Page 1
UNIVERSITEIT GENT
FACULTEIT ECONOMIE EN BEDRIJFSKUNDE
ACADEMIEJAAR 2008 – 2009
The use of the Balanced Scorecard: linking different enabling theories
Masterproef voorgedragen tot het bekomen van de graad van
Master in de Toegepaste Economische Wetenschappen
Kevin De Ruyck
onder leiding van
Prof. Dr. W. Bruggeman
Page 3
UNIVERSITEIT GENT
FACULTEIT ECONOMIE EN BEDRIJFSKUNDE
ACADEMIEJAAR 2008 – 2009
The use of the Balanced Scorecard: linking different enabling theories
Masterproef voorgedragen tot het bekomen van de graad van
Master in de Toegepaste Economische Wetenschappen
Kevin De Ruyck
onder leiding van
Prof. Dr. W. Bruggeman
Page 4
PERMISSION
Ondergetekende verklaart dat de inhoud van deze masterproef mag geraadpleegd en/of
gereproduceerd worden, mits bronvermelding
Kevin De Ruyck
Page 5
I
Preface
Writing this master dissertation was probably the biggest task I’ve faced so far. It took a lot of
work and a healthy amount of creativity, but I think the result is worthy of the name master
dissertation. I am grateful to Professor Bruggeman for his help in my search for a suitable
subject matter and his guidance on the path to the problem’s solution. Professor Gerrit Van
Daele, Katrien Van Daele, Karen De Visch and David Van Laethem I would like to thank for
their revisions and corrections of my work. Finally I offer a word of sincere thanks to my
family for their support and patience.
Kevin De Ruyck
Page 6
II
Index
1 INTRODUCTION ............................................................................................................ 1
2 LITERATURE REVIEW .................................................................................................. 3
2.1 Developmental approach to enabling formalization .................................................. 3
2.1.1 Enabling versus coercive formalization ................................................................ 3
2.1.1.1 Repair ............................................................................................................... 4
2.1.1.2 Internal transparency ....................................................................................... 5
2.1.1.3 Global Transparency ........................................................................................ 5
2.1.1.4 Flexibility ......................................................................................................... 6
2.1.2 Development of enabling formalization ................................................................ 6
2.1.2.1 Experience-based development ........................................................................ 7
2.1.2.2 Experimentation ............................................................................................... 8
2.1.2.3 Professionalism ................................................................................................ 8
2.1.3 Attitude towards performance measures............................................................... 9
2.2 Levers of control framework ................................................................................... 10
2.2.1 Levers of control items ........................................................................................ 10
2.2.2 Impacts of interactive and diagnostic use ........................................................... 11
2.3 Advantage of the Balanced Scorecard ..................................................................... 12
2.4 Hypothesis development .......................................................................................... 12
3 RESEARCH METHOD .................................................................................................. 15
3.1 Participants .............................................................................................................. 15
3.2 Construct development ............................................................................................ 16
4 RESULTS ...................................................................................................................... 19
4.1 Association degree of diagnostic and interactive use
with degree of enabling formalization ..................................................................... 19
4.2 Regression on attitude towards performance measures ........................................... 20
4.3 Regression on the degree of enabling formalization ............................................... 21
4.4 Econometric issues .................................................................................................. 22
4.4.1 Normality of residual terms ................................................................................ 22
4.4.2 Heteroscedasticity ............................................................................................... 22
4.4.3 Multicollinearity.................................................................................................. 22
4.4.4 Endogenity .......................................................................................................... 23
Page 7
III
5 CONCLUSION .............................................................................................................. 24
List of appendices
Appendix A: Construct Items ................................................................................................ VIII
Appendix B: First Regression on attitude towards performance measures.............................. XI
Appendix C: Regression to test the impact of the degree of enabling formalization .............. XII
Appendix D: Heteroscedasticity tests ................................................................................... XIII
Appendix E: Multicollinearity .............................................................................................. XIV
Appendix F: Survey ................................................................................................................ XV
List of tables and figures
Fig. 1: Developmental model (Wouters & Wilderom, 2008) ................................................. 9
Fig. 2: Capability model (Henri, 2006) ................................................................................. 12
Table 1: Participants’ industries .............................................................................................. 15
Table 2: Participants’ functions ............................................................................................... 15
Table 3: Participants’ responsibility level ............................................................................... 15
Table 4: Construct analysis ..................................................................................................... 18
Table 5: Description data of degree of use and formalization for t test .................................. 19
Table 6: Independent-samples t test of degree of use and formalization ................................ 19
Table 7: Regression on attitude towards performance measures ............................................ 20
Table 8: Regression on degree of enabling formalization ........................................................ 21
Page 8
IV
Used abbreviations
PMS: Performance Measurement System
BSC: Balanced Scorecard
OLS: Ordinary Least Squares
VIF: Variance Inflating Factor
Page 9
1
1 INTRODUCTION
Performance Management Systems (PMS’s) have been around for a long time. The currently
applied models are widespread and diverse. However, the Balanced Scorecard (BSC) has
become an important player on the market of PMS’s (Kaplan & Norton, 1992). The BSC can
motivate breakthrough improvement on its four areas (Kaplan & Norton, 1993). In a BSC the
performance is measured using not only data from the past (Financial), but also data which
can measure future profits (Customer, Internal Business Processes, Learning and Growth)
(Kaplan & Norton, 1992). Studies showed that a company which uses the BSC outperforms a
company with a traditional PMS on financial measures (Davis & Albright, 2004). The BSC is
a measurement system which is weighed. The used indicators on a small operational basis
should be aligned with the overall strategy. The BSC translates the strategic objectives into a
coherent set of performance measures (Norton & Kaplan, 1993).
A study showed that a good way to motivate management is introducing strategic alignment.
The process of introducing strategic alignment, has a positive impact on the motivation of
management (Decoene & Bruggeman, 2006). This study revealed that the translation process
of the strategy towards the different levels in the organization is important to motivate
management. In addition, another study, performed by Hall, has proven that PMS’s have an
effect on management performance. He found that there is a positive, yet indirect impact.
PMS’s strengthen the role clarity, which is the job’s expectations, and the psychological
empowerment, which is the intrinsic task motivation. These concepts influence managerial
performance in a positive way (Hall, 2008).
The modern organization often needs to abandon the strong hierarchical structure and place
more power with the employees. Empowering leadership has an indirect positive effect on
performance, through the variables knowledge sharing and team efficacy (Srivastava, Bartol
& Locke, 2006).
