Wi JN Hoven 95 Organizational
Wi JN Hoven 95 Organizational
Wi JN Hoven 95 Organizational
Information Systems
The Case of Monitoring Information and Control
Systems in Machine Bureaucratic Organizations
Fons Wijnhoven
CIP-GEGEVENS KONINKLIJKE BIBLIOTHEEK, DEN HAAG
Organizational learning and information systems : the case of monitoring information and
control systems in machine bureaucratic organizations / Alphonsus Boudewijn Jacobus
Maria Wijnhoven. - Enschede : Universiteit Twente, Faculteit der Technische
Bedrijfskunde
Proefschrift Universiteit Twente Enschede. - Met index, lit. opg. - Met samenvatting in het
Nederlands.
ISBN 90-365-0732-4
Trefw.: management informatiesystemen
PROEFSCHRIFT
door
Alphonsus Boudewijn Jacobus Maria Wijnhoven
geboren op 29 oktober 1957
te Sint-Anthonis
Dit proefschrift is goedgekeurd door de promotoren:
Preface ................................................................................................................................xiii
References ......................................................................................................................293
Index ................................................................................................................................307
This book is the result of a six-year part-time study on organizational learning and
information systems conducted at the School of Management Studies of the
University of Twente. The study was enabled by the research position I received in
the Department of Information Management.
Before the start of this project, I taught government organization at the School
of Public Administration of the University of Twente. Specifically I would like to
acknowledge prof. Chris L. Menting, head of the Government Organization Group
for introducing me to many aspects of organization analysis. In that group I
acquired my interest in organizational learning.
In 1987, I started work as assistant professor in information management at
the School of Management Studies. In 1988, I started this project under the
supervision of professors P.A.E. (Lisa) Van de Bunt and Ronald K. Stamper. Prof.
Van de Bunt recommended to me various papers and books about organization
studies from the large diversity of perspectives that exist in that field. To these he
also added his own practical experience. This led to a broad perspective for this
project, and a close connection of the subject with actual trends in world economy,
as discussed in chapters 1 and 2 of this book. Prof. Van de Bunt also convinced me
to focus the study on organizations with the greatest learning problems (machine
bureaucracies), and on the information systems that are most criticized from the
organizational learning perspective (monitoring information and control systems).
Arguments for this choice are explained in chapters 5 and 6. Prof. Stamper's
erudition together with an interesting and unique approach to the field of informa-
tion management formed excellent additions. One of his major ideas, semantic
analysis, is applied in chapters 4 and 7, while trying to organize the many and
confusing concepts and ideas related with organizational learning. His view on
semantics has been guiding the conceptualization and operationalization of MICS.
Prof. Stamper is also acknowledged for the considerable effort he took to manage
the EMIR (Effective Management of the Information Resource) research pro-
gramme, within which this project was carried out. I particularly appreciated his
effort in being my harshest critic.
Many other colleagues and friends should be mentioned here. I will keep the
list short as the book has become long enough already. Specifically I want to
mention dr Tinus van Drunen, Ir Cees van Slooten, drs Arjen Wassenaar with
whom I had many discussions about the concept of information system and the
importance of my study for management studies and information management. In
contrast to what is most common in computer science, I decided on basis of these
discussions, to define information systems as technical and social systems, and thus
not limited to hardware, software and data.
During this project organizational learning was becoming increasingly popular
as a theme in organization and information science. This was particularly the result
of work done by Peter Senge at MIT and A. De Geus at Shell Planning. Both
authors stressed the importance of constructing 'The Learning Organization'. I felt
very uncomfortable with the concept of 'The Learning Organization' because of the
unconditionality of the knowledge thus gained, like a carpenter who thinks that the
world exists of nails only. Therefore, I developed the concept of organizational
learning needs, to introduce a contingency element in my approach of
organizational learning. This idea is briefly expressed in chapter 4, and is described
in more detail in chapter 5, where the organizational learning of four types of
organizations is linked with the learning needs of these organizations. This idea was
also strongly supported by two of my M.Sc. students, Mark Hafkamp and Stephan
Kordelaar, who contributed substantially to this project by helping me with the data
collection in the five cases studied in chapter 8. These cases enabled me to further
work out the contingency approach to organizational learning.
I also want to thank other M.Sc. students who created important contacts with
business companies, which later on where willing to participate in this project.
Thanks to my friends dr Geerten Schrama (now at the Center of Environmen-
tal Management at the School of Public Administration) and dr Timo Saarinen from
the Helsinki School of Economics for commenting on a previous manuscript. I also
want to express my appreciation for professors B. Hedberg (Stockholm Business
School), C.L. Menting, E. Spoor (Free University Amsterdam) and O.A.M.
Fisscher (School of Management Studies, University of Twente) for acting as
members of my promotion committee and reviewing the work. Thanks also to Ms.
K. Emmett, who carefully checked my English.
Finally, I want to thank my wife Carolyn Karthaus, and my children Kim,
Jules and Armelle. They all gave me the support needed to continue.
Before starting to read this book, the reader might find the following notes
useful. Chapters 1, 2 and 3 are about the design of this project and its research
methodological basis. A reader interested solely in the concept of organizational
learning can skip these three chapters, and start reading chapters 4, 5, and 6.
Readers interested in the empirical case studies should carefully read chapters 1 to
7. Chapter 8 might be incomprehensible without reading all the preceding chapters,
because these demonstrate the basic concepts, hypotheses and goals of the case
studies. Chapter 9 is a concluding chapter, based on all insights gained in the
previous chapters. The basic conclusions might be read, however, without reading
the preceding chapters.
Minutes of Effort to Assemble Late Craft Production, Mass Production, Percent Reduction
Fall 1913 Spring 1914 in Effort
Engine 594 226 62
Magneto 20 5 75
Axle 150 26.5 83
Major Components into a
Complete Vehicle 750 93 88
Source: Womack et al., 1990, p. 29, figure 2.1
Table 1.1: Craft Production in 1913 versus Mass Production in the Assembly Hall: 1913 in
1914.
The emphasis on cost reduction was the basis for investing in more machinery.
According to Zuboff (1988), at the start of this mechanization process in the 18th
2 Organizational Learning and Information Systems
and 19th centuries, it was not at all clear that manufacturing would be cheaper than
craftsmanship. According to Ashworth (1975), coal consumption and steel
production, both essential for mechanization, did not increase to a level suitable for
manufacturing before the 20th century.
The second half of the 20th century, however, led to a new era, often called the
post-industrial society (Bell, 1979), in which fabrication was replaced by the
processing and recycling of services as the dominant mode of production. The
manufacturing companies that stayed in existence had to tranform from output-
driven to market-driven organizations. The most successful machine bureaucratic
forerunners of this society are the Japanese manufacturers that capitalize on their
abilities to meet three market trends: quality, flexibility and innovation.
Often it has been suggested that better products should be more expensive. However,
the evidence seems to contradict this statement (Womack et al., 1990, p.92 and 93),
as low quality is in fact an important cost driver (especially restoration costs).
be produced in large series or bought from a supplier, and thus profit from the
economies of scale. The components could be assembled in a large variety of ways,
thus enabling some customization of the final products. Information technology was
also introduced for optimizing production and market demands (MRP, Aggerwal,
1985) and improving the flexibility of production devices (for instance with flexible
manufacturing systems and CAD/CAM, cf. Kerr, 1991). This situation differs
considerably from the philosophy of Henry Ford, who stated that clients could buy
any car, so long as it was a T-model and black. Nowadays the number of car types is
much greater than that available in the heyday of mass production, as is illustrated in
table 1.3.
1
JIT is the acronym for Just-in-Time. JIT-deliveries imply that inventories can be kept extremely low,
without the risk of being out of stock.
4 Organizational Learning and Information Systems
In the 1980s, companies also started competing on the basis of their innovation
potential. Especially in the 'high tech' line many new products were launched (cf.
Hamel and Prahalad, 1990). At that time, information technology opened the way to
new forms of competition, that especially affected information-intensive service
industries such as banking and insurance (Cash et al., 1992). Many product
innovations like Automatic Teller Machines and home banking emerged in Western
Countries. Mergers among insurance companies and banks aloowed for a broader
services portfolio, leading to synergies2, and improved client services (so-called total
financial service, to optimize the customers' added value). Also, flexibility was
improved by providing clients with a 24 hour access to services. At the end of the
1980s many I.T. competitive differentiation opportunities were exhausted and I.T.
became a necessary evil. Many financial service companies had financial troubles and
started a low price strategy to increase their share of the market (and thus reduce
overhead costs per client) (Porter, 1985). Low cost was thus introduced after quality
and innovation in the service sector. Mergers continued with the aim of reducing
production and marketing costs by the principle of economics of scale. At the same
time, many services were 'industrialized' to lower the production costs through
increased mechanization (Grönroos, 1991).
Innovation of products and the innovation potential of automobile factories
can be illustrated by measuring characteristics of the production development
process, measuring development lead times, and the number of product models that
are launched within a time period. Evidence on both issues is given in table 1.5.
2
In the U.S.A. these collaborations in one business sector were not permitted under anti-trust legislation.
Changes in Machine Bureaucracies 5
Womack et al. (p.120) also gathered data about the number of models and average
age of models. Clearly, Japanese producers outperform American and European
manufacturers on this aspect of innovation potential.
Third World countries have invested heavily in their intellectual capital, by investing
in higher education for the local population (Porter, 1990, p. 466). Additionally, they
are developing their own internal markets, by making very cheap products that may
not qualify for Western standards, but suit local needs. Many Third World countries
have competitive advantages over for instance Japan: they have more space and
cheaper energy resources. They can learn from Japan that the main advantage to be
gained when use is made of both body and mind. This opportunity has never been
made use of by the colonial industry, which only sought to exert control, thus
keeping the minds of the workforce lazy as in a command economy.
Organizational learning was already an essential feature of ancient civilizations.
The Egyptians could not have constructed their pyramids without a well-developed
body of knowledge about construction and organization. For instance, the pyramid of
Cheops covers thirteen acres and contains 2,300,000 stone blocks, each weighing an
average of two-and-a-half tons. Their construction is estimated to have taken over a
hundred thousand men a period of twenty years (George, 1972, p.4). The Chinese
Empire, the Roman Empire, the Dutch colonial company (East Indies Company, a
multinational founded in 1602) all required large bureaucracies to control and
coordinate activities among the many people involved. On other words,
organizational learning in large organizations was connected with the development of
bureaucracies. Bureaucracies are often dominated principally by control, leading to
an internal command economy with the dysfunctional impacts as mentioned earlier.
The post-industrial society, also called the information age (Bell, 1979),
introduces new ways of management based on:
1. Organizational perestrojka: the introduction of organization internal market
principles and democracy (Ackoff, 1992; Peters, 1992).
2. Removal of many middle management positions whose tasks could be carried
out more efficiently by a computer or a datacommunication network (Leavitt
and Whisler, 1958).
These organizations are better learners because of their increased ability to connect
and process ideas and data, and their improved efficiency in communication (Douma
and Schreuder, 1991; Gurbaxani and Whang, 1991). Computers can make a
significant contribution to learning, because they can reduce the costs for distributing
ideas and data, and can reduce the costs and capacity for data storage and analysis.
The added value of computers is, however, not always obvious. Often there are cases
of bad use, non-use and negative impacts of computers (Jayaratna, 1990; Williams,
1991). Under what conditions then are computers useful and under what conditions
are they damaging to organizational learning? The answer to this question requires a
broader perspective of information systems than computer-based systems (Stamper,
1973), because organizational learning makes use of more ways of data processing
than that carried out by the computer (cf. Nonaka, 1988).
Let us study these questions by focusing on bureaucracies that have developed
large stores of knowledge in the past by developing good working principles. Let us
Changes in Machine Bureaucracies 7
also focus on organizations that apply the simplest computer tool for management
learning, called Monitoring Information and Control Systems, systems that give
regular reports for developing insights about how things are going.
Information systems are often defined as:" ...an integrated, user-machine system for
providing information to support operations, management, and decision-making functions in an
organization. The system utilizes computer hardware and software; manual procedures; models
for analysis, planning, control and decision making; and a database (Davis and Olson,
1984, p.6). Management information systems can only realize their purpose, i.e.
informing managers so they can make the best decisions, when the organization is
changed as well. Research and practice in the area of management informations has
therefore shown that it is much wiser to define an information system as a social
system consisting of models for analysis, and rules for information handling, that are
possibly served by information technology in the form of computers and data
Changes in Machine Bureaucracies 9
Lee and Guinan (1991) tried to find an answer to the question: what are MRSs
contributions to the managerial task. More specifically they tried to measure the
influence of an MRS on the management's ability to control and innovate the
organization (two basic factors also of organizational learning) in one specific
company. In operationalizing control, they discovered two types of control that were
linked with competing theories. The first, called managerial control, is about a
manager's effectiveness in influencing organization members to behave according to
organizational directives. The second type, called self-control, is about organization
members' own ability to increase organizational effectiveness. Self-control is
particularly important when task complexity increases (Hersey and Blanchard, 1982;
Galbraith, 1973). The researchers found several variables that define scores for
control and innovation in the organization, listed in table 1.7.
Variables of control and innovation These variables were used to measure
Managerial control subjective beliefs about the relation
• Task Clarification between control, innovation and IT-use.
• Work Assignment
• Procedural Specification (like standard
The data showed that especially the
operating procedures) planning and control scores were believed
Changes in Machine Bureaucracies 11
The study indicates that IT impacts are closely related to the organizational
norms. An identical system studied by Lee and Guinan would probably show a
different impact if the high amount of self-control that was measured in the
organization was replaced by managerial (hierarchical) control. In that case IT would
be used as an instrument to support the authority and power of managers, and thus
be mainly used for planning and control, decreasing the amount of self-control of the
employees (the Pierce-Delbecq and Hage-Aiken hypothesis). This study thus suggests
that, in that case, organizations would not profit from all the opportunities of MRS.
could indicate some learning, when for instance the users' understanding of
reality had increased. This could be checked by looking at the number of
alternatives considered.
2. Hypothesis 2: "DSS aided groups will show less variance in profit performance than
non-DSS aided groups (p.146)". This means that there is more certainty about the
outcomes of decision processes, which is an indication of learning. "On the other
hand, less variability in performance also may not be desirable. The decision aid may be
limiting risk taking by encouraging uniform decision processes and outcomes. Given the
significant differences in net earnings between the two groups, however, more stable
performance would appear to be preferable (Sharda et. al., p. 153)".
3. Hypothesis 3: "DSS aided groups will take less time to reach a decision than non-DSS
aided groups (p.146)". This is an indication of learning and probably also
speeding-up of communication. Sharda et al.'s study however did not find
significant statistical differences among DSS users and non-DSS users regadring
time.
4. Hypothesis 4: "DSS aided groups will consider a greater number of alternatives than
non-DSS aided groups (p.146)". This is also a very substantial indication for
learning. However, even though DSS supported groups reported that they had
investigated more alternatives than the non-DSS supported groups, the
differences were not statistically significant at a .05 probability level.
5. Hypothesis 5: "DSS aided groups will report greater confidence in their decisions than
non-DSS aided groups (p.146)". This could indicate learning, i.e. improvement of
memory contents and retrieval. Sharda et al.'s study however did not show
statistically significant differences between the DSS users and the non-DSS
using group.
Van Schaik (1988) carried out a second study on the impact of DSS, in which he
assumed that a DSS does not impact on decision-making quality, but that a decision-
making strategy does. This statement is often implicit in DSS-usage and DSS
research. To test this statement he created four experimental groups, among which
he varied the use or non-use of a DSS, and the use or non-use of a decision-making
strategy. His findings are summarized in table 1.8.
No DSS difference DSS
No Strategy 4.80 (1.33) not significant 4.84 (1.42)
difference significant at p < 0.05 significant at p < 0.06
Strategy 6.55 (1.09) not significant 6.57 (1.48)
Mean scores of experimental groups (standard deviations in brackets), after 7 trials. Source: Van Schaik,
1988, exhibit 4-11, p. 121.
Table 1.8: Research Results of DSS-impacts.
Table 1.8 clearly reveales that decision-making improvements are more the result of a
Changes in Machine Bureaucracies 13
well chosen decision strategy than the result of DSS usage. These findings imply that
Sharda et al.'s results could possibly be a consequence of research artefacts.
From neither study may it be deduced that a DSS contributes to organizational
learning in non-laboratory environments, because, for instance, management style,
interpersonal behavior and organizational norms were not modelled.
Markus (1984) identified and described some information systems from the
perspective of their role in organizations, as shown in table 1.10.
Information Sys- Design features Examples
tems Type
Changes in Machine Bureaucracies 15
MICS is the focus of our study, and will be treated in the following subsection.
Operational Systems are not systems for learning. Four other types of information
systems, not investigated in this study, could support organizational learning. These
are: knowledge-based systems, planning and decision systems, communication
systems and interorganizational systems.
Knowledge-based systems are intended to routinize knowledge available in different
parts of the organization so that one could do without the expert or support the
expert on routine matters. MICS should support the development of new insights for
efficiency and control or fundamental improvements, not by experts but by managers
and employees, and thus realizing a learning loop. According to Coats (1992),
knowledge-based systems are most valuable during the period of their development
because this requires the elicitation of knowledge (a very interesting learning
exercise), and for the distribution of codified knowledge in and between
organizations.
Planning and decision systems are important as well for realizing effective
organizations (JIT etc.). They could contribute significantly to organizational learning
when they allow experimenting without disturbing reality (Senge, 1990a). They can
help to make design teams more effective by facilitating the connection of the
members' knowledge, insights and ideas: a true example of organizational learning.
Communication systems can support organizational learning and are now creeping
16 Organizational Learning and Information Systems
The classic machine bureaucracy seems to inhibit innovation, but might be quite
good at making the existing process more efficient. Monitoring Information and Control
Systems (MICS) are very important in signalling issues for efficiency improvement. This
implies a restricted way of organizational learning, within the constraints of pre-set
targets. Lean machine bureaucracies seem to be intrinsically interested in
improvement, which suggests the possibility of using MICSs to provide incentives for
more fundamental problem solving. Because of their relatively simple character (a
database and some predefined reports), these systems are widely spread. Their
opportunities for control as well as for the fundamental analysis of problems make
them fit closely to the management style and culture of machine bureaucracies (a
further explanation for this statement is given in chapters 4 to 6). An essential feature
of MICSs is that they provide data for critical evaluation by means of data that are
compared to fixed targets. Depending on the organizational context this critical
evaluation can lead to three types of behavior:
1. Doing nothing, as when:
• people lack the means to make sense of the data.
• people lack the behavioral opportunities to act (lack of time, money,
motivation, difficulty with problem solving etc.).
• the data show that everything is on target, so there is no reason to act (of course
one should hope that the system is not misdirecting attention in this
case).
2. Giving rewards and punishments. The system then acts as an extension of the
internal competitive environment. It can easily lead to lack of learning because
people are not willing to share knowledge. If these political problems cannot be
Changes in Machine Bureaucracies 17
This study is based on three approaches for the study of organizations and
information systems. The first approach is cybernetics, that introduced the term
organizational learning (Bateson, 1972) to improve understanding of organizational
control. The second approach is organization development, that reformulated
organizational learning by emphasizing the human, interpersonal and political
aspects and preconditions for organizational learning. The third approach is
semiotics, that defines the nature of information and therefore gives insight into the
features of MICS. These approaches are used to define the major scientific problems
associated with information systems and organizational learning. A more detailed
theoretical analysis of the approaches is given in chapters 4 and 6.
The term cybernetics originates from the Ancient Greek word 'kubernetes' meaning
'steersmanship'. It was introduced into the modern language by Norbert Weiner
(1952) to describe the study of steering missiles to hit moving targets (like planes in
World War II). The problem this application of cybernetics initially wanted to solve
was that missiles are not able to catch planes without the constant input of
information about the target's movement and factors that could influence the
movement of the missile itself (like wind direction and speed). On the basis of this
information, deviations from the target are measured and corrective actions taken.
An everyday example of a cybernetic system is the house thermostat, by which the
temperature of a room can be kept at a constant (the target) despite changes in
temperature of the environment.
Garreth Morgan (1986, p. 86-87), summarizes the features of cybernetic systems in
relation to communication and learning as follows:
" First, that systems must have the capacity to sense, monitor, and scan significant aspects
of their environment. Second, that they must be able to relate this information to the
operating norms that guide system behaviour. Third, that they must be able to detect
significant deviations from these norms. And fourth, that they must be able to initiate
corrective action when discrepancies are detected."
These four features define an organizational learning process in its most elementary
form. The corrective actions initiated from learning can be reactive response,
behavioral adaptation, or organizational changes to improve the organization's
20 Organizational Learning and Information Systems
An example of the working of the adaptation and selection principles is the Roman Catholic Church, an
organization that has already existed for about 2000 years. To some extent it adapts to its environment, because
it connects its activities and policies to new social problems. For instance in the 1950s the Dutch Roman
Catholic Church preached high birth rates because this was important in the social and political emancipation
process the catholic part of the population was trying to achieve against the then dominating protestant part.
Nowadays the Pope tries to influence the attitudes of catholic people against issues such as homosexuality, and
divorce, and has an ethical view on subjects such as unemployment, social security and many other politically
and socially important issues of the 1990s. The Church nevertheless does not want to adapt a number of its
more basic forms of identity, such as male priesthood, despite the fact that in The Netherlands the Church is
suffering from declining interest in celibatarian male priesthood. The philosophy behind this policy is strongly
related to ideas about celibacy and asceticism and the place and role of women in our society. It is evident that
the inertia of basic identity principles on the one hand lends strength and solidity, which are important for the
continuation of the organization, but on the other hand it also can lead to people leaving the Catholic Church
in the Netherlands (cf. Lijphart, 1968; Windmuller, 1976 for interesting further reading on Dutch politics and
the role of the Church).
Single-loop learning is about adapting to changing inputs, within the existing pre-set
norms. Double-loop learning is about the change of the pre-set norms themselves, as a
reaction of the ineffectiveness of the existing norms. In our example of the Catholic
Church a double-loop learning process could result in the removal of celibacy for
priests. There is always a tension between single-loop and double-loop learning, and
the management must decide how many single-loop or double-loop learning efforts
are required. Besides, the management must support a successful single-loop and
double-loop learning process. Both these management issues are about organizations
learning to learn, called deutero learning (Argyris and Schön, 1978). Very often
changing environments are difficult to understand, and knowledge gained soon
becomes obsolete. This means that in highly dynamic environments organizations
must be particularly good at double-loop learning. Organizations that do not allow
for this high level of double-loop learning, such as many classic machine
bureaucracies, will be selected out. This has motivated some authors incorrectly, e.g.
Swierenga and Wierdsma (1990), to equate organizational learning with double-loop
learning. Single-loop learning has also led to important competitive advantages
without inhibiting double-loop learning, as will be shown later on in the case of lean
organizations.
From the cybernetic paradigm, I have found some important issues to investigate
about organizational learning:
1. An essential characteristic of organizational learning is the reception of signals
about the organization's environment, so that organizational actions can be
initiated that make the organization head straight for its (moving) targets.
2. MICS should be developed and be available, so that (negative) feedback loops
are established that make critical evaluation and steersmanship possible.
Additionally, I think that although many traditional management information
systems were of this monitoring and control type, they were not effective because they
lacked the behavioral connections that are important in order to obtain the correct
interpretation of the monitoring data and allow the convergence of the conclusions to action
(Ansari, 1977). These behavioral issues of organizational learning are discussed in the
following section.
making and coordination, division of space, design of production flows, and the
rational use of technical instruments. This means that organizational learning from
this perspective is of a technical, substantive kind aimed at increasing existing
technical knowledge and skills. Information systems could be useful for shortening
lead times, making problems visible and analyzable. Machine bureaucracies are good
examples of human machines, as they are constructions of rules and procedures that
should work independently of whoever applies the rules. This approach links up
perfectly with the cybernetic approach, because cybernetics also applies machines for
achieving goals. These cybernetic machines have information systems as an essential
component. In fact the bureaucracy does not differ on this issue, because it precisely
describes procedures for communication and information supply (cf. Weber, 1964).
The focus on socio-emotional relations emphasizes the fact that most organizational
learning is done by a group of people. Effective communication between the
participants is essential for increasing the group's learning potential. For effective
communication it is necessary that members of the group be loyal to each other, and
disclose their minds so that everyone profits from the group's success. According to
Argyris and Schön (1978), however, many organization members do not disclose
themselves and have egoistical attitudes that in the end are counterproductive for the
group. Much consultancy in this area is about teaching team members to disclose
themselves, accept criticism, and training people to have effective confrontation
meetings. This means that organizational learning can be inhibited when groups do
not have effective socio-emotional relations. Information systems can provide data
that give opportunities for instrumental learning, but will not be of any use when the
group is not able to discuss these data and find acceptable solutions that lead to new
behavior.
Power relations are mostly difficult to observe in organizations, because they have a
longer term perspective and demand a tacit way of operating. The aim of these
relations is to improve opportunities of influencing the behavior of other
organization members. Organizational learning can be blocked by the existing power
relations when some people have the power to not allow certain issues to be
discussed. When trying to implement new ways of working, it is necessary to find
commitment for this project among the people who have access to the required
money, time and motivation. Organizational learning, therefore, is restricted or
supported by commitments that are based on the use of power in the organization.
Negotiation relations are about decisions in organizations with respect to scarce
resources. This means that negotiation relations are much more explicit than power
relations, because it is necessary during negotiations to specify claims, to give
arguments and motivations, and respond to proposals. Organizational learning can
be restricted by negotiations because the group of bargainers must decide on what to
do and put priorities on the right actions. However, organizations can also learn to
negotiate, so that negotiations can be performed more quickly, more efficiently, and
the chance of dissatisfaction with the decisions is reduced by using certain
Investigating MICS, MB and Organizational Learning 23
unlimited choice among technological options and almost unlimited control over the
consequences. The third approach, called the emergent perspective, states that the
technological and organizational imperatives are both incomplete, because it is the
interaction between the systems and the organizational features that cause certain
impacts. The latter approach emphasizes the inseparability of organizations and
systems in leading to joint impact. Impact is itself not a static moment but a learning
process in itself that could lead to inertia/reinforcements of the status quo (single-
loop learning) as well as important changes in mind (double-loop learning) or even
the development of a culture of continuously searching for improvements and
innovation (deutero learning). The best term to use as independent variable is
therefore socio-technical monitoring information and control system. This reasoning is
also consistent with the literature on MICS that emphasizes a 'broad sense' systems
definition which includes computer-based systems (Flamholtz, 1983; Lawler and
Rhode, 1976; Ansari, 1977).
It is obvious that the third approach is very similar to our concept of information
systems. The emergent perspective also does not have unrealistic expectations of
information systems and IT. It also does not say that information systems are useless
in learning activities. The emergent perspective poses the question of under what
conditions MICS-type systems can be useful for supporting organizational learning,
and states that the creation of MICS and its conditions is a learning process in itself.
The adherence of the emergent paradigm leads to a causal structure that deviates
from the one-directional relations that one would expect when using the concept of
impact, see figure 2.1.
difficult, as the researcher must venture among the many complications of power
relations that dominate at this management level. The impact of tactical and
operational systems on effective organizational learning depends not only on systems
and organizational features, but on environmental conditions that impose learning
needs as well.
Before completing this section on MICS and impact research, it is useful to mention
some alternative ways in which one could study organizational learning and
information management, namely by:
1. ... studying developments in other computer applications than MICS, that
provide managers with opportunities to understand their business much better
(e.g. Decision Support Systems, Expert Systems; cf. Huber, 1991);
2. ... redefining systems development processes as organizational learning
processes (e.g. Vennix, 1990; Business Reengineering, cf. Hammer, 1990;
Davenport, 1993);
3. ... applying systems impact studies, which study the problems in the systems
development process for learning from them, and learning from actual use
(Chew et al., 1991; Saarinen and Wijnhoven, 1994; Wijnhoven, 1992b).
The value of organizational learning as part of information management, is that it
provides an integral approach to information management and business
administration, and gives emphasis to knowledge development, the prime business
mover in post-industrial, knowledge-intensive service oriented, society (Quinn, 1992).
As a consequence, information managment must change its traditional perspective
on managing data, in managing knowledge. For information management this will
require the acquisition of new competences and a change in its organizational
structure (Burrows, 1994).
2.3 Problems in Organizational Learning and Monitoring Information and Control Systems
One of the main reasons why we know so little about the I.T.'s role in relation to
organizational learning is the obscurity of the term 'organizational learning'.
Additionally, information, information system, information technology, management
and organization should be clarified in their mutual relationship. Knowledge in these
areas is missing because of:
1. The ambiguity of the concept of organizational learning, which is caused by its many
roots in disciplines such as psychology, economics, management science, social
science, and cybernetics. This means that communication is difficult. Even
within one paradigm or research field there is much ambiguity about the
meaning of this concept. To improve this situation, a working definition of
organizational learning should be introduced. This definition is that
organizational learning is learning about organizational problems by individual
organization members via the interpretation of data about these problems in connection
26 Organizational Learning and Information Systems
Alternatively, one can also state that the time for machine bureaucracies is over as a
result of new competitive environments. This is probably correct to some extent,
nevertheless, large organizations contain an enormous potential because so many people
contribute labor and knowledge. The problem is, as stated before, how to make optimal use of
them. When well organized and managed, especially by managing its knowledge
carefully, a big organization can profit substantially from its size.
These statements about research topics refer to some important scientific problems,
which should be solved. In this study, I will select the following question:
This question should result in data for the construction of an explanatory theory that
might also be valuable in evaluating a MICS on its contributions to organizational
learning. Chapter 3 discusses the consequences of this approach.
relation is almost 100%. Therefore, the basic organization principle is related to the
optimized use of expensive production capacities, which means that all materials,
supplies and people should be there in the right quantity, shapes, quality, and
moment. Formal planning is necessary to accomplish this. Also, a clear definition of
tasks, responsibilities and authority is necessary to have an effective well-organized
social system related with the technical production process.
Since the second world war, machine bureaucracies have had to face increasing
demands on quality, flexibility and innovation of their organizations. To cope with
these demands the organizations had to increase their internal and external
complexity. Competition on quality for instance demanded regulations for preserving
the quality standard required. This led to the production of quality handbooks and
certification procedures (Garvin, 1987; Evans and Lindsay, 1993). Behaving
according to these rules is supposed to be a necessary condition for realizing quality
standards. At the same time this increased the demand for bureaucratization via the
development of formal communication routines, inventory procedures, machine
maintenance rules etc.
Allowing for flexibility in the production process required in many instances the
implementation of flexible manufacturing systems. The effective use of these systems
required again many rules about the availability of data and resources at precisely the
right time (Kerr, 1991). Besides, machine bureaucracies had to think-over their
production or value-addition technology. Dependent on market characteristics
(especially specificity of customer and dynamics in demands), continuous production
(leading to stock sold) is not always optimal (stock has large opportunity costs). Many
machine bureaucracies, therefore, are series producers, but some are even batch and
discrete unit producers. This has large consequences for the managerial knowledge
needed (especially at the tactical and operational decision levels) and thus for the
learning problems as well (cf. Van Rijn, 1985; Hill, 1982). Besides, as a consequence
of modern competition, service industries are forced to optimize the relation between
output versus input, but must especially focus at optimizing their relations with
clients and markets (cf. Kotler, 1988). These issues, therefore, must be considered in
detail in the rest of the study.
Innovation requires a certain amount of de-bureaucratization, because it demands
the free flow of ideas among people (Hamel and Prahalad, 1990). The impact of the
formalization process described earlier led to a significant problem in adapting to
changing market environments. This means that many a company, in trying to be a
perfect company from an efficiency, quality and flexibility perspective, dug its own
grave at the same time (Lammers, 1985).
Another problem with bureaucratization was the reinforcement of complexity. The
larger the organization, the more complex its internal processes and the more
difficult it is to manage as a whole. Hence, in the 1980s a reconsideration of the use
of being a single organization was initiated and many organizations introduced
vertical decentralization and even introduced outsourcing of departments and
divisions (e.g. Philips Electronics; also cf. Wissema, 1987). Also, external complexity
Investigating MICS, MB and Organizational Learning 29
grew enormously because many new products and new suppliers entered the market.
Liberalization of trade (GATT-discussions, European Market etc.) complicated
competition even more.
The issues relating to the growing internal and external complexity and market
dynamics changed the ways in which organizations could form a strategy. In
bureaucracies, people improved learning in a specialized way, but lost the
opportunity to create synthetic views. These increasing learning problems were met
by cognitive complexity reduction and an increase of organizational learning
capacities (cf. Galbraith, 1973). Cognitive complexity reduction techniques consist of
the creation of slack resources and the creation of independent tasks. Improvement
of organizational learning capacities can be realized by:
• ... reorganizations consisting of the development of matrix and project
organizations, and specialized intelligence groups and/or,
• .... the use of computer-based information systems for developing new ideas
(especially DSS) and analyzing existing ways of operating (MICS).
Both strategies for increasing organizational learning are dealt with in this study.
The results of these considerations are summarized in figure 2.2.
The main focusses of this study are on the dependent variable 'organizational
learning' and the independent variable 'socio-technical learning system' (MICS). In
chapters 4, 5 and 6 these theoretical constructs are described in more detail and
made accessible for empirical research. Chapter 3 describes the methodological
problems involved.
