1
Amsterdam Business School
Master Thesis
A study about the relationship between knowledge intensity and a
professional’s individual performance
. The partial mediating effect of management control systems in a professional service firm context.Name: Luke N. Vella Critien Student number: 10853936
Thesis supervisor: Prof. Dr. D.M. Swagerman Date: April 2016
Word count: 19479
MSc. Accountancy & Control, specialization Control
2
Statement of Originality
This document is written by student Luke N. Vella Critien who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
3
Abstract
This study investigates the relationship between the degree of knowledge intensity in
professional service firms and the professional’s individual performance, taking into account the use of different types of management controls systems and the degrees of tightness to manage and influence employee’s behavior and performance. This study controls for organizational size and employee working experience. The final theoretical model, presented in figure 2, was empirically tested with survey data containing 371 responses gathered and facilitated by a research project at the University of Amsterdam. Primarily, this study draws on current literature concerning knowledge intensive firms and their performance outcomes. In addition, this study uses previous findings in the management accounting literature concerning the use and application of the contingency theory. This study provides a rather new perspective on how this contingency approach can or should be used in the context of the professional service firm. To test the formulated hypotheses a multiple regression analysis with simultaneous mediation was performed. The results indicated that (1) knowledge intensity significantly and positively influences the individual performance of professionals, (2) knowledge intensity significantly influences the design of the organization’s management control system, (3) management controls systems, namely positively and negatively via behavioral controls, and results controls, respectively, mediate the relation between knowledge intensity and individual performance, and (4) when controlling for organizational size and working experience the results of the hypothesized relationships stay approximately the same in significance and coefficient weight (with the noteworthy exception that when excluding the control variables size and working
experience from the regression analysis, explicit personnel controls have a significant positive mediating effect on the relation between knowledge intensity and individual performance).
4
Preface and acknowledgments
This thesis was written in order to finish and obtain the master’s degree in Accountancy & Control at the University of Amsterdam. Without the help of many the completion of this thesis would be not possible all together. I would therefore like to offer my sincere gratitude to all of my family members, friends, fellow students and teachers whom participated or sustained me during this period. I would like to thank all members of the research project, whom gathered the empirical data, for participating and generating such a strong survey response. I also emphasis my gratitude towards the project coordinator: Helena Kloosterman for being such a kind and helpful individual. I am everlasting grateful for the immeasurable patience, support and above all, the love that my mother Irene Vella Critien-Vassallo has given me during these past months in which I could write and finish my thesis. Also I would want to thank my thesis supervisor: D.M. Swagerman for his sustaining contact and positive mindset. Last but not least, my praise goes to God, without whom nothing was and will never be possible. I humbly thank you for providing me with the physical and mental strength and capacity to successfully complete this phase in my life.
Luke Nicholas Vella Critien Almere, the Netherlands April 2016
5
Contents
Abstract ... 3
Preface and acknowledgments ... 4
1. Introduction ... 6
2. Theory, hypotheses & framework development ... 8
2.4. Hypotheses and research framework ... 12
2.4.1. Hypotheses ... 12
2.4.2. Theoretical framework ... 17
3. Research method ... 18
3.1. Survey procedure and data collection ... 18
3.2. Data screening ... 19
3.3. Exploratory factor analysis ... 20
3.4. Response analysis ... 33
4. Results ... 34
4.1. Multiple regression analysis ... 34
5. Discussion and conclusions ... 46
5.1. Conclusions ... 46
5.2. Limitations study ... 50
5.3. Future research ... 50
References ... 52
Appendix A. Summated Survey Questions ... 61
6
1. Introduction
This paper tries to answer the research question; to what extent does knowledge intensity or human capital influence the professional’s individual performance and how is this relationship affected by the degree of management controls system tightness (hereafter MCS) in a professional service firm context (hereafter PSF). Specifically this paper tries to determine the causal relationships between the presence of knowledge intensity and the individual performance of the professionals. During the investigation of this relationship an account is taken of the mediating effect which the degree of MCS tightness has on the above hypothesized relationship. This study tries to contribute and increase the rather limited body of knowledge and understandings in the management accounting literature, concerning the phenomena of how knowledge intensity and PSF’ performance interact and how MCS mechanisms fit in this picture. This study is driven by the fact that the PSF sector is one of the strongest growing and most profitable industry in the global economy (Walker, 1985, Winch & Schneider, 1993, Empson et al., 2015). When looking at the recent past one can clearly draw a conclusion that there is an economical shift visible which is characterized by knowledge-based services whom are gaining considerable ground or even becoming an important actor into diminishing the industrial-based manufacturing firms (Walker, 1985, Powell & Snellman, 2004, Anand et al., 2007, Winch & Schneider, 1993). The economy is becoming more convincingly and strongly characterized as being a knowledge based one (Nordenflycht, 2010, Empson et al., 2015, Crook et al., 2011). Supported by the fact that knowledge is one of the most prevalent or
representative characteristics of a PSF, it is clear that these PSF are becoming more and more seen as representatives or so called role models of this new economy (Nordenflycht, 2010, Empson et al., 2015, Winch & Schneider, 1993, Grant, 1997). This increased perceived importance of these ‘new’ PSF is the driving force behind this research, which will be aimed at enhancing the current
understandings in this specific research field.
This paper draws on the contingency theory to interpret the findings regarding how the PSF’ characteristic of knowledge intensity acts as an essential contingency variable to the design and shaping of the organizational MSC. This research paper will base its findings and conclusions on an empirical analysis which draws data from a questionnaire taken among a wide array of PSF’s active in multiple service industries across multiple countries.
The remainder of this research paper is structured as follows. First, section two gives insight into the current management accounting literature concerning the theory of knowledge intensity, the
7 use, design and the so called ‘fit’ of MCS. Lastly, the contingency theory and its importance, which this paper makes use of, is discussed. Consequently the developed theoretical framework is
presented, which is followed by the formulation of the hypotheses. Next, in section three, the methodology of the empirical research process is provided. Consequently in section four the
empirical findings are presented. Finally in section five, the discussion, conclusions and the research limitations and possible future research ideas are discussed.
8
2. Theory, hypotheses & framework development
The first three parts of this section include an overview of the current findings in the
management accounting literature concerning knowledge intensive firms, the design and use of MCS, and the applicability of the contingency theory. In the fourth part of this section the hypotheses, the theoretical framework and a visualization of the expected relationships are presented.