The previous theories clarify motivation at the level of management. The motivation of the
entire workforce is explained by Nohri, Groysberg and Lee (2008). The motivation in their
model is explained by four drives. Firstly, there is the drive to acquire: this need can be met
by an objective reward system. The second drive is the drive to bond: a strong culture should
Page 10
2
be in place to keep employees motivated. Next is the drive to comprehend, which can be
satisfied by designing jobs so they have meaning and are challenging. The final drive is the
drive to defend: people get to know and express ideas of the institution. This enhances
feelings of security and confidence. Performance management and resource-allocation
processes should be fair and transparent. Fulfilling these fundamental needs can boost
motivation. This study examines the defense drive suggested by Groysberg and Lee, namely
getting performance management supported by the employees. Adler and Borys (1996) found
that rules, procedures and instructions can be considered as a form of formalization. Coercive
formalization aims to force employee compliance, while enabling formalization makes
employees feel motivated and helped by the system. Lillis (2002) states that there are so many
dimensions a PMS should measure, that it is virtually impossible to include them all. Wouters
and Wilderom (2008) built their case on this inherent completeness of a PMS, stating that a
PMS should be seen as a form of formalization. They created a framework of development to
achieve enabling formalization in an organization.
Simons (1995) built a framework for control systems. Important forms of data usage of the
PMS are diagnostic and interactive use. While diagnostic use refers to top-down control,
interactive use refers to the discussion about the performance measures. Henri (2006) found
that interactive use has a positive impact on organizational capabilities, while diagnostic use
has a negative impact on organizational capabilities.
The next part of this paper will study the different trends in research on how organizations
should set up a system to create a workforce which is enabled. Members of an organization
are “enabled” when they experience the organization as an environment in which they are
motivated and stimulated to perform. These enabling theories are applied to the PMS. At the
end of the second part, these theories are compared and linked to form hypotheses. The third
part describes the adopted research design. The fourth part discusses the analytical results of
the research. The final part contains the concluding words of this paper.
Page 11
3
2 LITERATURE REVIEW
2.1 Developmental approach to enabling formalization
2.1.1 Enabling versus coercive formalization
The core features of a bureaucratic form are the workflow formalization, the specialization
and the hierarchy. Formalization is the whole of written rules, procedures and instructions. A
bureaucracy is seen in a positive and a negative view. In the negative view, creativity, work
satisfaction and motivation decrease. The reason for this decrease is the abrogation of a great
part of the freedom. Workers in the bureaucracy are coerced into complying with the
overlying hierarchical level. Studies showed that formalization leads to work dissatisfaction,
absenteeism, physical and psychological stress, feelings of powerlessness, self-estrangement
and alienation (Rousseau, 1978) (Arches, 1991) (Bounjean & Grimes, 1970). The positive
view states that more guidance is given to the members of the organization. Responsibilities
are written down in clear and explicit words, hence there is lower role conflict and ambiguity,
resulting in a more positively perceived effectiveness (Adler & Borys, 1996). So there is a
positive and a negative opinion on formalization, but in the past no research has been done
about under which conditions these rules and procedures are accepted by the workforce and
under which there is an adverse feeling towards this. In organizational theories the types of
this formalization have often been neglected. People resent what they consider “bad” rules,
but “good” rules are often taken for granted (Perrow, 1984). In the past, little research has
been performed about when rules are considered “good” or “bad”. Adler and Borys have a
theory concerning these different types of formalization (1996). According to Adler and
Borys there are two main types of formalization: Coercive and enabling. The goal of coercive
formalization is forcing compliance. However, enabling formalization motivates and helps
employees to perform their tasks. These are the two contrasting types of formalization which
can be distinguished. Formalizations differ in three areas: the characteristics of the system, the
process of designing the system and the process of implementing the system. The
development process and these characteristics will lead to an enabling formalization. They
Page 12
4
state that four generic features concerning formalization exist: repair, internal transparency,
global transparency and flexibility (Adler & Borys, 1996, p. 70).
2.1.1.1 Repair
Often in highly bureaucratic environments, managers separate the routine production tasks
from non-routine repair activities. They do so because they don’t value potential contributions
of workers. An example of this is the locking of a machine’s control panel, in order to prevent
workers from fiddling with the parts. They need to rely on a special workforce to fix the
problems. The workers will notice the management’s lack of trust: breakdowns will be
stimulated and possible suggestions for improvement won’t be given. Highly usable systems
are characterized by the fact that breakdowns in the system can be mended by the workers
themselves instead of them having to call a specialized workforce (Adler & Borys, 1996, p.
70). In this manner, repair processes are integrated in the routine operations. In the enabling
logic, operations are not fully programmable. Procedures are generated which facilitate
response to work problems. Accounting is a formal control system used to express operational
rules and standards. Most information like standard costs is only understandable for
accounting experts. If this information is broken down in parts relevant for employees, this
could be a great help to deal with operational problems. This information can also be of
service to repair and update the accounting structures (Ahrens & Chapman, 2004).
Breakdowns have a potential to signal work problems and become an opportunity for the
performance of the organization. This paper will define repair as follows: repair means that
users can mend and improve the work process themselves rather than allowing breakdowns
and other non-programmable events to force the work processes to a halt (Wouters &
Wilderom, 2008, p. 492).
Page 13
5
2.1.1.2 Internal transparency
The internal transparency refers to the degree of intelligibility of the processes for the
organizational members. In this way it is related to repair (Adler & Borys, 1996, p. 72).
Internal transparency means that users have a good understanding of the logic of a system’s
internal functioning and they have information on its status (Wouters & Wilderom, 2008, p.
492). In coercive logic, reliance on users’ skills is low: there is no insight in internal
workings. Equipment status will often only be provided in case of breakdown. The language
will only be interpretable for the technical staff, not for the person who operates it. In the
enabling logic however, there is a sound grasp of the internal processes, by explaining key
components and providing understandable best-practices. The major key is layered
information: an information overload should be avoided (Ahrens & Chapman, 2004). So
internal transparency means that users have a good understanding of the logic of a system’s
internal functioning and they have information on its status. Enabling formalization provides
users with a clear understanding of the underlying rationale why certain control mechanisms
are in place. Such formalization also codifies best-practice experiences, and users are
provided with feedback on their performance (Wouters & Wilderom, 2008, p. 492).
2.1.1.3 Global Transparency
Global transparency checks whether employees know about the broader system in which they
are working (Adler & Borys, 1996, p. 72). In coercive formalizations, employees of different
task-forces are separated from each other and communication between them is minimized. If
employees go beyond their function, they’re told it’s not their job (Heckscher, 1994). In an
enabling approach, employees are given a wide range of contextual information and an
understanding is given how they fit in the whole. The insight in the budgets of the
departments is often only visible for the top management. Budgets for a department could
also be made available for other departmental managers. Lateral coordination can be achieved
because key targets of important departments can be communicated to make sure other units
prioritize their tasks accordingly (Ahrens & Chapman, 2004, p. 280). So global transparency
refers to the intelligibility for employees of the broader system and context in which they
operate. Controls are designed to afford employees an understanding of how their own tasks
Page 14
6
fit into the whole. Information from outside one’s specific domain is available (Wouters &
Wilderom, 2008, p. 492).
2.1.1.4 Flexibility
Users must be able to make decisions with the information provided (Adler & Borys, 1996, p.
74). In coercive logic, machines take controlling decisions. Steps are to be explicitly followed
and exceptions must be authorized by the supervisor. In enabling logic, the machines give
suggestions and employees choose in the end. Deviations are not seen as risks but as learning
opportunities. A description made by Ahrens and Chapman refers to the organizational
members’ discretion over the use of the PMS. By technical development, personal computer
applications allow differentiated reports, composed of measures which are all stored in the
same database. In this way, different aggregations of performance information are created.
