’ is an example of an item included in the measuring instrument.

Moderating variable: Role Ambiguity. Individuals who are confronted with role ambiguity,

due to fulfilling multiple roles, might face tension due to differences in goals achievement and prioritizing (Floyd & Lane, 2000). Individuals with high level of role ambiguity are

(27)

21 uncertain about their contribution to align their activities and balance exploitative and

explorative activities. Furthermore, they also face difficulties how to initiate and

communicate constructive changes. Role ambiguity exists when individuals lack information about their responsibilities within the organization. Since this research focus on the level of individual’s ambidexterity, role ambiguity as a perceptual construct could be relevant.

To measure the level of role ambiguity the 8-item scale from Fuller, et al. (2006) is used. This measure is chosen, because it captures the construct, focused on lack of information, which is related to both BI usage and knowledge sharing. ‘I have to buck a role or policy in order to carry out an assignment’ is an example for a measured item.

Control variables: To control for inconsistent effects the variables, educational level,

hierarchical level and frequency of usage have been included in the questionnaire.

It is likely that the degree of education affects the cognitive abilities, which influences the level of ambidexterity. Additionally, it can be argued that higher educated individuals can cope with complex situations, interpret large amounts of information and are willing and open to share this with others (Lin, 2007; Bititci, et al., 2012). This item has been measured as: (1) Master degree, (2) Bachelor degree, (3) Other degree.

Secondly, the hierarchical position of an individual within an organization might influence an individual’s behavior and choices with respect to knowledge sharing and

balancing exploitative and explorative activities. Raisch & Birkinshaw (2008) assume that an individual’s position within a firm (manager – no manager) affects the opportunities to

socially interact with other organizational members and is able to delegate activities to others. Therefore a question have been incorporated in the questionnaire to verify if an individual (1) has a managerial position), (2) has no managerial position.

Last but not least, the frequency of BI usage has been measured for as control variable. Firstly, this variable measures if the participant is suitable to include as sample in the analyses. Secondly, it also enables to identify and check participant’s intensity of using BI tooling. This item has been measured on a 4-item scale (1) Never, (2) On a daily basis, (3) On a weekly basis, (4) On a monthly basis.

3.3 Key informant check

To check if participants match with the aim of this research, a key informant check has been executed. Kumar, et al. (1993) described the term ‘key informant’ as, an individual who is particularly qualified to answer questions regarding the topic of investigation. Since the unit of analyses of this study is at the individual level and is conducted through a diverse

(28)

22 population, it makes sense to check for the frequency of BI usage. All of the participants in the sample size answered to the question ‘How frequent do you make use of BI tooling?’ Four participants answered they never used BI tooling and were therefore excluded from the sample size.

3.4 Common method bias

The method used to collect the data for this research is subject to common method bias (CMV), because all variables are measured in a single survey at the same time (Campbell & Fiske, 1959). In order to reduce this bias, the survey has been designed in such a way, that both dependent and independent constructs have been asked in various sections. In order to verify if no CMV occurred, the Harman’s single-factor test was conducted after collection of the data. This test assumes that common method bias is present, when a factor analyses shows that there is a single factor that accounts for >50% of the variance. The test result shows that the strongest factor explains 13.83% of the variance, whereas 18 items with an Eigen Value >1 explained 79.79% of variance. These results indicate that CMV is not present for all items in the data set.

4. DATA ANALYSES & RESULTS

This chapter provides an overview of the results based on various analyses. Based on the outputs, the suggested hypotheses were either accepted or rejected. First, the descriptive statistics are reported, including the process of data validation and data screening. The next section reports the reliability of all computed variables. The third section contains an overview of the correlation analyses and reports the most relevant significant correlations. Fourth, results of normality analyses are reported and finally the results of multiple hierarchical regression and post-hoc analyses are provided.

4.1 Descriptive statistics

An overview of the descriptive variables, including skewness and kurtosis values is shown in Table 2. As mentioned in previous section there was a dropout rate of 16%. Since a

convenience sampling technique with snowball effect is used to collect the data, a response rate cannot be calculated. The sample size for executing analyses is for all items measured N=138.