2.1. Knowledge intensity
The current body of literature covering the complex system of MCS and the related elements of human capital-knowledge intensity in the context of PSF is still rather undeveloped. According to many researchers, organizations are becoming much more intensive and knowledge-driven (Walker, 1985, Powell & Snellman, 2004, Anand et al., 2007, Winch & Schneider, 1993, Nordenflycht, 2010, Empson, 2015, Grant, 1997). A knowledge-intensive firm is an organization where human capital and thus human knowledge and experience is the primary and most essential input needed to provide a service (Anand et al., 2007, Starbuck, 1992, Crook et al., 2011). PSF’s do not sell any tangible goods like finished products or materials, rather they sell their know-how, their expertise (Treem, 2012). Knowledge-intensity or human capital can be described as knowledge, skills, capacities or abilities that reside within an individual (Schultz, 1961). Anand and others (2007) claim that a firm can be labeled as knowledge intensive if the use of knowledge, in providing services or products, is the most important input for the firm in comparison to any other input, either being a psychical or non-tangible resource that is available to a firm. Other researchers stress the rather ambiguous character of the term and phenomenon which is knowledge (Anand et al., 2007, Starbuck, 1992, Hinings & Leblebici, 2003, Alvesson, 2001). The concept of knowledge in the context of PSF should therefore be made tangible by further investigating this phenomenon. It is essential to
comprehend the importance of knowledge when trying to enhance the understandings about the relationship between the presence and role of knowledge in a PSF. The presence of knowledge or human capital in organizations can lead to sustainable competitive advantages and can stimulate a service firm’s performance (Morris & Empson, 1998, Wright et al., 1994, Hitt et al., 2001, Crook et al., 2011, Coff, 1999, Park et al., 2003). There is an economical shift visible, which is driven by the growth of new industries, where knowledge-based services are gaining ground or even replacing manufacturing or industrial-based production firms (Walker, 1985; Powell & Snellman, 2004; Anand et al., 2007, Treem, 2012, Ekstedt, 1989). These PSF are becoming role models or representatives of
9 this growing knowledge-based economy (Nordenflycht, 2010; Empson, 2015, Winch & Schneider, 1993, Grant, 1997). According to Powell & Snellman (2004) an essential or characterizing
component of this so called ‘knowledge economy’ is the weight and importance that is placed on the intellectual capacities of humans rather than the use of physical inputs or other natural resources. Morris & Empson (1998) mention the relevance and importance of specific human capital or knowledge in the existence of and performance delivering capabilities of PSF. In the current literature there is some evidence available which contains a rather mixed point of view concerning the role that knowledge plays in a firm. Whilst there is current evidence on the relationship between human capital and organizational performance, which was found to be positive (Youndt & Snell, 2004), the current literature mostly looks at other relational aspects of knowledge. For example Erden et al. (2014) found evidence supporting the claim that an enhanced flow and transfer of knowledge between parties positively affects the firm performance. Whilst other researchers claim that
knowledge intensity can act as a positive mediating actor between human capital information
disclosure and firm performance (Lin et al., 2012). Yet others like Hitt et al. (2001) state that human capital has both a direct and indirect effect on a firm’s performance. Also Germain et al. (2001) and Skaggs & Youndt (2004) found evidence stating that knowledge or human capital could act as a mediating variable, with a consequently positive effect on the organization’s performance. But there is also contradictory evidence available in the literature. Shrader & Siegel (2007) claim that a negative relationship exists between (previous) obtained industry experience and firm performance.
From the above mentioned findings it becomes comprehensible that, currently, there is no clear and specific evidence regarding the relationship between knowledge intensity and the
professional’s individual performance, in specific, in a PSF context. This paper tries to contribute to the literature by adding, more, and in depth knowledge in this specific research area.
2.2. MCS design, use and fit
The literature puts forward the question how PSF should be managed and organized (Morris & Empson, 1998). When looking at the relationship between PSF performance and knowledge intensity, we have to take into account the managing and controlling of these professionals with the use and design of MCS. According to Coff (1999) human assets in service firms will most probably cause management control issues, due to management not having the right (complex) knowledge to control their professionals. This paper draws on the four types of controls, which were developed by
10 Merchant & Van der Stede (2012), which make up the MCS of an organization. They categorize the management controls into four distinct categories, namely: the so called ‘hard controls’: behavioral or action controls and results controls, and the so called ‘soft controls’: cultural and personal controls. There are many different interpretations on the meaning or definition of tightness of a MCS. Some state that a MCS can be regarded as being tight when there is a frequent monitoring of the employees activities (Anthony et al., 1992). Whilst others claim that controls tightness is represented by the degree to which there is a good presence of goal congruence; the scenario where employees act in the organizations best interest and prioritize the achievement of organizational objectives (Merchant & Van der Stede, 2012). Amigoni (1978) views the tightness or looseness of controls as the style of control. He states that a more dominant presence of the administrative perspective of controls and a weak or low participation in target setting are regarded as being characteristics of a relatively tight control system. Whereas loose control systems are characterized by an high employee target setting participation and a more dominant presence of the social and individual perspective of controls. These views are very similar and in line to the definition of MCS tightnesss of Hopwood (1973). Finally, Whitley (1999) interprets tight controls as the situation when decision rules are very accurately defined whereas loose controls are characterized as being more autonomy giving or accepting deviations from the norm when making decisions. The above mentioned authors all have their own interpretation concerning the definition of tightness of a MCS. Although it is clear that there is a spread in the opinions concerning the definition of tightness of a MCS, these interpretations also have some overlap as can be seen above. This research paper sets its own definition for the term tightness of a MCS. In this paper tightness is defined as the degree of flexibility in the MCS.
Tightness can be presented in two ways; firstly via an increased covering or scoping of the MCS and secondly via an expansion of the acceptance and tolerance of deviations from the MCS norm or standard. The degree of flexibility of the MCS is represented by the implicit and explicit tightness. Firstly, implicit tightness can be described as a form of tightness that is created by decreasing the level of tolerance for deviations from the MCS. In other words, a more implicit tightness is
represented by the control system as being made less forgiving. Whilst, secondly, explicit tightness is characterized by a stronger presence of more rules, procedures and routines and more controls in general, and thus results in a MCS that has a lower flexibility. Prior studies have proven that the design and consequently the MCS’s tightness depends strongly on the organizational structure, its
11 human capital or intellectual asset intensity and should be adjusted to fit accordingly (Coff, 1999, Ditillo, 2004, Herremans & Isaac, 2005, Herremans et al., 2011).
2.3. Contingency theory
This paper draws on the contingency theory which will function as a tool and framework to interpret and understand the findings concerning the effects that the shaping and design of the MCS has on the relationship between the contingency variable; knowledge intensity and the outcome variable; the professional’s individual performance. The contingency theory is a specific approach to an organizational control system application and design, which, in general, states that there is no universally appropriate management accounting or control system for all organizations. The appropriate control system will depend on specific circumstances and specific aspects of the
organization (Otley, 1980). Gerdin & Greveb (2004) investigate the so called ‘fits’ of MCS’s which are currently available in the management accounting literature. They find that there are several distinctions or approaches to applying the contingency theory in the design and choosing of a
management accounting systems strategy. There are two main approaches to the forms of ‘fit’ which are the cartesian approach and the configuration approach (Gerdin & Greveb, 2004). These two approaches can be characterized by either being seen as traditionalists whom view the fit between the organizational structure and context as a constant changing factor that provides a mechanism through which incremental changes or movements of organizations are accepted (Donaldson, 1996 in Gerdin & Greveb, 2004). On the other hand the contrasting proponents advocate the configuration approach. This approach claims that there are only a few states of MCS ‘fit’. Organizations have to make significant changes or advances to change from one state to another (Mintzberg 1983). This paper draws on the contingency theory which is developed and used by Mintzberg (1979). Mintzberg (1979) describes several contingent variables which are regarded as independent variables which influence and determine the design and structure of the organization. The design of these variables is contingent on the specific organizational situation. He mentions several contingency variables like organizational size, age, ownership, and other actors which influence the structure and composition of the organization and consequently its management accounting system (hereafter MAS). According to Mintzberg (1979) there is no universally acceptable or general ‘one size fits all’ MCS framework or set of controls. He states that the MAS or MCS should be designed and based on several
12 or an addition to the current body of literature concerning the applicability of the contingency theory. I would want to expand this theoretical framework by claiming that the contingency theory, when used by an organization, specifically when applied in a PSF research context, needs to include the contingent variable knowledge intensity when looking at the design and shape of an organization’s MCS. Prior research has also found evidence for this so called contingency fit relationship with MCS and the performance of a firm. Pernot & Roodhooft (2014) found evidence stating that when the MCS fit is low, the organizational performance will suffer.