These represent differentiated but interrelated mental maps, which prepare the organization
for changing circumstances. (Ahrens & Chapman, 2004). ‘‘Flexible systems encourage users
to modify the interface and add functionality to suit their specific work demands’’ (Adler &
Borys, 1996, p. 74). So flexibility means that users can make controlling decisions after
enabling systems have provided information.
2.1.2 Development of enabling formalization
A main theory on how a PMS can get strong support by the workforce was published in 2008
by Wouters and Wilderom. This study relied on the Adler and Borys framework on sorts of
formalization. Enabling and coercive formalization were used specifically on a PMS. They
stated that the last two dimensions (process of designing and the process of implementing the
system) have great importance on the final outcome. However, design and implementation are
hard to distinguish (Adler & Borys, 1996). They decided to combine them under the name
Page 15
7
“development process”. The way a the development process is implemented, will impact the
extent to which employees will perceive the PMS as enabling. According to Wouters and
Wilderom, this process should be characterized by being experience-based, supportive
towards experimentation and using the employees’ professionalism (Wouters & Wilderom,
2008, p. 492).
2.1.2.1 Experience-based development
Experience-based development of a PMS involves “the identification, appreciation,
documentation, evaluation and consolidation of existing local knowledge and experience with
respect to quantitatively capturing and reporting relevant aspects of performance” (Wouters &
Wilderom, 2008, p. 493). A PMS must pay attention to experience and user involvement.
Aside from the top-down goals, there must be attention to local knowledge. Often PMS’s only
have five characteristics (Wouters & Wilderom, 2008, p. 493–494). First, the firm’s
objectives were cascaded to lower levels. Secondly, underperforming parts of the company
are signaled. Thirdly, there is an ability to get more details when needed. Fourthly, there is
transparency and uniformity in the overall PMS. Finally, there is one information system for
the whole organization. Often external experts can be involved to implement a standardized
PMS. However, top-down PMS’s are less likely to succeed. A major conclusion by Wouters
and Wilderom, linked to a PMS that should be experience-based, is that local initiatives for a
PMS are often more successfully implemented (Wouters & Wilderom, 2008, p. 512). Often
informal measurement systems appear locally. So it is best to build on existing knowledge
and accumulated experience, as it will improve organizational learning (West & Iansiti, 2003)
and lead to organizational change and improvement (Zollo & Winter, 2003).
Page 16
8
2.1.2.2 Experimentation
Experimentation involves “the first development of a performance measure and the
subsequent testing and refinement (in several rounds) of its conceptualization, definition,
required data, IT tools, and presentation, together with employees (whose performance is
going to be measured), to result in a measure that is a valid, reliable and understandable
indicator of performance in a specific local context” (Wouters & Wilderom, 2008, p. 493). A
performance measure will not be successful from the first time, but only after a few attempts.
Tinkering with the data should be possible. A study by Lowe and Jones (2004) showed that
the phase of making general goals into specific measureable goals is a process of trial-and-
error. This process needs input by the employees. West and Iansiti (2003) found evidence that
experience accumulation and experimentation are important drivers of performance.
2.1.2.3 Professionalism
Professionalism involves “an orientation by employees toward learning for the purpose of
improving work practices” (Wouters & Wilderom, 2008, p. 493). This orientation makes it
possible to rely on the experiences and experimentation by the workforce. Caldwell, Herold
and Fedor (2004) found that if employees are inclined to improve their work practice, they
will have a positive attitude towards performance measures. Professionalism in combination
with experience-based development and experimentation make a PMS more enabling. In
coercive formalizations, the procedures are designed by superiors, whose main goal is the
power to control the employees. Enabling formalizations’ procedures aren’t made top-down,
their goal is to improve the performance of the tasks of the employees. Adler and Borys
(1996) suggest that employees have a more positive attitude when they are enabled in their
tasks, but if they feel management tries to coerce their effort and make them comply, their
attitude will be more negative. Lillis (2002) stated that it is challenging to design the perfect
PMS, if not impossible. In that view, it’s important to foster the secure support from the
employees to improving performance. Important in a good BSC is that the measures should
Page 17
9
be objective. Cheating should not be possible, if so the credibility of the BSC completely fails
(Malina & Selto, 2001).
Fig. 1 Developmental Model (Wouters & Wilderom, 2008)
2.1.3 Attitude towards performance measures
In the empirical part of the paper by Wouters and Wilderom (2008), these independent
variables were studied in a case study in the logistics department of a medium-sized beverage
company. Professionalism was measured by surveys in this department. Enabling
formalization was measured on the level of the individual as attitude towards performance
measures from the department. Attitude towards performance measures reflects the perceived
usefulness of the performance measures. In the framework of Adler and Borys enabling
formalization is characterized by a more positive attitude towards the PMS by the members of
the organization. He stated that in an environment with room for experimentation and
attention to experience of employees, professionalism will drive the perceived usefulness of
the performance measures. He found proof for this statement in an experimentation- and
experience-rich environment. However, he did control for a number of constructs: leadership
style (Bass & Avolio, 2000), team trust (Baer & Frese, 2003) and work pressure (Stanton et
al., 2001). Wouters also created job satisfaction as a percentage, directly asked to participants.
Page 18
10
2.2 Levers of control framework
2.2.1 Levers of control items
In companies operating in a business-environment that requires flexibility, creativity and
innovation, control systems should consist of four parts to perform the best: a beliefs system,
a boundary system, diagnostic controls and interactive controls (Simons, 1995). According to
this study, it is important to have empowered employees, but there is a real need to control
them. Otherwise scandals can occur. A beliefs system is “the explicit set of organizational
definitions that senior managers communicate formally and reinforce systematically to
provide basic values, purpose, and direction for the organization” (Simons, 1995, p. 34). A
boundary system “delineates the acceptable domain of strategic activity for organizational
participants” (Simons, 1995, p. 39). A boundary system contains the actions which employees
should avoid. By only saying what employees shouldn’t do without telling them what they
should do, employees will be more creative and entrepreneurial, within limits. The boundary
system and the beliefs system are considered to create dynamic tension. The diagnostic
control system provides the common data, which is used to check whether the workforce
complies with the company goals. Interactive control systems focus on information based on
decisions made by subordinates, so there is contact between the different layers of
management. Interactive systems want to motivate workers by involving them in the process.
There are four characteristics of interactive control systems. First, they focus on constantly
changing, potentially strategic information. Secondly, information is significant enough to
pay attention to it on all the levels of the organization. Thirdly, this data should be discussed
face-to-face. Finally, it is a catalyst for continuous debate about relevant data. This framework
appears to be often used in PMS functioning (Abernethy & Brownell, 1999). Nevertheless,
the interactive and diagnostic systems are often the only ones that are used. The boundary
systems and the belief systems might be too generally formulated (Ahrens & Chapman,
2004).