30 Organizational Learning and Information Systems
Methodology and Research Design 31
This study aims at providing clear concepts and a theory. This implies two activities:
concept formation and theory construction.
assumptions yet in the field. Many complex theoretical constructs are used to define
organizational learning, often exacerbating the problems of comparing insights and
research results. Some authors for instance regard organizational learning as
confirmation or rejection of organizational theory-in-use. These discoveries are then
embedded in organizational memory (Argyris and Schön, 1978). Other authors stress
the importance of creating scientifically valid action-outcome theories for
management (Duncan and Weiss, 1979), whereas others think that the coherence of
views among organization members (shared mental models) is even more important
(De Geus, 1988). One other major discussion has been the question of whether
organizational learning can be more than just the sum of the learning of the
individual organization members. Hedberg (1981) for instance states that
organizational learning indeed occurs through the individual members, but that in
some way organizations preserve certain behaviors, mental maps, norms and values so
that organizational stability is maintained despite the fact that organization members
come and go and leadership changes. Alternatively, Kim (1993) proposed a stronger
link between individual mental models or maps and organizational mental models.
He stated that despite some formal changes in an organizational mental map,
organization members could keep behaving in the old way, following their personal
mental models. But when a substantial part of the organizational membership leaves,
it is most likely that the organization's mental model will also change.
Often the ideal of a learning organization is discussed. A learning organization should
facilitate the learning of all its members and continually transform itself (Garratt,
1987, p. 77). Peter Senge uses the terms 'organizational learning' and 'the learning
organization' in a much broader sense than the cognitive psychological one that
dominated the previous definitions. He explicitly regards organizational learning as a
cognitive (theory, memory, and views) and a behavioral issue. The following quote is
illustrative of his view:
" ...in everyday use, learning has come to be synonymous with 'taking information'. (...)
Yet, taking in information is only distantly related to real learning. It would be
nonsensical to say, 'I just read a great book about bicycle riding - I've now learned that.'
Real learning gets to the heart of what it means to be human. Through learning we re-
create ourselves. Through learning we become able to do something we never were able to
do. Through learning we reperceive the world and our relationship to it. Through learning
we extend our capacity to create, to be part of the generative process of life. [...] This,
then, is the basic meaning of a 'learning organization' - an organization that is
continually expanding its capacity to create its future. For such an organization, it is not
enough to survive. 'Survival learning' or what is more often termed 'adaptive learning' is
important - indeed it is necessary. But for a learning organization, 'adaptive learning'
must be joined by 'generative learning', learning that enhances our capacity to create"
(Senge, 1990a, pp.13-14).
I shall not summarize all these definitions to a single all-inclusive definition that
satisfies all these authors. Such a definition is dangerous in that it may exclude some
Methodology and Research Design 33
sciences has been made by Burrell and Morgan (1979). These authors stated that the
main discussions about the nature of social science (and therefore the development of
social theory as well) can be organized along the subjective-objective dimension.
Typically, scientists adhering to an extreme subjectivist approach emphasize the
nominal nature of 'reality', meaning that 'reality' as such does not exist outside the
mind of people and their language. They therefore believe that one cannot
objectively measure and investigate 'reality', but should try to understand it by
developing insights into the mind-set and culture of the people. On the contrary,
extreme objectivists are convinced of the fact that the world is real and factual. They
therefore think that scientific research could lead to objective and universally true
laws about the nature of society. Subjectivists state that this is not possible because
people have their own free will, which can lead to undetermined actions, and
conclude that people's behavior therefore should be explained and studied in its
idiosyncratic conditions. Both extreme positions probably have many variants. It will
suffice here to state that the subjectivist vision is typical of organization development
and the objectivist view is typical of cybernetics.
The objectivist and subjectivist approaches are mostly presented as polar extremes
between which a researcher has to choose (cf. Burrell and Morgan 1979). My
epistemological position is however pragmatic: science and theories must meet the
functions they should have for their user. Depending on the specific context,
objectivism and subjectivism are both useful. Some basic functions of social theory
are explanation, understanding, prediction, description, construction and assessment
(cf. Bernstein, 1976). Let us explain these functions, and choose what is required to
fulfill our research objectives.
• Explanation. The resulting theory must provide explanations for the success and
failure of MICS in concrete business organizations. Because no validated theory
in this area exists at the moment, it is not the generalization of the theory over a
large sample that is our concern here. What we aim at is an understanding of
MICS's influence on organizational learning in concrete business organizations
so that a theory can be generated.
• Understanding. Understanding is a type of explanation that emphasizes the
possibility of the scientist or analyst having empathy with the subject(s) under
investigation. The explanation therefore is not primarily based on
measurements, but on communication and interaction between the informant
and the investigator. Understanding is essential for creating organization
theory, especially in under-researched areas like the intersection of
organizational learning, MICS and machine bureaucracies. The reason for this
is that a language that deviates from everyday life must add considerably to
knowledge, otherwise it is a nuisance. This means that in an early stage of
theory development, concepts must be close to everyday language. This is a
major strength of grounded theory that in its substantive phase uses everyday
concepts. The value of a theory, however, increases when it has a broader
36 Organizational Learning and Information Systems
application area than just the idiosyncracies of the observed case. Therefore
generality of insights must be strived for. This is achieved via the formalization
of substantive theories, gradually introducing more academic jargon (Glaser and
Strauss, 1967).
• Prediction. According to positivism, theories must have universal validity,
meaning that the relations found are valid in the past and present (explanation)
and future (prediction) (Hempel, 1965). This statement is of course only valid
under identical (ceteris paribus) conditions and therefore hard to test
empirically (Kieser and Kubicek, 1978, part I). Prediction is also a more
ambitious target of theory construction than explanation, because many yet
unknown and maybe even non-existing facts can falsify a prediction, as is often
the case (cf. Etzioni-Halevy, 1985). It is therefore useful to distinguish between
the explanation and the prediction function. Of course many theory users
might be more interested in predictions, so that they can predict 'what would
happen if...' An example is an investigation of the impact of the lowering of a
bank discount rate on inflation in a country. Other theory users might be more
interested in explaining why certain things happen. For instance an
organization might be interested in the question why some projects were a
failure and others were a success, while developing new knowledge about
project management.
• Description. Some theories have a primarily descriptive function. This means
that they have a consistent set of issues to describe a case or phenomenon for
other purposes (explanation, understanding, prediction etc.). Often ideal types
are constructed for this purpose. Some examples of these types are Max Weber's
typology of 'Herrschaft', Jung's classification of psychological types, and
Mintzberg's organizational configurations. Description and its possible
functions are, however, closely connected, as Jung's typology can serve to
improve a therapist's understanding of his client, and Mintzberg's typology can
aid in the diagnosis and design of organizations.
• Construction and assessment. If we can describe, explain and predict phenomena,
then we also have opportunities to find out whether certain policy proposals
will work or not. This means that one can assess the effectiveness and
ineffectiveness of policies (such as using MICS for management learning) and
offer suggestions for improvement (e.g. finding leverages to augment the
effectiveness of MICS).
In this study, cases were used to obtain reliable data for the generation of an
explanatory theory, which could also be used for the evaluation of MICS. The
analysis will use principles of grounded theory construction, emphasizing the
importance of analyzing evidence from cases that, from a theoretical point of view,
are significantly different but comparable. The objective is not to find statistical
regularities, but to find new concepts and how they are related in order to reach an
explanation.
Methodology and Research Design 37
3
This way of selecting cases is called 'Theoretical Sampling' by Glaser and Strauss, 1967.
Methodology and Research Design 39
4
This heuristic is also called 'analytic induction' by Glaser and Strauss (1967). The reader is kindly
referred to these authors for further clarification of 'analytic induction'.
40 Organizational Learning and Information Systems
When applying the case study strategy, some important general methodological problems
must be dealt with: construct validity, internal validity, external validity and
reliability. Construct validity is about establishing correct operational measures for
the concepts being studied. Internal validity is about establishing a causal
relationship, whereby certain conditions are shown to lead to other conditions, as
distinguished from spurious relationships. External validity is about establishing the
domain to which a study's findings can be generalized. Reliability is about
demonstrating that the operations and analysis of a study (such as the data collection
procedures) can be repeated and will lead to the same results.
These requirements can be met in several ways in the case studies. Table 3.2. column
3 describes how this study treats these methodological problems.
This means that most of the recommended tests are explicitly part of the research
strategy followed here.
Chapter 1 and 2 have explained the motivation to this study and the major problems
and questions. Chapter 3 discussed the research design and some methodological
problems involved. The following step is to describe the concept of organizational
learning, in chapter 4. Chapter 5 clarifies the term machine bureaucracy. This
concept has been studied extensively in the existing literature on organizations, and
has led to a conceptual concensus. Additionally, the way machine bureaucracies learn
is examined and some hypotheses are defined that indicate a distinction in learning
among four types of machine bureaucracies. Chapter 6 relates organizational
learning, specifically in machine bureaucracies, with the role and influence of MICS.
These theoretical chapters are then followed by empirical chapters. The theoretical
and conceptual findings first need some further elaboration to an operational
language, so that it is easier to guide the data collection and analysis of findings. This
means that the hypotheses need very precise formulations. The conceptual
ambiguities that possibly still remain must be solved. This is part of chapter 7.
Chapter 8 presents the results of the five case studies. In this chapter the results of
the first case study is taken as input to the analysis of the second etc. In this way the
analysis proceeds by accumulating insights and further formalizes the theory. Chapter
9 finally puts together all the results and discusses a further elaboration of the theory
and observation instruments.
42 Organizational Learning and Information Systems
Concept of Organizational Learning 43
pp. 55-59). A substitute for the term knowledge in a management context is the
term 'management theory' which is a combination of science and judgement.
Management theories thus contain goals, purposes and the way managers think
they could achieve them, possibly formulated in some hypotheses about means-
goals relationships, explanations and predictions of events. A management
theory, for instance, could state that decentralization leads to better motivated
personnel who process information from the environment more effectively,
which leads to a higher organizational performance. This theory consists of
several hypotheses that are open for refutation. The concept of management
theory looks simple; however, in practice it is difficult to observe. This is
because of the often hidden and tacit aspects of management theories. Argyris
and Schön (1978) therefore distinguished espoused theories, containing a
person's public explanation of why he does what he does, from theories-in-use,
which give a 'genuine because reason' for his behavior. The latter is often only
partially espoused for political and cognitive reasons (Schutz, 1939, for a classic
account of the methodological problems involved in observing these). Tacit
knowledge is particularly important when there is a close connection between
knowledge and action. In that case, concrete experience is a more important
motivator to action than the application of explicit abstract models. Craft
technology is based on the application of implicit knowledge because of the low
task analyzability (Perrow, 1967; Mintzberg, 1983). Sometimes a layman can ask
very fundamental questions, and initiate organizational learning by elicitating
everday used organizational knowledge (cf. Coats, 1992). Another example is
strategic planning, that often results in documents that never lead to concrete
actions. The authors of these plans often lack sufficient understanding of the
practical situation and such intangibles as socio-political involvements (Ansoff,
1988). An organization has many concrete experiences as well as abstract
conceptualizations. The learning process should be developed to connect both
and to guide the organization's energy and resources in the right proportion to
the development of prehension.
2. The concept of 'process of learning' is also vague. Kolb suggests distinguishing
between two processes: reflection and experimentation. These processes are most
interesting from a managerial point of view because they correspond with two
basic management activities: analyzing the situation and developing new ideas
(reflection), and testing ideas (experimentation) and learning from that
experience (a new reflection). Management learning, however, is not divorced
from political processes, which even sometimes take more energy than the
learning processes. For instance, Kumar (1990) found in his survey of IT-project
evaluations, that only in 18% of the cases was evaluation motivationed by: "The
use of evaluation results, as a feedback device for improving future development and
project management methods and for evaluating (and improving) the systems
46 Organizational Learning and Information Systems
development project personnel..." (Kumar, 1990, p. 210). The main motivation for
evaluation, according to Kumar's findings, was 'project closure', by
demonstrating that the objectives of the project are achieved and that appraisals
can be provided to members of the project group.
Kolb developed a learning styles inventory, that measures a person on the dimensions of
the organizational learning construct: concrete experience (apprehension), abstract
conceptualization (comprehension), active experimentation (extension) and reflective
observation (intention). This results in four dialectically opposed forms of adaptation
to the world. Mostly, people tend to specialize in a certain kind of knowledge and
learning activity, called a learning style. The convergence learning style relies
primarily on the learning abilities of abstract conceptualization and active
experimentation. The divergence learning style has the opposite learning strengths,
emphasizing concrete experience and reflective observation. The assimilation
learning style has the dominant learning abilities of abstract conceptualization and
reflective observation. Finally, the accomodation learning style emphasizes concrete
experience and active experimentation. A person's learning style is measured via two
uni-dimensional variables: AC-CE (the abstract conceptualization minus concrete
experience score) and AE-RO (active experimentation minus reflective observation
score). According to Kolb these two-dimensional variables can be reduced to uni-
dimensional variables, because of the high correlation between AC and CE, and
between AE and RO, thus possibly measuring two underlying constructs (Kolb, 1984,
p. 75). Weisner's study (1971) of a Midwestern division of a large American
industrial corporation is particularly interesting here. Weisner applied the learning
style inventory to the five major functional groups (marketing, engineering,
personnel, finance and research). About 20 managers of each group were rated. The
results are pictured in figure 4.1.
It is evident from these data that organizational learning can easily lead to
organizational differentiation, which allows for specialization, meaning the
development of specific expertise by specific people. The close connection between
organizational learning and differentiation was already mentioned by Lawrence and
Lorsch who defined differentiation as:"...the difference between cognitive and emotional
Concept of Organizational Learning 47
The sociology of regulation is concerned with The sociology of radical change is concerned with
The status quo Radical change
Social order Structural conflict
Consensus Modes of domination
Concept of Organizational Learning 49
6
O.L. is short for Organizational Learning
50 Organizational Learning and Information Systems
7
To be explained in section 4.6.
Concept of Organizational Learning 51
The first explicit account of organizational learning in the literature known to me,
was given by Cyert and March in their classic volume: 'A Behavioral Theory of the Firm'
in 1963. The authors described organizational learning as a part of the decision-
making process in an organization, and as consisting of four major activities:
• Quasi-resolution of conflict. Goal conflicts can be solved by constructing one
consistent set of goals (the organizational objective). Often, however, it is not
possible nor required to do so. Different and conflicting goals can coexist in
one organization by keeping them separated in discussions, and by keeping the
defenders separated locally, or by paying sequential attention to the different
goals.
• Uncertainty avoidance. Organizations often cope with uncertainty by avoiding
"...the requirement that they correctly anticipate events in the distant future by using
decision rules emphasizing short-run reaction feedback rather then anticipation of long-
run uncertainty events (p. 119)" and by avoiding:"...the requirement that they
anticipate future reactions of other parts of their environment by arranging a negotiated
environment. They impose plans, standard operating procedures, industry tradition, and
uncertainty-absorbing contracts on that environment" (p.119).
• Problemistic search. Searching for s solution is a motivated search for solving a
specific problem. It is usually based on a simple model of causality; however,
the learning process can increase its complexity. Finally, it is biased by the
experience, education and goals of the participants in the search process.
• Organizational learning. Organizational learning is: "...adaptation with respect to
three different phases of the decision process: adaptation of goals, adaptation of attention
rules, and adaptation of search rules. We assume that organizations change their goals,
shift their attention, and revise their procedures for search as a function of experience
(p.123)".
52 Organizational Learning and Information Systems
These four activities are assembled into a model of decision-making by describing the
information flows among them. This is illustrated in figure 4.2.
P = 19 + 8.92I + 10X
dI = -(P-P')/30
(Where: dI is the additional money for incentives; P' = target number of policies sold;
P = actual number of policies sold.)
When economic factors change, the number P might decline, requiring new
incentives to keep the system within the target range, which is important for covering
fixed costs. This means that from the previous equation a new optimal incentive
system must be found. This is typically single-loop learning: error-controlled regulation.
It might be that the impact of the new incentive system is slow, and leads to mis-
steering. This can have negative influences on the whole system. To solve this
problem, the meta-system requires information about these impacts, and has to look
for new principles that might improve the working of the single-loop. The related
54 Organizational Learning and Information Systems
8
De Raadt does not make a distinction between double-loop and deutero learning. This distinction is
discussed in the section 4.5 (on organization development). It suffices here to state that the term double-loop
learning refers to learning about the basic assumptions of the management theory used in single-loop
learning processes. Deutero learning is about the way the organization facilitates learning by e.g. the
development of openess and creativity, encouragement of innovations, quality circles etc. The deutero
learning is specifically an organization development subject, because it is about issues like interpersonal
relations, power relations, and cultural change.
Concept of Organizational Learning 55
Why can people be better learners within organizations than without this social
context? This question is basic for motivating people to learn in organizations, and
very close to the subject of organization design (including the development of
structures, systems, procedures and policies). Herbert Simon (1976, pp. 102-103)
answers this question by describing five principles (premises) that transform
individual behavior to organizational behavior. These principles are: the division of
work, the establishment of standard practices and work procedures, the transmission
of decisions via systems of authority and influence, the provision of channels of
communication, and the training and indoctrination of organization members.
The division of (learning) work allows people to concentrate on specific problematic
topics, analyze them and try to find solutions for the organization. It is also possible
that people are connected to jobs that are the input for the learning process (for
instance data gathering and storage), and manage this in a very careful and
professional way. For instance, it is quite unlikely that someone can be an excellent
problem analyst, solution constructor, implementor of the solutions, and manager of
a department all at the same time (applying all relevant abstract conceptualizations).
Therefore, the division of work and allocation of learning tasks are essential for
having effective learning systems in organizations.
Because of the division of work, people need clear standard practices and procedures so
that the separate tasks in the learning process are well connected. A classic problem
here is that data providers use different meanings (semantics) for data than the
information system's end users. Multinationals, for instance, often cannot tell how
well or badly they are doing internationally, because the data are defined in different
ways in the separate countries. They therefore require expert studies to give the
CEOs a useful, consolidated, body of knowledge. Executive Information Systems
projects therefore require (re)formulation of the data definitions, so that the
interpretation of the data can be done automatically. Via this standard practise and
procedure, the delay between problem occurrence and problem identification is
shortened, and the CEOs have tools to make intelligent analyses themselves.
Transmission of knowledge and ideas occurs via systems of authority and influence.
These systems can be authoritarian, which means that the manager thinks for the
organization, and organization members only have to obey or act accordingly. In the
case of a paper mill studied by Zuboff (1988), changing the information transmission
processes implied a change in the authority and influence structure. This change was
resisted by the middle management, who feared the loss of any reason for existence
of their jobs. Nevertheless as a consequence, feedback cycles were shortened, leading
to less loss and higher performance. Also, more people were engaged in problem-
solving by adding insights from their own specialization and attention focus. For this
heterogeneous group to become effective, personal power was replaced by skills of
communication and group interaction.
Channels of communication are very important in the cybernetic paradigm for starting
learning. It is not only the computer-based information system that supports these
56 Organizational Learning and Information Systems
communication channels, but also the formal and informal communication systems
that must be made easier.
Training and indoctrination aim at the internalization of basic norms and
knowledge. The double-loop learning process challenges these internalized norms
and knowledge, which is very important for not getting an organization of
unthinking people.
A Method for Learning: Using Learning Curves
The learning curve describes the costs of a product unit through time. The
assumption of this perspective is that in doing a certain job recurrently, a learning
process is started up by which the cost
per unit product decreases. (See figure
4.5).
Knowing the precise shape of this slope
is extremely important for business,
because it improves the cost estimation
of a product considerably, and is an
indication of the price competitiveness
of a company. According to Yelle
(1979) the first decades of learning
curve research (1935-1969) were
dominated by a 'classic industrial
engineering' perspective, with as the main topics: shapes of the learning curve,
parameter estimation, industrial engineering applications such as setting time
standards and incentives, classic cost control, and purchasing and bidding functions.
Since the 1970s, topics have moved to business policy-making, and public and
service-related issues, which indicates a double-loop learning process. One of these
modern issues is e.g. the relation between the learning curve and the product life
cycle.
Argote, Beckman and Epple (1990) also posed the question: how does learning
happen through time, and how can knowledge be transferred between organizations
and departments? With respect to the first issue, the researchers found that after a
period of steady decreases, production costs start to increase. This is explained from
the fact that organizational knowledge is often not well adapted so that it depreciates.
This explanation is also consistent with results in psychology on the lack of
persistence of individual learning. Concerning knowledge transfer or distribution,
Argote, Beckman and Epple found in their empirical study on World War II US
navy ship building, that:
" The initial gain in production may have been due to learning by doing in the design and
construction of shipyards and the equipment used in them as well as to learning by doing
in the construction of ships. Once shipyards began production they did not benefit from
learning at other yards (p. 151)" (whereas the ships they built were almost
Concept of Organizational Learning 57
identical).
In a case study in a multinational, multi-plant electronics firm, Adler (1990) was
more successful in finding evidence for a learning curve as a result of knowledge
transfer among departments. This empirical study analyzed knowledge sharing
between the Development and Production departments of the company, knowledge
transfer from an initial location to a newly set up plant, and ongoing knowledge
transfer between different production locations. Adler concluded that from this case,
called Hi-Tech, that apparently much learning curve research has been on the wrong
track, because of the focus on capacity utilization. Adler concluded:
" It is primarily 'learning,' the accumulation of knowledge in the form of manufacturing
knowhow, rather than capacity utilization, that accounts for the rapid productivity
growth rates by Hi-Tech" (p. 939).
To create effective knowledge transfer it is important to use a communication
medium with appropriate richness (Daft and Lengel, 1986). As many organizations
are strongly differentiated, a very rich medium is required or otherwise a strong
codification of the knowledge transferred must be realized (Boisot, 1986). Some cases
illustrate the effectiveness of this last perspective (c.f. CSC Index, 1990).
Many problems can easily arise in learning processes according to the cybernetic view.
Eight of these are listed below.
1 Role-constrained learning. This occurs when people have discovered new insights
but are not allowed to change their behavior according to these new insights.
This means that the relation from insights to action is blocked. This is very
common in organizations after training sessions (Van der Vegt, 1973).
2 Audience learning. This happens when the coupling between individuals' actions
and the organization's actions are weak. This happens for instance when some
individuals initiate a change that is not taken over by other organization
members. This is often the case in situations where excellent ideas are
obstructed by other organization members who are afraid of losing power.
3 Superstitious learning. This occurs when individual actions are followed by
organizational actions, but for which the coupling between organizational
actions and environmental responses are ambiguous. For instance, the success
of a company can be attributed to the choice of a correct management theory,
but it could also be good luck. Also, organizational failures cannot always be
blamed solely on mismanagement.
4 Learning under ambiguity. This relates to problems with coupling environmental
responses to individual beliefs. The problem is that often there is not one single
and objective explanation of an outcome. Different people might all have a
different view about success and failure. A management theory might help in
discussing the interpretation of reality, but everyone could still in principle
58 Organizational Learning and Information Systems
The cybernetic perspective also has some limitations that are listed below:
1. The cybernetic perspective mainly focuses on the quantitative aspects of
information: but there is in principle no reason why information systems
should not contain qualitative and subjective information as well. Mathematics,
however, cannot be as easily applied in that case, and another epistemology
must be integrated in the approach.
2. Information systems are sometimes difficult to describe in terms of TPS, MIS,
DSS etc. Sometimes it is difficult to define and recognize the systems that
manage the organization. Informal information systems exist as well, that can have
changing collections of individuals participating in the discussions and thus
contributing their knowledge and information. It is then hard to find out
which people contributed what information in such a 'garbage can'-like
decision-making process (Cohen, March and Olson, 1972).
Concept of Organizational Learning 59
Argyris and Schön (1978) developed hypotheses about the prehension and
transformation of knowledge in organizations, as well as about the development of
learning, which therefore neatly complements Kolb's insights9.
The prehension part is discussed in terms of theory of action, clearly linking cognition
with action. From their experiences in observing organizational learning, they stress
the importance of the distinction between theory-in-use and espoused theory. Theory-
in-use is frequently mainly tacit, and sometimes even tacit for the individual that uses
it. This type of theory is what really motivated individuals' actions. Argyris and Schön
(p.16) note that every organization member can have another theory-in-use. During
interactions people can adjust their theories-in-use, and even form shared knowledge.
Organizational knowledge can be presented in public maps, and therefore becomes
overt and espoused theory. In contrast to the theory-in-use, the espoused theories
frequently lack a clear connection to the individual's actions. Sometimes well-written
policy statements diverge strongly from actions.
Learning is described as follows by Argyris and Schön (1978, p. 18):
" When there is a mismatch of outcome to expectation (error), members may respond by
modifying their images, maps, and activities so as to bring expectations and outcomes
back into line. They detect an error in organizational theory-in-use, and they correct it.
This fundamental learning loop is one in which individuals act from organizational
theory-in-use, which leads to match or mismatch of expectations with outcome, and
thence to confirmation or disconfirmation of organizational theory-in-use."
9
Argyris and Schön did not mention Kolb's vocabulary of transformation and prehension. The connection
between the two authors is made by myself.
60 Organizational Learning and Information Systems
Argyris and Schön propose (after the cybernetician Bateson, 1971) that organizations
learn in three ways: single-loop, double-loop and deutero. These terms were already
mentioned in the discussion about the cybernetic perspective, but the differences in
use are essential.
1. Single-loop learning:
" ...members of the organization respond to changes in the internal and external
environments of the organization by detecting errors which they then correct so as to
maintain the central features of organizational theory-in-use" (p.18). It is an important
feature of organizational learning that the "...learning agent's discoveries, inventions,
and evaluations must be embedded in organizational memory. They must be encoded in
the individual images and the shared maps of organizational theory-in-use from which
individual members will subsequently act. If this encoding does not occur, individuals will
have learned but the organization will not have done so" (p.19).
So no organizational learning will happen without individual learning and change of
organizational memory!
This type of learning is consistent with the way we have defined organizational
learning in the cybernetic sense, but we have added are the involvement of human
action and a description of organizational memory that is not only formal but also
contains tacit knowledge and memory as part of an organization's culture.
2. Double-loop learning
Single-loop learning is oriented towards effectiveness, meaning: how best to achieve
existing goals and objectives, and how best to keep organizational performance
within the range specified by existing norms. In some cases, however, it is required
that the organizational norms themselves be modified. This double-loop learning
process easily leads to conflicts between parties that still support the old theory-in-use
and parties that want a fundamental change in organizational norms. A good
example is Alvin Toffler's remark on Bell Company that they should no longer strive
for their main goal "every citizen a telephone, of any color as long it is black" (theory-
in-use) but replace it by the norm: "give the client what he really wants, telephones of
different types and colors, additional telecommunication products." When Toffler
presented this idea to the Bell Company, nobody responded for three years.
Nevertheless, people were busy changing their mindset (individual memory and
theory) and after three years the consequences were well enough understood and
people knew what had to be done. Argyris and Schön defined double-loop learning
as follows:
" We will give the name 'double-loop learning' to those sorts of organizational inquiry
which resolve incompatible organizational norms by setting new priorities and weightings
of norms, or by restructuring the norms themselves together with associated strategies and
assumptions (p. 24)".
Some further remarks are necessary to compare the concepts of single-loop and
Concept of Organizational Learning 61
double-loop learning:
" ...it is possible to speak of organizational learning as more or less double-loop. In place of
the binary distinction we have a more continuous concept of depth of learning" (p.26).
The change of norms, targets and theories thus implies political considerations.
According to Argyris and Schön, double-loop learning therefore is more difficult
than single-loop learning. The main problems for double-loop learning are lack of
openness in communication (because of a strife for uni-lateral control) which blocks
fundamental understanding of business problems (Argyris, 1970). The true reasons
for this attitude are often concealed, and are typical for the theory-in-use in most
western organizations. Argyris and Schön call the theory-in-use that emphasizes
domination of some people over others, win-lose situations and many tricks in
communication to protect one from being hurt and evaluated negatively, 'model I'.
This set of learning norms inhibits double-loop learning and therefore leads to
organizational ineffectiveness in the longer term. Argyris and Schön proposed to
replace model I by a model II theory-in-use that emphasizes the dispersion of valid
information (also when it can be negatively evaluated), free and informed choice
instead of control by a superior authority, internal commitment to choices made, and
careful monitoring of its implementation.
This problem statement and suggested solution differs widely from the cybernetic
view, in which learning inhibitions were supposed to be rooted in a lack of data,
restricted information processing and incomplete reasoning. Here we see sharply the
differences between the subjective and objective epistemologies and the order and
conflict views on organizations.
3. Deutero learning
Deutero learning, often also called second order learning or learning to learn, is
about organizations learning to carry out single-loop or double-loop learning.
" When an organization engages in deutero learning, its members learn about
organizational learning and encode their results in images and maps. The quest for
organizational learning capacity must take the form of deutero-learning; most particularly
about the interactions between the organization's behavioral world and its ability to
learn" (p.29).
Often organizations learn how to perform single-loop learning perfectly, but without
being at all capable of carrying out double-loop learning. Hopefully for them this also
reflects on the organization's learning needs.
According to me, these definitions of single-loop, double-loop and deutero learning
are confusing, because no concrete operationalization of learning activities is given.
The use of the term 'norm' in the definition of double-loop learning leads to further
confusion, because this term can mean almost anything in organizational situations,
including norms that form 'model I' and 'model II' and the operational norms that
are applied in single-loop learning. As a consequence it is often difficult to state when
an organization is single-loop learning, double-loop learning or deutero learning. The
62 Organizational Learning and Information Systems
The concept of organizational learning can be made more concrete by defining ways
in which it can be realized. Levitt and March (1988) describe four ways.
The first way of learning is called 'learning from direct experience'. With respect to
this learning type, a possible problem is the competency trap, which concerns the
situation in which organizations have invalid or incomplete knowledge which make
them believe incorrectly that they are on the right track. So, the organization might
be improving in the wrong direction. Another competency trap is that the acquired
knowledge is counterproductive to long term results.
Despite the dangers of competency traps, it is important to reflect in a systematic and
thorough way on one's experience. In Japanese management this has resulted in the
idea of Kaizen, meaning continuously thinking and improving by reflecting on
experiences. The Toyota company institutionalized this learning type by demanding
that employees think 5 times 'Why'. This means that everything that is experienced as
normal and fact, should be questioned. Answers to these questions should be
questioned as well, and so forth until what is actually happening or perceived has
been questioned to the fifth level10.
The second way of organizational learning is 'learning from the experience of others',
often also called vicarious learning (Chew et al. 1991), and was also emphasized in
Argote's study. Within one organization we could think that diffusion of knowledge is a
positive asset for the organization as a whole, because the costs of knowledge creation
are shared, and procedures, ideas and knowledge are standardized. Frequently,
however, knowledge is a strategic instrument and a source of power. The diffusion of
knowledge must then be controlled by the knowledge owner (e.g. via patents). Some
10
The problems involved are similar to the problems of learning under ambiguity, which stresses the
problematic nature of interpreting facts. It is also similar to superficial learning, which refers to the
incompleteness of knowledge.
Concept of Organizational Learning 63
Deutero learning is about learning to learn. In other words it is about the creation of
conditions that optimize single-loop and double-loop processes. Argyris and Schön
constructed Model II as a blueprint for organizations that are most successful in
organizational learning. It emphasizes the following values: least defensiveness and
public testing of theories. Model II also encourages, supports and rewards learning
through the provision of a culture of openness, a management style supporting
critical thinking, and an organization structure that facilitates the easy flow of ideas
and data at all levels in the organization. It accepts and appreciates disconfirmable
statements (new insights) and double-loop learning when needed. The concept of 'the
learning organization' as proposed by e.g. Senge, is a further operationalization of
Concept of Organizational Learning 65
Model II. An important variable then is the set of organizational norms that enables
organizations to learn effectively. The identification and implementation of these
norms is not at all an easy process. Two authors have become reknowned for their
proposed 'solutions', Tom Peters and Peter Senge.
Peters' Solutions
To demonstrate the practical way in which deutero learning processes work from an
organization development perspective, Tom Peters provides two interesting cases:
Electronic Data Systems (EDS) and Asean Brown Boveri.
EDS is a huge 'system integrator' with about 72.000 employees, operating in about 28
countries for over 7000 clients, organized in 38 strategic business units. A most
remarkable feature of EDS is:
" Boil down any SBU and you'll find projects. In fact, EDS is one big collection of project
teams. The number of people on a project can vary greatly throughout its life. The norm
is 8 to 12 EDSers, working together for a period of 9 to 18 months" (Peters, 1992, p.
24-25).
Because the greatest emphasis is placed on the project teams, the big problem is how
to connect people in EDS with each other, and specifically how to know which
people can participate in specific projects. Therefore, EDS has to learn to leverage its
skills (mainly connected with individual persons), and must develop a base
containing information about these skills. Additionally EDS has 'Centers of Service',
that set some people free for fundamental research on a particular new skill or
subject (e.g. imaging technology). These Centers of Service are temporary
organizational units, and have to earn their income by motivating project groups to
adopt the developed knowledge. EDS has thus explicitly been designed to augment
and store knowledge and make this knowledge into a shared organizational memory
via its procedures, systems and structure.