2.4. Hypotheses and research framework
This section of the study contains the developed hypotheses which are used to perform the statistical analysis of the research. This section also contains a visual representation of the
hypothesized relationships in the shape of a theoretical framework presented in figure 1. 2.4.1. Hypotheses
Since every organization is unique, its organizational structure will vary and therefore also its management control mechanisms needs (Waterhouse & Tiessen, 1978). For this research paper the following hypotheses were developed which represent the expected relationships between the
variables: knowledge-intensity, design, fit or tightness of the MCS and the individual performance in the PSF. This research papers aimed to investigate two very specific relationships: the first
hypothesis looks at the relationship between knowledge-intensity and the professional’s individual performance in a PSF. The second, third, fourth and fifth hypotheses are presented in sets of two and investigate the mediating effects of the cultural, personnel, results and behavioral controls tightness on the relationship between knowledge-intensity and individual performance.
2.4.1.1. Knowledge intensity and individual performance
This study regards knowledge intensity or human capital as an inseparable element that characterizes a PSF. Because these service firms have, human knowledge or capital, as their primary resource, it is clear that this resource element influences the PSF’s performance, either directly or indirectly. The current literature presents several findings which provide a rather clear consensus or agreement regarding the mechanisms of the relationship of knowledge intensity and a firm’s
performance. Prior studies found that knowledge has a positive mediating effect on the financial performance in organizations (Germain et al., 2001). Also Hitt et al. (2001) found that human capital
13 has a positive mediating effect on a PFS’s performance. There are also findings that state that an enhanced human capital disclosure positively influences the organizational performance (Lin et al., 2012). Yet other studies found evidence that well-made investments in human capital can generate better organizational performance (Crook et al., 2011). Taking into account the disadvantage of this rather costly type of capital, which is human capital, take for example the time and resources need to acquire, train and develop the employee’s skills, still it is expected that the return of this investment in human capital exceeds its costs. Crook et al. (2011) found evidence that human capital is strongly positively related to an organizations performance. They also found that human capital positively influences the operational performance which is a good indicator of the actual competitive
advantages that are created by the human capital that is available to the organization. This statement is supported by a study performed by Subramaniam & Youndt (2005) where they mention that intellectual capital, considering the fact that it requires investments like acquiring, hiring, training and retaining the human capital in the organization, it still leads to beneficial organizational characteristics and performances. Anderson (2012) found evidence stating that knowledge-based service firms in comparison to capital-based firms, which are firms that rely heavily on transferable assets like buildings or machines, have lower downside risk outcomes and have (significant) higher upside performance gains. Yet another study performed by Carmen Díaz-Fernandez et al. (2015) found evidence that the firm’s performance is positively influenced by the management’s educational background. I therefore come to the conclusion that human capital, based on the current findings, is one of the most crucial or essential elements to the success of an service organization in terms of performance (Crook et al., 2011, Anderson, 2012, Lin et al., 2012, Germain et al., 2001, Hitt et al., 2001, Haber & Reichel, 2007). I also conclude that the majority of the current research focuses on organizational level performance, or other derived indicator for firm performance, rather that the professional’s individual performance. Clearly, there is still an undeveloped research area in the specific PSF research context on how knowledge or human capital influences the professional’s individual performance.
Considering the above reasoning I currently expect the following relationship to be true:
14 2.4.1.2. MCS design and tightness
In the current management accounting literature there has been little attention regarding the mechanisms and the relationship between knowledge-intensive firms and their MCS design and related issues (Ditillo, 2004). A study performed by Ditillo (2004) investigated the effects of using different types of management controls within a knowledge-intensive firm, where three different projects, which had differing degrees of knowledge complexity, were analyzed. He mainly found that the degree of knowledge complexity determined the need for specific management controls or
required different types of management control mechanisms. Ditello (2004) concludes by stating that knowledge complexity is one of the crucial drivers in the design, shaping and use of the MCS. There are also other studies that prove the importance of this relationship between knowledge intensity and the design and use of the MCS. Herremans & Isaac (2005) found that the effectiveness of the MCS depends strongly on the type of knowledge and intellectual capital intensity that is available or present in an organization. They concluded that knowledge intensive firms should make use of a rather organic and flexible type of control. The control system should enable and encourage
employees to use their knowledge intensively, and to do so, the work environment has to be nurturing and fostering. As an example of informal controls they mention a strong value set and a code of conduct. There are also other researchers that sustain the above mentioned arguments. Adler & Borys (1996) did a research on the effect of having either an enabling type or a coercive type of
formalization of the workflow and business processes. They found that the use of more enabling procedures and processes stimulated the employees to perform their jobs and task more efficiently and enhance their job commitment. I draw on this finding since is provides a similar argument that proves that the use of more enabling types of MCS as for example cultural and personnel controls leads to positive organizational outcomes.
Based on this reasoning, I propose the following hypotheses:
H2a. There is a positive effect of knowledge intensity on the use of cultural controls. H2b. There is a positive effect of cultural controls on individual performance.
Therefore it is proposed that knowledge intensity has a positive indirect effect on individual performance through the use of cultural controls.
15 Selecting, acquiring, training and matching people or employees with work that is according to their abilities and capacities are forms of personnel controls (Merchant & Van der Stede, 2012). Prior studies have found significant evidence stating that investments made in human or intellectual capital will lead to organizational benefits, in specific to an enhanced firm performance (Crook et al., 2011, Subramaniam & Youndt, 2005, Huselid, 1995). Abernethy & Brownell (1997) performed a study where they found that when task uncertainty is high, when there is no standard or routine list of activities or tasks available, and when the analyzability of a task is low, and a stronger use is made of personnel forms of control this lead to a significant positive effect on performance. To reinforce both the underlying arguments of H2a and H2b, Pernot & Roodhooft (2014) also mention that a misfit in the design of the MCS will lead to a poor organizational performance. In specific they mention their preference and the importance of informal types of controls in manufacturing organizations.
Expected is that this relationship will exist even more strongly within service firms, taking into account the relative higher presence of human capital.
These arguments suggest the following hypotheses:
H3a. There is a positive effect of knowledge intensity on the use of personnel controls. H3b. There is a positive effect of personnel controls on individual performance.
Therefore it is proposed that knowledge intensity has a positive indirect effect on individual performance through the use of personnel controls.