Page 19
11
2.2.2 Impacts of interactive and diagnostic use
A different study, done by Henri (2006), showed empirical evidence concerning the impact of
the use of the PMS on performance. However, he added a mediator: capabilities. The different
uses of a PMS refer to Simons (1995). Diagnostic use of performance measures is used
mainly to control the workforce, to track and review the employees in a mechanistic way.
Interactive use of performance measures encourages the dialogue between the different
hierarchical layers of the organization. Adjustments can also be made from bottom to top.
Henri (2006) identified an interaction-term between these two uses, which is the ‘dynamic
tension’. If an organization applies the interactive use, the diagnostic use is utilized to create
dynamic tension which has a positive effect on capabilities. However, he didn’t find strong
empirical results which supported the existence of this interaction term. The Resource-Based-
View (Barney, 1991) states that resources will lead to a sustainable competitive advantage if
they are valuable, rare, inimitable and non-substitutable. Capabilities are necessary to use
these resources in a coherent way (Day, 1994). The capabilities the study refers to are
innovation, organizational learning, market orientation and entrepreneurship. These are
primary capabilities needed to create competitive advantage, which is only possible if they are
all present (Bhuian, Menguc & Bell, 2005) (Hult & Ketchen, 2001) (Hurley & Hult, 1998).
This view is consistent with the capability-view in the strategy research trends (Lengnick-Hall
& Wolff, 1999). Results show that interactive use has a positive effect on capabilities, while
diagnostic use has a negative effect - as expected - and the dynamic tension isn’t that clear in
the data (Henri, 2006). This is an important contribution to the value of the types of use for
the organization.
The paper of Widener (2008) included the whole ‘levers of control’ framework of Simons:
beliefs system, boundary system, diagnostic controls and interactive control. She found
impact on performance through attention and learning.
Page 20
12
Fig 2. Capability model (Henri, 2006)
2.3 Advantage of the Balanced Scorecard
The Balanced Scorecard doesn’t only have financial measures, but also indicators concerning
customers, internal processes and learning and growth (Kaplan & Norton, 1996). Malino and
Selto (2001) found indirect relationships between the BSC and organizational performance.
Davis and Albright (2004) researched the performance of BSC-driven financial companies
and non-BSC driven financial companies and found that the ones who did use BSC
outperformed the other group. Banker (2000) stressed the importance of the use of non-
financial measures. In a hotel environment he found a relationship between customer
satisfaction and future financial performance.
2.4 Hypothesis development
An interesting variable to see whether everybody is in favor of the PMS would be the attitude
towards the performance measures. A hesitant attitude towards the PMS should be avoided in
the organization. In this paper, attitude reflects the perceived usefulness of the performance
measures of their department. As a company, it's important to know how personnel are best
motivated to comply with the PMS. I will examine whether the variables that Wouters and
Wilderom propose in the development process have impact on the degree of enabling
Page 21
13
formalization of the PMS. As Adler and Borys (1996) propose, enabling formalization leads
to a more positive attitude. So the characteristics will also impact the attitude of the
employees. This basic assumption will also be tested.
The theory of Adler and Borys uses the formalization of the organization, opposing to the
theory of Simons. The formalization movement in research is much wider and it can be
applied to the whole organization, but in this paper I focus specifically on the PMS. The
levers of control framework can only be applied to the control system. I will zoom in on the
interactive and the diagnostic use, which only provide the data usage of the performance
measures. The following propositions are in the alternative form, aligned with the
expectations (H1).
Propositions
A. The degree of interactive (diagnostic) use of the performance measures will be
positively (negatively) associated with the degree of enabling formalization
Comparing Wouters and Wilderom with Adler and Borys, there are some similarities
in these two trends of research. An enabling formalization focuses on making
procedures simpler for employees, which is similar to the interactive use that promotes
a dialogue between the top-level management and the other employees. In a coercive
formalization, data is used to make the employees comply with the rules while
diagnostic use consists of tracking the effectiveness and efficiency (Simons, 1995). So
there are obvious similarities in these concepts. However, in the characteristics of
formalization there is no mentioning of the data usage. Hence this could be an
interesting addition to the formalization theory.
B. The degree of professionalism has a positive impact on the degree of enabling
formalization
This is one of the three developmental variables of the PMS suggested by Wouters and
Wilderom (2008) which drives enabling formalization. In an environment which
encouraged experiences and experimentation by the workforce, this variable had a
positive impact on attitude towards the performance measures of the department.
Page 22
14
C. The degree of experience-based development has a positive impact on the degree of
enabling formalization
This is one of the three developmental variables of the PMS suggested by Wouters and
Wilderom (2008) which drives enabling formalization. It is stated that local
knowledge is very important to create enabling formalization.
D. The degree of experimentation has a positive impact on the degree of enabling
formalization
This is one of the three developmental variables of the PMS suggested by Wouters and
Wilderom (2008) which drives enabling formalization. According to Wouters and
Wilderom, the possibility to tinker with the performance data is important to create
enabling formalization.
E. The degree of enabling formalization has a positive impact on the attitude towards
performance measures
When employees feel the PMS helps them to perform better, their attitude towards it
will be positive. In contrast, if the PMS is only in use to coerce the employees, their
attitudes won’t be positive. (Adler & Borys, 1996, p. 66).
F. The degree of interactive (diagnostic) use has a positive (negative) impact on attitude
towards performance measures
Interactive and diagnostic use show similarities with enabling and coercive
formalization. Interactive use has a positive relationship with capabilities while
diagnostic use has a negative relationship with capabilities (Henri, 2006). This leads to
the expectation that interactive use will be perceived as more useful, while diagnostic
use will be perceived as less useful. There is no association between these variables,
but a causal relationship. This stems from the fact that the attitude towards
performance measures of the organizational members is the result of the degree of use
of these measures and the degree of enabling formalization.
Page 23
15
3 RESEARCH METHOD
These hypotheses were researched by means of a questionnaire, which was distributed online
to random people in various organizations. For the body of research material to be both
extensive and representative, not only companies which use the BSC were contacted, but all
types of PMS’s were allowed. 1440 companies from Flanders and Brussels were contacted
through email to participate in this online survey (see appendix F). A reminder was sent after
2 weeks. Afterwards, an additional 1460 companies were contacted and a reminder was sent.
In total there were 87 participants. Their email addresses were selected on Belfirst 2008, made
possible by Bureau Van Dijck. With this data, I have tested the hypotheses.
3.1 Participants
Participants should give a fair representation of the corporate life, which means people of all
levels and sectors should be included.
Industry Frequency Function Frequency
Processing industry (textile, food, drinks) 11 Direction 28
Construction industry 8 Sales and marketing 7
Metal industry 12 Human Resources 33
Wholesale and retail 6 Accounting and finance 2
Catering industry 3 Operations 12
Chemical industry 5 ICT 5
Energy and water 4 Total 87
Agriculture and forestry 0 Table 2:
Transport and communication 6 Participants’ functions
Services 32
Financial sector 0
Total 87
Table 1: Participant’s industries
Responsibility Frequency
Top Management 36
Middle Management 33
Lower Management 9
Non Management 9
Total 87
Table 3: Participants’ responsibility level
Page 24
16
A few remarks are in order. Not a single participant from the financial sector seemed to
submit the survey. This is a possible consequence of the financial crisis of 2008. There is also
a large part of HR-people who completed the survey. This is no problem in such. However,
36 top managers entered in this survey. This could be a problem, in particular in combination
with diagnostic and interactive use. The survey defines these uses as the way top management
uses performance measures. However, if the participant is part of the top management, that
could cause distortion when explaining for the attitude. Instead of filtering out these
observations, a dummy variable will be used to indicate the top managers.