Asean Brown Boveri is a huge company operating in power plants, power
transmission, power distribution, transportation, environmental control, financial
services, and other types of business such as metallurgy, process automation, robotics
and superchargers. In 1991 they booked $28.9 billion in revenue in 140 countries.
CEO Sune Karlsson of the $1 billion revenue Power Transformers Business Area
gives an illustration of how this giant organization is managed:
" Our most important strength is that we have 25 factories around the world, each with its
own president, design manager, marketing manager, and production manager. These
people are working on the same problems and opportunities day after day, year after year,
and learning a tremendous amount. We want to create a process of continuous expertise
transfer. If we do, that's a source of advantage none of our rivals can match" (quoted
from Harvard Business Review, in Peters, 1992, p. 51).
Additionally Karlsson tries to create internal competition by providing detailed monthly
information on the performance of all 25 units.
66 Organizational Learning and Information Systems
" But Karlsson is well aware that such competition must be constructive; he insists (...)
that the key task is creating a 'culture of trust and exchange' (p. 51).
But..
" Sharing of expertise does not happen automatically (...) People need to spend time
together, to get to know and understand each other....People must also see a payoff for
themselves (....) We have to demonstrate that sharing pays - that contributing one idea
gets you twenty-four in return" (according to Karlsson, quoted in Peters, 1992,
p.52).
Not only knowledge storage and dissemination are important for organizational
learning, but people must be motivated to participate and create open
communications as well. Both cases clearly have developed model II characteristics.
Information technology has a prominent role in the construction of these modern
organizations, by the creation of electronic highways, knowledge and skills databases.
The following motivational and leadership features are even more basic than the
construction of these (infra-)structural arrangements:
1. Internal market principles, that give people direct feedback to their
performance and provides strong incentives for high performance.
2. Decentralized organizations and reduction of bureaucracy. This empowers
people and does not frustrate initiatives and creativity.
3. Developing expertise as never before. (Something that might look contradictory
to the decentralization trend, but isn't when the organization connects smart
people together, e.g. by means of EDS's Centers of Service, or connecting with
research institutes).
4. Management's support staff must contribute to these three principles. They
must change their bureaucratic sense in which knowledge is power. Top
management must not use its support staff for control and the support staff
must not slow down the decision-making.
5. "The essence of an effective KMS11 is advertising, marketing, packaging, incentives, big
travel budgets, and the psychodynamics of knowledge management12. The crux of the
issue is not information, information technology, or knowledge per se. It's how, for
example, you get busy people in those miniature ABB units to want to contribute to the
KMS. The answer turns out to lie more with psychology and marketing (...) than with
bits and bytes" (Peters, 1992, p. 384).
Senge's 'Solutions'
Senge states, in line with Argyris and Schön, that organizational learning is not only a
11
Accronym for Knowledge Management Structures, Peters' term for learning organization.
12
Italics from Tom Peters.
Concept of Organizational Learning 67
cognitive activity but often requires a change in power relationships and attitudes.
Many of the norms that inhibit learning are tacit and private, and must be well
analyzed for their organizational impacts, so that true learning can occur. Senge
describes five disciplines as basic for effective organizational learning (Senge 1990a, p.
5-13):
• Systems thinking by which people learn to understand the patterns and linkages
that exist among phenomena.
• Personal mastery: Defined by Senge as: "...the discipline of continually clarifying and
deepening our personal vision, of focussing our energies, of developing patience, and of
seeing reality objectively. As such, it is an essential cornerstone of the learning
organization - the learning organization's spiritual foundation" (p. 7).
• Mental models:"....deeply ingrained assumptions, generalizations, or even pictures or
images that influence the world and how we take action" (p.8). Many of these
theories-of-action are tacit and obstruct organizational change. By
understanding these models we can change them and provide basic conditions
for innovation.
• Building shared vision. This is the discipline of translating individual visions
about a wishful future to a joint sense of destiny in an organization.
• Team learning. The strength of a team is that it can discover insights that none
of the members would have discovered alone. This can only happen, however,
when a free-flow of meaning is created among the members. This is not
something that happens by chance but must be learned and teams can learn to
become more effective by a constant process of improvement in this discipline.
It is essential, according to Senge's theory, that these five disciplines are developed as
an ensemble, which he calls 'The Fifth Discipline'. Senge presents a substantial number
of ideas to support his fifth discipline concept and is extremely normative. Senge
does not concretely explain why an organization requires these five disciplines and
does not describe how organizations in specific contexts must learn these disciplines.
The theory underlying his thought is therefore obscure and not open for academic
research. Nevertheless, it is part of the organization development tradition because of
its strong emphasis on tacit processes, personal and interpersonal development and
change. In fact Peters' ideas are more concrete because he describes concrete and
measurable organizational arrangements that should be made, given the context of
large organizations. His approach is however weaker than Senge's on the
psychological aspects of learning.
completely reduced to the deutero process, the construction of new norms for
organizational learning. For the deutero learning process not much more knowledge
exists than what is gained from organizational change studies. Precise prescriptions of
what the organization should look like exist, but how to change organizations to
reach these ideals is a blind spot. This means that the informative content of this
perspective is very limited. Learning to learn is an interesting term, but how should it
proceed and what should be the result? Additionally, no operationalization of the
connecting thought between an organization's environment and the way an
organization should be designed to facilitate learning is made. Thus, a major issue of
the organization literature, termed contingency theory, seems not to have affected the
organization development perspective. Finally, the most dominant learning approach
in organizations seems to be single-loop according to the authors. Nevertheless, this
remark must be taken with care. There is no survey study that validates this opinion.
No attempt has yet been made to measure for instance the learning activities and
efforts in organizations on single-loop, double-loop or deutero learning via a survey.
This book will not provide the data to qualify statements about single-loop learning
effort and the learning needs. It will, however, provide some basic conceptualizations
and suggestions for measurement.
The soft systems perspective does not exclude the possibility of objectivism. For
instance, a group can find sources of ineffectiveness and the means for a more
effective control of an organization. The results of this analysis can be used as input
for the design of a MICS. In principle the soft systems perspective only provides a
tool that can be used for modeling objective (cybernetics) or subjective (OD) reality.
The technique when used in the case of order must lead to one model that relates all
important issues. In the case of the conflict view it leads to several subjective models.
For further reading in the soft systems perspective we recommend: Rosenhead et al.
(1989) and the special issue in the European Journal of Operational Research on
modelling for learning (Morecroft and Sterman eds., 1992). Both publications use a
subjective epistemology, but differ in their approach. The first places emphasis on
conceptualizing how we talk about problems. The second stresses the importance of
describing major variables and their interrelations in systems dynamics models. In my
opinion conceptualizing preceeds modeling, however, several interations between
modelling and conceptualizing will improve both. Soft systems modelling does not
always have to be followed by system dynamics modelling. Many other techniques
can be used to further define vaguely defined problems (for instance PAM, Stamper
et al., 1988; Kolkman, 1993) When reality is clearly defined, soft systems lose their
value, and hard modeling techniques are more relevant (Wijnhoven, 1992a). Also
when major conflicts about problems exists, soft modelling will not solve them. It
then depends on how the actors in the political arena want to regulate their conflict.
Precise data about possible outcomes are more valuable then and hard models are
also required (Coleman, 1972; Teich, 1991).
Concept of Organizational Learning 71
Motion studies movement) has been expanded with a new method, called business re-
engineering. Business re-engineering is about the redesign of business processes. This is
done first via an accurate analysis of existing processes, especially in terms of costs
and benefits, and by comparing the processes with business bench marks. After this,
a design of a new process is made. In this design transaction processing information
systems often have a vital role as business levers (Hammer, 1990; Davenport, 1993).
The business re-engineering movement therefore basically aims at reducing business
costs and optimizing the process-demands relationships in technical ways. The
authors on business re-engineering state that human involvement in the re-
engineering process is vital, but do not clarify how it should be organized (Kennedy,
1994). Business re-engineering is regarded by the latter as a necessary evil for
organizational survival to which employees have to adjust. Some authors regard
business re-engineering as a revolutionary process (Hammer and Davenport), others
regard it as an incremental process (Davenport and Short). Business re-engineering
however often implies more than changing the organization's transformation
processes. The possible changes that could be made are listed in table 4.4.
Subject From To
Work units Functional departments Process teams
Jobs Simple tasks Multi-dimensional work
People's roles Controlled Empowered
Job preparation Training Education
Focus of performance measures Activity Results
Advancement criteria Performance Ability
Value Protective Productive
Managers Supervisors Coaches
Organizational structure Hierarchical Flat
Executives Scorekeepers Leaders
Source: Personal conversation with prof. P.A.E van de Bunt, October 1993
Table 4.4: Possible Organizational Changes as a Consequence of Business Re-engineering.
When order is the dominant philosophy, models of reality will not conflict. This
means that generally accepted variables are measured and accepted norms applied. A
cybernetic system thus can be constructed, with a MICS that creates accumulating
insights into reality when confusion about variables, norms and measures are
removed. However, when the conflict model is valid, then these models will be
disputed. If a MICS is implemented, its output will be object of political discussions
instead of it resolving disputes about reality. The limitations of scientific
74 Organizational Learning and Information Systems
management and the time-motion studies are therefore not that they only support
single-loop learning, because they can indeed detect major problems in working
efficient and effective. Their limitations are rather that they are disconnected from
the social requisites to make them successful. Taylor himself fought a continuous
battle against his opponents not because his method was considered incorrect, but
because the social and political consequences were not effectively disputed.
The semantic analysis technique was developed for disentangling the conceptual
confusion that often appear when using theoretical constructs. The technique is
described in the following products of semantic analysis (Stamper et al., 1988 and
Stamper, 1987):
• Describing generic-specific structures. For instance: the generic 'components'
can have 'parts' and 'materials' as specifics. Specifics are placed in a box with the
generic term at the top of this box. This kind of relationship can also be
represented by an arrow from the specific to the generic, as in the case of a
component of a good, which is itself a kind of good.
• Describing part-whole dependencies. For instance a good consists of several
parts. This relation is graphically represented via a line with a dot in the middle
between the part and the whole.
• Role names can be given to a line for referring to a term with a meaning that is
about the relation between the two connected terms.
• Some terms can be described in more detail. In order not to make the scheme
too complex, a number is allocated before the term, referring to the number of
the chart describes the term in more detail.
• When the connected term is a sign, this can be expressed by placing it between
brackets, or connecting it via a broken line.
• Particulars (for instance European Community, an organization name) are
Concept of Organizational Learning 75
Two major terms are explained in the psychological perspective: knowledge and
learning process.
Knowledge is regarded as the subject that is transformed via reflection and
experimentation in the learning process. Three types of knowledge are described:
science, judgement and potential knowledge. Science is public and accepted
knowledge. Judgement is uncertain, often private and not tested knowledge. Science
and judgement both constitute 'management theories' which are explanations or
means-goals theories used for action in organizations. These theories are to some
extent espoused and to some extent tacit. They can be more or less abstract when
measured in terms of Kolb's AC-CE variable. Potential knowledge is about experience
and data. This does not mean that they always speak for themselves. Often a large
amount of abstract knowledge is required to make an interpretation of these data, as
for instance in case of interpretation of macro-economic figures that requires some
knowledge of macro-economics and politics.
A learning process is a set of activities that improves or adapts knowledge. According to
Kolb this can occur basically in two ways. The first way is reflection, which is thinking
over experience for constructing models, theories and concepts. The second way is
experimentation by which people test the quality of the gained abstractions in
practical situations. Via their education, socialization and work experiences
individuals develop a learning style in favor of a certain type of knowledge (abstract
or concrete) or learning process (experimentation or reflection).
The soft systems perspective uses the subjective epistemology. It provides some
techniques by which people describe a theory of their perceived reality. The critical
success factor concept can be used to acquire the basic concepts. System dynamics
provide a logic in which the relations among these concepts can be defined. System
dynamics also supports the detection of deeper lying systemic structures, that explain
unexpected and dysfunctional impacts of current behavior. Soft systems techniques
are especially useful when a group of people want to define a shared mental model.
Sometimes differences of opinion become more clear and can be solved by finding
common opinions. It is, however, more problematic when the models are
incompatible, which means that they cannot be reconciled. In this last case, soft
systems cannot do more than map the different views.
The generic term 'learning processes' has three specifics: single-loop, double-loop and
deutero learning. Single-loop learning consists of the following: use or re-use of
existing management theories and data, the dissemination of such knowledge, and its
storage and adaptation. The storage process is a logical antecedant for organizational
memory. Another antecedent for organizational memory is the existence of
organizational knowledge and learning norms that can be stored. Double-loop
learning also includes the removal of obsolete theories, which is a synonym for
unlearning. Deutero learning is an antecedent for learning norms. This also implies
that the concept of management theory is distinguished from the concept of learning
norms (but both are organizational knowledge). Some authors are not clear about
this distinction, but if it is not made, the distinction between double-loop learning
and deutero learning becomes irrelevant.
Organizational memory stores contain three types of knowledge: data (potential
knowledge), management theories (science and judgement), and learning norms.
Management theories are distinguished according to purpose: explanation and
78 Organizational Learning and Information Systems
technology (knowledge about means and goals). Explanations and technologies can
be tacit and/or explicit. Four types of organizational learning norms are
distinguished: learning policy (including infrastructure for learning), procedural,
action and responsibility norms. Individual learning norms are called learning style.
The generic 'learning field' has not been clearly defined so far. Learning fields are,
however, essential for knowing what learning is about and what it must achieve.
Policy norms define what are regarded as important fields of learning. Because
solving learning problems should add to organizational effectiveness, learning fields
are defined in terms of organizational effectiveness. Quinn and Rohrbaugh found
three dimensions of the organizational effectiveness:
• Structure. This dimension has two extreme values: flexibility and control.
• Focus. The extreme values here are: organizational internal and external.
• Means-goals relationships. The extremes are an emphasis on means and an
emphasis on goals.
Output quality was found to be a possible dimension on its own, but the authors are
not very clear about it. The results confirm existing distinctions among paradigms of
organizational analysis, as presented in figure 4.8.
The learning effort concept is mentioned infrequently in the literature, but should be
defined as a combination of learning fields and learning activities. This means that,
when more fields are the subject of learning, more learning activities are undertaken
and the greater the learning effort. In chapter 7, this definition is explained further.
It is also particularly interesting to note that the Human Relations and the Open
System Models emphasize differentiation, spontaneity and flexibility (Quinn and
Rohrbaugh, 1983, p. 374), which fit neatly into the Organization Development
perspective of organizational learning. The Rational Goal Model and the Internal
Process Models emphasize integration, formalization and control (Quinn and
Concept of Organizational Learning 79
points out the main issues, but does not give measures yet.
4.9.1 Deutero Learning: Designing Norms that Govern Single-loop and Double-loop
Learning
open debates among its members, and encourages people to think creatively
and take initiatives. People should be empowered to do so, by being given
responsibilities and authority if necessary, and the organizational incentive
system should reward people that behave according to organizational
perestroika.
• A learning policy finally requires from CEOs that they take charge of major
projects to redesign the way business is done, and create major procedures. This
frequently requires management revolutions, that also are known under the name
of business re-engineering projects (e.g. Davenport and Short, 1990; Walker, 1992).
The learning policies must be implemented in more concrete learning norms, about
responsibilities, actions and procedures.
computers and telematics is important here, but also the way data are obtained, the
management of data quality, and the need for some specialists to extract information
and knowledge from the data.
The issues of deutero learning are listed in table 4.5 in the form of questions that can
be used when wanting to start up a deutero learning process.
Dimension 1. Development
This dimension is about activities by which a management theory is constructed by
scanning the internal and external environment of the organization. Important in
this step is the additional use of frames of reference, consisting of a mixture of tacit
knowledge (beliefs, norms) and explicit knowledge (obtained by internal or external
Concept of Organizational Learning 83
training, education, and by experience in the same or a related problem field). The
result is an explicit theory, containing goals and methods for achieving them. In this
area soft modelling is widely acknowledged as an effective method. Some computer
supported tools are also available (e.g. 'PAT', Kolkman, 1993; 'CSFmatrix',
Wijnhoven, 1993; Acar and Heintz, 1992; 'I Think' from High Performance Systems
Corp.). The output of soft modelling consists of a collection of yet imprecisely
defined terms, by which an understanding can be made of a yet under explored area.
Because of the volatility and complexity of many managerial situations, especially
those faced by the strategic apex, it is often not worth the effort to make these models
more precise. In less volatile situations, for example in the case of route planning, it
can be extremely valuable to make the model mathematically and logically precise
and to test the hypotheses that underly the models with the data available (e.g.
Wijnhoven, 1992a). So in the theory development process two sequentially related
phases can be defined: the development of a soft model and the development of a
hard model. Additionally, the models must be implemented, which requires the
explicit knowledge that was acquired, becomes integrated with tacit knowledge of the
practice field (Hedlund, 1994).
Dimension 3. Dissemination
It is important that the management itself believes in what it thinks, but it is also
essential that other people are convinced as to why things must be done in the way
prescribed. Effective communication in the semantic and the pragmatic sense is vital.
The first sense concerns understanding the relations among the variables in the
management theory. Large bureaucratic organizations mostly have a large amount of
differentiation among departments and organization members, which complicates
unambiguous communications, and leads to more successful intragroup learning
than learning of the organization as a whole. The semantic problems involved lead to
misunderstandings about some concepts. The resulting subcultures and professional
groups with different jargons is a well-known problem in larger and complex
84 Organizational Learning and Information Systems
Dimension 4. Adaptation
During the process of plan formulation, the business reality may have changed.
Hence, there is a basic need for adjustment of the management theories, and this
should be based on a test of elements of the earlier theory (e.g. the actions
prescribed). Managers could make use of a range of insights from scientific
methodologies. The adaptation can result in a refinement of the theory, but also in
the detection of the need for a complete revision. Thus a double-loop trigger could
also be the result of adaptation. From learning norms, concrete management theories
for learning processes can be derived. For instance, as part of the procedural norms,
management could decide to develop a decision support system. This information
system is a field for adaptation when its efficiency for the learning process is
evaluated. Changing the procedural norms, however, can also involve the
replacement of the decision support system by a group-decision support system to
encourage interactions of learners in a learning lab under the support of a facilitator.
The change of this norm is in fact a deutero learning process and must be evaluated
in relation to the organization's abilities to meet organizational learning needs.
Concept of Organizational Learning 85
Dimension 5. Storage
This issue is not well treated in the general discussion on management life cycles, but
yet is a basis for organizational learning. Knowledge can easily become forgotten or
distorted when not well documented (cf. Yelle, 1991). In order to realize a learning
loop in complex environments it is essential to write down what the operational
goals are, and the performance criteria for people involved in the project and work
process. Only then can they become effective guidelines. If they are not well
documented it is not clear how evaluations should take place, and the learning loop
is not closed (cf. Peffers and Saarinen, 1993).
Traditionally, the storage of knowledge was achieved through individuals' memories,
organizational myths and stories, symbols and many tacit norms that form the basis
part of the organization's culture. The development of administrative organizations
led to a first instrument for systematic storing management theories, especially in
terms of finance and resources. Formal planning techniques can also be used as tools
for storing organizational knowledge. Administrative organization and formal
planning techniques were instruments for management used in all great projects in
history. Nevertheless, the rise of the large machine bureaucracies (beginning of this
century) would not have been possible without the development of organizational
memory and knowledge storage procedures and techniques. Universities had an
important role in these developments, as is illustrated by the rise of business schools
at universities.
Information technology led to a larger and more powerful organizational memory, by
developments in:
1. Databases that allow for the efficient and reliable storing and retrieval of data.
2. Modelbases that allow for the management of several models represented in a
mathematical way. These models can be stored, accessed, developed, used and
improved when required.
3. Knowledge-based systems and expert systems provide opportunities for storing
qualitative knowledge, by means of the use of formal logic. This is possible
when the management theory is made explicit and represented in logical
chains. Knowledge rules can be connected to simulate a human reasoning
process.
4. Integration of these IT-opportunities (e.g. Kerr, 1991).
The flow diagram only describes activities that can be formally described and
managed. One should however not underestimate the importance of informal
processes in organizational learning. Brown and Duguid (1991) give an account of
how to conceive informal organizational learning. The first point they made is that in
informal learning practices, knowledge is seen as understanding practice and may
never be detached from practice. Abstract knowledge (called canonical practice by the
authors) can blind the organization to the fact that it is the practice of the
organization members that determines the success of the company. In many cases,
abstract knowledge is written down in handbooks and documentations, but is such
Concept of Organizational Learning 87
The term bureaucracy has its origin in the writings of Max Weber, who defined it as a
legal way of excersizing command and control over people. The basic categories by
which bureaucracy exists are summarized by Weber (1921/64, pp. 331-332) as
follows:
" (1) A continuous organization of official functions bound by rules.
(2) A specified sphere of competence. This involves (a) a sphere of obligations to perform
functions which has been marked off as part of a systematic division of labour. (b) The
provision of the incumbent with the necessary authority to carry out these functions. (c) That
the necessary means of compulsion are clearly defined and their use is subject to definite
conditions. (...)
(3) The organization of offices follows the principle of hierarchy; that is, each lower office is
under the control and supervision of a higher one. There is a right of appeal and of statement of
grievances from the lower to the higher (...).
(4) The rules which regulate the conduct of an office may be technical rules or norms.13 (...)
(5) In the rational type it is a matter of principle that the members of the administrative staff
should be completely separated from ownership of the means of production or
administration.(...)
(6) In the rational type case, there is also a complete absence of appropriation of his official
position by the incumbent. (...)
(7) Administrative acts, decisions, and rules are formulated and recorded in writing, even in
cases where oral discussion is the rule or is even mandatory."
The rules led to routinization and standardization and served to make output
predictable and reliable. This is important for modern government and business. It
also became a major issue in the quality movement in the 1980's (Garvin, 1988;
Evans and James, 1993).
13
Henderson and Parsons, the translaters of Weber's original German book, state (1964, p. 331): "By a
'technical rule' he probably means a prescribed course of action which is dictated primarily on grounds
touching efficiency of the performance of the immediate functions, while by 'norms' he probably means rules
which limit conduct on grounds other than those of efficiency. Of course, in one sense all rules are norms in
that they are prescriptions for conduct, conformity with which is problematical."
Organizational Learning in Machine Bureaucracies 89
Many of Weber's ideas for efficient control were taken over in business
administration, with legitimacy based more on the efficient handling of resources
than on its legality. Henry Fayol pleaded for a rational setup of management, by
separating it from the primary functions of the company (technical, commercial,
financial, security, and accounting activities). The management function is then
responsible for realizing optimal principles: division of work, authority, discipline,
unity of command, unity of direction, subordination of individual interests to the
general interest, remuneration, centralization, scalar chain (line of authority), order,
equity, stability of tenure of personnel, initiative, and esprit de corps (Fayol,
1916/1949, p. 19-20).
Researchers saw many negative effects of bureaucracy. For instance, Merton (1948)
states that bureaucratic organizations give higher priority to organizational structure
and processes than to organizational goals and their social-economic function. This
can lead to a very inflexible organization, which loses a sense of its environmental
function. Bureaucracy also leads to under-utilization of human potential, because
people become slaves of the routines that are prescribed for the work process
(Maslow, 1970; Mayo, 1945). It is particularly important for nongovernment
bureaucracies that often machinery is applied that demands the strict application of
certain rules. It therefore requires the suppression of informal organizational
processes. The importance of informal processes and informal communication
networks was however made evident by research of the so-called human relations
school of management (e.g. Mayo and Herzberg, 1938). Additionally, modern
researchers found out that the growing complexity and dynamics of organizational
environments demanded less formalized and constrained organizational behavior
(Argyris, 1970; Beer, 1981; Galbraith, 1973; Simon, 1947). The introduction of
machinery, however, often increases the need for formality in organizations. The
resulting classic machine bureaucracies therefore have large problems with
formalization, under-utilization of human resources and inflexibility in coping with
changing environmental business conditions (Woodward, 1965). This trend was
reversed in the 1950s by the development of lean machine bureaucracies in Japan.
Classic machine bureaucracies are supposed to have large problems with
organizational learning, because they have strategies, structures, cultures and
information systems as defined below.
• Classic machine bureaucracies have longer term, reactive and cost-leadership
oriented strategy. This is because of the long pay-back period of machine
investments, and a strong focus on optimization of designed processes for
realizing lowest costs. Increasing dynamics are difficult to match with these
machine bureaucratic features, because they require longer and shorter term
orientation, active and pro-active strategies (cf. Miles and Snow, 1978; Zahra
and Pierce, 1990).
• The organizational structure of classic machine bureaucracies is not developed
for organizational learning innovations (changing organizational knowledge),
90 Organizational Learning and Information Systems
but only for conserving the way of working. This is also a form of organizational
learning, but it can mismatch with learning needs in dynamic and complex
environments. Machine bureaucratic structures result in a very formal way of
collaboration, strongly institutionalized in departments, and strict authority and
responsibility allocations that govern actions. The organization usually has an
efficiency objective, but can have an expensive hierarchical coordination
mechanism (Galbraith, 1973; Gurbuxani and Whang, 1992). This is again
mainly a problem of dynamics, because the complexity can be dealt with. The
effective handling of increased dynamics often conflicts with the slow
procedures of classic machine bureaucracies. Hence, a more organic
organization type is required (Burns and Stalker, 1961).
• The organizational culture of classic machine bureaucracies is dominated by
'uni-lateral' control and defensiveness. It is the boss who thinks and decides,
and his employees act (Argyris and Schön, 1978). This goes against demands for
handling high levels of complexity, because a single manager has only one set of
brains. The thinking and innovative process therefore must be a joint effort of
many organization members.
• The increasing complexity of products, production processes, production series
and variants, requires a more rational treatment of organizational knowledge.
This is also done in classic machine bureaucracies, but mostly too slowly, as was
shown by the lean car manufacturer examples in chapter 1. The environmental
dynamics and complexity demand information systems that match perfectly
with the working flexibility required. Additionally, these systems must be easy
to adapt to frequently changing user requirements (Land, 1982).
We are not interested in the machine bureaucracies themselves, but in organizational
learning and information technology. Machine bureaucracies are used as a source of
case material to develop a substantive theory, that later can be formalized into a more
general theory (cf. chapter 3). We will avoid using the buzz word 'the learning
organization' because it suggests that there is one best type of learning organization14.
This study therefore describes several ways of learning in machine bureaucracies.
These learning types are further linked with environmental needs (learning needs)
and conditions (learning norms). A normative viewpoint is evident here stating that
effective organizations have a learning needs-learning norms match. This reasoning is
analogous to contingency statements in organizational design, which state that
organizational environment influences the effectiveness of organizational structures
and processes.
The factors in this study now are: organizational environment, organizational norms,
14
It seems as if writers on 'The Learning Organization' (like Senge, 1990; Swierenga and Wierdsma,
1990; Garratt, 1987 and Garvin, 1993) have not understood the lessons from the contingency approach that
propose the equifinality of organizational forms (cf. Mintzberg, 1983 and Dotte, Glick and Huber, 1993 for
excellent reviews).
Organizational Learning in Machine Bureaucracies 91
and transformations. These are described for the organization in general and for the
organization's learning subsystem in particular in table 5.1.
Factors Machine Bureaucracy Organizational Learning
Environment Complexity and dynamics Learning needs
Norms Structure and culture learning norms (policy, action, procedural and
responsibility)
Transformation Transformation of goods Transformation of knowledge via single-loop,
and values double-loop learning, and of learning norms via
deutero learning activities
Table 5.1: Conceptual Relation Between Machine Bureaucracy and Organizational
Learning
From the study of Womack et al. (1990), lean and classic (car) manufacturers are
distinguished on the basis of 10 items described below.
• Attitude to quality
The lean organization has 'Kaizen' as a basic mind set. Kaizen is a Japanese term
meaning an intrinsic motivation to improve quality. This means that learning about
quality should come from within. In the classic bureaucracy, quality improvements
start from a need to comply to market pressures. For instance, many companies start
up quality programmes to receive a certificate, because no certificate often simply
means no business. This sometimes leads to a further formalization and
bureaucratization of the existing classic bureaucracy. In lean organizations these
quality costs are far lower than the profits.
• Level of decentralization
Lean organizations allocate many decision responsibilities to the shop floor. This is
done via vertical decentralization to improve the reaction speed when problems
occur, and via horizontal decentralization to reduce communication difficulties
among the experts and the shop floor. This means that the shop floor should be
equipped with expert knowledge, and with a motivation to take responsibilities. Staff
groups with expert knowledge are then small and mainly have coordination and
facilitation tasks. In the traditional bureaucracy, large staff groups exist that do the
thinking for the shop floor and to some extent for the management as well. People
on the shop floor then only require a low educational level, and have no managerial
responsibilities. In that case, there is a strong distinction between 'we' (the workers)
and 'them' (the management). Research in the area of machine bureaucracies has
shown that the number of people in administration and management versus people
engaged in production (so-called administration/production-ratio) has been
increasing relatively (Anderson and Warkov, 1961; Child, 1973; Indik, 1964;
Parkinson, 1958). Parkinson even believes that this trend is based on immanent laws
of machine bureaucracies, that can be described as follows: (1) an employee of a
bureaucracy aims at reproducing subordinates and tries to avoid competitors and, (2)
employees in a bureaucracy provide each other with work, mainly for coordinating
and controlling each other's jobs. The lean organization starts from the opposite
view, which could be described as: (1) regard each other as equals so that you can
profit from each other's knowledge and experience and (2) try to reduce coordination
costs by creating lateral relations instead of the slow and costly hierarchy.
• Availability of lateral structures
The classic machine bureaucracy manages its communication processes via
hierarchies of command. When a problem occurs that a person cannot handle, he
goes to his superior to ask for advice and support. When the superior cannot give
this support directly, he contacts some of his subordinates. When the problem
cannot be solved in that way, the superior goes to his superior etc. When the
Organizational Learning in Machine Bureaucracies 95
problem is rather complex and needs the collaboration of several people for a certain
period, a project or task group is created, under the authority of the superior of all
the people involved. This means that a number of people are taken out of the daily
working process, to work for some time on the project. Two problems can now occur
that are solved differently in classic and lean organizations:
1. When the project work and routine work come into conflict, the lean
organization gives priority to the project work. The project leader (called 'shusa'
in Japanese) has the authority and means to enforce the collaboration of the
people involved. In the classic organization, the routine work mostly has
priority above the project work.
2. When the problems are incidental, ad hoc contacts might solve them. In the
classic bureaucracy it is very difficult to make these contacts because it requires
the approval of superiors and the internal differentiation even makes it difficult
to find out if the required expertise is in fact available within the organization.
In the lean organization the hierarchical chain is very short and people are
expected to find the required internal contacts themselves.
• Relations with suppliers
In traditional machine bureaucracies, suppliers are members in a negotiation process.
The idea dominates that what you pay for supplies increases production costs and
lowers profits, and hence you need a strong negotiating position and should apply all
kinds of negotiation tricks (Mastenbroek, 1987). In lean organizations, the supplier is
regarded as a partner in the value chain. This means that the quality of deliveries are
essential for your own success. The supplier is therefore provided with information
and expertise with which he could improve himself. In the classic bargaining
situation suppliers and buyers will seal important information off or supply
misinformation. In the mutual partnership relation co-makership is very likely. The
buyer might even take shares and thus take responsibility for the supplier's success
because he needs a continuous relationship.
• Relations with clients
In the classic organization, clients are regarded as the end of the production process.
In the lean organizations they are regarded as the starting point of everything. To
satisfy their needs is the ultimate goal of the lean organization. The classic
organization strives for maximization of profits, minimization of costs, and increase
of market share. It does not develop a clear picture of the environment, and has an
internal focus. Market research is done occasionally, but the results are not always
clearly communicated within the organization. Besides, market research mostly
describes the existing trends in the market. The lean organization tries to create the
market by having excellent contacts with buyers, constantly following them, so that
new ideas can evolve in direct contact with clients.
• Relations with employees
In classic organizations, employees are regarded as sellers of labor, and laborers and
management relations are mainly negotiations of the win-lose type. This can lead to
96 Organizational Learning and Information Systems
hostile relations. An illustration of this is in the USA where a position is mostly seen
as temporal and frequent job change is regarded as positive, because it demonstrates
the sales value of a person's human capital. In lean organizations, workers are seen as
essential for the company's success. They must be highly motivated and dedicated to
give an excellent performance. The organization invests strongly in its employees by
providing training and giving people life-long employment. In Japan the relation
between the organization and the employees even goes so far that the families of the
organization members are also regarded as part of the company (which gives these
organizations a traditional flavor).
• Financial decision-making structures
In the classic organizations, banks invest on the basis of profits and other financial
data, which are possibly compared with bench-marks from other organizations. In
lean organizations financial data are not that important. Longer term expectations
on qualitative issues such as the quality of the employee-management relations are
valued much more. A strong 'we'-feeling in the organization will possibly contribute
to the absence of strikes and the higher contributions of employees than can be
expected from the hostile relations of the classic organization. In Japan, lean
organizations are mostly part of larger conglomerates with an internal bank, called
'Keiratsu'.
• Human resource management ideas
In the classic organization an employee is attached to a specific position and may be
promoted or receive another position, although this is not common. He is also
expected to carry out uncritically whatever the management asks him to do. He must
adapt well to the existing organizational culture and setting. The lean organization
expects an open mind of its employees, and motivates people to bring in
unconventional ideas.