There is some evidence available in the current management accounting literature that brings up several questions regarding the effect that behavioral and accounting controls actually have on the performance of an organization. On the one hand Abernethy & Brownell (1997) found that the use of accounting and behavior types of controls, like budgets and financial targets; lead to an enhanced firm performance, but this effect on firm performance is not as strong as the effect of using personnel controls. Whilst on the other hand Abernethy & Stoelwinder (1995) found evidence that
organizations which have a strong professional orientation, and where dominantly, a results based form of management control is used there resides a resistance among the professionals, a sense of offensiveness towards these limiting controls. This negative attitude could and will most likely have a negative effect on the professional’s performance. Therefore this valuable argument is taken into account when proposing the hypothesized relationships. According to Raelin (1985) professionals
16 have a tendency to disregard and resist conformity or standardized procedures, since they are of nature individualistic. Therefore it seems that professionals prefer an environment with less rules and targets. He also mentions that professionals accept action types of controls like routine procedures but as long as these behavioral controls do not limit or constrain their activities or practices. Bucher & Stelling (1969) conclude that professionals typically build their own role in the organization; they prefer to differentiate themselves from other professionals. These characterizing elements provide us with a view on how these professionals should act on these action and results focused type of
management controls. From the above discussion it is clear that there are several arguments that are both in favor of and against the use of results and behavioral or action type of controls in professional service organizations.
This discussion therefore leads to the following two sets of hypotheses:
H4a. There is a negative effect of knowledge intensity on the use of results controls. H4b. There is a negative effect of results controls on individual performance.
Therefore it is proposed that knowledge intensity has a negative indirect effect on individual performance through the use of results controls.
H5a. There is a negative effect of knowledge intensity on the use of behavioral controls. H5b. There is a negative effect of behavioral controls on individual performance.
Therefore it is proposed that knowledge intensity has a negative indirect effect on individual performance through the use of behavioral/action controls.
17 2.4.2. Theoretical framework
In figure 1 the (simplified) theoretical framework of this research paper is depicted. The control variables organizational size and working experience are not included in the theoretical framework but will be used in the regression analyses in section three of this study.
18
3. Research method
3.1. Survey procedure and data collection
The empirical study was carried out as a research project by staff members and several MSc. students who aim to achieve their master’s degree in Accountancy & Control at the University of Amsterdam. Before the final survey was sent out, it underwent a pre-test. This was done in order to enhance the ease of use and to improve the content validity of the questionnaire items. Appendix A includes all the survey questions that were used in the statistical analyses of this specific research paper (note that several questions are reverse coded, see Appendix A). Survey responses were gathered from November 2015 till the end of January 2016. The respondents were given a website link on which the participant could digitally fill out the survey.
3.1.1. Sample population
The population of this research is the professional service firm in a rather broad sense. The sample was taken, without limiting the selection based on country of origin or place of settlement, amongst all organizations that could be characterized as a PSF. Yet selecting only those PSF’s which are according to the criteria set by the research project coordinator. The criteria for acceptance of participation were the following: the organization had at least 50 employees in service, across the whole business concern. The respondent, the professional, would need to have at least a working experience of 3 years in their current field. The respondent could not be, in any way possible, an owner, partner or a member of the firm´s board. The survey was performed under the understanding that a PFS is an organization which provides a specific, knowledge intensive service by an educated or highly skilled employee who works in any of the following fields: accounting, actuarial services, advertising, architecture, biotechnology, consulting engineering, consulting human resource, consulting information technology, consulting management/strategic, consulting technology, engineering, fashion design, financial advising, graphic design, insurance brokerage, investment banking, investment management (hedge funds, VC, mutual funds), law, marketing/public relations, talent management/agency, media production (film, TV, music), medicine/physician practices, pharmaceutical, project management, real estate, recruiting – executive, research/R&D, risk management services, software development and talent/management agency. The survey was conducted among organizations which were settled in the following countries: Netherlands, United States, Aruba, Australia, Austria, Belgium, Canada, China, Germany, Israel, Kyrgyzstan, Malta,
19 Netherlands-Antilles, Norway, Qatar, Saint-Martin, Slovenia, Switzerland, United Arab Emirates, and United Kingdom.
3.2. Data screening
To ensure that the data is useable, reliable and valid to perform statistical analysis and to test the defined hypotheses, data screening measures have been performed. The data screening process assessed the presence of missing data, univariate outliers; by investigating the standard deviation in responses to the Likert type scale questions, the normality of the data, and finally multicollinearity.
From a total of 371 received responses, 45 responses were eliminated due to having more than 20% of the total values missing. All the variables that were used in further statistical analysis did not contain more than 2% of the values missing. Outliers in the data were also investigated. Since almost all research items were measured on a 5 point Likert-scale, practically there were no actual outliers present. To assess the presence of unengaged respondents I calculated the standard deviation of every response. 10 respondents had a standard deviation smaller than 0.5, which indicates an entire unengaged response without any meaningful variation, which were eliminated from the data. The remaining and usable 316 responses were used to perform the statistical analysis.
To assess the normality of the data the kurtosis scores of the questionnaire items were analyzed. A threshold level for the kurtosis scores of 2.2 was used (Sposito et al. 1983). None of the items displayed any kurtosis issues (i.e. all items had a kurtosis value < 2.2). To further support the assumption that the data is normally distributed the Central Limit Theorem is applied. The sample size was initially N=371, after data screening N=316. The sample size of 316 can therefore be deemed as normally distributed. Finally the independent variables were analyzed for
multicollinearity. Two methods were utilized to determine if multicollinearity was present. Both the Variance Inflation Factor (VIF) and a Tolerance test were performed. These tests, which were run multiple times, show that there is no presence of multicollinearity among the independent variables in this sample. On average the Tolerance levels of all the independent variables were all greater than 0.50 which is above the threshold level of 0.10 (Kleinbaum et al. 1978). And all the VIF values were smaller than 2 which are below the acceptable threshold level of 10. All in all the independent
variables do not display problems with multicollinearity (the collinearity results are presented in table 1, appendix B).
20 3.3. Exploratory factor analysis
According to Williams et al. (2010) there is not uniformed consensus concerning the suggested sample sizes that are needed to complete a factor analysis. According to Tabachnick & Fidell (2001) at least 300 responses are needed, whilst Hair et al. (1995) conclude that the sample size would have to be greater than 100 to suffice. MacCallum, Widaman, Zhang & Hong (1999) state that such rules of thumb can lead to bad conclusions because no account is taken of the complex elements present in a factor analysis. I made use of a clean sample size of 316 respondents, after filtering out respondents according to the mentioned data screening steps. This sample size is large enough to undertake meaningful statistical analysis.