3.2 Construct development
I measured the items of the constructs on a 7-point Likert scale. All items are reformulated
into statements which the test persons should agree or disagree with. Then they were graded
on a scale from 1 to 7.
The degree of interactive and diagnostic use, the way top management uses these performance
measures, was constructed by Henri (p. 551).
The degree of professionalism (p. 514) and attitude towards performance measures (p. 513 –
514) has been tested by Wouters. The constructs ‘team trust’, ‘work pressure’ and ‘leadership
style’ (p. 501) were defined by Wouters.
The other variables were constructed for the use of this paper, in accordance with the
definitions given in the literature study (see appendix A). Constructs are calculated by taking
the average of these items. Thus, for all of these constructs values range from 1 to 7. 1 stands
for a very low level of this construct and 7 for a very high level of this construct.
One item of repair was removed, and repair was constructed together with internal
transparency, since there are strong similarities in these concepts (see appendix A).
Afterwards also the degree of enabling formalization was constructed, which is comprised of
repair, internal transparency, global transparency and flexibility. They are all coded in such a
way that a high degree of enabling formalization is a high number. All construct reliabilities
can be evaluated as acceptable to very good, the common cut-off level of 0,70 is maintained
with all constructs.
Correlation analysis gives a number of significant relationships, but no problems should be
expected. For example, repair and internal transparency, global transparency and flexibility
Page 25
17
are all strongly positively correlated with each other and with the overall degree of enabling
formalization, but this is trivial. There also appears to be a positive association between the
degree of interactive and diagnostic use, which is possible, but can give an unexpected result.
Other significant correlations are much lower, but these correlations are mainly a problem due
to sample size. This does not influence parameter estimation in the regression models.
Page 26
18
Construct
Diagnostic
use
Interactive
use Professionalism Attitude
Team
trust
Work
pressure
Leadership
style
Repair & internal
transparency
Global
transparency Flexibility Formalization Experimentation
Experience-
based dev.
Cronbach Alpha 0,892 0,897 0,886 0,959 0,721 0,801 0,949 0,729 0,823 0,81 0,889 0,895 0,826
Items 4 7 25 10 7 14 10 10 5 5 20 7 6
Mean 5,7356 4,9343 5,7255 5,6391 5,2841 3,7939 5,1885 4,6828 5,0759 5,1356 4,8943 4,5107 4,8544
Std. Deviation 1,03735 1,06522 0,52849 0,97484 0,81061 0,70731 1,07643 0,71743 0,90129 0,95015 0,72443 1,25765 1,1116
Minimum 1,5 1 4 1 3 2,21 2 2,4 2 1 2,2 1 1,5
Maximum 7 7 6,64 7 6,71 5,36 7 6,6 7 6,6 6,2 6,71 7
N 87 87 87 87 87 87 87 87 87 87 87 87 87
Pearson Correlations
Diagnostic use 1
Interactive use ,739** 1
Professionalism ,385** ,385** 1
Attitude ,706** ,532** ,495** 1
Team trust ,288** 0,205 ,442** ,435** 1
Work pressure -0,105 -0,107 -0,125 -0,175 -0,199 1
Leadership style ,376** ,344** ,435** ,543** ,294** -0,102 1
Repair & internal
transparency ,472** ,458** ,412** ,521** ,340** -0,117 ,215* 1
Global transparency ,698** ,631** ,328** ,670** ,342** -0,077 ,417** ,691** 1
Flexibility ,644** ,608** ,404** ,705** 0,187 -0,118 ,406** ,605** ,692** 1
Formalization ,662** ,623** ,439** ,698** ,336** -0,12 ,369** ,909** ,880** ,843** 1
Experimentation ,450** ,351** ,334** ,598** 0,2 -0,118 ,472** ,344** ,467** ,548** ,496** 1
Experience-based
development ,514** ,335** ,347** ,618** ,278** -0,141 ,318** ,441** ,552** ,505** ,556** ,770** 1
**. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed).
Table 4: Construct analysis
Page 27
19
4 RESULTS
4.1 Association degree of diagnostic and interactive use with degree of enabling
formalization
Using the correlation matrix proposition A can already be commented. There appears to be a
positive association between the degree of enabling formalization and the degree of
interactive use, as predicted, but also between the degree of enabling formalization and the
degree of diagnostic use. The Pearson correlation between the degree of diagnostic use and
the degree of enabling formalization is 0,662, which is significant at the 2-tailed 0,01 level.
This correlation for the degree of interactive use and the degree of enabling formalization is
0,622, which is also significant at this level.
If this use is tested in 2 groups, one with a high level of enabling formalization (> 5) and one
with a low level of enabling formalization (<= 5), this can be confirmed.
Formalization N Mean Std. Deviation
Diagnostic use Lower than 5 46 5,3152 1,17548
Higher than 5 41 6,2073 0,57256
Interactive use Lower than 5 46 4,4783 1,08304
Higher than 5 41 5,446 0,78259
Table 5: Description data of degree of use and formalization for t test
t Sig. (2-tailed)
Diagnostic use 4,574 0,000 (Equal variances not assumed)
Interactive use 4,725 0,000 (Equal variances assumed)
Table 6: Independent-samples t test of degree of use and formalization
The independent-samples t test confirms that both types of uses are positively correlated with
the degree of enabling formalization. This nuances proposition A: not only the degree of
interactive use, but also the degree of diagnostic use, is positively associated with the degree
of enabling formalization.
Page 28
20
4.2 Regression on attitude towards performance measures
A regression equation that explains attitude towards performance measures is estimated. First
the control variables Wouters and Wilderom suggested were included but afterwards they
were omitted (see appendix B).
Coefficients Test
Dependent variable: attitude B
Std.