Since all questions have been asked on a seven point Likert scale from (1) Strongly agree to (7) Strongly disagree, all items measured in the constructed variables have been

(29)

23 recoded reversely to interpret the data more easily. The negatively-keyed items technique provides the opportunity to detect acquiescence bias, for this research none have been included since most items measured were used in prior studies and haven’t been negatively-keyed either. The new developed questions are set up in line with all other questions. But more important, the completeness of surveys has a higher change in case of convenience and user friendliness for the participant. Because the population of which data could be collected is limited, no negatively-keyed items have been incorporated.

To check for missing data values, frequency analyses have been performed. For 6 items missing values have been identified. For each of the 6 items, less than 7 values were missing, which is in line with the threshold of <10% missing values as identified by (Myers, 2011). This allowed them to be replaced with new values using Hotdeck. Hotdeck is an imputation method with replaces a missing value with the value of a similar ‘donor’ in the dataset that matches the ‘donee’ (Myers, 2011). However for all above mentioned missing values no suitable donor deck could be identified and therefore the missing values were replaced by the mean of each unique item. A downside of this method is the fact that the means could suppress the true value of standard deviations and errors. In order to retain a sufficient sample size and the number of missing values is small compared to the actual size, it seems justified to use this method (Field, 2013).

Overall, all variables show a relatively high mean, which indicates a positive attitude towards the measured, constructs. Especially the construct knowledge sharing (mean = 5.73) indicates that individuals are open and willing to exchange and share knowledge.

Table 2: Descriptive Continuous Variables

Besides the continuous variables, Table 3 shows the frequencies of the control variables. The majority of the participants 93% finished at least a Bachelor degree or higher and 53% reported to use BI tooling on a daily basis. Concerning the participant’s current position it’s almost equally divided.

Variables N Min Max Mean Std.

Deviation Skewness Kurtosis Individual Ambidexterity 138 14 40 29.03 6.13 -.39 -.68 Exploration Activities 138 3 7 5.38 0.70 -.68 .56 Exploitation Activities 138 3 7 5.38 0.82 -.71 -.19 BI Usage 138 2 7 5.2 0.74 -1.19 3.23 Knowledge sharing 138 3.11 6.67 5.73 0.64 -1.76 3.83 Proactive Personality 138 4.11 7.00 5.76 0.53 -.62 .75 Role Ambiguity 138 2.57 6.71 5.38 0.85 -.86 .30 Descriptive Statistics

(30)

24 Table 3: Frequencies Control Variables

4.2 Reliability analyses

The items for the different construct variables where chosen from existing literature and checked for reliability. To check for reliability the Cronbach’s alpha (α) were calculated in SPSS. Table 4 provides an overview of the results of the reliability analyses. From the 67 questions as asked in the survey, 4 control variables and 10 items to measure environmental dynamism have been left of this research.

Table 4: Reliability Statistics

Since all constructed variables, except exploration activities, have an (α > .70), I choose not to remove any items from these constructs to achieve higher reliability rates. However, what is quite surprising are the lower Cronbach alpha values for exploration activities. This construct has been proven to be reliable in the research performed by Mom, et al. (2009), in where respectively (α = .90) have been reported. Therefore an exploratory factor analyses have been performed to check for interdependence among the constructed variables.

Variables Level Frequency % Valid % Cumulative %

Master degree 66 48 48 48 Bachelor degree 62 45 45 93 Other degree 10 7 7 100 On a daily basis 73 53 53 53 On a weekly basis 41 30 30 83 On a monthly basis 24 17 17 100 I'am a manager 62 45 45 45

I'am not a manager 76 55 55 100

Educational level Frequency BI usage Hierarchical level Variable Cronbach's Alpha N of Items Individual Ambidexterity .707 14 Exploration Activities .648 7 Exploitation Activities .739 7 BI Usage .788 10 Knowledge Sharing .785 9 Proactive Personality .734 9 Role Ambiquity .755 8

(31)

25 Since the Sample size N = 138, and number of items is 7 as a minimum of N = 70 is suitable for performing an Exploratory Factor analyses test (Nunnally, 1978). Although the

correlation matrix shows for some items (p < 0.3), the Kaiser-Meyer Olkin Measure reports a value of (KMO = .519), which is above the boundary as identified by Kaiser (1974), but still on the low side of the spectrum. In Appendix B the output’s for these analyses are reported. Additionally, all items have been checked if removal increases the constructs

Cronbach’s alpha. This is not the case, therefore it has been decided to continue with all items in this construct. Even though this study reports (α < 0.70) for exploration activities, Mom, et al. (2009) reported sufficient construct validity, unidimensionality, convergent and

discriminant validity to have confidence in this construct. A possible reason for the found discrepancies could be the difference in sample size. As mentioned previously, this research focusses primarily on individual ambidexterity and this construct reports an acceptable Cronbach’s alpha value.