• People's motivation base
In the classic bureaucracy people have an extrinsic work motivation, meaning that
they work to receive a monetary retribution, and search for other jobs with better
payment. The lean organization aims at optimizing intrinsic work motivation,
meaning that people come into work because they are interested and gain great job
satisfaction from the work intself. The life-long employment, offered by many lean
organizations, also discourages looking for other jobs.
• Sources of new ideas and R&D
The classic bureaucracy suffers from a 'not-invented-here-syndrome', whereas the lean
organization actively searches for all kinds of ideas that might be interesting to
explore (Leonard-Barton, 1992).
The differences between lean and classic machine bureaucracies are thus summarized
under the 10 items mentioned. Chapter 7 describes a proposal for an index of
leanness of organizations based on these items.
Similar theorizing on lean organizations is also done by Hedlund (1994), while
defining his N-form and M-form organizations, where N stands for 'new' and
Organizational Learning in Machine Bureaucracies 97
'novelty', and M stands for 'multidivisional'. His publication, however, was done after
our operationalization and case studies were completed. For completeness, we
present one of his tables that gives a summary of his conceptualizations.
N-form M-form
Technological Combination Division
interdependence
People interdependence Temporary constellations, given Permanent structures, changing pool
pool of people of people
Critical organizational level Middle Top
Communciation network Lateral Vertical
Top management role Catalyst, architect, protector Monitor, allocator
Competitive scope Focus, economies of depth, Diversification, economies of scale
combinable parts and scope, semi-independent parts
Basic organizational form Heterarchy Hierarchy
Source: Hedlund, 1994, p. 83, tabel 1.
Table 5. 1: N-form vs. M-form
Most interesting is Hedlund's statement that 'new' is not always better. Table 5.2 lists
several issues on which the traditional, M-form, organizations perform better than
the N-form organizations.
Fundamental, radical innovation not achieved by Radical innovation through specialization, abstract
(re)combination and experimentation only articulation, and investment outside present
competences
Long time to acquire fundamental new knowledge Rapid infusion and diffusion of drastically new
because of restrictions on senior recruitment and prespectives through people, acquisitions, and spin-
acquisitions offs
Difficulty in coordinating very large projects because Large systems design capability through complex
of reliance on small groups articulation and tightly controlled complexity
'Competence traps' through too constrained Risk management through 'competence portfolio'
development path
Bias for internal exploitation of ideas Freedom to use most effective mode, internal or
external
Difficult to change overall vision because of internal Change of basic direction and culture through
management promotion external recruitment of top management
Strategic vulnerability through strong focus and Strategic robustness through quasi-independent
98 Organizational Learning and Information Systems
interrelationship parts
We will not further discuss which form of organization is the best, but we will
emphasize that the success of an organizational form depends on its environmental
conditions. The conditions of the most interest here are the organization's learning
needs (see further section 5.4).
The discussion about classic and lean organizations showed that machine
bureaucracies are moving to new, more competitive forms that match with new
environmental demands. At the same time, machine bureaucracies are changing in
their products and transformation technology. Two trends are particularly interesting
in this case. The first trend is industrialization of services (Grönroos, 1990). Services
lose their classic interpersonal nature by lowering the labor intensity and degree of
interaction (Schmenner, 1986, pp. 28-31). Cash dispensers are typical examples of
industrialized services because they enable clients to take money from their bank
accounts without direct interaction with a bank employee. The second trend
concerns manufacturing organizations that increase their supply of services (Kotler,
1988, p. 476-493). For instance, a car manufacturer may develop a car lease service as
a new business with synergy to manufacturing, and which is profitable in itself. Other
examples are: machinery manufacturing companies that also sell consultancy, and
computer manufacturers developing and selling software, supplying educational
programs and free communication services via a Value Added Network that is
otherwise used for user support and remote maintenance.
The importance of the distinction between services and manufacturing in our study
on organizational learning, is that both types of business have different types of
transformations and products that lead to different ways of organizational learning.
Some evidence for this proposition was found by Mills and Moberg (1982, reprint
from Bateson, pp. 152-153) in an overview of major research about the relation
between technology (which is treated as a synonym of transformation in their study)
and organization structure. Of the 26 studies reviewed, out of the 11 studies on
manufacturing 10 seemed to have found a clear relation between transformation and
structure. Of the 8 studies with a service sector sample, 5 showed relations between
transformation and structure. Of the 7 studies with a mixed population, only 1
showed a relation between transformation and structure. This finding is especially
important when organizational learning processes are regarded as a subtype of
organizational transformation, and when organizational learning norms are regarded
as a subtype of organizational norms. This would predict that the distinction between
Organizational Learning in Machine Bureaucracies 99
Source: Ginzberg and Vojta, 1981. Reprinted Table 5.5: Percent of Gross National Product in US
in Bateson, 1989, p. 25 economy.
Table 5.4: Percent of Labor Force in US
economy.
Also in terms of gross national product, services outnumber goods-production (despite the many problems with
measuring the value, of for instance, government services). See table 5.5.
The non-profit services are excluded from our research objective. Interestingly enough, many of these services
are now becoming profit services, because of government retrenchment policies (for example, health care,
pension funds, state computer facilities, railways). Because many additional services are now priced, and their
volume can be more easily measured, the percentage of services in gross national product will rise in future
statistics.
Ginzberg and Vojta also mention that services are becoming increasingly organized in machine bureaucratic
configurations as is stated in next quote (Ginzberg and Vojta, 1981, pp. 33-34):
" Since services for consumers have to be provided where the consumers are, economists have long assumed that the
100 Organizational Learning and Information Systems
economics of scale characteristic of manufacturing could not be achieved in service enterprises. Services cannot be
produced for inventory and cannot be shipped. That, however, is not the entire story. Improvements in
communications, particularly in processing and transmitting numerical data, facilitated the growth of large service
companies in the postwar decades by linking together in single enterprises large numbers of small service
establishments. Major banks were among the first to develop worldwide systems of branches. Now multi-unit hotel
chains, automobile-rental companies and fast-food-franchise enterprises have followed the example set by the banks.
The economics of these arrangements are based on the gains that the large service company can achieve through integrated
planning, financing, accounting, marketing and similar functions. Even large producer-service firms in law and accounting
have increasingly expanded overseas through the establishment of branches, partnerships or franchises. This development
helps to explain the surprising fact that legal services have recently emerged as the largest export industry in New York City,
outranking its apparel industry."
M. B. Service Manufacturing
Item
Materials and Knowledge Machines, physical materials and
Equipment labour
Involvement of client Client is ego-involved Client has contact after production
in production (sales) and sometimes before
production (design and contracting)
Information High, accurate and timely Planned work
processing information from client is needed
Responsibility for Client has joint responsibility for Responsibility for success lies with the
success success producer
Description of process Input, conversion and output are Clear distinctions between input,
phases hard to distinguish conversion and output (related with
logistic stream)
Stocks and buffers Stocks are impossible. Buffers are Stocks are possible (under certain
made by selection of clients, conditions) and buffers are created by
routinization of service and rationing planning of the production stream
Organizational Learning in Machine Bureaucracies 101
Systems boundaries Operating core is open system Operation and administration are
(involvement of client), both closed systems.
administration is closed system
Professionalism Can be high or low. Low (except in engineering)
Based on Mills and Moberg, 1982, pp. 154-161.
Table 5.6: Eight Items for Describing Service and Manufacturing Organizations.
On the basis of these ideal typical considerations, chapter 7 defines a scale of the
extent to which an organization can be called service or manufacturing.
Three major criticisms relating to machine bureaucracies can possibly affect this
study, and therefore must be commented on here. The first criticism is that, in
contrast to Mintzberg's statements, machine bureaucracies15 do not match the simple
and stable environments in an effective way. The second criticism states that machine
bureaucracies are not a relevant case to generate insights for a theory on
organizational learning. The third criticism states that machine bureaucracies are
unpractical in all cases and therefore are only the bad examples that must be removed.
Let us comments on these statements before we proceed.
Comments on Match
15
Doty, Glick and Huber (1993) make a most valuable distinction between organizational design
configuration (structure and processes) and contextual configuration (described in terms of dynamics and
complexity). Here, by the term machine bureaucracy we mean the organizational design configuration.
Organizational Learning in Machine Bureaucracies 105
Doty, Glick and Huber (1993) stated that few theories have received so much
attention with such meager empirical evidence as Mintzberg's organization theory.
The authors especially reject Mintzberg's statements about the match between
organization design configuration and organization environment. Table 5.8 shows
results from a sample of 128 organizations. They found a very low correspondence
between organizational design and environment.
Contextual Configuration:
Design Configura- Simple Machine Professional Divisiona- Ad- Total for
tion structure Bureaucrac Bureau- lized Form hocracy Design
y cracy Config.
Simple 3 1 6 2 0 12
Structure
Machine 1 5 10 21 1 38
bureaucracy
Professional 2 1 11 16 2 32
bureaucrcay
Divisionalized 0 0 0 10 1 11
form
Adhocracy 7 1 19 4 4 35
Total for 13 8 46 53 8 128
Contextual
Config.
Source Doty, Glick and Huber, 1993, p. 1217
Table 5.8: Doty, Glick and Huber's findings about Contextual and Design Configurations
This evidence would falsify the statement that design and environment should
match. Mintzberg however does not believe in a deterministic relation between
environment and organizational design. In fact, he states that more frequently
deviations of these statements exist. These findings also support our suggestions that
machine bureaucracies are often confronted with more dynamic and complex
environments than in the classic situation (low dynamics and low complexity). For
Mintzberg's theory the situation however worsens when organizations that fit
according to the theory do not perform well. The authors scored their sample on six
effectiveness criteria (derived from Quinn and Rohrbaugh, 1983), and correlated
these findings with the classification of each case in an ideal type, contingent ideal
type and contingent hybrid type16. In all these cases the correlations are very low,
16
The ideal type model is conceptualized as consistency across the relevant dimensions and is modeled as
the lack of deviation from the one type. The contingent ideal type model defines a finite number of ideal
106 Organizational Learning and Information Systems
types of contexts and a single ideal type of organization structure or strategy that is appropriate for each
ideal-type context. The contingent hybrid type model allows hybridization among the initial ideal types and
defines continua of contexts. A single hybrid type should then match a specific context (Doty, Glick and
Huber, 1993, p. 1202-1203).
Organizational Learning in Machine Bureaucracies 107
However, although many criticisms can be directed at the study of Doty, Glick and
Huber, for instance that they used a non-random sample of only 128 cases, the study
clearly indicates some problems with the Mintzberg theory. Because not many other
tests of Mintzberg's theory have been carried out, and the theory is based on a
volumious amount of research, summarizing most of the main stream of organization
analysis from the beginning of this century till the beginning of the 1980s, it would
be unwise to throw away this baby with the bath water. Nevertheless, important
amendments to the theory are required. In this study for example we explicitly study
lean machine bureaucracies.
Comments on Relevance
A hypothesis in this study is that many machine bureaucracies are confronted with
increasing complexity and dynamics, which would remove them from the machine
bureaucratic configuration. This is in many instances not possible, because of
typicalities of the production process in that it requires a strong technostructure that
defines standard work processes. The lean organization for instance cannot do without a
technostructure, but makes better use of knowledge and skills available in the
operating core. This leads to a closer relationship or integration between
technostructure and core. Empirically this can look like decentralization. Many
service organizations have made serious attempts to get away from the expensive
professional way of service supply, and demand more standardization in work
processes (Grönroos, 1990; Schmenner, 1986). The demand for more flexibility is
realized by the extensive use of information technology and providing service
specialists with easy access to client and product data.
The move of machine bureaucracies to simple structures is not relevant, because the
simple structure lacks the knowledge, the division of labor, and the infrastructure to
supply services and goods on a large scale for low costs. The move to divisionalized
forms often occurs when the machine bureaucracies become part of larger
organization conglomerates. This does not necessarily affects the stability and
simplicity of the environment. It could however increase the organization's
opportunities of access to knowledge and information from other business units that
are part of the divisionalized form. Research has indicated that this does not happen
often. Adhocracies are irrelevant for our research, because they are only viable in very
dynamic and complex environments, and have only limited abilities for large scale
production. Missionary forms, that use indoctrination as a major coordination
mechanism are likely for religious organizations and political parties. Some machine
bureaucracies, especially Japanese manufactures, also put much emphasis on
indoctrination. This is also very relevant for lean organizations in addition to the other
coordination mechanisms. By indoctrination, management costs can be kept low,
which keeps the organization lean. This leads to the insight that organizational
configurations are not only determined by their environmental context, but that
108 Organizational Learning and Information Systems
17
Here obviously the machine bureaucracy is impied.
Organizational Learning in Machine Bureaucracies 109
• Job enlargement, and asking people to find a way of organizing this themselves.
• Flatter structures, with more decision responsibilities at lower levels (job
enrichment). This led to so-called organic organizations, that demanded a
considerable sense of responsibility of people on the shop floor, though it was
still very formal in its working methods (professional bureaucracies). Some
organic organizations also removed formal ways of working and thinking, and
are mainly designed to solve unique problems (called adhocracies).
Many of these suggestions were adopted in the conceptualization of lean
organizations (Womack et al., 1990) and 'The Learning Organization' (Senge, 1990a).
These are discussed under the headings of responsibility norms, procedural norms
and action norms.
Complexity and dynamics, are both determinants of the need for organizational
learning. Both are individual or shared perceptions of internal and external
environmental reality. After Duncan (1972), an organization's environment is split
up into 8 components, further divided into several factors (see table 5.9).
Internal Environment
1. Organizational personnel component
a. Educational and technological background and skills
b. Previous technological and managerial skill
c. Individual member's involvement and commitment to attaining systems's goals
d. Interpersonal behavior styles
2. Organizational functional and staff units component
a. Technological characteristics of organizational units
b. Interdependence of organizational units in carrying out their objectives
c. Intra-unit conflict among organizational functional and staff units
d. Inter-unit conflict among organizational functional and staff units
3. Organizational level component
a. Organizational objectives and goals
b. Integrative process integrating individuals and groups into contributing maximally to attaining
organizational goals
c. Nature of the organization's product service
External Environment
4. Customer component
a. Distributors of product and service
b. Actual users of product and service
5. Suppliers component
a. New material suppliers
b. Equipment suppliers
c. Product part suppliers
d. Labor supply
110 Organizational Learning and Information Systems
6. Competitor component
a. Competitors for suppliers
b. Competitors for customers
7. Socio-political component
a. Government regulatory control over the industry
b. Public political attitude towards industry and its particular product
c. Relationship with trade unions with jurisdiction in the organization
8. Technological component
a. Meeting new technological requirements of own industry and related industries in production of
product or service
b. Improving and developing new products by implementing new technological advances in the
industry
Source: Duncan, 1972, p. 315.
Table 5.9: Factors and Components Comprising the Organization's Internal and External
Environment.
The simple part of the simple-complex dimension deals with the degree to which their
are only a few factors of relevance in the decision unit's environment, and these
factors are similar to another in the sense that they are located in the same or a few
components. The complex part indicates that the factors and components in the
decision unit's environment are large in number (Duncan, 1972, p. 315). In complex
situations the development of knowledge (e.g. in terms of action-outcome relations) is
difficult, and needs highly specialized and educated people. Thus there is a direct
impact of the complexity dimension on the needs for knowledge development
activities, the dissemination of the relevant knowledge and the actual use of the
knowledge. Duncan proposes to measure the amount of complexity by multiplying
the number of decision factors with the square of the number of components.
The static-dynamic dimension refers to the unpredictability of the environment. It is
measured via the frequency by which the factors of the decision unit's internal and
external environment remain basically the same over time or are in a continual
process of change. This dimension contains two subdimensions:
• The degree to which the factors identified by decision unit members remain the
same over time, or are in a process of change.
• The frequency with which decision unit members take into consideration new
and different factors in the decision-making process.
This obviously indicates learning needs because high dynamics requires the ability to
update frequently and remove obsolete knowledge. Re-use of knowledge in this case
will be small because conserved knowledge depreciates easily in highly dynamic
environments, except when connected with more fundamental knowledge (double-
loop learning). In more dynamic environments 'quick fixes' (single-loop learning) are
not sufficient for an effective organization in the longer run.
Dynamics and complexity together determine the score on organizational learning
needs. Especially the dynamic dimension contributes to the learning needs score,
because complexity can be solved by developing a correct theory that is implemented
Organizational Learning in Machine Bureaucracies 111
in practices and machinery, and thus considerably reduces uncertainty and the need
to learn. The dynamic factor constantly and directly increases the need for learning.
In very high dynamic environments learning is very difficult, as Hedberg (1981)
stated, and decisions must be based on incomplete knowledge. The organization in
that environment requires a very flexible, team-oriented configuration, that is closer
to an adhocracy or a professional bureaucracy than to the machine bureaucracy.
The organization literature (Buchko, 1994) has criticized this operationalization of
dynamics and complexity as a measure of environmental uncertainty. It has, however,
high construct validity as a measure of organizational learning needs because of its
emphasis on generating insights in factors and their relations (management
knowledge) (also cf. Duncan and Weiss, 1979).
1. Learning Policy
The following issues for mission and policy norms were determined in chapter 4:
• The adaptation of organizational learning in mission and identity statements of
the company.
• The development of an organizational learning infrastructure.
• The development and management of core competencies.
• Basic organizing principles that support learning..
• Management motivation for business re-engineering.
As shown in sections 5.2.2 and 5.2.3, machine bureaucracies can differ significantly
on the lean-classic (organizational norms) and service-manufacturing (organizational
transformation) dimensions. Additionally it was assumed that these dimensions
strongly link to the way organizational learning occurs. Table 5.10 therefore describes
how lean and classic machine bureaucracies differ on mission and policy norms.
The differences between both machine bureaucracy types are therefore huge on the
learning policy side. Norms and transformation technology are however not
unrelated. For instance a service organization puts much effort into an infrastructure
for supporting communication, because communication is an essential technology
for generating added value in services (Schmenner, 1986). This infrastructure is
however not necessarily used for organizational learning. Because service
organizations are mostly dependent on organization members' knowledge and skills,
they will probably place more emphasis on the development and management of core
competencies. They also require open systems, because it is the clients themselves
who are the subject of the production process. Business re-engineering is an
important issue in service organizations, because the number of services and service
varieties is exploding, and the market demands dramatic reductions of production
costs. Information technology has an important leverage potential here (Hammer,
1990). One of the most important reasons why services can more easily re-engineer
than manufacturing organizations is the fact that most of their norms are
implemented via software (in the broad sense of the term including computer
programs, organizational written or tacit rules), whereas manufacturing organizations
mostly have substantial large investments in machinery that is difficult to change or
replace (hardware constraints).
These considerations suggest that it would be easier for service organizations to
become lean than manufacturing organizations. There are however no data that
clearly support my opinion. This study wil further explore some of the statements
made about the service-manufaturing distinction in chapters 8 and 9, while
comparing service and manufacturing companies.
hierarchical levels, and span of managers' control (Daft, 198918). Classic machine
bureaucracies are characterized by centralized authority, and some vertical
decentralization of authority to the technostructure. Because of the increase in
learning needs some interesting transformations of learning responsibilities can be
predicted.
In situations of increasing complexity, decentralization is required because
management is not able to understand and decide about all issues (Galbraith, 1973).
Increasing dynamics requires fast decision-making lines as well, which are only
realistic in situations of decentralization and delegation of authority.
In classic machine bureaucracies, division of labor is along functional lines, meaning
that people are grouped together in departments by common skills and activities.
This can lead to high expertise within separate departments and persons, but to
severe coordination problems as well, when a more dynamic environment requires a
flexible combination of skills and effective interchange of knowledge.
18
Daft also mentions formal reporting lines and systems for effective coordination as part of organization
structure. I prefer to see them as part of the organizational processes that are closely related to procedural
norms to be discussed later on.
114 Organizational Learning and Information Systems
The classic machine bureaucracy, with its functional organization, must move to
divisional, matrix and even to team-based organization principles in a situation of
growing learning needs. The lean organizations have already achieved a strong team-
oriented organization. The service organizations can easily be transformed to the
team-like structures, because they are often involved in highly interpersonal
technology (except with the back office) and higher market dynamics provide
stronger incentives to move in that direction. For manufacturing, the strong coupling
between activities in the production process make it more difficult to create
autonomous teams. The technostructure will have an important say in what happens.
out with explicit reward expectations. For instance communication about defects and
problems is obscured when it leads to negative evaluations. In situations where
interpersonal trust is very high, these kinds of group dynamics do not work. Argyris
(1970) and Argyris and Schön (1978) however attribute most learning errors to a lack
of interpersonal trust in organizations, so that the real causes of the learning
problems of the organization are not discussed.
Knowledge and information are also often connected with the status people have.
Removing the knowledge then leads to a change in power distribution, and therefore
easily leads to resistance of change. Action norms also include statements about the
importance of learning and change in relation to existing work. Simon (1976)
therefore states that organizations do not try to achieve omniscient knowledge, but
that organizations (and people) try to achieve a satisfactory situation. Simon is right
here, however, this principle is difficult to evaluate because many different goals and
values motivate people at any one time. As a consequence, evaluation of
organizational learning needs should be based on a broader view of organizational
priorities.
Finally, action norms concern the motivation of organization members to develop
and adapt knowledge, or an organization's willingness to not do so, by buying the
knowledge and skills elsewhere. This issue is an implementation of an organization's
view of what must be regarded as its core competence in terms of concrete actions
that must be taken to acquire the knowledge.
Table 5.12 sketches how action norms relate to organizational leanness.
4. Procedural norms
Procedural norms relate to the way information is handled in organizations.
Complexity and dynamics have a direct impact on the contents of these norms.
The traditionally stable environment of machine bureaucracies led to the
Organizational Learning in Machine Bureaucracies 119
19
Has recently changed. For instance Philips Electronics is now aiming for survival, and is taking part in a
very competitive and risky business.
120 Organizational Learning and Information Systems
might be very ineffective, and a high relationship (low task) style is required.
Organizational learning in this case involves not only learning the job (task), but
learning to manage persons and relationships as well (Hersey and Blanchard, 1982;
20
This study does not investigate the deutero learning process, which is about the shaping of the
122 Organizational Learning and Information Systems
organizational learning norms. It takes its results (learning norms) as an internal contingency factor for
analyzing the effectiveness of the single-loop and double-loop learning processes and the role and value of
MICS. The reason for this limition is that MICS has no role in the deutero learning process, because MICS is
part of the (procedural) learning norms itself.
Organizational Learning in Machine Bureaucracies 123
This subsection explores the role of machine bureaucracies in the development and
removal of management theories, their storage, use, dissemination and adaptation.
These learning activities are labelled double-loop and single-loop learning.
Stable and simple environments do not require much double-loop learning. The
development of knowledge is then often delegated to specialists in the
technostructure (frequently organized in a Research and Development department),
sometimes with the involvement of the management as well. For instance, in an
insurance company I visited21, the principle trend for managers was to delegate all
knowledge development processes to
specialists, even the most simple ones.
Development of knowledge is done
infrequently and has a low priority in
simple and stable environments.
Feedback comes slowly and irregularly,
mostly organized in market research. In
these studies potential customers
express their opinion of products and
services and their future demands when
requested. The results of these studies
lead to decisions about developing,
manufacturing and selling new products. Because of the low complexity and
dynamics the results are very precise, so that good calculations of costs and benefits
can be made. The environment is low risk, and therefore demotivates the search for
innovations.
21
Not the one that is described in chapter 8.
124 Organizational Learning and Information Systems
When the environment becomes more dynamic and complex a more active
knowledge development approach is required. Besides, it becomes difficult for
specialists alone. Often in machine bureaucracies problems are encountered in the
manufacturing of new products because members from the production department
were not involved in the product design (Hill, 1984). The risk also increases as the
environment becomes more complex. Processes of critical evaluation must be
speeded in order to gain relevant knowledge, because knowledge depreciation also
speeds up. Knowledge development must also become systematic (e.g. explicitly
asking clients about satisfaction, and actively searching for problems to be solved).
When the dynamics of the environment is very high, knowledge depreciation is faster
than the knowledge development process. In this case the value of knowledge
declines rapidly, and management is either left over to good luck or beome able to
avoid uncertainty by creating a negotiated environment (Cyert and March, 1963).
Figure 5.3 shows that the increase of the value of knowledge is directly related with
the complexity of the environment. Figure 5.4 demonstrates that the accumulation of
knowledge is effective until the environment passes a certain level of dynamics. After
that moment the depreciation of knowledge goes faster than its accumulation of
value. Speeding up the feedback process can increase the value of knowledge, as is
shown in fig. 5.5.
In case of high complexity and
dynamics, a delegating style is
appropriate, because very short
communication lines are required and
much knowledge is decentralized. The
decentralization also leads to a
restriction of the area that must be
understood, and thus simplifies the
problem. This can of course lead to
suboptimization and dysfunctional
effects in the longer run. It is typical of double-loop learning that it detects these
suboptimization problems and solves them by generating an awareness of limitations,
and the new insights that are required. Action norms (motivation to rethink the
management theories especially in a broader perspective) and procedural norms
(creating communications and activities to detect limitations and discover a wider
perspective) must be set so that double-loop learning activities emerge.
Organizational Learning in Machine Bureaucracies 125
Storage of knowledge
When dynamics increases, the storage of knowledge is useful for creating continuity,
but can also act as a brake on innovations. Stored knowledge can be used for learning
from the past. A more interesting opportunity is that of innovation by connecting
different core competencies. This requires a matrix organization structure or task force,
because core competencies are generally not shared among departments or SBU, and
an inspirational leadership that stimulates ideas and activities in the organization for
connecting competencies that are seemingly very different and unconnected.
These considerations are generalized in table 5.16 that summarizes the use of
conserved knowledge in lean and classic machine bureaucracies.
MB: Lean Classic
Storage:
Acquisition Much knowledge and data Much knowledge and data
Retention Less, because much removal when needed Much
Closely connected with mind and motivation In archives, formal rules and procedures
Retrieval Much. Applied to problem solving Less. Connected with procedures,
indirectly linked with problems
Table 5.16: Differences in Knowledge Storage Between Lean and Classic Machine
Bureaucracies
Use of knowledge
Dissemination of knowledge
Adaptation of knowledge
Low complexity and a rather stable environment lead to slow and incremental
changes in organizational knowledge. Actions in this situation are strongly motivated
by action plans that have undebated models containing means-goals theories as their
foundation. Not using these models can lead to actions that are not-legitimated. This
can lead to severe sanctions when with hindsight these actions seem to have been
ineffective. Adaptation of knowledge is mostly done by the person responsible for the
knowledge. In the highly differentiated structure of machine bureaucracies this
implies that knowledge adaptation is a specialist (technostructure) activity. Because
feedback about mistakes is a slow process, adaptation is a time-consuming business,
often leading to the implementation of knowledge that is already invalid.
Additionally, the low risk of the environment demotivates organization members to
start adaptation processes.
Performance control for evaluation and knowledge adaptation is often not done and
not enough attention is paid to it. For instance in the Dutch high-tech company
mentioned in chapter 4, at an assembly unit in 1992 production norms were used
Organizational Learning in Machine Bureaucracies 129
that had been developed in 1968. In this organization (with fast moving dynamics
because of rapid products and process innovations, this led to norms that are not
applicable for the effective steering of the units. Strangely enough the organization
has not changed its production norms in 14 years. Now that the company has come
into a very hostile and competitive environment, it is being forced to reconsider its
norms. This study is not being carried out by its own technostructure or
management, but by one of my M.Sc. students! The rules, norms and procedures
have reified the organization in such a way that it is not capable even of realizing a
single-loop process.
The adaptation process can provide important triggers for double-loop learning. This
happens when the complexity of the environment increases, so that the management
theory cannot give a valid explanation or offer effective proposals, or when the
dynamics has increased to such an extent that a theory must be found that improves
understanding. Often uncertainty avoidance strategies are regarded as more effective
than learning. In the longer run this could be untrue, as was shown in the case of
lean production.
The rating of learning activities can be done by noting the number of learning fields
an organization is concerned with, and the number of activities undertaken as was
done for the double-loop learning score. A further description of a scale for single-
loop learning is presented in chapter 7.
5.5 Summary
In this chapter we presented the major factors describing the lean-classic and service-
manufacturing dimensions of machine bureaucracies. The lean-classic distinction is
defined by 10 items: attitude to quality, level of decentralization, availability of lateral
structure, relations with clients, relations with employees, financial decision-making
structures, human resources management ideas, people's motivation basis, and
sources of new ideas. The manufacturing-service distinction is described by 8 items:
material and equipment, involvement of client in production, information
processing, responsibility for success, description of process phases, stocks and
buffers, systems boundaries, and professionalism.
The term learning needs has become more concrete via the application of Duncan's
list of environmental components and factors by which the dynamics and the
complexity of organizations can be scored. In chapter 7 we will investigate how an
index of learning needs can be defined.
The term deutero learning is made more operational in this chapter. Deutero
learning can be described by its leanness and service-manufacturing nature.
Responsibility norms can be described in 8 organization structures: functional,
divisional, matrix, task groups, networks consisting of independent companies,
networks with expertise centers, volvos, and parallel learning structures. Action
norms are about incentives, interpersonal trust, attitude to knowledge removal,
learning priority, and source of knowledge. Procedural norms are about feedback
frequencies, but are not very concretely defined yet. This will be done in more detail
in chapter 6 (about MICS).
Some hypotheses were defined about the value of knowledge given a certain degree of
complexity and dynamics. One can state that increasing complexity leads to a higher
value of knowledge. The value of knowledge increases until a moderate level of
dynamics is achieved. Higher dynamics leads to a decrease of the value of knowledge.
Single-loop learning is linked with both the lean and the classic characteristics of
organizations. For storage specifically, the focus is on acquisition, retention and
Organizational Learning in Machine Bureaucracies 131
retrieval of knowledge, in all cases very different in lean and classic machine
bureaucracies (table 5.16). For the dissemination of knowledge, the focus is on
distribution, mutual understanding, and synchronization of understanding. The
adaptation of knowledge seems to be different for action planning and performance
evaluation. In the following chapters we will often use the terms problem
anticipation and critical evaluation instead of action planning and performance
evaluation, to stress the importance of learning in these activities. Further arguments
for the use of these concepts are also given in section 6.3.2.
In chapters 6 and 7 the dimensions of organizational learning will be further defined.
132 Organizational Learning and Information Systems
Role and Value of MICS for Organizational Learning 133
6.1 Introduction
The organizational learning literature is full of contradictions about MICS' role and
value. In her study of Texas Instruments' planning and control information system,
Jelinek (1977) concluded that information systems contributed considerably to
(institutionalizing) organizational learning. These results have been severely criticized
by Mintzberg, because according to him the systems did not capture any knowledge
and they failed soon after her book was published. Mintzberg therefore states (1989,
p. 350):
" Texas Instruments' own fancy planning system was subsequently believed to discourage
innovation. In fact, there never was any evidence that the company's success stemmed
from anything more than a capable leader who knew how to learn and whose own energy
and enthusiasm enabled him to attract good people and to invigorate them. Good people,
of course, make for good organizations. They also design good systems, at least systems
that are good for them. But remove the good people and the systems collapse. Innovation,
it turned out, could not be institutionalized."
In another publication Mintzberg directly links MICS' abilities with machine
bureaucratic environments:
" ...in the tall administrative structure of the Machine Bureaucracy, information must pass
through many levels before it reaches the top. Losses take place at each one. (...) The fact
that the transfers are vertical - between people on different status levels of the hierarchy -
means that intentional distortions of information also occur. (...) Probably a greater
problem is the MIS's emphasis on 'hard' (quantitative), aggregated information. A good
deal of evidence suggests that it is not this kind of information top managers need to
make their strategic decisions as much as it is soft, specific information. (...) Often the
MIS data are too late as well" (Mintzberg, 1983, p.184).
Mintzberg's conclusion is that MICS contributes nothing.
Mohrman and Cummings (1989) are more optimistic about MICS use:
" High-performing organizations employ multiple systems for gathering relevant
information, making appropriate decisions, and communicating responses to specific
groups and departments. They often supplement sophisticated management information
systems with less quantitative devices to ensure that information flows in all directions.
For example, these include coordinating councils, customer focus groups, employee sensing
meetings, cascading information sessions, and weekly videos of 'messages from the
president'. These information-processing systems help organizations better scan their
environment and integrate their subparts so they respond to complex and changing
134 Organizational Learning and Information Systems
22
Argyris uses the term MISS for monitoring information and control systems.
Role and Value of MICS for Organizational Learning 135
Many times, the term 'information system' has been connected specifically to
computer-based information systems. Although it is obvious that computer
technology has contributed substantially to the creation of modern information
systems, the major developments in information systems were not computer-based
but social and organizational. A restriction to computer-based systems does not:
1. Include the many formal and manual systems that do precisely the same work
and are often required in addition in order to achieve effective computer-based
systems.
2. Include the many informal systems and norms that are also engaged in
information processing.
Number 1 is exemplified by a case on flexible manufacturing systems. In order to
manufacture a certain steel pressed product, a small company bought a flexible
manufacturing unit costing over $200,000. This advanced system was, however,
hardly used. People continued using the traditional construction tools, because the
organization was not able to provide the required data input to manage the
manufacturing process via the computers. As a consequence, an elaborate formal
administrative organization had to be developed that was not yet in existence in this
small informal organization.