Error t Sig. (2-sided)
(Constant) 0,818 0,476 1,718 0,090*
Diagnostic use 0,405 0,108 3,743 0,000*
Interactive use -0,131 0,097 -1,359 0,178
Repair & internal transparency 0,059 0,133 0,444 0,659
Global transparency 0,199 0,131 1,521 0,132**
Flexibility 0,353 0,111 3,198 0,002*
Top management member ( 0 = No, 1 = Yes) 0,114 0,142 0,808 0,421
* Significant at 2-sided 1% level R² = 0,631
** Significant at 2-sided 13,2% level
Table 7: Regression on attitude towards performance measures
A total R² of 63.1% is achieved, which indicates a model which explains a lot of the total
variation. There is an F value of 22,766. So the null hypothesis that all the coefficients are not
significantly different from 0 can be rejected with a p-value of 0,000. A dummy variable top
management member is included, which captures the effect of the people which indicated
they are top manager. However, a positive impact from the degree of diagnostic use is found,
against expectations. If the degree of diagnostic use goes up by one point on the 7-point scale,
the attitude towards performance measures goes up by 0,405 points on the 7-point scale. The
degree of interactive use does not appear to be significant. Flexibility is significant at the 2-
sided 1% level, global transparency is significant at the 2-sided 13,2% level. If global
transparency goes up by 1 point, attitude towards performance measures goes up by 0,199
points. If flexibility goes up by 1 point, the attitude towards performance measures goes up by
0,353 points. The variable repair and internal transparency is not significant. However, there
is a large positive correlation between these 4 characteristics of enabling formalization. The
null hypothesis that the residuals are normally distributed, can not be rejected, according to
the Kolmogorov-Smirnov test for normality. The p-value here is 0,20. The dummy variable
which was added is not significant.
Page 29
21
To check whether it is right to add the degree of interactive use, the degree of diagnostic use
and the dummy variable top management member in this regression model, another regression
equation explaining attitude was created, this time using only the characteristics of enabling
formalization suggested by Adler and Borys (see Appendix C). The same variables (global
transparency and flexibility) were significant, but only an adjusted R² of 0,545 was achieved.
In the regression that includes diagnostic use, interactive use, and the dummy top
management member, the adjusted R² is 0,603. So by adding these variables, the regression
stays efficient, so it’s a good addition to this model.
These findings partly explain proposition E: only the characteristics global transparency and
flexibility of the degree of enabling formalization have a significant positive influence on the
attitude towards performance measures. Proposition F has to be nuanced: only diagnostic use
has an impact on the attitude towards performance measures, however a positive impact.
4.3 Regression on the degree of enabling formalization
Then a regression is made to explain the degree of enabling formalization.
Coefficients Test
Dependent variable: formalization B
Std.
Error t Sig.
(Constant) 1,292 0,682 1,894 0,062
Experimentation 0,07 0,078 0,895 0,373
Experience-based development 0,24 0,089 2,69 0,009*
Professionalism 0,371 0,127 2,924 0,004*
* Significant at 2-sided 1% level R² = 0,383
Table 8: Regression on degree of enabling formalization
The F-value of the equation is 17,203. So the null hypothesis that all the coefficients are not
significantly different from 0 can be rejected with a p-value of 0,000. The degree of
experimentation is not significant. However, the degree of experience-based development and
the degree of professionalism both have a positive sign and are significant at a 1% level. If the
degree of experience-based development goes up by 1 point on the 7-point scale, the degree
of enabling formalization goes up with 0,24 points on the 7-point scale. If the degree of
professionalism goes up by 1 point, the degree of enabling formalization goes up by 0,371
Page 30
22
points. The null hypothesis that the residuals are normally distributed, can not be rejected with
a p-value of 0,195, according to the Kolmogorov-Smirnov test for normality. Again, this is
good news for this model. This model supports proposition B and C, but no evidence is found
that supports proposition D.
4.4 Econometric issues
4.4.1 Normality of residual terms
The null hypothesis that the residual terms are normally distributed, can not be rejected in
both equations, which means the constructed model still holds.
4.4.2 Heteroscedasticity
In both equations, there seems to be no constant variation in the residual term (see appendix
D). Especially the second equation tests positive on a White-test on heteroscedasticity.
However, problems aren’t severe. The exact residual term sorted for an independent value is
unknown. A White-correction can make the estimations of the standard error consistent,
which only applies in case of many observations. Since there only is a limited amount of
observations, there is no indication this correction would make standard errors of the
parameters more accurate. For that reason I chose not to make these corrections.
4.4.3 Multicollinearity
As stated before, there is a certain degree of multicollinearity in these equations, due to
correlations of independent variables. The VIF’s are calculated in Appendix E, but they
appear to be sufficiently low, so multicollinearity in the equations is rather low. The only
right measure against this would be adding more observations, but since the multicollinearity
problem isn’t severe, this isn’t necessary.
Page 31
23
4.4.4 Endogenity
Since I am dealing with 2 different equations, questions rise about endogenity and whether
there is a need to use other estimation models like Lisrel. Here however there is no need to
use structural equations modeling, since there only are two simple equations and no reciprocal
relationships in the formed theory. In the regression on attitude towards performance
measures, the variation of the characteristics of enabling formalization are not only explained
by the developmental variables, but also by externalities. So I conclude by stating that two
OLS regressions are in order here.
Page 32
24
5 CONCLUSION
Although there are similarities between interactive use and the way the rules are organized in
an enabling formalization and contrasting features between diagnostic use and enabling
formalization, there appears to be a positive association between the degrees of enabling
formalization, interactive and diagnostic use. The more the users of a PMS find it enabling,
the more data is used, whether it is in an interactive way or a diagnostic way. The attitude
towards these measures will be more positive if the data is used in a diagnostic way. A
possible explanation for this is that data must be used to follow up goals. Otherwise the
system will not be taken seriously. There is no support in the data that interactive use can
boost the attitude of the employees.
The data supports the developmental approach of Wouters and Wilderom (2008) towards
creating an enabling formalization. In regression analysis the degree of professionalism and
the degree of experience-based development have a significant positive impact on the degree
of enabling formalization, which proves this theory. For the degree of experimentation, no
evidence is found. I can conclude that, in the development process, experience-based
development and professionalism are the most important elements to create enabling
formalization.
An important basis for the enabling formalization theory is that employees will have a more
positive view towards this enabling formalization or “good rules”, and a more negative view
towards coercive formalization or “bad rules”. These characteristics of formalization were
split out and checked to see if they have an impact on the attitude towards the performance
measures. Although only a significant relationship was found with global transparency and
flexibility, there is a large correlation between these four concepts. These four concepts are
important to create a high degree of enabling formalization. However, according to this data, I
can state global transparency and flexibility are the most important drivers of the attitude
towards performance measures.
An important result of this paper is that in a firm which organizes itself in such a way that
there is a high degree of enabling formalization and combines this with a diagnostic way to
measure results, the attitude towards these performance measures will be most positive. I.e. it
will be valued more. Therefore, the system will be supported by the organizational members.
Page 33
V
References
Abernathy, M. A., & Brownell, P. (1999), The role of budgets in organizations facing
strategic change: an exploratory study, Accounting, Organizations and Society, 24(3), 189–
204.
Adler, P.S., Borys, B. (1996), Two types of bureaucracies: enabling and coercive,
Administrative Science Quarterly, 41, 61-89.
Ahrens, T. A., & Chapman, C. S. (2004), Accounting for flexibility and efficiency: A field
study of management control systems in a restaurant chain, Contemporary Accounting
Research, 21(2), 271–301.
Arches, J. (1991), Social structure, burnout and job satisfaction, Social Work, 36(3), 202-206.
Baer, M., & Frese, M. (2003), Innovation is not enough: Climates for initiative and
psychological safety, process innovations, and firm performance, Journal of Organizational
Behavior, 24, 45–68.