4.3 Correlations

The bivariate correlations between all subjected variables were computed in order to measure the strength of the relationship between the variables. Although the control variables could be indicated as categorical variables, they have been measured on a scale level in SPSS, to be able to classify all correlations. The coefficients, which were significant at the 1% and 5% levels, are marked with stars and shown in bold font. To quantify the intensity and meaning of the relationships between the variables under investigation the mean, standard deviation and Pearson correlation coefficients are reported in Table 5.

In Table 5, we can observe strong positive relationships between dependent variable Individual ambidexterity and both moderators, role ambiguity (r = .468; p < .01) and proactive personality (r = .574; p < .01). Additionally a positive significant relationship is observed between mediator knowledge sharing and moderator proactive personality (r = .283; p < .01) and role ambiguity (r = .382; p <.01. This indicates that both moderators are strongly correlated with the constructed variable knowledge sharing.

Another interesting finding is the significant negative correlation between educational level and proactive personality (r = -.242; p < .01).

(32)

26

4.4 Normality analyses

To test to which extend the variables are normally distributed, I have verified the skewness and kurtosis values as reported in Table 2 and the Kolmogorov-Smirnov statistics as shown in Table 6 on the next page.

For all constructed variables a negative skewness has been found, meaning the mean is less than the median, which indicates that most respondents score higher on agree. For BI and knowledge sharing skewness is less than -1, which indicates highly skewness. For these last two variables a kurtosis higher than 3 is reported and can therefore be rated as leptokurtic, indicating strong peaks at the mean of the distribution. In Appendix C, histograms and Q-Q plots are reported. Additionally Kolmogorov-Smirnov tests have been assessed to verify the normality distribution. In case the results of these tests are significant (p <. 05) the normality assumption is violated (Field, 2013). As shown in Table 6 all constructed variables report values (p < .01), which means all variables are significantly non-normal distributed.

However, since all variables are negatively skewed, there is uniformity for all constructs. All normal Q-Q plot’s show a relatively straight line, which indicates a relatively normal

(33)

27 Table 5: Means, Standard Deviations, Correlations (2-tailed)

Table 6: Normality analyses – Kolmogorov-Smirnov Statistics

Variables Mean SD 1 2 3 4 5 6 7 8 9 10 1. Individual Ambidexterity 29.03 6.13 (.707) 2. Exploration Activities 5.38 0.70 .693** (.648) 3. Exploitation Activities 5.38 0.82 .803** .136 (.739) 4. BI Usage 5.22 0.74 .294** .236** .225** (.788) 5. Knowledge Sharing 5.73 0.64 .277** .127 .292** .238** (.785) 6. Proactive Personality 5.38 0.85 .574** .227** .623** .302** .283** (.734) 7. Role Ambiquity 5.76 0.53 .468** .297** .416** .318** .382** .403** (.755) 8. Educational Level 1.59 0.62 -.146 .015 -.198* -.141 -.042 -.242** -.122 --9. Frequency BI usage 2.64 0.76 .202* .144 .173* .025 .318** .112 .139 .002 --10. Hierarchical Level 1.55 0.50 -.077 -.153 .026 .005 -.166 -.037 -.113 .160 .057

--**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

Statistic df Sig. Individual Ambidexterity .112 138 .000 BI Usage .112 138 .000 Knowledge Sharing .206 138 .000 Role Ambiguity .154 138 .000 Proactive Personality .106 138 .001 Variable Kolmogorov-Smirnov a

Examining the level of individual ambidexterity : the role of business intelligence usage, knowledge sharing proactive personality and role ambiguity