To assess the construct validity I performed a exploratory factor analysis which revealed the relationships among the survey items and provided support for the construct’s unidimensionality. Before extracting the factors I performed several tests to determine the usability and suitability of the sample data for the factor analysis. Two tests were performed mainly the Kaiser-Meyer-Olkin
(KMO) test which is a measure of the sampling adequacy and the Bartlett’s test of Sphericity. The KMO test should show a result within the range of 0 and 1 and should preferably have a value above 0.5 (Williams et al., 2010). The results show a KMO measure of sampling adequacy of 0.828 which shows that the data is suitable for factor analysis. The result on the Bartlett’s test of Sphericity should present a significance smaller than 0.05. The findings show a test result with a significance of 0.000; this means that the null hypothesis is not rejected. Both tests confirm the suitability of the sample data. To determine the amount of factors to extract from the items I used the Kaiser’s criteria: factors with an eigenvalue greater than 1 should be extracted (Williams et al., 2010). After running the exploratory factor analysis multiple times the final iterated factor analysis resulted in the extraction of a total of 16 factors as can be seen in table 2. These 16 factors explained a total variance of 67% (rounded off to zero decimals). The amount of variance that is explained by the model is above the required threshold level of 60% (Field, 2013). The median (for Likert-scale items) and mean (for scale items) were used to complete the missing values. Table 2 presents the individual loadings of the survey items on the factors. The factor loadings which were smaller than 0.4 were suppressed to enhance the readability of the table. There were no cross loadings present when suppressing loadings of 0.4 and smaller. The factor analysis was iterated, by eliminating cross loadings or loadings smaller than 0.4, until a clean rotated matrix was displayed.
21 To determine the convergent validity I looked at the loadings of the questionnaire items on the factors. All items loaded, on average, in excess of 0.5 on their respective factor which is in support of the convergent validity. Al loadings, as can be seen in table 2, have an absolute loading in excess of 0.7 with an occasional loading in the range of 0.4-0.6 on a very few items. The discriminant validity was certified by investigating the amount and significance of the cross loadings on the
factors and by analyzing the factor correlation matrix that was generated by running the exploratory factor analysis using a Promax with Kaiser Normalization rotation. This resulted in zero cross loading above the threshold of 0.5. The factor correlation matrix does not show any correlation coefficients between the factors in excess of 0.7 or smaller than -0.7. Based on the lack of significant cross loadings and coefficients greater and smaller than 0.7 and -0.7, respectively, the discriminant validity is assessed and supported.
The reliability of the factors is determined by analyzing the Cronbach’s Alpha values. Nunnally (1978) suggests that a minimum Cronbach’s Alpha of 0.6 is deemed sufficient. Factors which have a Cronbach’s Alpha greater than 0.6 are deemed as being adequately reliable for statistical analysis. The Cronbach’s Alpha scores for each factor are presented below (see table 2).
Table 2 Construct validity* Factors Supervisor evaluation Self-evaluation continuous improvement Self-evaluation career success Self-evaluation work performance Individual performance
Essential duties performance 0,84 Job/task responsibility fulfillment 0,839
Expectation performance 0,816
Formal job requirements performance 0,813 Job/task duties completion 0,814 Performance related activities engagement 0,728 Acceptance of obligatory job aspects 0,712
Creativity or new ideas 0,794
Finding improved ways to work 0,788
Implementation of new ideas 0,78
Improvement processes and routines 0,67
In search of career opportunities 0,808
Career progress 0,734
Obtaining personal career goals 0,735
Skill development for future career 0,598
Accuracy of work 0,604
Quality of work output 0,544
Quantity of work output 0,486
22
Table 2 (continued)
Factors
Explicit tightness Implicit tightness
Results controls
Performance measure for everything 0,678
Frequent checking of attainment of goals/targets 0,767 Large number of performance goals/target expectation 0,668 Frequent supervisor checks of attainment goals/targets 0,778 Frequency dialogue about results of performance measures 0,531 Meeting pre-established goals/targets with no exceptions according to
expectation
0,465
Goals/targets are essentially a guideline rather than a true commitment 0,667 Responding to new, unforeseen opportunities is considered more
important by my supervisor than achieving pre-established goals/targets
0,619 Explanations of deviations from pre-established goals/targets are
strongly considered by the supervisor
0,559
Cronbach's Alpha 0,80 0,41
Factors
Explicit tightness Implicit tightness
(1)
Implicit tightness
(2)
Personnel controls
Many steps/procedures before hiring 0,779
Extensive hiring process 0,784
Multiple interviews before hiring 0,687
Hiring candidates with similar types of job experience 0,837 Hiring candidates with similar types of education and training 0,786
Little consistency in the type of professionals that get hired 0,798 Strong variation in the competence of employees within a job title 0,736
Cronbach's Alpha 0,77 0,68 0,54
Factors
Explicit tightness Implicit tightness
Behavioral/Action controls
Existing processes, procedures or rules are present for every possible scenario
0,736 Established processes, procedures and rules cover all of my job tasks. 0,763
There are rules for everything 0,718
Primarily established processes, procedures and rules give broad guidelines concerning activities
0,646 Frequently supervisor monitoring of the extent to which established
process, procedures and rules are followed
0,53
Encouragement using procedures flexibly 0,81
Freedom to adjust rules to best perform job tasks 0,76 Encouragement in adjusting procedures to suit the situation 0,75
23
Table 2 (continued)
Factor
Explicit tightness Implicit tightness
Cultural controls
Planned team-building events 0,796
Regular hosting of social events 0,782
Company sponsored teams for sporting events and fundraisers/volunteer events
0,673 Communication of organizational core values to employees 0,579 A sense of “ownership” for the organization rather than just being an
employee
0,567
Weak colleague friendship 0,655
Socializing with colleagues outside of work 0,633
A growing similarity in personal and organizational values 0,504
Cronbach's Alpha 0,80 0,63
Factor
Knowledge intensity-human capital
Highly skilled employees 0,778
Experts in their particular jobs and functions 0,761
Creativity and brightness 0,724
The best employees in the industry 0,689
New ideas and knowledge 0,711
Cronbach's Alpha 0,86
Factor
Working experience
Working experience in current organization(years) 0,842 Working experience in current field(years) 0,827
Cronbach's Alpha 0,80
Factor
Organizational Size(control variable)
Entire company size(employees) 0,828
Organizational unit size(employees) 0,82
Cronbach's Alpha 0,69
*This table presents the final results of the exploratory factor analyses. To extract the factors I used the principal component analysis. The rotation method is the Varimax with Kaiser normalization. Factors with an Eigen value of > 1 were extracted and presented. The table reports the Cronbach's Alpha of each factor. The total variance that is extracted from the factor analysis amounts totally to 67% . To enhance the table's presentation and readability I suppressed all factor loadings that were < 0.40. A total of 16 factors were extracted from the exploratory factor analysis.