Banker, R.D., Potter, G., Srinivasan, D.(2000), An empirical investigation of an incentive
plan that includes nonfinancial performance measures, Accounting Review, 75(1), 65–92.
Barney, J. , Wright, M., Ketchen, D.J., The resource-based view of the firm: 10 years after
1991, Journal Of Management, 27(6), 625-641.
Barney, J. B. (1991), Firms resources and sustained competitive advantage, Journal of
Management, 17, 99–120.
Bass, B. M., & Avolio, B. J. (2000), MLQ multifactor leadership questionnaire (2nd ed.),
Redwood City: Mind Garden (Technical Report).
Bhuian, S. N., Menguc, B., & Bell, S. J. (2005). Just entrepreneurial enough: the moderating
effect of entrepreneurship on the relationship between market orientation and performance,
Journal of Business Research, 58(1), 9–17.
Bonjean, Charles M,, and Michael D. Grimes (1970), Bureaucracy and alienation: A
dimensional approach, Social Forces, 48, 365-373.
Caldwell, S. D., Herold, D. M., & Fedor, D. B. (2004). Toward and understanding of the
relationship among organizational change, individual differences, and changes in personal
environment fit: A cross-level study, Journal of Applied Psychology, 89(5), 868–882.
Davis, S., Albright, T. (2004) , An investigation of the Balanced Scorecard implementation on
financial performance, Management Accounting Research, 15, 135-153.
Day, G. S. (1994), The capabilities of market-driven organizations, Journal of Marketing, 58,
37–52.
Page 34
VI
Decoene, V, Bruggeman, W, (2006), Strategic alignment and middle-level managers’
motivation in a balanced scorecard setting, International Journal of Operations & Production
Management, 26, 429-448.
Hall, M. (2008), The effect of comprehensive performance management systems on role
clarity, psychological empowerment and managerial performance, Accounting, Organizations
and Society, 33, 141-163.
Heckscher, Charles (1994), Defining the postbureaucratic type, In C. Heckscher and A.
Donnellon (eds.), The Post-bureaucratic Organization: New Perspectives on Organizational
Change, 14-62, Thousand Oaks. CA: Sage.
Henri, J-F. (2006) , Management control systems and strategy: a resource-based perspective,
Accounting, Organizations and Society, 31, 529-558.
Hult, G. T. M., & Ketchen, D. J. (2001). Does market orientation matter?: a test of the
relationship between positional advantages and performance, Strategic Management Journal,
22 , 899–906.
Hurley, R. F., & Hult, G. T. M. (1998), Innovation, market orientation, and organizational
learning: an integration and empirical examination, Journal of Marketing, 62, 42–54.
Kaplan, R. , Norton, D. (1992), The Balanced Scorecard: Measures That Drive Performance,
Harvard Business Review, 70, 71-79.
Kaplan, R., Norton, D. (1993), Putting the Balanced Scorecard to Work, Harvard Business
Review, 71 (5), 134 – 142.
Lengnick-Hall, C. A., & Wolff, J. A. (1999), Similarities and contradictions in the core logic
of three strategy research streams, Strategic Management Journal, 20, 1109–1132.
Lillis, A. M. (2002), Managing multiple dimensions of manufacturing performance—An
exploratory study, Accounting, Organizations and Society, 27, 497–529.
Lowe, A., & Jones, A. (2004), Emergent strategy and the measurement of performance: The
formulation of performance indicators at the microlevel, Organization Studies, 25(8), 1313–
1337.
Malina, M. , Selto, F. (2001), Communicating and controlling strategy: an empirical study on
the effectiveness of the Balanced Scorecard, Journal of management accounting research, 48-
90.
Nohria, N., Groysberg, B., Lee, L-E. (2008), Employee Motivation: A Powerful new model,
Harvard Business Review, 78-84.
Page 35
VII
Perrow, Charles (1983), The organizational context of human factors engineering,
Administrative Science Quarterly, 28, 521-541.
Rousseau, Denise M. (1978), Characteristics of departments, positions and individuals:
Contexts for attitudes and behavior, Administrative Science Quarterly, 23: 521-540.
Simons, R.(1995) , Control in an Age of Empowerment, Harvard Business Review, 80-88.
Srivastava, A. , Bartol, K.M., Locke, E.A. (2006), Empowering leadership in management
teams: effects on knowledge sharing, efficacy and performance, Academy of Management
Journal, 49(6), 1239-1251.
Stanton, J. M., Balzer, W. K., Smith, P. C., Parra, L. F., & Ironson, G. (2001), A general
measure of work stress: The stress in general scale, Educational and Psychological
Measurement, 61(5), 866–888.
West, J., & Iansiti, M. (2003), Experience, experimentation, and the accumulation of
knowledge: The evolution of R&D in the semiconductor industry, Research Policy, 32(5),
809–825.
Widener, S. K.(2008), An empirical analysis on the levers of control framework, Accounting,
Organizations & Society, 32, 757-788.
Wouters, M. , Wilderom, C. (2008) , Developing performance-measurement systems as
enabling formalization: A longitudinal field study of a logistics departement, Accounting,
Organizations and Society, 33, 488-516.
Zollo, M., & Winter, S. G. (2002), Deliberate learning and the evolution of dynamic
capabilities, Organization Science, 13(3), 339–351.
Page 36
VIII
Appendix A
Construct items
Formalization:
Repair
Employees can mend work processes themselves.
There is no need to contact superiors when dealing with problems with performance
measures and performance data.
I can mend control processes myself if necessary.
Problems in the control system are quickly mended by employees.
If there is a problem with performance measures, only top management can change
them (omitted).
Internal Transparency
There is a good understanding about the key performance indicators in your domain.
The reasoning behind the performance management system in your domain is
understandable.
It is understandable why certain control measures are taken in your domain.
There is information about the current status of the key performance indicators in your
domain.
Best-practices experiences about the control system are made available.
Feedback is given on my performance.
Global Transparency
There is a good understanding about the key performance indicators in the whole
organization.
The reasoning behind the performance measures of the whole organization is
understandable.
It is understandable why certain control measures are taken in the whole organization.
There is information about the current status of the key performance indicators of the
whole organization.
The link between your own tasks and the broad organization is clear.
Page 37
IX
Flexibility
You can make controlling decisions personally after information is given by the
performance measurement system.
You can take initiatives to improve the key performance indicators of the performance
measurement system.
Your own decisions are very important when trying to improve key performance
indicators.
The interface of the performance measurement system can be altered.
Functionalities can be added to suit specific work demands.
Experimentation
Management and employees worked together to develop the used measures.
Performance measures were altered several times during development.
Employees had the possibility to experiment with different measures during development.
I was able to experiment with different measures in general during development.
I was able to alter tables and graphs during development.
I had impact on the procedures for data gathering during development.
I was able to alter qualitative descriptions during development.
Experience-based development
I was asked to give my opinion towards the performance measures during development.
I was able to share experiences that could be relevant during development.
My input was used in the development process of the performance measures.
BSC was developed exclusively by the top management.
Opinions about performance measures given by employees were appreciated during
development.
Local knowledge was documented during development.