Number 2, I encountered in a large insurance company that had developed an
executive information system. The management problem was however not the lack of
accurate information, but the organizational culture, not used to top executives
having such an active approach to management and information gathering. This new
way of management by open communications between executives and directors of
the departments, meant a large change in work and culture.
An assessment of information systems thus not only requires an evaluation of their
technical operation, but also an evaluation of their social function. This means that
all six layers of information are necessary to evaluate the physics, empirics, syntactics,
semantics, pragmatics, and social aspects (Stamper, 1973; also see section 1.4). The
first three layers are about information technology, the other three layers are about
social and organizational aspects of information systems.
One of the basic reasons for starting this study, and many other IT impact studies, is
that information technology can change traditional ways of management and
organization. This statement however, does not mean that one should use a
technology deterministic view on impact, because impacts are a result of the
interaction of organizational features and information technological opportunities
(cf. chapter 2). This means that one must consider the formal and informal
information systems as a broader environment in which computer-based information
systems impact on the organizational environment. The basic computer technologies
136 Organizational Learning and Information Systems
These technologies are based on the application of digital computers that consist of
internal and external memories, control units, registers, and logical and arithmetic
units (Tanenbaum, 1984). They serve as replacements for archives in the shape of
paper files, personal files and memory that can have exactly the same function. The
main question at the technical level therefore becomes: What is most cost effective:
computers or the traditional technology? The answer to this question depends on the
application field. In purely technological terms the answer depends on the cost of
purchase and maintenance of the physical devices, the complexity of creating the
software and procedures and technical problems with coding and storing. Each time
of use would mean some (fictional or real) payback of the investments involved. If
use is restricted to a specific period the payback is often not sufficient.
Information systems can have several functions (roles and values) in organizations.
Several approaches are valuable in describing these functions, such as:
1. Economic principles, for instance via description of the value chain and IT's
contributions (Porter, 1985).
2. Organization design principles, by describing different coordination mechan-
isms and IT as an additional coordination mechanism (Mintzberg, 1983).
3. Management functions and IT's role. This is especially useful when trying to
define user requirements for specific management functions (Davis and Olson,
Role and Value of MICS for Organizational Learning 137
1985).
4. Organizational transformation, which closely resembles the previously
mentioned value chain, however with emphasis on the construction of products
and delivery services as a transformation of inputs to outputs (e.g. Perrow,
1967; Hill, 1983). In that case information systems can have an important role
in controlling, planning and checking production runs.
5. Planning principles by identifying stages in the development of the information
supply by constructing a hierarchy of systems (Anthony, Daerden, Bedford,
1984; Porter, 1988).
Organizational learning focuses on management in the organization's context and
therefore requires the organization design and coordination perspective. The management
approach might be interesting as well, but requires the elaboration of a different
conceptual framework.
The coordination approach states that a viable organization has an operating core, a
middle line, a strategic apex, a support staff (including administrative services), and a
technostructure. Organizations can differ tremendously in the number of people and
amount of money spent on these five groups in the organization (cf. chapter 5).
Information systems have organizational functions when supporting or replacing
these groups. These systems generally have specific names (cf. Markus, 1984; and
McKeown and Leitch, 1993), and are grouped in table 6.2.
This study is about the MICS systems (often also called MIS or Management
Reporting Systems). At the semantic level MICS systems must have data and models
that are unambiguous for different users. This presumes a common understanding of
reality described in shared management theories. The receiver of these insights and
data is not a passive consumer of these messages, but checks its validity in relation to
other insights and data. This involves an adaptation process, which requires a specific
responsibility structure. At the pragmatic level the insights gained from new theories
138 Organizational Learning and Information Systems
and data must result in new behavior. It is therefore important that action norms are
created that induce people to use new data and insights. The receivers of these data
and insights also have opportunities for creating their own knowledge, so that the
received messages are checked for their relevance. At the social level MICS' role must
fit in a set of learning responsibilities and procedures. When for instance MICS leads
to a possible change of power (as was the case with Markus' financial information
system and Zuboff's paper mill, cf. Markus, 1983 and Zuboff, 1988) this must be
foreseen and be part of its design. Also ways of disseminating messages, ways of
communicating and decision-making can change and must be planned and
implemented when required. MICS distinction from other information systems (such
as operational systems, DSS, knowledge-based systems, and CSCW) is however only
typological, and in practice not always easy to make. It does however focus attention
on issues of importance for investigation.
An additional way of looking at IT can be mentioned: IT-architectures. IT-architectures
are configurations of systems. This proposes interconnections between systems,
which makes it sometimes difficult to talk about systems as having one function or
one technology. The architecture concept also suggests the view of separate systems
that profit from each other's data, models, and features (e.g. MICS using the database
of the order processing system that is connected with an Electronic Data Interchange
system using telematics, Earl, 1989). Because of increasing IT-integration, this is
becoming more realistic and will pose major organizational learning questions in the
near future.
Figure 6.1 gives a quick summary of the information systems typology found so far.
In the following subsections, the technical and organizational dimensions of MICS
are further described.
1. Physics
Role and Value of MICS for Organizational Learning 139
2. Empirics
From an empirics point of view, MICS's require a specific kind of information flow.
Depending on the way these data flows are organized, one can state that the
organization has a different level of
control.
A basic feature of all MICS is that data
about operations' results are created and
made available in the organization. This
first level of control, as described in
figure 6.2, leads however to
interpretation problems, because:
• Norms are unknown, which
makes it impossible to make
comparisons with plans and this
means that data about results are used as arguments for and against, depending
on the interests at hand.
• Ways of measurement are unreliable, because no systematic measurement tools
have been developed that are relevant from the managerial perspective.
Both problems are solved by developing clear standards of performance in the
problem anticipation process and by developing measurement instruments. Often
only one of these solutions is implemented leading to second degree control (cf.
figure 6.2). Option 1 is the case of feedforward control, and option 2 concerns
140 Organizational Learning and Information Systems
feedback control. Both are ineffective when not combined. The feedforward system
does not evaluate and check its assumptions. The feedback system does not know
why the operations were executed, and thus does not know the assumptions. When
both systems of feedback and feedforward are combined a closed control loop is
realized called third level control (Flamholtz, 1983; cf. figure 6.3). This closed control
loop becomes a single learning loop when an evaluation and reward component is
added, which motivates people to draw conclusions from the data (figure 6.3). In this
case the technical information system and the human motivation system are
connected, which is a basic requirement of any human learning system and the
fourth level of control. The single-learning loop becomes a double-learning loop
when the evaluation, measurement and planning subsystems become a field for
feedback. This corresponds to the fifth degree of control. IT can support closed
learning loops, because of its efficient, quick analysis, and fast dissemination of data.
Lean organization possibly profit more from IT for learning, because of their higher
awareness of data value (lower power
play and more explicit mental models).
3. Syntactics
A fully developed MICS thus consists of
six basic elements that form the MICS-
structure (Flamholtz, 1983; Ansari,
1977):
• Standards and objectives.
• Measurement instruments.
• Source of data and order of
presentation.
• Timing and frequency of data.
• Route of data flow.
• The extent of information sharing among potential users.
The cybernetic approach to organizational learning defines MICS in terms of its
information flows (empirics) and its structure (cf. the discussion of De Raadt's
cybernetic organizational learning concept in chapter 4). They do not bother about
complications with understanding a system's output and how new insights should
change behavior or organizational structures. This last concern is stressed by
organization development authors, to which we owe much of the following
description of semantics, pragmatics and the social aspects of MICS.
At the syntactic level MICS also should have some important features, as otherwise
its usableness (and success) will be low. These features are:
• Man-machine interface quality
• Flexibility of databases, providing many entrances and opportunities of
adaptation
• Quality of the administrative organization. The chance of poor data can be
reduced by introducing rigorous procedures, and methods of checking quality
Role and Value of MICS for Organizational Learning 141
of data.
• Comparability of data structures, so that the chance of effective exchange of
data is high.
4. Semantics
Weick (1985, pp. 52-54) describes these problems as sense-making processes, with the
following elements:
1. Effectuating: "People learn about events when they prod them to see what happens. (...)
People find out what's going on by first making something happen. (...) Since action is the
major source of human perceptions and intuition, any assessment of the potential for
sense making must pay close attention to action." The involvement of the people
where the data are about in interpretation processes is therefore essential in
sense-making.
2. Triangulating: "People learn about an event when they apply several different measures
to it, each of which has a different set of flaws. (...) These various 'barometers', each of
which presents its own unique problem of measurement, begin to converge on an interpre-
tation".
3. Affiliating: "People learn about events when they compare what they see with what
someone else sees and then negotiate some mutually acceptable version of what really
happened". People not only want to have several sources of data (triangulating)
but also want to discuss with other people how they perceive reality and want to
bargain about what 'really' happened.
4. Deliberating: "People learn about events through slow and careful reasoning during
which they formulate ideas and reach conclusions". This simply means that people
need time to make up their minds about reality by means of interpreting data.
5. Consolidating: "People learn about events when they can put them in a context". The-
refore people need more than simple data, but also a view by which they can
relate data to interpretations of what happened.
Information systems, particularly MICS, are supposed to support making sense,
therefore to support the creation of information or mental models instead of information
processing capabilities (Nonaka, 1988). Weick is however not very optimistic about
the potentials of information systems to support the sense-making process:
" People using information technologies are susceptible to cosmology episodes because they
act less, compare less, socialize less, pause less, and consolidate less when they work at
terminals than when they are away from them. As a result, the incidence of senselessness
increases when they work with computer representations of events" (1985, p. 56).
In my perception the cosmology problem has its roots in a poor match of MICS with
individual and organizational interpretation systems.
Individual interpretation systems concern the individual's way of understanding the
142 Organizational Learning and Information Systems
world. Four types were identified on the basis of Kolb's prehension and
transformation dimensions in chapter 4 (figure 4.1): accommodator (concrete
experience and active experimentation), diverger (concrete experience and reflective
observation), converger (abstract conceptualization and active experimentation), and
assimilator (abstract conceptualization and reflective observation). MICS systems
match well with the 'diverger', because these systems improve the sensing process by
supporting the process of data gathering in organizations. The 'assimilation' learning
type is mainly interested in developing and applying abstract models. In fact MICS
develops models only in an incremental way. The data can improve insights in reality,
but the MICS-application is not explicitly designed for improving or generating many
alternative models. MICS contains a model of reality itself, that is difficult to change.
EIS-applications have more opportunities here. Particularly Decision Support
Systems are designed for developing models (c.f. 'I Think'; Wijnhoven, 1992b). E-
mail systems, teleconferencing and Group Decision Support Systems are particularly
designed to 'converge' insights that are available at different places and persons in the
organization.
Especially in small companies, and in situations in which managers work
independently, these psychological traits are important. If the decision-making
process is of a more collective kind, it is more useful to stress the importance of
culture and organizational structures and processes in understanding the
interpretation of data. According to Daft and Weick (1984) organizational
interpretation systems should be designed in line with the complexity of the
environment, they call it analyzability, and the extent to which organizations actively
search for interpretations themselves (learning effort in our terms). By applying both
variables dichotomously, four interpretation systems are defined, presented in table
6.3.
Computer systems could be usefully applied when the situation is analyzable and
when large amounts of data must be processed. MICS is often used in organizations
for supporting 'conditioned viewing'. This means the monitoring of processes by
means of internal and impersonal (objective) data. This implies that the information
output must lead to clear action proposals, which is only possible in closed loop
learning situations. In lean organizations, the data from monitoring processes are
however also used for a more active 'discovering' of what is at stake, and therefore a
start to learning. Executive information systems are often mentioned as tools for
supporting the 'enacting' process, which requires opportunities for processing data of
an undetermined format. 'Undirected viewing' is only possible with very informal
information systems (Hedberg and Jönsson, 1978). Computer-based information
systems will possibly not pay off in that case.
The analyzability also requires different types of control. Hofstede found four basic
questions that determine organizational control:
1. Are objectives unambiguous, or can ambiguity be resolved?
2. Are outputs measurable, or can acceptable surrogate measures be found?
3. Are the effects of management interventions known?
144 Organizational Learning and Information Systems
Political control situations have incompatible models of reality, which means that
data can be interpreted in different ways to support conflicting interests. All other
types of control, with the exception of expert control, refer to situations in which
shared mental models are created, and therefore data lead to interpretations that
support common interests.
As a conclusion, three types of scores of MICS can be given at the semantic level:
• Learning styles of users and implicit learning styles of systems. The thesis was that
MICS mainly supports divergence and accommodation knowledge. If the
organization (or situation) requires convergence and assimilation, MICS cannot
cope.
• Analyzibility and intrusiveness of organizations and match with MICS. The thesis is
that MICS supports 'conditioned viewing' and 'discovering' (both require shared
mental models to make unambiguous interpretations), but is poor on 'enacting'
and will not support 'undirected viewing'.
Role and Value of MICS for Organizational Learning 145
• Control types and the extent of incompatibility of mental models that are the conceptual
basis for MICS. When outputs cannot be scored (also not with surrogate
measures), MICS has little use. MICS cannot support situations that are
characterized by political control. MICS will be particularly relevant at 'trial and
error' and 'routine' control, because people can learn from data about processes
that are repetitive. This can lead to a situation of knowledge saturation, thus
'expert' control.
5. Pragmatics
Pragmatics is about the question: what actions are evoked by the information made
accessible to people? Studying this question it is necessary to understand that with
MICS, managers and employees are involved in a negotiation and possible learning
process. Lawler and Rhode (1976) suggested that some of MICS' elements have a
severe impact on organization members' intrinsic and extrinsic work motivation.
Their hypotheses are summarized briefly in table 6.4.
23
Based on Lawler and Rhode, 1976, p. 81, table 5-1
24
Source: Lawler and Rhode, 1976, p. 64, table 4-3.
25
Meaning that scores should reflect a person's efforts.
146 Organizational Learning and Information Systems
L. High Variety
8. Source of M Improvement in Job Capabili- K. Rewards that are important
Motivation ties
Source: Lawler and Rhode, 1976
Table 6.4: Values of MICS Elements that Produce Intrinsic and Extrinsic Work Motivation.
The MICS-elements list describes how the cybernetic approach would define a
control system. The organization development approach would however emphasize
how these information streams impact on actual behavior, motivations,
understanding and possible conflicts.
Lawler and Rhode warn for possible dangers of dysfunctional behavior that can result
from inproper application of control information systems. Dysfunctional behavior of
control information systems is defined as:
" ...employees [that behave] in ways that look good in terms of the control system measures
but that are dysfunctional as far as the generally agreed upon goals of the organization
are concerned" (p. 83).
This has been researched in the sociology of organizations since the 1930s (Michels,
1925/66; Merton, 1940; and Gouldner, 1950), and now has received considerable
attention by Senge under the heading of 'systems thinking', meaning a way of
conceiving the indirect and non-obvious impact of decisions in the longer run. A
classic example is described by March and Simon (1958) in figure 6.5.
Lawler and Rhode describe four types of dysfunctional behavior in relation to MICS,
namely:
• Rigid bureaucratic behavior. Figure 6.5 illustrates this type of dysfunctional
Role and Value of MICS for Organizational Learning 147
relation between MICS-usage and performance, i.e. organizational norms, were not
included in the research design (cf. our discussion of the study of Lee and Guinan in
chapter 1).
6. Social
According to Mintzberg one should distinguish between two types of systems for
monitoring and control:
• Performance control systems:
" The purpose of performance control is to regulate the overall results of a given unit.
Objectives, budgets, operating plans, and various other kinds of general standards are
established for the unit, and its performance is later measured in terms of these standards
and the results fed back up the hierarchy by the MIS" (Mintzberg, 1983, p.75).
This means that performance control influences decision-making and action only
indirectly by establishing the targets the decision-maker must achieve. Performance
control information systems can contribute to understanding the company's activities
when the organizational units rated are more or less independent. When units
influence each other's performance a separate measurement system per unit could
bias the results because some of the performance is not the result of the unit's own
operation. The choice then can be to disentangle the units, or to develop a market-
like pricing principle for inter-unit transactions, or to develop one performance
measurement instrument for all the units together!
• Action planning systems:
" Action planning emerges as the means by which the nonroutine decisions and actions of
an entire organization, typically structured on a functional basis, can be designed as an
integrated system. All this is accomplished in advance, on the drawing board so to speak.
Behavior formalization designs the organization as an integrated system too, but only for
its routine activities. Action planning is its counterpart for the nonroutine activities, for
the changes. It specifies who will do what, when, and where, so that the change will take
place as desired" (p. 78).
Whereas performance control systems cannot really cope with the interdependences
of functional units, action planning is typically used to solve this problem.
In relation with Mintzberg's organization types, planning and control information
systems have different roles. This finding is summarized in table 6.5.
Table 6.6 summarizes the items that can be observed in MICS in empirical research.
Leaness: Lean Classic
MICS:
150 Organizational Learning and Information Systems
Technical
Physics • Coupling of systems, via net- • Functional systems. Islands of automation
work and databases
Empirics • On-line systems • Offline systems, with period reports
Syntactics • High quality user interfaces • Hard copy reports
(easy understandable • Change of database on request and when
structure of software) feasible
• Flexibility of databases • Inconsistent data
• High quality administrative • Incompatible data structures
organization
• Compatible data structures
Organization
Semantics • Consistency with possible • Inconsistency with control type
control type • Mental models are diverse and
• Shared mental models incompatible. They reflect stake holders'
positions
Pragmatics • Decisions are implemented in • Many complications in translating decisions
high speed and trust to actions
• Action based on theoretical • Action based on past experience (routine)
understanding of practical or command
problems
Social • Social networks of problem • Separation of problem anticipation and
anticipation and critical critical evaluation networks
evaluation are closely • MICS service problem anticipation, or
connected punish-reward
• MICS serves problem
anticipation and critical
evaluation
Table 6.6: Differences of MICS in Lean and Classic MB's.
The following section discusses in more detail the roles of MICS for organizational
learning activities (single-loop and double-loop), and what this implies for the value
of MICS in learning processes.
In the preceding chapters a distinction was made between the role and values of
MICS. The role of MICS concerns the way MICS aids and changes learning
processes. The value of MICS in organizational learning terms is related to the
contributions of MICS in single-loop, double-loop and deutero learning processes.
26
CASE-tool is short for computer-assisted systems analysis, and commonly used in the business analysis
phase of information systems analysis and design. One widely spread CASE-tool is System Development
Workbench, a trademark of CAP Gemini Pandata.
152 Organizational Learning and Information Systems
files, waste analysis and variance analysis are techniques to augment process and
product quality. Additionally, complaint files are sources for the improvement of
market performance. Market analysis results in insights into markets and their
developments. Finally, the PIMS analysis results in insights into product, process and
market innovations. These methods all require performance and planning data that
can be provided by a MICS. Table 6.7 gives a summarizing list of the roles of MICS
in the double-loop learning process.
Table 6.7 only provides insight into roles for theory development. Unlearning or
knowledge removal is a social process that cannot be supported in a direct sense by
MICS. MICS can only show what would happen if no changes were made in all these
fields.
MICS in Adaptation
Single-loop learning consists of changes of the theory while keeping the basic norms
unaffected. Adaptation of knowledge is about the assessment of existing theories and
management principles, so that its working is improved. It is basically an application
of the empiric cycle, which states that hypotheses should be tested and changed on
the basis of empirical evidence. The adaptation process leads to minor changes in the
theory, but also could initiate a request for a double-loop learning process.
The basic idea of the single-loop learning process is that predefined norms are used
to control existing behavior. Many means for control exist that correspond to this
idea:
• Training. People are taught to behave according to some principles in a specific
154 Organizational Learning and Information Systems
The problem anticipation and critical evaluation control systems also have different
owners, the technostructure and middle line respectively. This leads to two different
social networks for learning. In machine bureaucracies these networks are socially
separated and meet infrequently. The lean organization has procedural norms that
connect them, thus enabling a closed learning loop.
MICS in Storage
Chapter 5 described storage as storing knowledge and enabling its re-use. More
precisely, three organizational knowledge storage processes were described:
acquisition, retention and retrieval (after Walsh and Ungson, 1991). These processes
can be assisted by MICS.
• Acquisition of knowledge and MICS. By acquisition we mean the way knowledge
(data, experience, judgement and science) enters into the knowledge storage
activity. Frames of reference play an important role here because issues that
cannot be given a place within them are kept out of the attention focus, and are
thus regarded as irrelevant for storage. A double-loop trigger would recognize
the inadequacy of the existing frame of reference, thus requiring its redesign by
theory development. The sensory parts of MICS can play an important role in
storing data that are relevant for creating science and judgement. It is important
to note that the data themselves do not contain knowledge, but that a clear
classification scheme related with a frame of reference can lead to easy
interpretations that do result in knowledge.
• Retention of knowledge and MICS. Many media can be used for knowledge
retention: individual memory and files, cultural elements (language, symbols,
sagas etc.), rules and principles of transformation processes, organization
structures, organizational physical layout and external archives. MICS can be a
particularly good instrument for organizational knowledge retention, when the
acquisition facility is well-designed. The problem is often however that much
knowledge is based on heuristics that are difficult to elicit (Kerr, 1991), or that
are not repetitively used and therefore not worth storing. This is not so with
156 Organizational Learning and Information Systems
with shared data (which gives information about a specific moment in time),
synchronization of understanding can occur. MICS can provide data that are more
timely than the traditional manual bureaucracy. Automatic sensors could even lead
to on-line information. The question is whether on-line information is required. This
of course depends on the number of changes in the processes, and the consequences
(risks) of reacting too late. In the process industry, where for instance chemical
processes are controlled by computers, a late reaction can lead to a catastrophe,
whereas in the banking industry late reactions to performance decline can lead to
mismanagement but do not necessarily lead directly to a disaster.
Knowledge use has two big problems: relevance and applicability. Relevance is about
the potential of knowledge for solving problems. Applicability is about the problems
related to using the knowledge, for instance its complexity in relation to the organiz-
ation members' ability to handle it. MICS makes knowledge easily available but is
poor in the distribution of complex knowledge.
Knowledge re-use suffers particularly from two problems: reapplicability and validity.
Reapplicability concerns the opportunities of re-using knowledge for solving
recurring problems. The main problem is that problems do not always recur
frequently enough to make knowledge storage cost-effective (cf. Hofstede, 1981).
Another problem with reapplicability is that situations and problems sometimes
might look the same but are not. Regarding one problem as being the same as a
previous one might lead to a bias in perception that could lead to serious mistakes.
The validity problem concerns the fact that theories and insights are only valid for a
specific period. For instance, economic theory of the 17th century is probably invalid
for understanding business in the post-industrial society. The solution to this validity
problem is a constant validity checking of theories in the adaptation process.
Connected with MICS, theories are mostly the basis for the definition of the norms
and the measurements that are part of a MICS. This means that each time MICS is
used, the underlying management theory is reapplied. MICS should be audited on
the understanding of its basic principles by its users, and the validity of its underlying
theory.
Although MICS' power in storing and retrieving data and experience has not yet
been explored, it is important here to state some of the limitations of MICS as a
computer-based information system. The following quotation of Kim defines these
limitations:
" ... the mental models in individuals' heads are where a vast majority of an organization's
knowledge (both know-how and know-why) lies. Imagine an organization in which all the
physical records disintegrate overnight. Suddenly, there are no reports, no computer files,
no employee records, no reporting manuals, no calenders - all that remain are the people,
buildings, capital equipment, raw materials, and inventory. Now imagine an
158 Organizational Learning and Information Systems
organization where all the people simply quit showing up for work. New people, who are
similar in many ways to the former workers but who have no familiarity with that
particular organization, come to work instead. Which of these two organizations will be
easier to rebuild to its former status?" (Kim, 1993, p. 44).
Kim's answer is the first. This answer provides an important insight into the relative
importance of the social and technical side of the learning system. The question is
however not what is most important, because hopefully the situation as described by
Kim will never happen. What is important is what MICS can contribute to the
broader social system, and what the social system cannot do without it!
Table 6.9 summarizes MICS' roles in single-loop learning.
Deutero learning is about the description of norms that govern learning processes.
These norms are about learning policy/identity, responsibilities, and procedures of
communication and message handling. MICS can play a role in deutero learning,
because it is an implementation of some of the procedural norms. For instance, when
an organization wants to change its informal way of learning to a formal one, MICS
might play an important role because MICS requires clearly defined responsibilities
(output expectations). MICS also makes it possible to measure and control people by
defining measurable performance. This can lead to an organization type that
emphasizes single-loop learning. The possible negative effects of this kind of control
have been discussed earlier, and must be well considered while formulating policy
norms. The impact of MICS on procedural norms can be described in technical and
social terms:
• The technical parts of MICS are interesting from the deutero learning
Role and Value of MICS for Organizational Learning 159
The model of organizational learning used so far, is a descriptive model. On the basis
of this model alone it is not possible to state how good or bad MICS is from the
organizational learning point of view. If we want to make such statements, it is
necessary to define the values of MICS in learning processes. Table 6.10 describes
MICS' values in general.
As deutero learning is not studied further, only double-loop and single-loop learning
values are further defined here.
Double-loop learning is rated by adding up the scores (-1, 0 or +1) on each cell
intersecting learning fields (human resources, processes, markets and products) and
learning activities (development and unlearning). MICS can add value to the eight
thus recognizable learning incidents as follows.
• MICS' additions to human resources. MICS can provide data about human
160 Organizational Learning and Information Systems
MICS' role and value in double-loop learning is mainly as a quick data supplier. Lean
organizations have many and well-organized information sources to make the many
connections between relevant databases. These databases may be spread among
several organizational units, such as marketing, product engineering, manufacturing
engineering and strategic business units. This means that classic organizations will
have many problems putting the data together at the physical (access to databases),
empirical (connections between databases and systems), syntactic (consistent data
definitions) and semantic (creating interpretations from data between separate groups
in the company) levels. The value of MICS therefore differs significantly between
lean and classic machine bureaucracies. The classic machine bureaucracy will regard
MICS as overhead and is not clearly aware of MICS' contribution. Lean
organizations have lean information systems that are directly connected with the
generation of value. Systems or system parts that are regarded as non-contributing are
removed (cf. Van Nievelt, 1992). The problem of increasing information overloads
that happen in complex environments with large databases and information
processing (Ackoff, 1968), is managed in lean organizations by developing an explicit
view of the business problem and information systems that are directly connected
with these problems (Nonaka, 1988). This means that the chance of irrelevant and
uninterpretable data is much smaller than in the case of classic machine
bureaucracies. Additionally, the lean learning norms enable a higher learning speed
because of their many lateral structures and decentralized learning procedures that
relieve communication channels in the organization. The resulting value patterns of
lean and classic machine bureaucracies in dynamic and complex environments are
drawn in figure 6.7.
6.5 Summary
This chapter first presents the key statements in the theory about organizational
learning, MICS and machine bureaucracies that must be operationalized and tested
for explorative purposes. Secondly, the theoretical concepts calling for observation
are extracted from these statements and then it is shown how they are
operationalized, a step which requires further elaboration of the theory.
The theoretical model summarized here consists of statements and conclusions. With
a statement we mean a proposition. Some statements are hypotheses and thus are
subject to empirical confirmation or falsification. To reduce the number of
statements to be investigated, most statements are assembled via a syllogism to a
conclusion. Some of these conclusions are hypotheses for the empirical research. In
this chapter an 'S#' is placed before a statement, and an 'Con#' in front of a
conclusion. For reasons of parsimony (Leege and Francis, 1974, p. 35), the number
164 Organizational Learning and Information Systems
Evidence for both statements was presented in sections 1.3 and 5.1., where it was
shown that most (classical) machine bureaucracies are very slow in adapting to
environmental changes. The following conclusion can be drawn from both:
This study does not enable a test of S1, because only machine bureaucracies are
studied here. We can however make observations of control mechanisms in machine
bureaucracies and see how they affect innovation and inertia, but we can make no
comparison with other types of organizations. We also link this statement with
organizational learning by assuming (on basis of Argyris and Schön 1978 and De
Raadt, 1992) that:
• the degree of control achieved results only from single-loop learning, and
• the degree of innovation achieved results from double-loop learning.
The literature is however very vague about the relation between single-loop and
double-loop learning and how these activities relate to the learning norms in
organizations. Two definitions of a degree of organizational learning are optional
here, with considerable consequences for the theory as a whole. These are:
1. Organizational learning is a single variable, where double-loop learning (DLL) is
a deeper form of learning than single-loop learning (SLL) (e.g. Senge, 1990;
Hammer, 1990). In this hierarchical model of OL, the presence of DLL scores
higher (say 2) than SLL (say 1). It might be argued in justification that, although
SLL is possible without DLL the contrary is impossible. This does not seem
entirely reasonable because DLL is not always more important or more
complicated than SLL. For instance in any heavily capitalized manufacturing
process only the incremental changes brought about by SLL are possible until
investment in a new expensive plant is possible, so most of the time DLL is very
limited while SLL is highly sophisticated. So we prefer to leave this hierarchical
model aside.
2. Organizational learning is composed of a mixture of SLL and DLL which
probably co-vary in ways that depend on the learning norms but can mostly be
treated as independent. We could try to rate the effort devoted to each kind of
organizational learning and treat them as a vector measurement of organizational
Operationalizations and Method of Analysis 165
learning or perhaps attempt to justify some way of combining them (by simple
addition, possibly). It may be appropriate to introduce a limit to the total
learning effort so that more SLL leads to less DLL and vice versa. Learning norms
however also could be such that the total amount of organizational learning
increases.
The vector model of the organizational learning concept is chosen in this study and
implies the following statement:
S3.1: Classic manufacturing MBs put less effort into double-loop learning than lean
manufacturing MBs.
S3.2 Classic and lean manufacturing MBs do not differ regarding the amount of
learning effort put into single-loop learning.
S3.3 Classic service MBs put less effort into double-loop learning than lean service MBs.
S3.4 Classic and lean service MBs do not differ regarding the amount of learning effort
166 Organizational Learning and Information Systems
Statements S3.1 to 3.8 are based on our previous treatment of machine bureaucracies
(chapter 5). The main conclusion drawn about the classic-lean distinction was that
both machine bureaucracy types are organized for cheap production by applying
many control mechanisms (single-loop learning), but that the lean MB was not only
cheap but also much more innovative than the classic case. Additionally, all MBs are
expected to invest substantially in single-loop learning via the construction of control
mechanisms, rules and procedures. Service organizations are expected to invest more
in these SLL-activities than manufacturing organizations, because of their shorter
learning cycles and more direct contact with clients who can provide the organization
with feedback signals. At the same time, however, it is expected that service
organizations have more problems with constructing systems for SLL, because as their
output is intangible it is more difficult to measure their performance. Therefore,
service MBs may be expected to have more double-loop learning activities. According
to these assumptions, the following patterns of values are predicted among the four
types of MBs (see fig. 7.2).
Learning norms are supposed to influence the single-loop and double-loop learning
processes. Learning norms themselves are however influenced by environmental
Operationalizations and Method of Analysis 167
S6: Single-loop learning efforts counteract low environmental complexity and dynamics.
This statement is based on theoretical and empirical research about error correction,
which has been treated in section 5.4.2. A most important feature is that errors are
detected via a performance measurement tool (Juran, 1964) and that the existing
management theory can explain and handle the existing variety in simple and stable
situations. The single-loop learning process is, however, restricted to solving problems
that can be managed within the management theory and its related tools. Probems of
huge complexity however often require the development of new management
theories by specific research of a (interdisciplinary) project team, the R&D group or
some outside consultants, that are typical of double-loop learning.
S7: Double-loop learning efforts counteract high environmental complexity and dynamics.
Evidence for this statement was presented in sections 5.4.1 and 5.4.2. This statement
seems to be intuitively correct, and has been propagated widely by organizational
development authors. A highly dynamic environment requires frequent major
changes in the business; high complexity calls for high levels of knowledge, and the
combination of high dynamics and high complexity multiplies the efforts required.
An excellent example is the situation in which a television set producer is confonted
by situations of increasing competition (dynamics), that demand reduced costs,
improved quality and a greater variety of client's choices at the same time (increasing
complexity). In that case the production process does not require simple
improvements in existing procedures and technology, but also reconsiderations of
what should be produced and what kinds of technologies are needed. If, however,
these innovations are carried out in low dynamic and simple environments, the
168 Organizational Learning and Information Systems
resulting learning process may lead to many inefficiencies by creating too many
changes.
S8: Double-loop learning involves reorganization and so entails higher risks than single-loop
learning.
Hannan and Freeman (1976 and 1984) suggested that double-loop learning also
carries 'Reorganization Risks', which lowers the survival chances of an organization.
This is obvious because it creates instability and the chance of dysfunctioning, which
can lead to a loss of assets, loss of clients and markets, quality problems etc. This
problem can be solved or reduced by developing procedures and norms that
accompany organizational double-loop learning and that can guide the choice for
single-loop or double-loop learning activities (Garratt, 1987). Many organization
analysts and consultants therefore propose methodologies of change for reducing the
risks of reorganization (c.f. Bushe and Shani, 1989 for an excellent example relating
to machine bureaucracies). Also, project management methods have been developed
in many cases for this purpose (c.f. Rogers, 1964, Franke, 1987). Evidence for S8 has
also been presented in sections 4.9.1, 4.9.2 and 5.4.3.