24 3.3.1. Individual performance
To measure the individual performance of professionals amongst the research sample the definition of performance in Podsakoff & MacKenzie (1989) and Welbourne, Johnson & Erez (1998) were used. The survey measured two aspects of professionals individual performance, namely the performance based on their self-evaluation and the performance based on the perception that the professionals have of their performance were they in the place of their supervisor to evaluate. This method of evaluation clearly creates limitations for the research findings. This research limitation will be elaborated further on in section five of this research. The variable supervisor evaluation (SUPEVA) was measured through seven questionnaire items which asked respondents to rate themselves if they were in the position of their supervisor on the following points: completion of job/task duties, performance according to expectation, performance on essential duties, fulfillment of job/task responsibilities, performance based on the formal job requirements, performance on the engagement in related activities and acceptance of obligatory job aspects. I measured the self-evaluation performance (SELFEVACONT, SELFEVACAR; SELFEVAWORK) using eight
questionnaire items. These items asked the respondents to rate their own (individual) performance on the following points: career progress, obtaining personal career goals, skill development for future career, in search of career opportunities, improvement of processes and routines, implementation of new ideas, finding improved ways to work, creativity or coming up with new ideas, accuracy of work, quality and quantity of work output. The final exploratory factor analysis revealed that the eighteen questionnaire items, which were used to measure individual performance, loaded on a total of four factors. Seven out of fifteen questions loaded on the first factor which was labeled SUPEVA (supervisor-evaluation). The next four questions loaded on the second factor which was labeled SELFEVACONT (self-evaluation continuous improvement). Consequently, four out of fifteen questions loaded on the third factor which was labeled SELFEVACAR (self-evaluation career success). And finally three items loaded on the last factor which was labeled SELFEVAWORK (self-evaluation work performance). The Cronbach’s Alpha for supervisor (self-evaluation, self-(self-evaluation continuous improvement, self-evaluation career success and self-evaluation work performance is 0.91, 0.84, 0.85 and 0.76, respectively. The variables SUPEVA, SELFEVACON, SELFEVACAR and SELFEVAWORK will be used as proxies in further statistical analysis for the measurement of the professional’s individual performance. These three factors together capture 26.3% of the total explained variance of 67%
25 3.3.2. Knowledge intensity-human capital
The definition of knowledge intensity or human capital intensity is adapted from Subramaniam & Youndt (2005). They state that human capital is determined by the degree of education, training and experience which is available in an organization and which is necessary for service provision. Knowledge intensity is measured with the use of five questionnaire items which asked respondents to assess the degree of knowledge that is present in their organization. The items focus on the following points: the skill level of employees, the job and function expertise, the degree of creativity and brightness, developing new ideas and gathering (new) knowledge and finally being the best in the industry. The exploratory factor analysis revealed that the five questions load on one single factor. This factor was labeled KNOWINT (knowledge intensity). The Cronbach Alpha of this factor is 0.86 and it captures 8.5% of the total explained variance of 67%.
3.3.3. Control systems
To measure the type and tightness of the control systems that were used in the sample firms the questionnaire contained a broad set of questions which were design to assess and determine the specific design and use of the management control systems. This study makes use of primarily, Ouchi’s (1979) definition of control measures but also adding knowledge from; Perrow’s (1970), Merchant’s (1982), and Snell’s (1992) interpretation of an organizations control responses. Mainly the items which describe and measure the four control system types: results controls, personnel controls, behavioral/action controls and cultural controls, were used. The respondents were asked assess the presence and degree of intensity or in specific terms the tightness of the organizational control measure from four different perspectives.
3.3.4. Results controls
The final exploratory analysis revealed that nine questions, that were used to measure the results controls tightness in a firm, loaded strongly on two factors. Six out of nine items loaded on the first factor whilst the remaining three items loaded on the second factor. The first factor was labeled EXPLRESCONTIGHT (explicit results control tightness) which was partially derived from Jaworski et al.’s (1993) definition. Explicit results control tightness is defined as the degree to which goals or targets are part of the control system. A tight control system would be characterized by one that
26 displays many results or target measures and controls in terms of amount and reach. To measure this definition the following five measures were assessed by the respondents: allot or a compendium presence of performance measures, frequent checking of attainment of goals and targets, a large number of performance goals and target expectation, frequent supervisor checks of attainment goals, and targets and meeting pre-established goals and targets with no exceptions according to
expectation. The second factor, which consisted of three items, was derived from the definition of Hage & Aiken (1968), Simons, (1987) and Van der Stede (2001). This factor was labeled
IMPLRESCONTIGHT (implicit results control tightness). Implicit results control tightness is defined by the degree or extent to which deviations from targets or goals are tolerated, accepted or even encouraged. A tight control system is defined as a system where deviations are not wanted or accepted by management. Respondents were asked to assess the control system on the following elements: goals and targets are essentially a guideline rather than a true commitment and responding to new, unforeseen opportunities is considered more important by my supervisor than achieving pre-established goals and targets. The Cronbach’s Alpha for explicit and implicit results control tightness is 0.80 and 0.41, respectively. The second factor has an unreliable Cronbach’s Alpha of 0.41 and will not be used in further analysis. The two factors together capture 8.1% of the explained variance.
3.3.5. Personnel controls
Seven questions, that were used to measure the personnel control tightness, loaded on three factors. The first factor was labeled EXPLPERSCONTIGHT (explicit personnel control tightness) which determined the degree to which employee selection procedures are included in the
management control system of a firm. A tight personnel control system has an extensive employee selection procedure and sets outs specific demands for specific types of employees. Three items loaded strongly on the first factor. These three items measured the following elements: before being hired an employee undergoes many steps and procedures, the hiring process is extensive, before being hired an employee must undergo multiple interviews. The following two out of seven tems loaded on the second factor which was labeled IMPLPERSCONTIGHT1 (implicit personnel control tightness (1)). This factor measured the extent to which deviations from the employee procedures or protocol are tolerated or accepted. A tight control system would imply a system which doesn’t permit employees to deviate from the preset procedures and protocols. Respondents were asked to assess this type of control tightness by assessing the items: there is little consistency in hiring similar types
27 of professionals and there is strong variation in employee’s competence within the same job title. The third factor was labeled IMPLPERSCONTIGHT2, since it measures the same type of tightness. The reliability analysis revealed the following Cronbach’s Alpha for explicit and implicit personnel control tightness (1 and 2) are 0.77, 0.68 and 0.54, respectively. The first two factors have a reliable Cronbach’s Alpha in excess of 0.6, and are therefore suitable for future analysis. The third factor has a Cronbach’s Alpha of 0.54 which is too low (smaller than 0.60) and will therefore not be used in further statistical analysis. These three factors claim 6.4% of the total explained variance.
3.3.6. Behavioral controls
When an organization makes use of standardized processes, procedures, protocols, and specific behavioral rules and routines it can be noted that it makes use of action or behavioral control measures as a management tool. A tight explicit behavioral control system is determined by the extent to which the controls are present in amount and reach. A big amount and far reach of the controls present a tight system. To measure and assess the tightness and design of the action or behavioral types of controls in the sample group, the questionnaire contained nine questions which were presented to the respondents. The exploratory factor analyses revealed that five out of nine items loaded on the first factor which was labeled EXPLBEHCONTIGHT (explicit behavioral control tightness). This definition was derived from Hage & Aiken (1968), Van den Ven & Ferry (1980), and Cunningham & Rivera (2001). To measure the explicit tightness the following points were presented and had to be assessed by the respondents: existing processes- procedures or rules are present for every possible scenario, established processes, procedures and rules cover all of my job tasks, there are rules for everything, primarily established processes, procedures and rules give broad guidelines concerning activities. The remaining three out of nine items loaded strongly on the second factor which was labeled IMPLBEHCONTIGHT (implicit behavioral control tightness). The
definition of implicit behavioral control tightness, which implies the extent to which deviations from established procedures, processes, protocols, specific behavioral rules and routines are accepted or even desired and encouraged, was adapted from Hage & Aiken (1968), Van der Stede (2001), and Bodewes (2000). A tight implicit behavioral control system is characterized by systems which prohibit deviations from the established procedures, processes, behavioral rules and routines. The following three items loaded on the second factor: encouragement using procedures flexibly, freedom to adjust rules to best perform job tasks, encouragement in adjusting procedures to suit the situation.