Page 38
X
Construct development of repair and internal transparency
Cronbach Alpha Items
Repair before omission 0,195 5
Repair after omission 0,227 4
Internal transparency 0,881 6
Repair after omission & internal transparency 0,729 10
Repair appeared to be a difficult item to describe. I opted to delete one item of repair, to get a
higher Cronbach Alpha. Repair and internal transparency were conflated, since these two
characteristics of enabling formalization are closest to each other. In doing so, a Cronbach
Alpha of 0,729 is achieved.
Page 39
XI
Appendix B
First regression on attitude towards performance measures
First a regression on attitude included the control variables Wouters suggested:
Model Coefficients Test
Dependent variable: attitude B
Std.
Error t Sig.
(Constant) -0,172 0,677 -0,253 0,801
Diagnostic use 0,355 0,099 3,607 0,001*
Interactive use -0,128 0,088 -1,459 0,149
Repair & internal transparency 0,043 0,125 0,348 0,729
Global transparency 0,09 0,121 0,741 0,461
Flexibility 0,343 0,103 3,345 0,001*
Top management member ( 0 = No, 1 = Yes) 0,084 0,129 0,656 0,514
Team trust 0,221 0,084 2,63 0,01*
Work pressure -0,051 0,087 -0,583 0,561
Leadership style 0,187 0,065 2,872 0,005*
* Significant at 2-sided 1% level R² = 0,710
The F-value of the equation is 20,939. So the null hypothesis that all the coefficients are not
significantly different from 0 can be rejected with a p-value of 0,000.
However, the residuals appear not to be normally distributed: the Kolmogorov-Smirnov test
for normality has a p-value of 0,001. So the null hypothesis that the residuals are normally
distributed is rejected.
It is chosen to delete these three control variables for a couple of reasons
• There is no research interest in these variables.
• No clear rationale was given why these variables influence the attitude towards
performance measures.
• Opposing to Wouters and Wilderom (2008), other broader variables are used to
explain attitude towards performance measures. There is no indication there is still
need for these variables.
• It is possible that the Kolmogorov-Smirnov test can not reject that residual terms are
normally distributed, after omission.
One objection could be that when there are fewer variables included, it is possible that
variables will be less significant. However, I opted to omit these variables from the
regression.
Page 40
XII
Appendix C
Regression to test the impact of the degree of enabling formalization
A regression on attitude towards performance measures was made, but only including the
characteristics of the enabling formalization.
Coefficients Test
Dependent variable: attitude B
Std.
Error t Sig.
(Constant) 1,282 0,489 2,623 0,01*
Repair & internal transparency -0,003 0,141 -0,021 0,983
Global transparency 0,38 0,124 3,073 0,003*
Flexibility 0,476 0,106 4,466 0,000*
* Significant at the two-sided significance level of 1%
R² = 0,561
We see that the same variables (global transparency and flexibility) are significant here. An
adjusted R² of 0,545 is achieved.
According to a Kolmogorov-Smirnov test, the null hypothesis that residuals are normally
distributed can not be rejected, with a p-value of 0,20.
Page 41
XIII
Appendix D
Heteroscedasticity tests
Coefficients Test
B
Std.
Error t Sig.
(Constant) 1,761 0,455 3,866 0
Diagnostic use -0,1 0,103 -0,967 0,336
Interactive use 0,032 0,092 0,342 0,733
Repair & internal transparency -0,091 0,127 -0,712 0,479
Global transparency -0,001 0,125 -0,009 0,993
Flexibility -0,104 0,106 -0,981 0,33
Top management member ( 0 = No, 1 =
Yes) -0,084 0,135 -0,618 0,538
a. Dependent Variable: Residues on attitude squared R² = 0,13
n* R² = 87 * 0,13 = 11,31
Chi² (6 degrees of freedom, 5% significance) = 12,5916 ( for n -> ∞)
The critical value is not surpassed.
According to the test, this equation is significant. So there is no heteroscedasticity (if,
however, a few variables are dropped, the degrees of freedom will make this model
significant).
Coefficients Test
B
Std.
Error t Sig.
(Constant) 1,039 0,621 1,672 0,098
Experimentation 0,004 0,071 0,049 0,961
Experience-based development -0,185 0,081 -2,279 0,025
Professionalism 0,028 0,115 0,247 0,806
a. Dependent Variable: Residues on formalization squared
R² = 0,127
n*R² = 87 * 0,127 = 11,049
Chi² (3 degrees of freedom, 5% significance) = 7,81473 ( for n -> ∞)
The critical value is crossed.
According to the test, this equation is significant. So there is heteroscedasticity to a certain
degree.
Page 42
XIV
Appendix E
Multicollinearity
To test the multicollinearity, the variance of the parameters is split out in a VIF and the basic
variance, to see to what degree collinearity inflates the variance of the parameters.
VIF Basic variance Total variance
Dependent variable: attitude
Diagnostic use 2,867 0,004 0,012
Interactive use 2,417 0,004 0,009
Repair & internal transparency 2,076 0,009 0,018
Global transparency 3,162 0,005 0,017
Flexibility 2,514 0,005 0,012
Top management member ( 0 = No, 1 = Yes) 1,122 0,018 0,020
Dependent variable: formalization
Experimentation 2,492 0,002 0,006
Experience-based development 2,517 0,003 0,008
Professionalism 1,151 0,014 0,016
Although there are some significant correlations, the VIF’s don’t give any indication of a
large multicollinearity problem. A rule of thumb is that there is a problem when the VIF is
larger then 5, there is no multicollinearity problem. This indicates there are no
multicollinearity problems. But also when looking at the VIF in combination with the basic
variance, we can state that the basic variance is sufficiently small. So when it is multiplied
with a larger VIF, the total variance will be still small enough. Here I can state
multicollinearity issues are not severe.
Page 43
XV
Appendix F
Survey
Participating request
Mail
Onderwerp: Onderzoek metingen performantie Universiteit Gent
Beste,
Ik ben een Masterstudent Toegepaste Economische Wetenschappen aan de Universiteit te Gent. Als
Masterproef doe ik een academisch onderzoek, met als onderwerp het ontwerp en gebruik van
prestatiemeetsystemen.
D.m.v. de vragenlijst , waarvan de link onderaan vermeld staat, verzamel ik gegevens over de wijze
waarop prestatiemeetsystemen tot stand komen en gebruikt worden.
Ik zou u zeer erkentelijk zijn, mocht u deze enquête willen invullen of deze mail willen doorsturen
naar een geschikt persoon (een managementfunctie) binnen uw onderneming? De enquête neemt
slechts 10 minuten in beslag.
Alle gegevens zullen anoniem verwerkt worden. Als u wil, kan u ook de resultaten van mijn
onderzoek vernemen. Hiervoor contacteert u me best op mijn emailadres.
Om de enquête in te vullen, gelieve door te klikken naar de volgende link:
Indien de link niet werkt, gelieve deze dan te kopiëren naar de navigatiewerkbalk van uw
webbrowser.
Alvast bedankt,
Met vriendelijke groeten,
Kevin De Ruyck