Complexity increases when new methods, rules, tools, and knowledge must be
applied in addition to the existing ones. Dynamics increase as a result of
reorganization until new practices have settled in. Change reduces certainty about
how to work and collaborate. Evidence for these statements has been treated in
sections 5.4.1 and 5.4.3.
An organizational learning paradox can be formulated as follows: learning is required
for survival, but learning may reduce survival chances as well. This paradox must be
solved by the development of learning norms that match learning needs and are
effectively implemented in learning activities to improve learning abilities.
S11: The more environmental complexity and dynamics, the lower the survival chance of an
organization which is unable to learn and adapt.
S12: The lower the survival chance of an organization, the greater its need for organizational
Operationalizations and Method of Analysis 169
learning.
Evidence for both statements was presented in sections 1.1, 1.2 and 1.3. Because of
S11 and S12, organizations must be careful not to embark too quickly on double-
loop learning strategies, because these could reinforce the trend to lower survival
chances. At the same time the choice for SLL or DLL or deutero learning must be
taken rationally, as far as the environment can be analyzed rationally. This requires
the creation and explanation of management theories. If not done so, only political
forces determine what will happen in the organization.
This conclusion results from S11 and S12, and has also been discussed in sections
4.8.3 and 5.4.1. The following two statements can be derived from this conclusion:
S12.1: Lowest learning needs exist in cases where low complexity and low dynamics exist
simultaneously.
S12.2: Highest learning needs exist when high complexity and high dynamics exist
simultaneously.
In these statements the situations of high complexity with low dynamics and low
complexity with high dynamics are not explained. Therefore we need a statement
about the relation between dynamics, complexity and learning needs.
Con 3: Lowest learning needs exist in cases of low complexity and low dynamics.
Moderately low learning needs exist in cases of low dynamics and high complexity.
Moderately high learning needs exist in cases of high dynamics and low complexity.
High learning needs exist in cases of high dynamics and high complexity.
170 Organizational Learning and Information Systems
Con 4: Learning needs determine the learning norms required for survival.
S14: Lean learning norms emphasize the critical evaluation and problem anticipation roles of
MICS, whereas classic learning norms emphasize the problem anticipation and
accounting roles of MICS.
S15: MICS contributes considerably to SLL-effort.
S16: MICS inhibits DLL-effort.
Evidence for the validity of S15 was given by Argyris (1980) and De Raadt (1992),
who both state that MICS is a useful tool for detecting errors and for error
correction. S16 is based on the research and findings of Argyris (1980), Hannan and
Freeman's inertia theory, Markus (1983), and classic theory about the impact of
control systems on motivation in organizations (Merton et al. as described in Lawler
and Rhode, 1976), described in sections 6.1, 6.3.2 and 6.4.5. The argument
essentially is that MICS includes a management theory that emphasizes uni-lateral
control in organizations, meaning the increase of power of certain people at the
expense of other people's power. This reduces the chance that new ideas will be
supported when they come bottom-up in the organization.
In order to score these contributions, we follow chapter 4, which proposes a score for
single-loop and double-loop learning effort in section 4.8.3, and section 6.4.5, which
states that the value of MICS should be assessed by looking at the intersections of the
learning field and learning activities dimension of organizational learning. Thus the
Operationalizations and Method of Analysis 171
Because this study is about the impact of MICS on organizational learning and, more
specifically, MICS' role and value for effective organizational learning, we will not
further elaborate the direct causal line between learning norms, complexity and
dynamics. Therefore, conclusion 5 is excluded as part of this investigation.
Two additional conclusions (con 6 and 7) can now be stated as well.
Con 6: MICS contributes to single-loop learning effort and inhibits double-loop learning
effort.
.
This conclusion is a conjunction of statements S15 en S16.
Operationalizations and Method of Analysis 173
This last conclusion is based on S6, S7, S10, S15 and S16.
As a summary to this section, first all major statements and conclusions are listed,
and figure 7.4 shows their connections with the conclusions. The figure also shows
which of these hypotheses are the subject of empirical research.
The major statements are:
S1: Machine Bureaucracies have strong controls.
S2: Controls inhibit innovation.
S3: Organizations have learning norms that determine the effort an organization puts into single-loop and double-loop
learning.
S4: MBs face increased environmental dynamics.
S5: MBs face increased environmental complexity.
S6: Single-loop learning efforts counteract low environmental complexity and dynamics.
S7: Double-loop learning efforts counteract high environmental complexity and dynamics.
S8: Double-loop learning involves reorganization and so entails higher risks than single-loop learning.
S9: Organizations create learning policies to reduce the risks of reorganization.
S10: Reorganization risks increase organizational complexity and dynamics.
S11: The more environmental complexity and dynamics, the lower the survival chance of an organization which is unable
to learn and adapt.
S12: The lower the survival chance of an organization, the greater its need for organizational learning.
S13: Dynamics contributes more to learning needs than complexity.
S14: Lean learning norms emphasize the critical evaluation and problem anticipation roles of MICS, whereas classic
learning norms emphasize the problem anticipation and accounting roles of MICS.
S15: MICS contributes considerably to SLL-effort.
S16: MICS inhibits DLL-effort.
The Conclusions are:
Con 1: Machine Bureaucracies are poor in innovation.
Con 2: A combination of complexity and dynamics determines the amount of learning need.
Con 3: Lowest learning needs exist in cases of low complexity and low dynamics.
Moderately low learning needs exist in cases of low dynamics and high complexity.
Moderately high learning needs exist in cases of high dynamics and low complexity.
High learning needs exist in cases of high dynamics and high complexity.
Con 4: Learning needs determine the learning norms required for survival.
Con 5: Learning norms decrease dynamics and complexity.
Con 6: MICS contributes to single-loop learning effort and inhibits double-loop learning effort.
Con 7: Depending on the Learning Norms, MICS contributes to or decreases complexity and dynamics.
Figure 7.4 summarizes the links between the Statements and the Conclusions.
174 Organizational Learning and Information Systems
Tabel 7.3 now lists the concepts of the statements and conclusions that are the
subject of empirical investigation. These concepts are operationalized in the following
sections.
On the basis of these reflections, the previously described model can now be
reformulated and simplified to the following research model (see figure 7.5).
Operationalizations and Method of Analysis 175
When going from theory to observations the following problems appear according to
the research methodology literature (Lee, 1989; Yin, 1984):
1. The theory must be described clearly and unambiguously in operational terms.
2. Measurement instruments must be correct operationalizations of the constructs
to be observed.
3. Often multiple items form one index to find a score for a case on a theoretical
construct. This requires a sound theory that combines the observations in one
score.
4. Observations among cases must be comparable.
The operationalization of the theory is partially realized. This is because of the
elicitation of the statements and conclusions. We must now add a correct
operationalization of basic theoretical constructs. The statements and conclusions
contain variables that must be observed, and by which empirical testing and further
exploration of theoretical notions are made possible. Some parts of the theoretical
model can be tested whereas others can only be described (e.g. Conclusion 4). In this
section some proposals are given for scaling variables that require observations of
multiple items. The validity of these scales will only be tested via reasoning about
observations made. Statistical testing of the reliability and validity is not possible,
because we lack the data to do so (for further insights into these methodological
problems see Kerlinger, 1986). Also, a test on the uni-dimensionality of the scales is
not possible here because the factor analysis technique that is required for this
purpose requires a large amount of statistical data. The reader must be well aware of
176 Organizational Learning and Information Systems
these limitations. The most important issue for empirical research, its meaning in
practical settings, is however closely guided. This is done via the application of case
studies and a research design that takes the two basic explaining variables as main
factors (lean-classic and service-manufacturing distinction).
Comparability of the cases is in principle difficult, because each case is in some sense
unique (even a statistical research approach cannot avoid this fact). Comparability is
achieved by constructing precise and standarized instruments by which we want to
observe the cases. Hence, much effort has been put into constructing the scores that
are described in the following subsections.
Conclusion 2 stated: "A combination of complexity and dynamics determines the amount of
learning need". What is required therefore are scores of environmental dynamics and
complexity with which a learning needs score can be assigned. The measures for
complexity and dynamics are based on a classic study of Duncan (1972) and Duncan
and Weiss (1978). Complexity was defined as the number of factors of relevance for
decision-making, and the number of components to which these factors refer. The
factors and components are issues that define an organization's internal and external
environment (cf. chapter 5, table 5.7). Dynamics was defined as the degree to which
relevant factors for decision-making remain the same over time or are changing, and
the frequency with which new factors are relevant (cf. section 5.4.1).
Duncan and Weiss' measures were developed for analyzing decision-making, and
were not applied to organizational learning. This study therefore applies the list of
factors and components to understand complexity and dynamics in the case studies
we will conduct. The observations will however not be made at the individual's
decision-making level, because this will make the data-acquisition and analysis too
laborious for our purposes. In the case studies we will ask interviewees for data
sources about the factors and components of the environment, and will try to
qualitatively assess the dynamics and complexity on the basis of these data. It will be
difficult to obtain data on some components and factors, which also means that not
all factors and components will be treated in detail. Only the most significant ones,
from our theoretical perspective, will be considered.
Conclusion 3 provides a starting point for constructing an index of learning needs.
Con 3 states: "Lowest learning needs exist in cases of low complexity and low dynamics.
Moderately low learning needs exist in cases of low dynamics and high complexity. Moderately
high learning needs exist in cases of high dynamics and low complexity. High learning needs
Operationalizations and Method of Analysis 177
exist in cases of high dynamics and high complexity". Thus four scores of the organizational
learning needs index are defined. See table 7.4.
Because this study does not make learning need scores independent from complexity
and dynamics, Con 2 and Con3 are assumptions for measurement.
Variable 2: MB-types.
Four types of Machine Bureaucracies were distinguished from the beginning of the
book, by two main dimensions: classic-lean and manufacturing-service. The classic-
lean distinction concerns the organizational norms (policy, structure, culture,
motivations) that govern the organization. The service-manufacturing distinction is
about transformation. Variable 2 thus has four values described in table 7.5.
Variable 2: Norms
MB-types
Classic Lean
Transfor- Manufacturing Classic-Manufacturing Lean-Manufacturing
mation
Service Classic-Service Lean-Service
Table 7.5: MB-types
Items for the process factor are found in the previously discussed paper of Mills and
Moberg (1982). The distinction we draw here is of course a simplification of reality as
for instance Schmenner has shown (1986). In fact many types of services and
manufacturing organizations will score on both values at the same time. Schmenner,
and also other writers on services like Grönroos (1991), however do not dispute the
relevance of the eight items listed in table 7.7.
Learning norms are operationalized in four dimensions: policy and identity norms,
responsibility norms, action norms and procedural norms (cf. chapter 4 section 9).
The scores on the four related variables (3.1, 3.2, 3.3 and 3.4) can have two extremes
that are related to the lean and classic nature of organizations. MICS is treated
separately from procedural norms, because it is the independent variable in this
study.
values apply.
Table 7.9: Scoring the Organization's Leanness Using Responsibility Norms.
Chapter 5 also mentions three other organization structures that were considered as
part of this index, namely: Network with independent companies, Parallel learning
structures, and Project groups. We have however no arguments why these three
structures would behave differently in the lean or the classic cases. Therefore they
would make no contribution to the index.
The main items describing action norms are given in table 7.10. This table gives an
index of action norms, with extremes related to lean and classic organizations. The
lean extreme emphasizes the team, the client, self-realization and fast reactions as the
main motivators for organizational learning. The classic extreme emphasizes money,
pain avoidance, and slow reactions as motivators for learning.
Table 7.11 describes the procedural norms index. The lean extreme emphasizes free
and continuous flow of data and information. The classic extreme emphasizes
discrete and constrained flows of data and information.
Variable 4: MICS-description
The double-loop learning process concerns the creation of goals for the learning
fields and the removal of goal definitions that are out of date. This process of double-
loop learning is therefore often called 'innovation'. Double-loop learning can be rated
via the application of the following questions:
• Human resources. How much money is spent on training? Is this training only
for developing skills to accomplish routine tasks, or is it also for learning to
develop new insights? Are newly acquired insights implemented in new
practices? Are people encouraged to think and create innovations?
• Tranformations. Much innovation is about changing the way products and
services are made, so that the organization is able to adjust more easily to
specific demands. For instance shorter delivery periods, more flexibility in
colors and product features. Much innovation is also about redesigning
processes to meet competition by decreasing costs of transactions and
coordination. This is not just marginal change but a revolution in thinking and
working to accomplish the same job.
• Markets. How many markets and market segments are served? It is important to
mention that markets for a company do not merely exist, but must be created
by improving communications to potential clients, developing a strategy and
plan to create a new profitable market, and adjusting products and services to
the specific needs and demands of these new markets or market segments.
• Products. How many new products and product series have been launched
within a period of time? Especially the series is an important observational unit,
because many organizations only produce a small number of products. For
184 Organizational Learning and Information Systems
instance car manufactures produce cars, and maybe also lorries, fork-lift trucks
and motor bikes.
A score for double-loop learning effort is obtained by finding evidence of learning
within the four fields of learning. A score of 0 is assigned when no learning and no
unlearning happens in the four fields. A score of 8 is attached when the organization
learns and unlearns on all four fields. The scoring principle is thus exactly the same
as the score that was defined for MICS' DLL-value in table 7.2, with the exception
that DLL-effort cannot be negative.
Section 6.4.3 detected two roles of MICS, namely: Problem Anticipation and Critical
Evaluation. MICS' role in one case can vary depending on the learning field and
learning activity it supports. Thus MICS can have the critical evaluation role at the
human resources field, and the problems anticipation role in the process
development field. The researcher can easily detect the role, when asking about the
purpose of MICS-usage.
Measuring MICS' value is important for investigating the most basic conclusions of
this study: "MICS contributes to single-loop learning effort and inhibits double-loop learning
effort" (Con 6). MICS' value is assessed by assigning a value (-1 for negative influence,
inhibit, 0 for no influence and +1 for positive influence of MICS) to the cells that
intersect learning fields and learning activities in table 7.2. This is done seperately for
single-loop and double-loop learning (var 6.1 and 6.2 respectively), because they were
regarded as specifically different. The minimum score for single-loop learning is then
-16 and the maximum score is +16. The minimum score for double-loop learning is -8
and the maximum is +8. The researcher can apply table 7.13, which gives indicators
for DLL- and SLL-values. This table is not intented to be complete. The learning
activities for DLL-values are omitted, because these activities are binary (development
versus removal). The researcher can easily ask himslef the question if the activity was
development or removal (or both). The learning fields for SLL-values are omitted in
the table, because they are easy to detect by asking about the application field.
Examples of Single-Loop Value Indicators
These scores are of course tentative and experiences with them in the separate case
studies must be documented. The final chapter must conclude about possible
adjustments and consequences for the information audit instrument.
Con 6 also requires the rating of Single-Loop Learning and Double-Loop Learning
effort.
in order to draw conclusions about the hypotheses (five conclusions and one
statement) listed in table 7.15. The comparative analysis of data from the cases must
result in conclusions about the validity of the Conclusions, and about possibilities to
further develop our understanding of organizational learning and information
systems (for instance by constructing a normative theory).
The following steps are undertaken in the separate case studies.
1. General description of the case in terms of contextual variables such as size, age,
organization chart, and learning fields.
2. Description of the organization in terms of learning needs, by scoring its internal and
external dynamics and complexity.
3. Description of the organization in terms of its lean-classic and service-manufacturing
nature. This for the classification of the case according to the main independent
variable machine bureaucracy type. Note that we do not assign a value in terms
of better or worse to the MB-type found.
4. Description of the organization in terms of learning norms. The description of
organizational learning starts with learning norms because this is similar to the
description made earlier of organization configurations, and sets the norms for
the two learning processes to be described additionally. At the same moment
the match between the machine bureaucracy type mentioned earlier and the
values on the learning dimensions is tested. Deviations between the theory and
the observations are of particular interest.
5. Description of MICS
6. Description of learning (single-loop and double-loop learning processes and
learning fields). Description of each learning step (development, storage, (re-
)use, dissemination, adaptation, removal/double loop trigger) and the role and
value of MICS therein.
7. Explanation of possible problems in learning steps from the role and value of MICS and
recommendations.
8. Conclusion regarding the validity of the main hypotheses.
Score sheets are used to summarize the findings. See tables 7.14 and 7.15.
Var 2: M.B.-type: 1: Classic- 2: Classic- 3: Lean- 4: Lean-Service
Org. Learning Machine Service Manufacturing
variables:
Var 1: Learning need 1 or 2 2 or 3 2 or 3 3 or 4
Var 3.1: learning Work harder Work harder Work smarter Work smarter
polcy & identity
Var 3.2: Res-ponsibil- Power-based and Power-based and Competence- Competence-
ity norms functional functional based based
Var 3.3: Action Money and slow Money and slow Team and fast Team and fast
norms
Var 3.4: Procedural Discrete and Discrete and Continuous and Continuous and
Operationalizations and Method of Analysis 187
Score sheet 7.14 contains a summary of expected values. Each case is related with
these values, so that a test of the predictive quality of the theory per case is made.
Because the purpose is to make a comparison between MB-types, the cases are
presented in order of their expected leanness. Thus first a classic manufacturer is
described, second a classic service company, third a manufacturer that is on the move
towards leanness, but, as was noted later, not lean yet, fourth a service company that
is at about the same stage of leanness, and finally a fully lean producer that also
places a high emphasis on service. The results of all the cases are compared,
188 Organizational Learning and Information Systems
specifically to find any patterns among MB-types and the organizational learning
variables. In the most optimistic sense, the discovered patterns will be interpreted
normatively, so that something can be said about the values the learning variables
should have, given a certain context. Score sheet 7.15 brings together the results
about the validity, or invalidity of certain Conclusions and Statements. This is input
for a further elaboration of the theory. The results from sheets 7.14 and 7.15 are
used to further elaborate on a theory on the role and value of information systems in
organizational learning.
Case Studies 189
This case is about Cardboard Co., which previously consisted of three independent small
manufacturers. At the end of the 1980s they merged under the heading of a packaging and paper
division of a multinational operating in the office supplies business. Since 1989 these three
companies still have separate production locations, but they share one management team that takes
responsibility for commercial and strategic policies and planning. The management team is located
at a separate town from the three production locations. The production locations are in Western
Europe and in towns, approximately 40 miles from each other. As suggested by the management,
two of these locations have significantly different organization cultures. This would explain
differences in success of an information system dedicated to the management of adhesive paper,
which is a relatively precious production component for finishing cardboard products. This fitted
well in our research objective of detecting learning norms and the interaction of these norms with
the learning process and MICS' use and value.
Cardboard Co. is established as part of the paper and cardboard division of the multinational
mentioned above and officially exists since 1989. One location of this company was already
established in the 1920s. The merger of the companies was intended to create more benefits from
large-scale economies. The three companies were acquired to reduce the company's dependence on
suppliers, to reduce oscillation of prices in the industry and to optimize mutual deliveries.
The organization has three production plants, of which two are studied in detail. These are
numbered 1 and 2 in table 8.1. The main differences between the plants are the product volumes,
products and processes, organization culture, and effectiveness of their MICS. MICS consists of an
adhesive paper management system (APMS) and a logistics management system (LMS). This case
study mainly concerns the APMS part of MICS, which provides information to the management
and employees about adhesive paper management in the (recent) past.
27
I am grateful to Mr St. Kordelaar, who contributed substantially in the data collection for this case.
190 Organizational Learning and Information Systems
A production location has an organization chart that looks like fig. 8.1.
The organization produces about 250.000 tons of solid cardboard a year. This
cardboard is a semi-manufactured product for the packaging industry. The Plant
mainly makes use of recycled paper, and applies several types of adhesive paper on
the cardboard. By changing the type of adhesive paper the cardboard acquires
different characteristics that are important for the particular packaging purpose. For
instance, the demands of the fruit industry with respect to waterproofing differ from
those of the toy industry. Adhesive paper is the main factor influencing the potential
variations Cardboard Co. can manufacture. The product variation is currently about
350 types. Adhesive paper is also the most costly raw material applied. For an average
type of cardboard, the adhesive paper makes up about 30% of the manufacturing
costs. The additional costs are: paper (10%), glue (1%) and fixed costs (personnel,
machinery, buildings etc.) (60%). Included in these cost estimations are maintenance
costs, technical support and overhead. Cardboard for tomato packaging, however,
requires a better quality of adhesive paper, leading to 64% of production costs.
Managing adhesive paper carefully is therefore a critical success factor.
Cardboard Co. delivers about 80 % of its products to other plants of the Cardboard
and Paper Division. Additionally it has about 250 external clients. Production is
Case Studies 191
mainly based on order processing principles, although plans have been made to
produce more on stock in the near future. As cardboard is a typical commodity, it is
difficult to follow a differentiation or focus strategy in this industry. The strategy is
therefore 'cost leadership' with some constraints on quality demands, depending on
the specific processing demands of the clients. The clients and their quality demands
are well known to Cardboard Co.
Obviously the efficiency of the production process is a major learning field. The
cardboard market is typically led by cost leaders. Human resource issues are
becoming increasingly important learning fields and training has been emphasized as
important for the company's success in several interviews. There is not much
innovation in products and most of the customers are known.
and quality demands are higher than in location 1. This leads to a greater complexity in scheduling,
logistics and materials handling. The result is a much higher spoilage of adhesive paper. Besides
the quality of monitoring adhesive paper usage, cultural differences also explain part of the higher
spoilage at location 2. When MICS was introduced, spoilage of adhesive paper at location 1 was
16% and at location 2 this was 30%!
Conclusion: the complexity and dynamics of CBM are very low. This means that the learning
needs score is the lowest possible (score 1), which confirms insights into classic manufacturing
machine bureaucracies.
The lateral structures are not strong because of the geographical distance and local
cultural differences. Additionally, the production processes are not easily comparable
because of some important differences in the processes. Despite these differences, the
large difference in spoilage of adhesive paper (16% at location 1 and 30 % at location
2) became apparent by MICS-data and led to an active search at location 2 for causes
and opportunities for improvement.
Because poor deliveries increased production costs, one location started to
systematically evaluate deliveries using MICS. Systematic problems were
Case Studies 193
Service-Manufacturing
Cardboard Co. is obviously not a service company. Its output is tangible and discrete,
thus easily measured in tons of each type of cardboard. Profits and (gross) costs are
detected easily as well. Only a few objective reference points are needed to determine
the performance of the Plant. This means that there are relatively many unambiguous
reference points.
The organization's output goals are defined in yearly plans, in terms of tons and
profits. Although this company is a clear case of a manufacturing machine
bureaucracy, its position in a consortium with no internal market feedback
mechanisms complicates learning from its output.
Cardboard Co.'s clients are only involved at the contracting stage and for the
definition of requirements. This is done at the general and divisional management
level. Work is a planned process, disturbed by only a few urgent orders. The final
success of the cardboard is partly dependent on what the clients (of the packaging
194 Organizational Learning and Information Systems
Plants) do with it. Phases in production can be clearly demarcated. Stock production
is avoided because stand-still costs less than stocking this relatively cheap but
voluminous product, and because stocks are risky as clients require specific types of
cardboard. The operation and administration systems are closed to the clients. The
informal communication in the Plant seems to be effective. Professionalism, also at
the managerial level at locations, is rather low. The merger introduced a trend for
more advanced management.
Conclusion: Cardboard Co. is a typical example of a Classic-Manufacturing machine
bureaucracy.
Learning policy
Cardboard Co.'s management aims at reducing costs, increasing volumes produced and sold, and
increasing returns-on-investments. Learning is not specifically an objective in this company. One
director for instance stated that Cardboard Co. is a company of do-ers rather than of theorists.
Cardboard Co. has no policy to explictly encourage communication and has no infrastructure for
organizational learning. Nevertheless, it places much emphasis on improving skills via training.
The training manager communicates with the quality manager about systematic errors, and
sometimes develops a course to improve the situation. Most production assistants also learn the job
of paper and cardboard making by a combination of working on the shop floor and studying at
school. Cardboard Co., because of its small size, has many possibilities for informal lateral
contacts. In this, Cardboard Co. clearly differs from the 'work harder' extreme. Nevertheless, self-
management of production groups is very limited. The planning department, logistics and
production management clearly decide about what has to be done on the shop floor. Business re-
engineering is a non-existing word at Cardboard Co. The management is however very keen on
opportunities to reduce costs. Motivation for business re-engineering therefore exist, at least in
principle.
Conclusion: this organization in most senses has 'work harder' learning policy norms. It has a
limited informality because of its small size.
Responsibility norms
Two learning processes can be detected in Cardboard Co. which I call project learning and
adhesive paper management learning.
Project learning concerns engineering, developing and discussing investment projects to improve
cardboard production. The initiative comes mostly from the project manager, who uses people
from Technical Services and Mechanics to find out the best solutions. No market orientation is
carried out, and responsibilities are clearly associated with the project manager and one or a few
additional support departments. The frequency of projects is quite limited (no process innovations
were mentioned during our visits).
Adhesive paper management learning can occur when shifts receive data about their performance
and receive possible instructions to improve. Particularly important here are:
• The Administration who makes statistical analyses via MICS and feeds these insights back
to the shifts in concrete suggestions. The difference between locations 1 and 2 in adhesive
paper management is frequently explained by location 2's lessor experience with MICS and
the fact that the administration there lacks understanding of the shop floor because none of
the administrators have ever worked there.
• The logistics manager, who analyses delivery data from APMS, and feeds these insights
back to the suppliers.
Case Studies 195
The second learning process has been especially successful in the past three years. Location 1 has
reduced its spoilage from 16 to 8%. Location 2 has just started to improve, now that reliable data
are available. They reduced spoilage in the last two years from 25 to 17%. One percent decrease of
adhesive spoilage leads to a cost reduction of about $175,000 a year!
Because the organization is in a stable and simple environment no type of active search via
networks exists. The volvos (shop floor groups, shifts) have some responsibilities to lower the
adhesive paper costs. They are supported in this by the management and the technostructure
(logistics manager and administration in particular and human resources for specific training).
Project groups do exist sometimes, but are restricted to solving technical problems. Task forces
with clear strategic intent are absent.
Conclusion: this case verifies the opinion that learning in classic machine bureaucracies is
basically power-based and functionally organized. The learning process is organized around the
technical production process. No divisional and matrix structures exist.
Action norms
Although we did not have the opportunity to interview workers, motivation is probably extrinsic
because the routine and often dirty nature of the work does not give much instrinsic satisfaction.
Extrinsic work motivations reduce the chance for creative learning and kaizen to null. People on
the shop floor have a defensive attitude to innovations because they do not have the training or
ability to get another job. Knowledge removal, when linked to the loss of jobs or requiring
additional training to keep a job are not favored. Our informants (project manager MICS, trainer,
location director) could not give exact data about training budgets, because these data were not
seperately registered in the organization. The locations do have their own full-time training
consultant. Some younger people worked in the organization as well, engaged in external training
to become professional 'paper makers'.
Conclusion: Cardboard Co. has 'money and slow' action norms. Its small size however enables
quick implementation of new operational insights.
Procedural norms
Feedback frequencies for a location as a whole (from divisional headquarters) are mostly once a
year, when the yearly plans are reconsidered and budgets are allocated to the locations. This is
slow, even for a classic manufacturer.
Data flows are discrete. For APMS, data are keyed in at the shop floor. Next day the
Administration checks the data and corrects errors. Additionally, Administration analyses the data.
Finally the shift can have its feedback and the logistics manager can take action if required. A big
problem in this long procedure is that the shifts are not present when the data become available.
Sometimes they stay away for the weekend, and sometimes even for a week. This makes
communication about what happened and what the sources of the spoilage are quite difficult.
Data access is not free. There are precise authority limits and defined information needs. It is not
clear how this hurts the company, but it certainly does not contribute to business awareness.
The number of issues measured is very limited. The main system, APMS, only measures adhesive
paper spoilage. Additionally a logistics information system includes data about materials, people
and machine hours. At Cardboard Co.-Headquarters, commercial data are also available. These
data sources are not connected in an integrated system that could give people at all levels access to
information. No-one seems to think that the dissemination of data could be valuable and people are
not interested in data other than those of direct relevance to their task.
The management style is a uni-directional communication from Cardboard Co.-Headquarters to the
location directors. The operators and paper makers only receive instructions to produce cardboard
in certain volumes and with certain characteristics.
Feedback is slow, because the information supply is slow. The environment does not require this to
be speeded up, although technically it would not be difficult to arrange. The impact of this on a
further reduction of adhesive paper spoilage would probably be less at location 1. This learning
196 Organizational Learning and Information Systems
cycle has already reduced spoilage to 8%; further reduction depends on production planning and
the commercial process (order acquisition), because the major spoilage source is 'loss caused by
width' (LCW). LCW is caused by the fact that each location has machines of different width. In the
commercial process, orders are made concerning the delivery of cardboard with a certain volume,
features and width. For instance, a client asks for a width of e.g. 2.00m. This order may be planned
on a machine of width 2.34m. The loss is .34m of production. The solution for this problem could
be:
• Install more machines of different or varying widths.
• Increase the price to compensate for losses, or refuse the order.
Because Cardboard Co. has a very stable environment the first option is realistic. The second
option is commercially unattractive and not considered by Cardboard Co.
Expertise is concentrated in departments. There are no interdisciplinary groups. Only occasionally
an 'interdisciplinary' project group is created, mostly consisting of the project manager and some
people from Technical Services.
Conclusion: the procedural norms are 'discrete and constrained', though more discrete than
constrained, because shifts have access to data about other shifts' performance.
MICS is a functional system. Data from the APMS branch can be transferred to its LMS-branch
(logistics management system), so that the quality of the data can be compared with registrations
from logistics and purchasing. It is however not yet completely integrated in the Logistics
Management System.
At the moment MICS is off-line. When it becomes part of a Manufacturing Resource Planning
system, which completely integrates APMS and LMS, it could become on-line as well, and the
learning process could speed up considerably. This is especially the case at location 2, where the
administrative process is much more complex because of the shorter product orders, larger variety,
and availability of two machines (instead of one at location 1). The use of an on-line system for
performance evaluation at the end of each day, could enable much learning in both locations.
Until now, however, data are often inconsistent between APMS and LMS, and interpretation is
done via hard copy reports. Changes in data supply must go through a bureaucratic request for
proposal, thus constraining flexibility in use.
There is some inconsistency with the control type that is possible in this situation. For instance
shifts could be set spoilage targets, and MICS could give data about performance. It is important
that a correct interpretation is made of the data, based on a concrete management theory. This
theory is not described explicitly, but easy to describe in a causal diagram (see fig. 8.2).
Case Studies 197
The theory visualized in diagram 8.2 is made concrete on the subject of adhesive paper
The problem anticipation network is centralized at Headquarters. Some further detailed scheduling
of activities is carried out at the locations. The performance control network consists of
administrators. At location 1 there was also a close connection between the local logistics manager
and the administrators. This was because the head of Administration at location 1 had a strong
affinity with production, and the administration and logistics people also work in the same room.
At location 2 this was both not the case, which hindered the understanding of APMS data!
MICS is thus used for logistics problem anticipation (LMS) or critical evaluation of performance
(APMS). These are however not yet integrated.
Conclusion: procedural norms are 'discrete and constrained', which is typical of classic machine
bureaucracies. Some deviations are important to note:
• Location 1 is clearly less bureaucratic than location 2 at the social level. This might explain
its better performance in adhesive paper management, because more complexity is involved
there.
• Absence of conflicts in management theories.
• Absence of problems of inconsistent data structures. The small size management
information system makes the problem of data-structure inconsistency easily manageable.
• High speed of decision-making.
Single-loop learning
Storage
Storage of knowledge is realized via:
• Training and handbooks that are used for that purpose.
• The management of delivery is carried out by Administration and the creation of delivery
performance files. The logistics manager has a precise historic overview per supplier which
is used for improving deliveries.
Case Studies 199
Dissemination
Dissemination of knowledge occurs by supplying data from MICS to the locations 1 and 2. The
locations therefore have bench-marks, that act as de facto standards against which to assess
improvements. MICS' role in this learning activity is obvious, and also has a positive value on it.
Dissemination is thus restricted to the transformation field and the field of human resources
(providing shifts with performance data).
(Re)-use
Planning algorithms have been developed to distribute orders to locations. This knowledge is part
of LMS and adapted and (re-)used by Cardboard Co.'s Headquarter logistics manager. APMS
generates historic overviews that are reused for performance control and motivation of personnel.
Here APMS' role is performance control. APMS' value is not restricted to process (efficiency)
improvement, but also for developing the human resource. APMS motivates people and also
clarifies some problems, which could be tackled by training.
Conclusion: MICS' role in Cardboard Co. is in the critical evaluation and problem anticipation
restricted to the learning fields of transformation and human resources. Cardboard Co. seems to be
learning in all the four areas of SLL. It scores on all cells intersecting SLL-effort activities and the
fields of transformation and human resources, leading to an SLL-effort score of 8. During our visit
no systematic SLL learning procedures about the other learning fields were detected. The reason
for this are the so-called 'authority limits', responsibility norms that restrict the opportunity for
learning in other fields. No negative impacts of APMS were found in the organization. In fact,
MICS contributes to all cells of SLL-learning in which Cardboard Co. is active. MICS' SLL-value
is thus +8. The only reason why it did not work in location 2 was because the administrative
procedures and evaluation experiences were absent.