28 These behavioral control factors capture a total of 7% of the explained variance. The Cronbach’s Alpha for the explicit and implicit behavioral factors is 0.78 and 0.77, respectively. Thus the analysis will make use of two proxies to capture behavioral or action control tightness.
3.3.7. Cultural controls
The exploratory factor analysis revealed that eight questions, which were used to measure cultural control tightness in an organization, load strongly on two factors. On the first factor, which was labeled EXPLCULTCONTIGHT (explicit cultural control tightness), four items loaded strongly. Explicit cultural control tightness measures the degree to which socializing events or employee procedures are part of the management control system. A tight control system would be characterized as one that strongly uses these socializing types of controls and cultural or relationship building tools to manage employees. Respondents were asked to assess the extent to which the following events took place or existed in their organization: hosting of social events for employees, hosting team-building events, company sponsored teams for sporting events and fundraiser or volunteer events. The second factor was labeled IMPLCULTCONTIGHT (implicit cultural control tightness). Four items loaded on the second factor. The definition of implicit cultural control tightness was partially derived and adapted from O’Reilly & Chatman (1986). This factor measures the extent to which deviations from the established culture, which consists of values, norms and beliefs are accepted or tolerated by management. A tight control system would show a strong similarity between the
employee and organizational culture and a high cultural congruence. To measure the implicit cultural control tightness in the sample group, respondents were asked to determine the degree to which there is: a sense of “ownership” for the organization rather than just being an employee, and a growing similarity in personal and organizational values. The reliability analysis presented the Cronbach’s Alpha scores of 0.80 and 0.60, respectively. These two factors together capture 5.7% of the total explained variance.
29 3.3.8. Working experience
According to Subramaniam & Youndt (2005) the degree of education is one of the elements that determine the presence of human capital in an organization. Working experience is measured with two questionnaire items. These items determined the respondent’s working experience (years) in their current field and current organization. The exploratory factor analysis revealed that these two questionnaire items loaded strongly on one single factor which was labeled WORKEXP (working experience years). The factor displays a Cronbach’s Alpha value of 0.80 which is adequately reliable. This factor claims a total of 2.5% of the variance explained. To limit the complexity of the statistical model the variable working experience was used as a control variable in the regression model.
3.3.9. Size
Organizational size is also used as a control variable in the statistical analysis. To measure the size of an organization two questionnaire items were used which asked the respondents to note their organization’s size in the number of employees in service. The following to items were used: total organizational size and organizational unit size. The exploratory factor analysis revealed that the two items loaded on a single factor which was labeled SIZE (Organizational size). This factor has a Cronbach’s Alpha value of 0.69 which is above the threshold value of 0.6 and therefore reliable and makes this variable acceptable to be included for further analysis. This last factor captures 2.3% of the explained variance.
Summary of constructs
All the variables, with the expectation of the variables size, working experience and item: frequency dialogue performance measures, are measured on a 5 point Likert-scale. Table 3 displays the
descriptive statistics. Table 3 shows that all survey questions range in between the value of 1 and 5 with either a minimum of 1 or 2 and a maximum of 5. The exploratory factor analysis revealed the unidimensionality of the variables. The constructs explain an acceptable variance of 67%. In
addition, the reliability analysis revealed that all the constructs, with the exception of the constructs: implicit personnel control tightness (2) and implicit results control tightness, have acceptable
Cronbach’s Alpha values. The Cronbach’s Alpha values, of the retained constructs, all range between 0.63 and 0.91. Both the convergent and discriminant validity are assessed and provide satisfactory results.
30
Table 3
Descriptive statistics for questionnaire items
Min Max Mean Median Std. Dev.
Supervisor evaluation
Job/task duties completion 1 5 4,14 4 0,79
Performance related activities engagement 2 5 4,04 4 0,81
Job/task responsibility fulfillment 1 5 4,22 4 0,74
Formal job requirements performance 1 5 4,12 4 0,74
Essential duties performance 1 5 4,23 4 0,77
Expectation performance 1 5 4,15 4 0,80
Acceptance of obligatory job aspects 1 5 3,94 4 0,91
Self-evaluation-continuous improvement
Finding improved ways to work 1 5 3,74 4 0,91
Implementation of new ideas 1 5 3,56 4 0,92
Creativity or new ideas 1 5 3,54 4 0,97
Improvement processes and routines 1 5 3,68 4 0,93
Self-evaluation-career success
Career progress 1 5 3,67 4 1,01
Obtaining personal career goals 1 5 3,60 4 1,00
In search of career opportunities 1 5 3,37 3 0,99
Skill development for future career 1 5 3,66 4 1,03
Self-evaluation-work performance
Quality of work output 1 5 4,08 4 0,69
Quantity of work output 2 5 3,88 4 0,73
Accuracy of work 2 5 3,97 4 0,82
Results controls explicit tightness
Large number of performance goals/target expectation 1 5 3,02 3 1,06 Frequent supervisor checks of attainment goals/targets 1 5 2,93 3 1,06 Frequent checking of attainment of goals/targets 1 5 2,94 3 1,06
Performance measure for everything 1 5 2,74 3 1,13
Meeting pre-established goals/targets with no exceptions according to expectation 1 5 2,98 3 1,09 Frequency dialogue about results of performance measures 1 6 2,98 3 1,13
Results controls implicit tightness
Explanations of deviations from pre-established goals/targets are strongly considered by the supervisor
1 5 2,66 3 0,87 Responding to new, unforeseen opportunities is considered more important by
my supervisor than achieving pre-established goals/targets
1 5 2,66 3 0,92 Goals/targets are essentially a guideline rather than a true commitment 1 5 2,81 3 1,12
Personnel controls explicit tightness
Extensive hiring process 1 5 3,11 3 1,07
Multiple interviews before hiring 1 5 3,30 4 1,26
31
Table 3 (continued)
Min Max Mean Median Std. Dev.
Personnel controls implicit tightness (1)
Hiring candidates with similar types of job experience 1 5 2,97 3 1,06 Hiring candidates with similar types of education and training 1 5 3,15 3 1,19
Personnel controls implicit tightness (2)
Little consistency in the type of professionals that get hired 1 5 3,19 3 1,05 Strong variation in the competence of employees within a job title 1 5 2,88 3 1,04
Behavioral/Action controls explicit tightness
There are rules for everything 1 5 3,18 3 1,19
Frequently supervisor monitoring of the extent to which established process, procedures and rules are followed
1 5 3,02 3 1,15 Primarily established processes, procedures and rules give broad guidelines
concerning activities
1 5 3,62 4 0,98 Established processes, procedures and rules cover all of my job tasks. 1 5 3,03 3 1,08 Existing processes, procedures or rules are present for every possible scenario 1 5 3,27 3 1,02
Behavioral/Action controls implicit tightness
Encouragement in adjusting procedures to suit the situation 1 5 2,79 3 1,06 Freedom to adjust rules to best perform job tasks 1 5 2,56 2 1,03
Encouragement using procedures flexibly 1 5 2,89 3 1,09
Cultural controls explicit tightness
Communication of organizational core values to employees 1 5 3,81 4 1,03 Company sponsored teams for sporting events and fundraisers/volunteer events 1 5 3,36 4 1,26
Planned team-building events 1 5 3,58 4 1,19
Regular hosting of social events 1 5 3,59 4 1,11
Cultural controls implicit tightness
Weak colleague friendship 1 5 3,91 4 1,16
A sense of “ownership” for the organization rather than just being an employee 1 5 3,47 4 1,15 Socializing with colleagues outside of work 1 5 3,42 4 1,15 A growing similarity in personal and organizational values 1 5 3,28 3 0,95
Knowledge intensity-human capital
Creativity and brightness 1 5 3,76 4 0,88
Experts in their particular jobs and functions 1 5 4,05 4 0,87
Highly skilled employees 1 5 4,07 4 0,82
The best employees in the industry 1 5 3,61 4 0,97
New ideas and knowledge 1 5 3,60 4 0,89
Working experience
Working experience in current field(years) 1 11 6,91 6 3,04 Working experience in current organization(years) 1 11 5,78 5 3,13
Size
Entire company size(employees) 1 4 2,84 3 1,08
32 Table 4 presents the Pearson’s and Spearman’s correlation coefficients matrix. The exploratory factor analysis revealed that the 16 constructs are distinct; this is further supported by the correlation matrix below. Table 4 displays both the parametric Pearson correlation coefficients and the non-parametric Spearman’s correlation coefficients which are rounded off to two decimals to enhance readability. The correlation coefficients range up to a coefficient of 0.53 with multiple significant correlations at both the significance levels of p. < 0.05, p. < 0.01, and p. < 0.001. The correlation matrix provides reasonable evidence to further ensure that there is no presence of multicollinearity between the constructs. There are no correlations greater than 0.7 between the constructs and thus no signs of multicollinearity (Tabachnick & Fidell, 2001).
Table 4
Pearson's & Spearman's Correlations a
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 SUPEVA (1) ,29*** ,34*** ,43*** ,26*** -,10 -,04 ,11 ,07 ,11* ,08 -,15** ,12* ,15** ,06 ,03 SELFEVACONT (2) ,22*** ,49*** ,50*** ,20*** ,10 -,12* ,09 -,01 -,05 ,06 -,23*** ,14* ,19*** ,13* ,04 SELFEVACAR (3) ,24*** ,52*** ,55*** ,28*** ,09 -,11 ,22*** ,11 ,11 ,19*** -,21*** ,23*** ,27*** -,05 ,12* SELFEVAWORK (4) ,37*** ,50*** ,53*** ,27*** ,03 -,10 ,12* ,03 ,00 ,16** -,20*** ,21*** ,22*** ,10 ,13* KNOWINT (5) ,19** ,24*** ,34*** ,31*** ,14* -,17** ,39*** ,21*** ,18** ,16** -,24*** ,42*** ,34*** -,01 ,08 EXPLRESCONTIGHT (6) -,08 ,09 ,09 ,04 ,16** -,16** ,23*** ,07 -,20*** ,46*** -,10 ,17** ,09 -,21*** -,03 IMPLRESCONTIGHT (7) -,00 -,13* -,12* -,13* -,19*** -,19*** -,01 -,16** ,08 -,08 ,16** -,13* -,08 ,04 ,07 EXPLPERSCONTIGHT (8) ,09 ,13* ,26*** ,11 ,40*** ,22*** ,01 ,16** ,07 ,19*** -,07 ,37*** ,21*** -,13* ,14* IMPLPERSCONTIGHT1 (9) ,07 ,02 ,11* ,07 ,22*** ,07 -,13* ,15** ,20*** ,17** -,02 ,15** ,10 ,08 ,11 IMPLPERSCONTIGHT2 (10) ,09 -,01 ,11* ,04 ,19*** -,15** ,08 ,09 ,22*** -,06 ,09 ,06 ,17** ,05 ,03 EXPLBEHCONTIGHT (11) ,11* ,05 ,17** ,18** ,17** ,49*** -,09 ,18** ,21*** -,04 ,05 ,16** ,11* -,01 ,13* IMPLBEHCONTIGHT (12) -,14* -,27*** -,21*** -,24*** -,25*** -,09 ,13* -,05 -,03 ,05 ,07 -,21*** -,16** ,01 ,11 EXPLCULTCONTIGHT (13) ,07 ,15** ,27*** ,20*** ,44*** ,18** -,14* ,40*** ,12* ,08 ,15** -,17** ,46*** -,12* ,19*** IMPLCULTCONTIGHT (14) ,07 ,20*** ,26*** ,22*** ,35*** ,08 -,08 ,21*** ,08 ,15** ,10 -,16** ,50*** ,01 ,09 WORKEXP (15) ,08 ,11* -,07 ,08 -,01 -,20*** ,06 -,13* ,08 ,07 -,01 ,01 -,10 ,02 ,06 SIZE (16) ,06 ,04 ,14* ,11 ,07 -,05 ,04 ,14* ,11 ,05 ,13* ,12* ,21*** ,11* ,05
a The Pearson correlation coefficients (rounded off to two decimals) are displayed on the lower left diagonal , whilst the Spearmans's correlation coefficients are displayed on the right upper diagonal.
33 3.4. Response analysis
To determine the non-response bias I performed an analysis on the difference in the mean score of a variable between early and late respondents. I compared the early and late respondents based on return date of the survey. I divided the survey responses, based on the mean return date in the sample, into early and late respondents. I created a dummy variable and performed an independent samples t-test. The results in table 5 show that the constructs, with a single exception, do not present any significant (p. < 0.05) difference between early and late
respondents. With the exception of variable: implicit personnel control tightness there are no significant differences between the means of early and late respondents. These results support the absence of significant non-response bias.
Table 5
Non-response bias*
Construct Early respondents (n=143) Late respondents (n=173) Significances (p.)
SUPEVA 4,13 4,11 0,712 SELFEVACONT 3,64 3,62 0,857 SELFEVACAR 3,57 3,58 0,895 SELFEVAWORK 4,02 3,94 0,234 KNOWINT 3,83 3,81 0,793 EXPLRESCONTIGHT 2,86 2,99 0,151 IMPLRESCONTIGHT 2,75 2,68 0,376 EXPLPERSCONTIGHT 3,24 3,11 0,222 IMPLPERSCONTIGHT1 3,08 3,04 0,719 IMPLPERSCONTIGHT2 3,16 2,94 0,024 EXPLBEHCONTIGHT 3,17 3,27 0,253 IMPLBEHCONTIGHT 2,76 2,74 0,844 EXPLCULTCONTIGHT 3,59 3,59 0,995 IMPLCULTCONTIGHT 3,52 3,52 0,948 WORKEXP 6,48 6,24 0,465 SIZE 2,76 2,59 0,108
*The construct means are not significantly different, with the exception of one construct (see sig., 2-tailed). The research sample was divided into early and late respondents based on the return date of the survey.