Double-loop learning
Double-loop learning is completely absent at the locations but is a specific task for Headquarters.
The tomato project is one example in which some product development happened to capture a
larger share of the fruit packaging industry. This project was a joint project of Cardboard Co.
together with some other parts of the division. Cardboard is a mature product making market
development difficult. New markets are created by taking over competitors or by fundamental
process innovations that could strongly decrease production costs. The main production factors are:
adhesive paper, used paper, personnel and (inflexible) machines. Personnel costs could be lowered
by further production automation. The costs of these innovations are however also substantial.
Automation of production could however be combined with flexibility improvement (introducing
Flexible Manufacturing Systems). Cardboard Co. is not seriously considering this.
Conclusion: Cardboard Co. has a DLL learning effort score of 0. Information systems are therefore
not used for double-loop learning. MICS reinforces the single-loop attention focus. This could
predict negative impact of MICS on DLL-value. The cause for this restriction to DLL learning is
however not MICS itself, but the existing learning norms (specifically the learning
responsibilities). Hence, the impact of these systems on the DLL-value is scored zero and not
negative.
• Location 2 has a decision-making network that seperates problem anticipation from critical
evaluation.
• Cardboard Co. locations (the Plants) are physically and mentally separated from their
Headquarters.
• Cardboard Co. has no systematic insight into other single-loop learning issues than the
adhesive paper management.
• Cardboard Co. has a strong functional separation of responsibilities, decreasing the capacity
for interdisciplinary thinking.
• Double-loop learning is not supported in Cardboard Co. (even not at Headquarters).
My advice is to develop a new MICS, and some new management structures, as follows.
MICS:
• The new APMS could easily be made part of an integarted MICS. It should not only prompt
suggestions for improved planning, but also systematic reviews of performance indicators.
The organization should develop a set of performance indicators and implement these as part
of the new LMS.
• The new LMS should be developed on the basis of an explicit management theory, which is
recognizable and understandable by all organization members.
• Organization members should be invited to participate in thinking about Cardboard Co.'s
success by combining data from LMS with the managerial theory.
Management structures:
• Authority limits should be less detailed and used more informally. To decrease the
communication problems, Cardboard Co. Headquarters should be closer to the locations.
• Cardboard Co.'s commercial department should consider innovations in product, markets
and transformations. The existing transformational technology is getting out of date, and
causes spoilage of expensive materials. It is quite certain that some competitor will introduce
solutions for the spoilage problem within a few years.
• Possibilities for reducing spoilage of adhesive paper with the existing machinery are now
exhausted. It is now time to consider new flexible manufacturing technology. Cardboard
Co's 'width loss' is substantially larger than its 'cut loss' (3.1 versus .4 in location 1). Width
loss can only be reduced by: not accepting non-standard orders, improving planning
processes by using an advanced planning module of LMS, or introducing flexibility in
machine width.
Some deviations from the expectations about learning norms and MICS are
commented on here:
• Cardboard Co. also uses project groups, though not frequently, to solve
problems. This improves the single-loop learning process. MICS contributes
sometimes as a system that signals problems.
• Cardboard Co. also learns by disseminating the experiences of volvos and
disseminating experience among the locations. Improvement projects often
involve the collaboration of logistics, volvos and training and human resource
specialists.
• New operational insights are quickly implemented. Because of Cardboard Co.'s
small size, management can easily check the effectiveness of the implementation
by just walking into the factory.
• The MICS provides data for budgeting and problem anticipation, but also for
critical evaluation.
• The most important reason for the high SLL-score is that MICS contributes to
two fields of learning, and does this on all SLL-activities.
The differerences on the learning efforts score are not very significant. It is more
significant that the negative impact (inhibiting) of MICS on double-loop learning
seems not to exist in this case. As far as DLL is constrained in Cardboard Co., the
causes are based on (responsibility) learning norms (called authority limits).
Con 4 states that learning needs determine the learning norms required. This
statement is a truism, but leaves open the important question of how learning needs
influence learning norms. The learning needs score for this case is 1 and this
correlates with the following learning norms:
• Absence of a learning identity and policy definition. Learning is not a major
issue, and therefore has not led management to define a learning policy.
• Because learning needs are low and the problems are simple, it is mostly not
necessary to make responsibility norms involving an interdisciplinary group.
• Action norms concern the implementation of concrete instructions and
suggestions for improvement (mostly found out as a result of an improvement
project, or as an idea of a manager or expert of the technostructure). The
organization likes concrete action suggestions, and disapproves of complex
theorizing.
• Procedural norms exist to disseminate information about performance between
locations 1 and 2, and between shifts.
Statement 14 says that classic learning norms only lead to problem anticipation
202 Organizational Learning and Information Systems
MICS roles, whereas lean learning norms lead to MICS that also supports critical
evaluation roles. This classic case however shows that statement 14 is not true
because APMS and LMS support both roles.
The validity of Con 6 in case 1 is not yet clear. MICS indeed increases SLL-effort. It
has however no impact on DLL-efforts. This means that Con 6 is invalid. The
suggestion therefore is to confirm Statement 15, and to modify Statement 16 to:
MICS has a negative or no influence on DLL-efforts.
Con 7 proves to be right in this case. The learning norms support SLL-effort that can
only reduce complexity and dynamics when they are already low. MICS contributes
to this SLL process by providing data useful for this process (Statement 6). In this
case, the critical evaluation role of MICS does not lead to theory development and
unlearning (DLL-efforts), but only leads to theory adaptation, because the
organization perceives neither high complexity nor high dynamics.
This case is about a major European Commercial Bank located in most European countries and
other continents. The case studied was its Branch in one European country. It has been particularly
successful in the business market segment, and now is entering into the less cyclically sensitive
private sector (especially the richer subsegment). The Bank aims at delivering full financial
services in both market segments. We stressed the functioning of the Branches in this case, because
these are the sites at which the 'moments of truth' occur when interaction with clients take place
(cf. chapter 5). A bank was chosen as a typical example of a classic machine bureaucracy. The
organization in this case is however larger than in the others (about 3000 employees). The Bank is
also interesting because of attempts by its senior management to improve organizational learning
and reduce its bureaucratic nature.
The bank originated from mergers among small banks under the leadership of a huge European
bank. In the country studied it has about 3000 employees. There is pressure to cut staff as the
payoff of information technology, and to meet price competition among banks in Europe. Data
about employment in this bank are given in table 8.4.
28
I am grateful to Mr M. Hafkamp, who contributed substantially in the data gathering for this case.
Case Studies 203
The Bank has a business plan for 1992-1996 listing these critical success factors:
• Actively approaching the target groups (upper 40% of the private market and
business companies).
• Stabilizing its market size in the state of operation.
• Full service to optimize relationships with clients.
• Reduction of costs (mainly personnel) by means of business process redesign
and information technology.
• Improving internal communication for improving core activities for the
business market, private market, and support and facilitation of these markets.
The influence of Headquarters should consequently decrease and some
structural organization changes are planned.
• Improvement of its image via new products and communication programmes.
• Improvement of the international character of the bank, mainly through the
development of an international educational center for its personnel and
managers.
• Improving performance stability by penetrating the less cyclical private market.
The bank has an explicit management theory stating that its success depends directly
on its public image, operational effectiveness and efficiency, and the stabilization of
cyclical performance. Figure 8.4 gives a causal diagram of this theory and some
indirectly related factors.
204 Organizational Learning and Information Systems
Striving for full service and the elite target group are not consistent with a cost
leadership strategy. The Bank is well aware of this fact and does not have a cost leader
but client satisfaction strategy.
The Bank is divided into two types of Directorates: Commercial Directorates (Stocks,
Corporate & International Affairs, Product Management & Marketing, Branch
Management and Liquidities) and Support Directorates (Credits, Payments, Human
Resources, Information & Organization, and Management & Planning). Most
employees work in the Commercial Directorates (about 2000). About 1600 of these
people work for the Branch Management Department, which consists of the
Branches that directly serve The Bank's clients. In the Branch Management
Department about 200 people have non-managerial jobs. Figure 8.5 gives a general
organization chart of The Bank.
A further detailed study was made of the
Branches. One reason for this choice was
that the Branches are the main locations
where The Bank can learn about its
performance. The other reason is that The
Bank is explicitly trying to make the
Branches more self-managerial. The
Department of Branch Management is
divided geographically into six regions,
managed by Regional Directors. In total
the Department consists of 78 Branches,
with 4 big Branches (containing more than 50 employees), 38 small Branches (7
employees) and 36 middle-sized Branches (between 7 and 50 employees).
Coordination between the Branches is hierarchical via the Regional Directors and
Case Studies 205
account management.
• To control and inspect the ways administrative procedures are performed.
• To assign tasks to Administration for performing internal controls according to
rules set by the regional director and the Internal Accounting & Control
Group.
• To authorize contracts and transactions within a certain authorization limit.
• Outside representation of the Branch.
• To contact important outside stakeholders such as solicitors and local
government.
These management tasks are to some extent shared with other sections of The Bank
(such as departments of Headquarters).
If we look at the organization charts, the four learning fields all are clearly allocated
to departments, as is shown in table 8.5.
on the private market (which up to that time had been contributing about 20% to The Bank's
revenues).
After the mergers that led to the existence of The Bank, the CEOs centralized authority and
introduced bureaucratic mechanisms to reduce internal instability. This period is now over and The
Bank is following a philosophy of decentralization and encourages self-management. However,
this increases internal instability. The instability differs with respect to administration and
commercial functions.
Because administrative tasks are clearly defined and easily measured, performance and planning
data for this part of Branch work are easily determined. The commercial functions however are
much harder to measure in terms of amounts and predictability of work-loads. As a consequence
only data about commercial performance are available with indirect relations to work-loads.
The bank is well aware of the risks involved in the commercial activities. Therefore it has
developed a specific approach to treat prospects, to estimate a relation and to develop a cost-
effective sales strategy. Nevertheless, work-load as well as changes in profitability are extremely
difficult to determine in advance, and risk is an inherent issue of the commercial services.
Table 8.6: The Bank's Branch Tasks and Activities Indicating Internal Complexity
What is remarkable is not only the amount of specialization but also of control to
check for possible mistakes, fraud and performance (the latter was only performed by
the management). Of the 24 managerial control activities only 2 were for
performance control. The managerial (76) activities were the task of the location
director (13), Chief of Administration (25) and several other managers at local,
regional and Headquarters levels.
Many procedures are described in administrative handbooks, overview lists for
208 Organizational Learning and Information Systems
has recently started moving to a more decentralized structure. The Bank could therefore better be
qualified as moving towards lean, but it is still fundamentally a classic machine bureaucracy. The
score on the 10 point leanness scale is 3.
administrative function more directly accessible to clients. Dual cores (see chapter 5) are however
still available. This can also be seen in the organization structure of the support and commercial
departments and the split between the Branches and the Headquarter departments. The Bank is
now in a process of reorganization and wants to reduce the barriers between the cores, and aims at
complete integration.
Professionalism is high in the commercial and management positions. Also at Headquarters,
professionals are organizing administrative functions. Professionalism is low in most
administrative positions.
Conclusion: The Bank is a service company, but there are still many manufacturing features,
particularly within the administrative function. The Bank is also a classic MB. Hence, The Bank
belongs to the classic-service MB type.
re-engineering. This is possibly also the result of the service transformation, which e.g. requires
less additional investments to create a network (cf. chapter 5).
Responsibility norms
The Bank is a prototype of a classic machine bureaucracy with a classic distribution of learning
responsibilities. Learning occurs along hierarchical lines with support from the technostructure at
Headquarters.
The Bank has a functional learning structure, thus dominated by the hierarchy (management and
technostructure). No use is made of organizational networks. The volvo principle is to some extent
available in the Branches to adjust work processes and communicate specific issues of interest.
Task groups have not been formed despite the fact that The Bank is starting a major organizational
change process. The change process is strongly supported by the organizational apex, but the
changes are implemented top-down with a major role played by the Branch Management
Department's Organizational Development group. Project groups are frequently started up to
accomplish technical and product changes. These are mostly initiated by Headquarter departments.
Conclusion: the competence-based structure in The Bank is restricted to operational working in the
Branches. The learning structure follows the principle of the command chain, and thus is of the
type 'power-based and functional'.
Action norms
Work motivation in the bank is primarily extrinsic and oriented towards personal success and
income. There is however no policy to pay people for success. This is also almost impossible in the
commercial function, because of the time lag between action and result, and the ambiguity about
this relation. The success of the administrative group is easier to measure, but again it is difficult to
assign efficiency gains solely to the efforts of the administrative group alone. The Department of
Information & Organization, regional director, account managers etc. all contribute in some sense
to the success of the administration. Despite this motivation, no indications of defensiveness were
found that would obstruct learning processes.
Learning priority is very high in the bank, although learning was not a basic part of its identity
definition. The reason for this conclusion is the large number of employees (443, making 15,6% of
the total work-force) in technostructure positions, which includes those responsible for the
standardization of work processes, personnel analysis, recruitment and training, analysts for
planning and control, budget analysts and accounting.
Knowledge sources for learning are internal as well as external. Many insights are gained from
professional education, and many are gained from experience and analysis of the business The
Bank is in.
Conclusion: despite The Bank's action norms falling mainly in the 'money & slow' extreme, it
invests substantially in learning (high learning priority).
Procedural norms
The operational feedback, client-relation interaction, must be very quick. A poor response of the
Branches to complaints could seriously harm The Bank. However, it is bureaucratic and reacts
slowly when problems about products are discovered. The problems are then communicated from
the bottom (account managers) to the Branch Director, who in turn communicates with the regional
director. The latter then communicates them to the director of Branch Management, who in turn
communicates with the Department of Product Management and Marketing. Recently, however,
the Branch Management Department and Product Management and Marketing moved into the
same building. Because both departments are rather small, communication at this level is quick and
effective. In our interviews with Branch employees and Headquarters nobody complained about
the existing communication procedures, which might indicate that feedback frequencies are in line
with demands. Larger banks would probably suffer more from this problem. The future
information network could be used as a lateral structure to support quick communication about
212 Organizational Learning and Information Systems
products.
In most cases the procedural norms can be qualified as 'discrete and constrained', which is typical
of classic machine bureaucracies. The information flow is discrete and not on-line. Many reports
are produced and disseminated on a periodic basis, e.g. once a week, once a month, once a year etc.
Data access is clearly connected to authorities and responsibilities. The management style is telling
and the technostructure and management are important knowledge creators in the organization.
Deviations from the classic case are the feedback time and the number of issues measured. The
feedback time differs strongly between operational and tactical-strategic feedback. Operational
feedback is sometimes rapid for administrative services, which have a very precise and limited
period. A service failure provokes quick feedback requiring direct action, because competing banks
can supply most administrative services as well. For commercial services the feedback period can
be long, because the impact of these services manifest themselves no sooner than after a few years.
The tactical and strategic feedbacks take a long time as well. Many communication nodes in the
hierarchy must be passed before problems from the Branches finally reach the attention of
Headquarters.
A most remarkable feature of The Bank's MICS is the large number of indicators measured. For
instance a Branch director receives the following reports, generated from several information
systems, all called MICS here:
• Monthly Overviews showing the profits of the bank related to several products, the costs
involved and the costs and profits per employee. Also, a comparison with the budgets is
made.
• Monthly Budget Comparison. About 193 items of cost and profit data are generated by
comparing the results to the budget, cumulative budget data from the beginning of the year
are given, and data about budget target realization are given.
• Status Interests Analysis, providing data on six interest products with respect to the capital
available, the interest margins and the interest margins related to the amount of capital
involved. It also compares these data with data from the last two years.
• Balance and Profit/Loss Accounts per product, generated monthly, quarterly and yearly.
• Human Resource Costs, on the implications of personnel changes on personnel costs.
• Absentee Overviews, on absenteeism through illness at all Branches in one region, for
commercial and administrative personnel.
• Acquisition Plan, which presents data about the money gained by account managers in
relation to periods in the past. The data only concern the six interest-generating products of
the bank. Data about acquired provisions (e.g. insurances sold) are registered elsewhere.
• Transaction Monitor, showing the average number of transactions (monthly, quarterly,
yearly), the work-force allocated to the Branch and the number of people required
(according to a general norm) to process these transactions. In total 196 transaction types are
identified in the reports!
• Overview of Transaction Bills. For each account manager and for each account and relation
a monthly report is made of the amount of money billed to a client.
• Total Overview of Added Value of an Account Manager. This is a report about the interest
margins, provisions received, amount billed, and costs of transactions per account manager,
quarterly and cumulative per year. This report is also made per relation per account
manager. As The Bank now has 6 interest-generating products and 8 provision-generating
products, and account managers have an average of 250 relations, reports can have a length
of about 3584 items!
• Number of Relations (accounts) in Relation to Commercial Work-Force. Per region and
Branch an overview is generated of the number of private and business relations, split up
into debit and credit relations, the size of the commercial work-force and number of
Case Studies 213
Single-loop Learning
Adaptation
When initiated by Branches, adaptation is sometimes a slow process. Problems with products are
first communicated to regional directors and the director of Branch Management. Secondly, they
are communicated to Product Management. Product Management also initiates adaptation itself.
This is done by asking account managers about previous experiences.
The Department of Management and Planning also initiates adaptation. Depending on past
performance and The Bank's policy and aims, this department develops new ways for internal
budgeting, tariffs and pricing. It develops new products, processes and markets, which identifies
this activity as adaptation.
Adaptation is also an activity of the Branch Management Department, mainly by suggesting new
parameters for budgeting. This statement explicitly uses the term 'suggesting' because The Bank
has changed its internal policy from a command (top-down) structure to a more participatory
approach that gives the Branches more freedom. The reason for this policy change was not only the
maturing of the organization (the initial years after the merger were completed and the internal
situation has become more stable), but also in order to widen the information processing
capabilities at the local levels, which was required because of the large variety of locations.
Finally, the Department of Information & Organization also has a task, in evaluating and
suggesting changes for administrative procedures and internal control, but also through changing
information supply and systems.
The role of MICS in adaptation is problem anticipation (scheduling and longer term) and critical
evaluation (of products, operations, costs etc). MICS' value is limited to the fields of processes,
products and markets. No specific human resource planning and performance measurement system
exists.
Storage
Administrative procedures are well stored in handbooks, courses and people's experience and
education. The Bank also has experience with storing the knowledge of the account managers. The
214 Organizational Learning and Information Systems
following systems were specifically developed for this purpose: Overview Prospecting, which
supports the account managers via expert knowledge about the prospecting process, Conversation
Notes, which document conversations and make them accessible to other account managers, and
the Calendering System that combines information from the two previously mentioned systems, for
making and archiving schedules of activities. The many other management reporting systems
mostly provide data in relation to past periods. This enables interesting learning from the past, but
it is not clear how it could be used for more than extrapolating the past. MICS' roles are thus
scheduling and monitoring progress. Its value in these activities is restricted to the process field.
Dissemination
The coordination between Headquarters and the Branches could be improved considerably via the
IT-network. The Bank has a project to develop this. At the moment the periodic reports are a basic
mechanism of dissemination. The Branch Management Department also organizes regular
meetings at one Branch, or region, in which interpretations are given of the data and the
conclusions are discussed. The Branch Management Department then serves as a facilitator in the
discussions. More operationally, the systems for the account managers also have an important role
in disseminating experience and knowledge.
MICS' roles are on the process, market and product fields here. MICS' value is positively related to
these three learning fields.
(Re-)use
The systems do not specifically lead to improved skills. It is also not clear how they contribute to
augmented knowledge. Nevertheless, the dissemination of data when accompanied by a discussion
facilitator seems a good way to update knowledge and improve behavior.
MICS' roles are problem anticipation and the critical evaluation of the fields of process, products
and markets. MICS' values are positively related to these three learning fields.
Conclusion: The Bank is actively engaged in single-loop learning in all four learning fields, and
scores 1 for all cells. Its single-loop learning effort score is therefore 16. MICS contributes to most
cells. It helps to adapt, store and disseminate knowledge in all fields, but it does not contribute
much to using this knowledge, because of the complexity in interpreting, combining and searching
through the many reports. No inhibitors in SLL were found because of MICS. This means that its
SLL-value is +12. Statement 15 (MICS' positive impact on SLL-efforts) is confirmed. The human
factor is however essential (Con 7). MICS' role is in problem anticipation as well as critical
evaluation.
Double-loop learning
There is a lot of training and indoctrination at The Bank. Training amounts to about 1% of its total
budget. It is, however, not explicitly aimed at innovation. Double-loop learning efforts in the
human resource field is therefore zero, and MICS has no specific role here.
The network project is an important issue for technological innovation, but it is not clear how the
change of technical infrastructure is related to new ways of providing The Bank's services. The
network project has already been going on for some years and implementation of this new
philosophy is slow. At the Branch no systematic re-engineering is taking place, despite the
feedback data available. Only the Department of Information & Organization seems to initiate
some process innovation. There is however no systematic business re-engineering process and
philosophy. DLL thus seems not to occur in the transformation field, and MICS neither contributes
towards, nor inhibits double-loop learning in this field.
The Bank is in a typical growth situation. Its percentage of the mature national market is still quite
low and they are working to get more. At Headquarters much energy is spent on how to achieve it.
This is why The Bank entered the private market, a segment in which they were not very
experienced. They approach the private market with the idea that customer relationships are more
Case Studies 215
important than short-term profit, so that longer term success can be achieved. This means that The
Bank displays double-loop learning on the market development field. MICS contributes to this
field by providing precise market information. It is however not clear how it unlearns. The director
for Organization Development mentioned this also as a typical problem in The Bank.
New products are being developed especially for the private market. At the same time, The Bank
wants to provide a full financial service. This requires the development of relationships with
suppliers (e.g. insurance) to sell their products. The Bank is very active in this product
development double-loop learning field. MICS contributes by providing accurate data about the
performance of products. Unlearning was not found.
Conclusion: double-loop learning occurs at the Headquarters level, where the experts are. Effective
double-loop learning however also requires bottom-up participation, involving the later producers
of services. The Bank seems to have no systematic process innovation. The Department of
Information & Organization, in collaboration with the Branch Management Department, should
consider developing a process innovation philosophy, policy and methodology. Most remarkable in
this case is the positive impact of MICS on The Bank's double-loop learning processes.
Headquarters seems to be particularly keen on MICS data for designing new products, and
initiating marketing strategies and process innovations. This is contrary to Statement 15!
In total, DLL-learning efforts are counted for one activity and two fields leading to a score of 2.
MICS contributes positively to both fields and the activity, and does not inhibit DLL learning.
MICS' DLL-value is therefore +2. MICS' role is in critical evaluation and problem anticipation.
As a conclusion we would state that The Bank must improve its procedural norms considerably.
This can be done more concretely at the technical and organizational levels as follows.
• The Bank measures its operations, but under-utilizes this potential, because not everyone has
direct access to the data. The system is batch and information dissemination is selective. An
electronic and on-line management information system seems most needed.
• The Bank could profit considerably from a clear managerial theory, so that the most
important issues to measure are identified. Specifically, management theories for Branch
Management are under-developed, leading to reactive behavior.
• Try to improve learning capabilities at the Branches, so that action and theory are more
closely connected. The coordination problem that this could generate should be solved by
quick network communications between the Branches and Headquarters. The long and slow
communication lines increase the chance of mutual misunderstanding. Solution: flatten the
organization. Develop organizational procedures that increase the learning speed in the
organization according to network principles that can be easily realized in The Bank (the
technical infrastructure is already available). The value of the regional director is also
obscure, and possibly superfluous.
• Knowledge creation is a centralized activity in The Bank. This leads to under-utilization of
intellectual capabilities at the Branches and increases the tension between theory and action.
Solution: empower Branches to think and make lateral structures that can act as
electronically supported project groups and task forces.
learning norms. The senior management for instance is experimenting with self-
management and empowerment (delivering data to the shop floor so that they can
carry out their own analysis, facilitated by some central expertise). After case 1, I
stated that Con 4 is a truism and requires defining learning norm profiles in order to
become informative. The following learning profile is found for The Bank.
The point we detected was that management indeed recognized the importance of
single-loop and double-loop learning, but that its learning norms are such that they
prevent employees, other than managers and business analysts at Headquarters, to be
involved. The bureaucratic hierarchical structure thus is imposed on the organization
of the learning processes. The 'determination' of these organization norms therefore
is a clearly political process of distributing learning responsibilities. Information
technology can enable a flatter organization structure with many lateral structures, a
high learning environment, to be efficient.
Statement 14 tells us that lean norms lead to MICS with critical evaluation and
problem anticipation roles, whereas classic norms lead to MICS with problem
anticipation and accounting roles. This classic MB however has a MICS with both
critical evaluation and problem anticipation roles. This means that statement 14 is
not true. It is interesting to note that the critical evaluation role of MICS in this case
is often ineffective, because MICS is too complex to handle and the learning norms
(especially procedural norms) limit the effectiveness of critical evaluation.
218 Organizational Learning and Information Systems
In 1991, one of my M.Sc. students integrated the information systems that were available at that
time in this Chemical Plant. These systems contained data for logistics, production planning, and
performance control. We investigated whether the organizational conditions were available to use
this management information system effectively. We concluded that many organizational changes
would be required. For instance, the existing cultural differentiation between the management and
the employees was regarded as a huge obstacle to effective communications and management.
After one year, we were invited to observe the changes. Meanwhile, the fibers industry was going
29
I am grateful to Mr St. Kordelaar, for contributing substantially to the data collection for this case.
Case Studies 219
through a crisis on the world scale and Western European producers suffered from fierce
competition from low wage countries. During our stay (November-February 1992-1993), over 100
employees out of the initial 240 were dismissed. Just after the completion of our data collection we
heard that the company had been sold to a competitor, and is now in a turnaround phase. All the
information we present here refers of course to the situation before the take-over.
The Chemical Plant is part of an international chemical concern and has over 200 employees. The
Plant produces chemical fibers used mainly in the carpet industry. It is particularly good at
producing a large variety of fibers, also for small orders and in a short time span. By a combination
of input materials and sequences of transformations about 40 different production streams exist. At
the top of the Plant's organization structure is the Plant manager. Organization charts are given in
Three types of policy problems exist, closely connected to specific organizational functions,
summarized in table 8.10.
Stakeholder Problem
Sales Satisfying important clients
Logistics Achieving planned delivery time
Production Achieving cost minimization
The reasons for these separate and sometimes conflicting issues, is the existing
performance measurement system, which evaluates these departments on their
handling of these problems. This results in very different management theories,
elicited on the basis of our interviews (see figures 8.10 and 8.11). We will restrict the
discussion to production and logistics, because sales is not actually part of the case
studied, but an external actor.
Case Studies 221
able to make a large variety of fibers in about one million color combinations. At the same time it
uses relatively obsolete transformational technology, because it makes the products in sequences,
whereas some competitors use parallel production modes. Because of the fierce price competition
in the market these process innovations are absolutely necessary. Substantial organizational
learning is therefore needed in order to stay in the market.
Lean-Classic
The company is seriously involved in quality programmes and has had an ISO 9000 certificate for
some years. Effective quality control demands decentralization of responsibilities. This has been
tried by the management, but is very ineffective because people at lower levels lack the skills and
authority to make independent decisions.
The Plant is organized via its sequential production process, and only aims at producing certain
amounts of fibers of certain types. For this no direct communication with clients is needed. The
sales people therefore disconnect the Plant from its clients. On the shop floor shifts divide jobs, and
some vague evaluation of the performance is made. At the Plant level, the managers form a
management team, which meets every three months. It is not clear what kind of communication
occurs in this way. Lateral structures will therefore be mainly informal.
The Plant is part of a larger division that in turn is part of a large multinational chemical concern.
Decisions about investments in the Plant are basically motivated by strategic concerns (product
portfolio and possible synergies), rather than the actual costs of investments compared to financial
interest rates or revenues. New ideas are to be gained from the consortium's research and
development organization. A small lab only supports quality control and inspection. The sources of
innovative ideas from inside the Plant are limited.
Career paths for lower level management (chiefs) and workers are very restricted. Unit and Plant
managers, however, are academically trained and have taken (or are taking) part in a management
trainee programme and have an international career path. As a consequence the relation between
the management and the employees is problematic because of the cultural segregation of both
groups (high versus low education, cosmopolitan versus local). The Plant manager emphasizes
improving these relations but this is still wishful thinking. The motivation of the workers and the
management is mainly extrinsic: keep the job, keep the salary or even increase it if possible, with a
minimum of effort.
The organization is a classic MB. Only its emphasis on quality and its being part of a larger
consortium could give it lean features. These observations lead to some discrepancies, because the
rather high learning needs require a more lean organization.
Service-Manufacturing
As the company produces fibers, the output is tangible and measurable in tons. Because of the lack
Case Studies 223
of direct contact with the clients it does not receive a clear and useful market feedback. The fibers
produced are high quality and custom made (order production). Clients only participate in the
design and planning of the production, which is Sales' final responsibility. The Plant's production
process is frequently affected by the specific demands of big and important clients, who require
delivery in a short period. This obstructs the scheduling of production. As a compensation, the
Plant hopes to acquire rebuy from these major clients.
The following points describe the production process:
• Machines, material, labor, and know-how (chemical knowledge, production organization
knowledge and experience) are the main production factors.
• Clients are not ego-involved and the sales group shut the Chemical Plant off from the
market.
• Production scheduling determines much of the Plant's efficiency and must be done very
accurately. This requires a high amount of information processing.
• If the Plant makes a poor product this is clearly visible and is the Plant's own responsibility.
• The production process has three clearly distinguished phases.
• Stocks are possible, but small because the production process is order oriented.
• Operation and administration (including management) are spatially and mentally separated,
but the production personnel is responsible for part of the data input to the MICS.
• Professionalism is low, except in engineering and management.
Conclusion: except for information processing, this case fits perfectly in the manufacturing class.
The Chemical Plant therefore scores classic-manufacturing on the MB-type variable.
Responsibility norms
The following responsbilities distribution exists:
1. Two unit managers are responsible for evaluating the materials used and the personnel.
224 Organizational Learning and Information Systems
2. Chiefs-of-shifts evaluate personnel and material, mainly on a daily basis. This does not
however lead to an evaluation of methods and techniques.
3. The Department of Control checks the targets. When it finds structural deviations, it first
tries to investigate possibilities for improving work methods. When this is not possible it
changes the target (single-loop and double-loop learning).
The chiefs-of-shifts and shift chiefs are responsible for implementing structural changes that are
designed by logistics, production management and unit management. They however lack the time
and the skills to implement incompatible 'solutions'. They are in a difficult position, because they
have several bosses and must maintain good contacts with the management as well as with the
workers. Line management is the most important coordination instrument. The shifts are volvo
teams and have a large autonomy of task allocation. The shifts however do not explicitly document
what they are doing to improve processes. They restrict themselves to ad hoc problem solving.
Projects do not fill in this gap, because after an initial enthusiasm, implementation problems are
usually experienced, leading to the termination of a project. The organization thus fits into the
'power-based and functional' extreme of learning responsibilities.
Action norms
The Chemical Plant has the following action norms:
1. Management is not directly involved in the production process, and expects workers to solve
problems themselves, whereas they in their turn lack the skills and motivation to do so.
2. The shop floor regards problems as incidents and applies only ad hoc solutions.
3. The shop floor feels that problems that recur frequently originate elsewhere (such as
production planning, logistics and sales).
It is important to note that the possible sale of the Chemical Plant is being prepared by a
managerial task group at divisional headquarters. This may possibly result in some further work
displacement at the Chemical Plant. Hence employees on the shop floor are more defensive and
less cooperative with the management. The Chemical Plant's action norms are thus according to the
'money and slow' extreme.
Procedural norms
The PLATO system contains all basic data to support the learning processes:
• Performance indicators and defined targets.
• Operational planning (scheduling) data.
• On-line access to data about the production process.
The system provides excellent capabilities for the on-line adjustment of processes, and gives access
to management as well as to operators. However, it lacks aggregated data, trends and overviews for
management (according to the Plant manager).
To generate management information via PLATO, the following additional responsibility norms
exist:
• A production assistant makes aggregated overviews.
• Input of weekly schedules and monitoring of realizations is carried out by the programme
office of Logistics.
• Input of raw data is the task of production personnel.
The frequency of the feedback is not a problem, but rather the organization that must connect the
data for effective use in the right culture and structure. At this moment, nobody takes the
responsibility for anything done with the data, and the production assistant is too junior to
influence what happens.
Except for data access and the number of issues measured, the procedural norms are according to
the 'continuous and free' extreme. The restriction of the number of issues measured to specific
targets is not always an indicator of a classic MB. It could be that these specific targets are the
result of an excellent understanding of the business, and that these are well monitored for their
relevance. At The Chemical Plant this is not the case, because mental models are incomplete and
Case Studies 225
conflicting.
Two reasons might explain this unexpected data pattern:
• Management is forced to create a participatory management style, because the shifts are
well-organized groups and form a counter-force against the management.
• The existing MICS is technically very sound, so data flows continuously and freely. There
are no technical restrictions to fast data flows.
• Learning is strongly horizontally decentralized. The procedural norms variable does not
measure the vertical decentralization of learning activities (this is done by the responsibility
variable). Looking at vertical decentralization of learning, the volvos are not well equipped
to participate.
Single-loop Learning
Sales at the divisional Headquarters emphasizes the importance of meeting delivery demands. The
CEOs at the business unit stress the importance of reducing costs because of increased competition
from low wage countries. Both are issues for single-loop and double-loop learning in the Plant.
Adaptation
The adaptation activities are dominated by cost reduction and quality improvement needs. The
logistics system and a performance control system have been established to support adaptation.
The logistics system is used for the problem anticipation role, by supplying a tool for scheduling
and rescheduling. The critical evaluation system consists of a set of standards for departments and
shifts. This critical evaluation system was only recently introduced and was regarded as a major
innovation. The intention is to monitor performance indicators and to adjust them on the basis of
evaluations of output in management employee communications.
Despite these learning needs and information systems, the organization has some important
problems with learning:
• Complexity of understanding the impact of actions that are interdependent and taken at the
same time. A solution could be the formulation of a coherent management theory, upon
which actions can be based that will be tested later for their impact.
• Urgency of orders (external dynamics) that disturbs the discovery of regularities and
theories.
The role of MICS here is therefore in critical evaluation and problem anticipation. MICS' value is
226 Organizational Learning and Information Systems
positive, as it contributes to learning processes of this kind. Nevertheless the effectiveness of the
learning processes is questionable. The effectiveness seems to depend on the availability of a
closed learning loop (c.f. our discussion of Hofstede in section 6.3.2).
Storage
The Chemical Plant stores knowledge as follows:
• Weekly predicted costs and output figures are stored and compared with results, based on
PLATO data.
• Frequently reports are made about garbage and out-turn.
• Monthly reports are made about aggregated data of weekly reports.
• Each year the Plant makes extensive reports about its activities, performance and plans.
• Audit checklists are made and are ready for re-use in case new audits are required.
MICS' role and value here is considerable.
Dissemination
Two main and incompatible management theories (the logistics management theory with its
emphasis on delivery times, and the production cost management theory emphasizing phased
production and the ability to produce buffers and stocks) exist that have been 'sold' by their
creators to the chiefs-of-shifts for implementation. This has resulted in a rather ineffective use of
both theories, because the chiefs-of-shifts are unable to connect them. Because of this conflict over
theories, MICS (PLATO) data cannot be interpreted unambiguously, and MICS data were not used
to test the validity of either of these theories.
Re-use
The logistics system has a logistics model that is re-used in scheduling processes. The performance
control system has standards and measures that are re-used in performance appraisal.
Conclusion: single-loop learning occurs for all four activities, but only in the process field. The
single-loop learning effort score is therefore 4. The role of MICS is substantial: problem
anticipation and critical evaluation. MICS contributes to all activities, and has no negative impact
on single-loop learning. Its value is therefore +4. The question about the effectiveness of the
learning process is related to responsibility, action and procedural norms. Given the existing
situation the norms negatively influence single-loop learning effectiveness. This is also strongly
influenced by the external instability (possible sale of the Plant, work displacement, world market
competition and production over-capacity)! A model about organizational learning should
therefore distinguish between learning effort and learning effectiveness.
Double-loop Learning
At the moment the management is vaguely considering investing in modern parallel transformation
technology. The chance that this will really happen is nil, because the implementation of a new
management theory is a slow and resistance-evoking process (according to several lower and
higher managers). At the same time all large investments have been frozen because of a possible
take-over of the company by a US-multinational. The double-loop learning score for The Chemical
Plant therefore is 0.
MICS has no positive or negative influence on this double-loop learning score, thus also leading to
a MICS' value of 0. It has only a critical evaluation control role in this area, for detecting the major
cost factors and providing internal data on which investments can be based.
The PLATO system seems to be technically and theoretically very sound. Also some
responsibilities have been created to extract knowledge out of the PLATO data. The major
Case Studies 227
problems in learning with MICS are caused by the unsuccessful integration and connection of the
existing management theories. In principle there is nothing against theoretical pluralism in a
company, but now the chiefs-of-shifts have the task to combine the theories, while they lack the
skills and authority to do so.
The following solutions are suggested:
• Raise the educational level of the chiefs-of-shifts so that they can carry out the theoretical
combinations, and give them the required authority (decentralization).
• Get unit management closer to the shop floor, so that they make decisions that are based on
concrete observations and experience of the shop floor. This suggestion could be combined
with the first one, and result in another type of unit manager and a career path from shift
chief to unit manager.
• Additionally, a much stronger integration of sales, logistics and production is required. The
geographical distance between Sales and the two other parts of the Plant inhibits this.
Discussions between logistics and production should also be based on the development of a
common management theory, and joint responsibilities for the whole.
• Regular meetings in a management team (sales, logistics and production) should be held
with the production assistant and logistics office to discuss the evaluation figures in order to
learn systematically from the Plant's performance. Of course the Plant's policy should also
be evaluated and adjusted on the basis of these figures.
Conclusions evaluation
Hypothesis Case 3
Con 4: Learning needs determine the learning norms required for survival. True
Statement 14: Lean norms emphasize the critical evaluation and problem anticipation roles of False
MICS, whereas classic norms emphasize the problem anticipation and accounting roles of
MICS
Con 6: MICS contributes to single-loop learning effort and inhibits double-loop learning False
effort.
Con 7 Depending on the Learning Norms, MICS contributes to or decreases complexity and True
dynamics.
Table 8.12: Evaluation Table for Cross-Comparative Assessment.
Comments
This case showed that this chemical manufacturing Plant, able to manufacture a large
variety of products, can handle its complexity very well. The big problem is handling
situations that require change, which are often induced from 'outside' (market
changes and divisional policies). This supports conclusion 3, that dynamics
contributes more to learning needs than complexity.
Concerning Con 4: after case 1, I stated that Con 4 is a truism and requires the
Case Studies 229
Concerning Statement 14: MICS has problem anticipation and critical evaluation
roles in this case, and is very advanced in lean terms. The other learning norms are
however very classic, which means that the potentials of MICS are under-utilized
from a learning perspective.
Concerning Con 6: MICS does not have much impact when the learning norms are
not appropriate. MICS only contributes to facilitate learning by providing data.
These data can only be interpreted via mental models. When mental models clash,
the learning situation can become political and disruptive. These insights
considerably influence our understanding of the relation between MICS and Single-
loop and Double-Loop Learning. Let us write this in a set of additional statements.
S17: MICS is a provider of data. The better MICS is, the more relevant are the data at any
time and place.
S18: The type of data that are available from a MICS depend on a mental model. This model
has semantic implications which are incorporated into the structure of the information
system.
S19: The availability of data and a mental model are prerequisites for an interpretation of
reality.
S20: In organizations, people or groups of people can share models, but also can have
unconnected and even incompatible mental models.
S21: When, in interpretation processes (the cognitive part of the learning processes),
incompatible mental models are applied (as in the Chemical Plant case), the result is an
increased double-loop learning effort if people are not deliberateley avoiding the problem.
S22: When, in interpretation processes, shared (non-conflicting) mental models are applied, the
230 Organizational Learning and Information Systems
30
I am grateful to Mr M. Hafkamp, who contributed substantially in the data collection for this case.
Case Studies 231
As a result of a students visit, the author received a lot of information about the innovation efforts
of this company, which was supposed to be lean in many respects. We contacted the company for
further information, which they were very willing to provide. The information showed that the
company had a large and complex MICS, but it was not clear how it was used. After an initial
discussion it became clear that the company was still heavily engaged in moving towards a more
innovative organization. It was clear that the MICS was not functioning properly and that the
organization was still not lean in all respects. The purpose of our study therefore was to find out
how lean this service organization was and what the problems were with using MICS for
organizational learning.
Health Co. is a health insurance company, founded in the 1930s, and recently confronted with
major problems because of its old-fashioned management style. As a result the management was
replaced and fundamental organizational restructuring took place, with innovative information
technology as one of its basic changes. The company could easily cope with the changes, because
most of its initial employees had retired and young employees had been appointed. In 1990 Health
Co. had about 324 employees. This number was 240 at the time we obtained our data (February
1993). In terms of its market share in the health insurance business, it belongs to the biggest in
Western Europe. It acts in the private sector as well as the business market.
The senior management had discovered two basic problems for the company:
1. The management had not enough control over business processes, which led to many
inefficiencies (cost problem) and long lead times (quality problem). Measuring quality and
operationalizing it was regarded to be of strategic importance.
2. Some tension between cost and quality problems also existed.
At the beginning of 1991, the senior management (assisted by an external consultancy firm)
developed a management theory consisting of five basic concepts, namely:
1. Planning and management on the basis of explicit targets in terms of volumes handled and
time required for handling the activities.
2. Development of an organization structure that supports high speed and quality: the so-called
volvo teams.
3. The development of an Incentive Bonus Scheme to motivate departments and units to
improve and find improvements, by rewarding them with a percentage of the profits the
solution provides.
4. Developing a training programme to improve individual and group performance.
Organization members should be rated on several indicators to find out what their training
needs are.
5. Use of information technology to boost individual productivity (like E-mail and image
processing).
The major learning fields therefore are process (efficiency, quality and IT) and human resources.
Also, some activities for market development exist in the construction of specific products for
specific market groups.
organization and its way of handling documents dramatically. Additionally, many new products
and product innovations are required to keep pace with competitors all working in a business with
declining profit margins. As a result the organization also puts a systematic effort into education
and training. In 1992 it spent an average of about 470 US dollars on personnel training per
employee. In general this led to an optimization at the individual task level. Interdepartmental
coordination is not well managed. The impact of the target setting explains this situation.
Interpersonal behavior style is informal but there is also a keen control on meeting standards.
The technological characteristics of the organizational units are very bureaucratic, meaning that
work is regulated by many formally written rules. The organization consists of six business units
that carry out the direct administrative interaction with a client and 12 service departments that
support the business units with specific knowledge and activities. The business units are:
• Corporate Business Division: responsible for the collective insurances of the employees of a
company.
• Individual Business Unit: services individual subscribers, professionals, and small business
entrepreneurs.
• Special Business Unit: responsible for special client groups for which specific rules apply.
• Personal Health Division: service the market of more wealthy client groups, not insured
according to minimal legal requirements.
• Health Saver Scheme: insurance for special medical treatments and supplies.
• International: to promote the company abroad, for traveling and world wide-coverages.
The twelve service departments support the business units with medical-technical knowledge,
information technology, finance (payments, accounting etc), marketing, information services for
dissemination of internal messages, human resources management, quality improvement, support
for brokers, telemarketing, client service teams for sales to large companies, contract storage and
maintenance, and a company secretariat for mainly legal issues. This organization structure is not
very stable because market developments and government policies (which influence the business
severely) often demand changes.
There is a considerable exchange of information between the service departments and the business
units, but not much between the business units and between the service departments themselves.
The Business Management Group (consisting of the general managers of the business units and the
managers of the service departments) meets only once a month with a standard agenda. This
discussion platform was started recently and is not yet well established. Problems are solved
informally when they occur. There is much more integration within the business units, where the
general manager, supervisors and volvo members (lowest shop floor level) meet each week for 1½
hour to discuss issues. The low frequency of interactions and the rather stable organization
structure indicate low internal dynamics.
Intra- and inter-unit conflict is low. When it arises, senior management interferes. The
competitiveness of the environment seems to strengthen the senior management's control over the
organization.
The objective and goal of the company is mainly to survive. There is no clear formal description of
the goals, but by employee training and the supervision of new managers, basic values (quality and
efficiency) are taught. New managers have an experienced manager as supervisor and mentor
during the first period. When the new manager is regarded as competent enough he is allowed to
'fly solo', i.e. act without supervison. With the training schemes the employees learn a lot about
activities that are carried out elsewhere in the company. This is done to achieve high personnel
flexibility (data about flexibility are documented on the flexibility chart that is part of MICS) and
mutual understanding.
The service is increasingly differentiated to meet specific client needs and to gain a larger market
share. The business units described earlier demonstrate the importance of trying to attract special
market segments.
Many new distribution channels are being actively developed, such as tele-selling and direct
Case Studies 233
marketing. In general, however, the networks that are made with brokers and existing client groups
are still very essential, making the business less dynamic.
New material for health insurance could be reinterpreted in terms of suppliers of health care.
Developments in medical science are important here, making the business more dynamic. The
suppliers of information technology, by which the effective and efficient handling of the huge
amount of documents is improved, are important. Labor supply changes all the time mainly
because younger and higher educated people are applying for jobs. High unemployment makes it
possible to recruit bright young people for administrative jobs. These people apply their greater
abilities to work that was previously thought to be merely routine. This means that many
improvements are made in these jobs.
Government policy has a huge influence on the insurance industry in Western Europe. It decides
about what treatments should be provided. Government also sets production capacities for health
care, and in some countries the governments have nationalized or nearly nationalized health care.
This makes the industry dynamically dependent on medical science developments that lead to
increasing health service costs and declining financial possibilities of governments. The public
attitude is that a minimum of health care should be accessible for all citizens. This puts a lot of
pressure on prices.
Conclusion: the dynamics is increasing in the business. Health Co.'s management is actively
engaged in making the situation more stable.
Although the organization borders on leanness, it does not behave in this way according to its
identity and policy norms. There are several reasons for this:
1. The organization has changed dramatically since 1988 and is still in a process of further
development of its lateral structures. This is done via the previously mentioned Business
Management Group, which was established just three months before our study started. The
Quality Improvement Team is best comparable to a technostructure and therefore supports
management in decision-making. The RTA processes (for MICS-maintenance) hardly work.
In general the learning in the company is carried out via traditional line management
structures. At the same time, informal communication (e.g. via E-mail) is well developed for
operational problems. Electronic data highways however are not implemented to support
information logistics. They are only used for operational processes and not for managerial
and learning purposes yet.
2. Health Co. lacks a learning policy and identity formulation. This could also be explained
because of its low administration/production ratio, by which management has not enough
time to formulate policy-identity goals. Health Co.'s management does not see the use of a
policy formulation because it wants to keep the management as flexible as possible.
Systematic investments in people are made, and performance control information systems
are used explicitly to detect training needs, which is important for increasing the operational
effectiveness of the organization.
3. The organization is not very decentralized. The influence of the hierarchical lines is strong,
which can be explained by the fact that the company is still busy redesigning itself and by
the strong competitiveness in the industry. Business re-engineering is an important issue in
the organization now, to improve service quality and decrease costs simultaneously.
Information technology applications are considered to realize these demands.
Because of these data, Health Co.'s learning identity and policy norms therefore fall in the 'work
harder' extreme.
Responsibility norms
The company combines the traditional classic functional and divisional coordination mechanisms
in a matrix structure. There is a weakly developed technostructure (Quality Improvement Team)
and small work-groups on the shop floor manage client problems at an operational level, but do not
have authority for self-management. Project groups are sometimes created (for instance for MICS
maintenance) but their authority is weak in relation to the authorities in the standing organization.
Service companies were expected to work with loosely coupled groups (chapter 5). This is not so
within the business units, although coordination among the business units is weak. The routine-like
nature of this service work bears a strong resemblance to manufacturing organizations.
Although Health Co. is seriously trying to change, its responsibility norms are still according to the
'power-based' extreme.
Action norms
The organization members' involvement and commitment to attaining the set goals were high.
Management tries to motivate people by means of the clear definition of targets. People are
measured daily for their time performance, and managers can thus carefully control the output.
Senior management holds the management accountable for performance and is very keen on
monitoring them. Managers who underperform are dismissed. Management explicitly tries to
create intrinsic motivation and openness. This is done via:
1. The strong emphasis on training that would make employees more connected with the fate
of the business and triggers interests in developing the whole.
2. Openness via the business unit is exercised in weekly meetings of 1½ hours with all levels in
the business unit together.
3. People are willing to remove knowledge when it improves practice. This is contrary to our
236 Organizational Learning and Information Systems
previous statement that MICS would make unlearning more difficult. The deutero norms
(specifically the action norms) make it otherwise. For the future, the organization can
achieve an important competitive advantage when MICS can have the same role for
innovation support, and simultaneously be used for efficiency control.
4. Learning priorities are high in the organization, which is indicated by a high emphasis on
training, and the institutionalization of improvement and innovation procedures. For instance
the procedures for MICS maintenance (called RTA procedures) are well documented. Also
other innovation projects are well guided by project support (so called Situation Review
Procedure).
5. Sources of knowledge are internal and external. An organization consultancy firm provided
the basic knowledge and insights for transforming the 1988 organization to what it is now.
Contacts with universities have been established leading to frequent joint projects.
This evidence places Health Co.'s action norms in the 'team and fast' extreme, which is typical of
lean organizations.
Procedural norms
Nevertheless, competition is fierce and clients feel no obligation to stay with the company. Thus
client dissatisfaction must be reacted to at once. This also counts for reacting to changes in
government policies and new medical services that are covered by competitors. Health Co.'s
management follows these trends intensively.
The existing MICS is still a periodic standard report. This makes the flexible use of data difficult,
and thus inhibits opportunities to learn from the large amount of data gathered. Improving people's
awareness of the organization's performance can be very motivating and increases people's
willingness to participate in creating improvements and innovation. This is excluded when the
management style remains of the telling form. This also explains the deviations from 'off-line
systems', 'standard reports', 'inflexible databases' and the fact that actions are based more on
tradition and command than business understanding, described in the following description of
MICS.
Health Co. has many features of a lean learner, but is not yet complete. This is clearly seen in the
following:
• Learning processes are not connected with the development of theoretical knowledge. The
so-called Situation Review Procedures are seldom effective and the RTA procedures are not
applied.
• Performance control is ineffective, because people lack the time to review and analyze the
huge amount of data that is created. Computer support for MICS could solve this problem,
but does not exist.
We had the impression that no explicit connection between problem anticipation and critical
evaluation was made for developing business understanding. This also indicates that Health Co.'s
MICS is according to the 'classic' extreme. To test this opinion we made a more detailed study of
single-loop and double-loop learning and MICS' role, as described in the following section.
Single-loop learning
We made some quantification of the role of MICS in developing, disseminating, and storing
efficiency knowledge consisting of production standard concepts. We compared these concepts
with alternative media like memos, vision statements (regular letters of management to its
Case Studies 237
personnel) and output measurements. Some of the following conclusions are based on this
investigation.
Adaptation
MICS use has led to an improvement in insights into the business for the management, particularly
on how to match work volumes and work capacities. According to the financial director this
resulted in a system's payback period of only 9 months, and has resulted in a total performance
improvement of 30%. There are however no indications that MICS increases people's feeling of
responsibility. Also no attempts were observed for increased self-management, except the use of
the term volvo and a not yet successfully working Business Management Group. The system
therefore has the role of a support of top-down management communications. The theories about
these types of systems (Lawler and Rhode, 1976) suggest that it can easily lead to dysfunctional
behavior. We have no evidence for this, but at this moment it is also unlikely to happen because the
standards are not too difficult to achieve and the company has the opportunity of dismissing less
motivated people under the influence of its survival crisis. No double-loop triggers were observed.
Storage
MICS is primarily used for reporting about operational tasks. It does not develop a major
management theory in the form of planning models that later on are tested for validity. This is
because this service industry has much difficulty with planning its processes, as the work-volumes
depend on unpredictable moments when clients approach the company with a request for service.
Particularly its quality policy, which demands speedy handling of claims and other services,
inhibits a longer term planning and the use of buffers and stocks. Additionally the activities consist
of many small handlings, whereas manufacturing environments have often large orders and project,
with detailed planning, pre-budgeting and post-evaluations. This will make the development of a
management theory much more complicated, but not impossible and very rewarding.
Dissemination
Data are quite easily accessible, and when used in connection with business awareness
programmes, might motivate doing things better than before. The business management group is
not yet working properly. The business units hold weekly meetings that could lead to important
dissemination of knowledge. How these meetings precisely work is not clear (possibly very top-
down again). E-mail and informal gatherings are often used to solve problems and disseminate
suggestions.
(Re-)use
Understanding of cost and quality sources seems to have increased and the developed knowledge
was explicitly used in departmental budgetting, which leads to a further reduction of the required
labor force (as mentioned under adaptation). Strong management and positive attitudes to the
company's survival, linked with the young age of the employees, enable the quick implementation
of progressive ideas.
Health Co.'s single-loop learning efforts are in all the four activities, but restricted to the
transformation and human resource fields. This means that Health Co. has a SLL effort score of 8.
MICS' role is in problem anticipation and critical evaluation. MICS contributes to (re-)use, storage
and dissemination of knowledge gained about transformations and human resources. It does not
contribute to adaptation, because the adaptation of norms that are used is a very underdeveloped
process by now. MICS has no negative consequences as far as can be mentioned. Its contribution is
thus +6.
Double-loop learning
238 Organizational Learning and Information Systems
Double-loop learning is typically an activity carried out by senior management, that does not use
much MICS data at the moment. MICS' value could increase at this level if:
1. Excellent user interfaces and interactive systems are made.
2. It is used to provide data about markets and products.
3. Decision proposals are well grounded on correct data and well discussed before they lead to
decesions. The E-mail system could have an important function here, because it can work as
a lateral structure when the people can react in an anonymous way to concept proposals.
At this moment the company seems strongly internally focused. This can be explained from the
fact that it is hard to become a market leader in this business and that much of the health insurance
business is regulated by government policies. Therefore the learning fields are transformation and
human resources, and not much product and market innovation is carried out. The double-loop
learning effort is development as well as unlearning. It is not clear how systematic and theory-
based these DLL-activities are. Implementation is quick in this organization, although DLL is a
centralized activity. This results in a DLL-effort score of 4 (2 activity and 2 fields). MICS role and
value in DLL is however not present yet, but also not negative, leading to a MICS value of 0.
Comments:
• Because the learning needs score is 3 (low complexity, high dynamics), it is
understandable that management tries to centralize the learning process.
Essential in this situation are action norms that realize quick implementations.
This was indeed the case at Health Co.
• Double-loop learning efforts are however lower than would be expected for a
lean service company. This is caused by Health Co.'s internal focus on quality
and efficiency and its lack of opportunities for product and market
development. The small size of Health Co. (about 250 employees) contributes
to the fact that Health Co. cannot create enough managerial knowledge and
skills for improving the double-loop learning process. This 'small-size' fact also
contributes to a management style that is strongly telling and centralized, and
without a clearly defined management policy.
Conclusions evaluation
Hypothesis Case 4
Con 4: Learning needs determine the learning norms required for survival. True and false
Statement 14: Lean learning norms emphasize problem anticipation and the critical True
evaluation roles of MICS, whereas classic norms emphasize the problem anticipation
and accounting role of MICS
Con 6: MICS contributes to single-loop learning effort and inhibits double-loop False
learning effort.
Con 7 Depending on the Learning Norms, MICS contributes to or decreases Learning True
Needs.
Table 8.15: Evaluation Table for Cross-Comparative Assessment
Comments
240 Organizational Learning and Information Systems
Concerning Con 4: learning needs are important, but the way an organization reacts
to learning needs also depends on issues such as internal power relations (compare
The Bank), being part of larger consortia (to have shared learning resources in e.g.
divisional R&D departments) and, for this case specifically, its size that prescribes the
effectiveness of centralized learning. After case 1, I stated that Con 4 is a truism and
requires the definition of learning norm profiles in order to become informative.
Table 8.16 presents the learning profile for Health Co.
About Statement 14. In this moving-to-lean case MICS has critical evaluation and
problem anticipation roles. The effectiveness of the critical evaluation role is however
still low, because the organization lacks the appropriate technical tools to make the
use of MICS cost-effective. Additionally, the reponsibility norms are not yet clearly
enough defined so that all data can be usefully exploited to improve SLL and DLL.
Critical evaluation and problem anticipation groups are not integrated.
Concerning Con 6: MICS has little impact when the learning norms and technology
are not appropriate. When mental models are shared, as in this case, the following
observations are important:
• Action norms are no limitation to implementation. Theories are quickly
transformed to action because their non-execution is clearly visible to the
powerful people who constructed the theory.
• Model incompatibility therefore has a strong impact on action norms as
supporters or inhibitors of theory implementation.
An additional statement therefore is required:
S23: The greater the model incompatibility, the more action norms exist inhibiting behavioral
learning (SLL and DLL). The more compatible the mental models, the more action
norms lead to quick implementation of new theories and insights.
Case Studies 241
This case concerns a high tech manufacturing Plant, named Hitec, located in Western Europe and
part of a division of a US multinational. It produces electronic apparatus for industry and the
military. In the beginning of the 1980s it had about 700 employees, gradually reduced to 200 in
1993 with a constant output of 75 million US dollars and the introduction of some new products. In
the mid 1980's the company faced possible closure. A new management team (consisting of
Europeans) was formed with a survival strategy, based on increasing worker commitment to the
organization's success, aiming at top quality products, just-in-time (JIT) delivery and excellent
internal communication. This resulted in an ISO certificate (1986), many other quality awards, an
MRPA+ certificate and increased labor productivity. It is now one of the best performing Plants of
the company. The Plant has one product line of its own and others manufactured under license
from Headquarters in the USA. Their main markets are: USA (21.4%), Germany (16.1%), France
(15.3%), Britain (10.8%), Italy (9.4%), Scandinavia (5.8%), and Japan (1.7%). Despite the large
budget cuts in the military in the recent years, sales have been constant and there is a large order
volume at the time of this study (second half of 1993).
8.5.2 General Description of Hitec
The parent multinational was founded in the first half of this century. The Hitec Plant was started
in 1961. Because of personnel reductions some hierarchical lines (middle management) have been
removed. The local Plant is headed by an the operations manager who is directly supported by a
Human Resources Department, a Public Relations Department, a Controller, a Quality Assurance
Department and the manufacturing operations manager. The manufacturing operations manager is
responsible for the work done in seven departments, named: Value Engineering, Manufacturing
Technical Support, Instrument Manufacturing, Order Processing, Purchasing, Manufacturing
Planning, and Warehouse, Customs and Shipping. This study will focus on the Instrument
Manufacturing Department, of which an organization chart is given in fig. 8.13.
The span of control of supervisors in the production organization is large: 8 for the 'Insertion and
242 Organizational Learning and Information Systems
Board Build' group (responsible for the final assembly and th production of electronic boards) and
30 to 40 for the two other groups. This is realized by the creation of semi-autonomous groups
(called work cells), which have many joint responsibilities with respect to:
• quality
• output (delivery time and volume)
• work coordination and distribution
• detailed planning and materials supply
• administration of issues directly related to production (such as failures of delivery and
quality problems)
• coordination with support staff (e.g. with quality and planning)
The work cells have a very mature way of collaborating, not requiring much supervision. The
'Insert and Board Build' cell has regular meetings on Monday mornings chaired by each of its
members in rotation. The supervisor is only present when special problems occur. Work cells and
the existing organization charts were introduced in the second half of the 1980s, after the
introduction of a new management philosophy called 'Manufacturing Excellence' based on four
main issues: Total Quality Commitment, People Involvement, Manufacturing Resource Planning
and Just-In-Time production. The major learning fields therefore are 'process' and 'human
resources'.
specialists, staff members etc. when required in order to carry out their tasks. Supervisors
can assist in this process, but in general are only facilitators and intervene when problems in
coordination are structural and absolutely require senior involvement;
• JIT philosophy. In planning and production the emphasis is on delivery targets, and not a
just-in-time approach to all parts of the logistic stream. The production system is certainly
not push, but pull and then some push when possible (cf. Aggerwal, 1985).
The external environment is increasing in complexity because customers demand more product
innovation. From 10 years the product life cycle was reduced to just a few years. At the same time
innovating the manufacturing process is important for producing the new products and keeping
pace with the cost reductions that are demanded by the industry. This means that management is
still faced with many uncertainties. Many projects are underway to develop new knowledge about
how quality can be improved and how innovations can be carried out.
Conclusion: complexity and dynamics are high and still increasing. The learning needs score
therefore is 4.
31
See Evans and James, 1993 for further information about ISO norms and Malcolm Baldridge Awards.
244 Organizational Learning and Information Systems
The relation with suppliers is regarded as one of mutual benefit. Hitec helps suppliers to improve
their quality by its Supplier Performance programme. Suppliers are precisely monitored on their
performance (delivery time, defects etc.) via a special information system. Problems are fed back
and suggestions for improvements are given.
Although the client-plant relation is indirect (mediated via sales, marketing and services), the Plant
appreciates systematic feedback from clients via service and sales reports which are analyzed by
the quality assurance department for possible structural sources of problems (in procedures,
equipment used, and skills of personnel).
Because the Plant is part of a larger division, essential financial decisions are made in the USA.
Hitec's management team does not have much influence on this decision-making process. The idea
of low interest rates in large consortia therefore is not true, but depends on top management's
perception of the Plant's profits in relation to investments. Therefore the Plant has an uncertain
future. The managerial skills and the quality of the organization however are unique in the larger
organization, possibly safeguarding it from tough decisions.
One of the main pillars of the new management philosophy is to increase the involvement and
skills of the employees. The Plant therefore has invested considerably in human resources by
technical training for job enlargement, and interpersonal and management training for realizing
autonomous work cells and establishing a new culture. The total training effort is about 5% of
annual wage costs. As a consequence, job rotation is now easy though with a negative reaction
from the more highly educated employees who feel they are given tasks not matching their
qualifications. However, this job mobility is confined to this Plant. The Hitec managers have not
been invited to become members of the American or international management team. But as a
benificial side-effect, Hitec's management gets on extremely well with local employees.
Major new ideas are not infused via labor force composition, but via the process of learning in the
quality programme and some external training and consultants.
Motivation is intrinsic and extrinsic. The intrinsic part is institutionalized via the quality audits and
diverse other learning norms (work cells, performance data monitoring by managers and
employees etc.). The US management has also instituted awards for extremely good performance
which some managers think are appreciated although interviews on the shop floor suggest these
awards and quarterly management 'appreciation' meetings are typically American and do not
belong to their culture. They prefer monetary awards.
New ideas are consistently searched for via the quality improvement processes. At the same time
management is very open for new ideas that emerge from academic research. This was one of their
motivations for offering me the opportunity to interview many people and study their organization.
In general they appreciated discussions about their management theory to preserve them from
possible business blindness.
Conclusion: the company is lean for all criteria of our scale of leanness.
department is now busy with a quality assurance programme for services, and makes systematic
assessment of field failures.
Business re-engineering is an enduring activity and motivated from quality improvement and
possible technological changes (in products and processes).
Conclusion: Hitec has 'work smarter' learning policy norms. Policy and mission norms are indeed
oriented towards the 'work smarter' extreme, but the 'work harder' norms (expressed in production
volumes) are also applicable to Hitec. Remarkably, Hitec has a large administration/production
ratio (123/7732). This could indicate a non-lean organization. The interpretation of the a/p-ratio is
however very complex. In situations where production automation is increasing (as at Hitec), it is
likely that the a/p-ratio will rise. Furthermore, a head count is not a good indicator of leanness
when personnel costs are only a minor part of production costs.
Responsibility norms
At Hitec, learning is an organizational responsibility, in which departments, management and work
cells participate. Work cells, responsible for and committed to quality, search for information to
support continual improvement. When the problems are interdepartmental, contacts with these
departments are made. A quality engineer may also be involved.
At Hitec the subdivision between a functionally and divisionally (market) oriented work force is
becoming irrelevant. Only the Insert and Board Build group is a functional 'work cell'. The other
two work cells are related to product and market lines. The work cells produce as a group from the
beginning to the end. Volvo teams are the basic organization principle in the production
department. Further automation in production and technical developments will decrease the
number of people involved and require closer connections between production and support groups.
This could lead to the development of one large work cell.
Project groups are created frequently to research possible errors, improve quality and search for
renewal. Professional project management exists here. The error detection and improvement
projects often have the same collection of participants and a routine-like procedure. The renewal
projects (that introduce a basically new way of working, e.g. Kanban), often use a completely
different way of working and other participants.
The Plant constructs partnerships with suppliers and service/sales companies, also to learn from
each other. The Plant also has a leading position in a national quality association, and via this
distributes ideas and picks up new interesting ideas as well. At this moment the Plant is regarded as
a production unit within the division. Divisional Headquarters centralizes design and engineering,
however some decentralization would be feasible because the expertise and technology is
available.
32
Only 77 of the 200 people employed are working in the manufacturing department (Instrument
Manufacturing Department).
Case Studies 247
Hitec combines functional specialism and market orientation in a matrix-like learning structure.
Learning is embedded to some extent in the volvos (work cells) that have substantially large
responsibilities and discretion in analyzing and solving problems. This resulted in a large span of
control for two production cells (about 30 to 40 employees per supervisor) and a very passive
management attitude of the senior coordinator of the Insert and Board Build cell. The importance
of task groups in this company is also interesting and indicative of its greater appreciation of
learning. The task groups also indicate interest in double-loop learning (major changes that
fundamentally influence the way of working and thinking. (Some examples were the task force
SMD33 and Kanban). These double-loop learning activities are restricted to transformational
innovations. The Plant is not allowed by headquarters to think strategically about product and
market development! The Plant's management team and divisional CEO's only discuss targets and
output performance (typically the divisional control type as described by Mintzberg). Double-loop
learning about products and markets therefore is a top-down activity.
Conclusion: distribution of learning responsibilities is competence-based, however restricted to
single-loop learning and the process field of double-loop learning! Major double-loop learning for
this Plant is done by divisional Headquarters, and results are communicated top-down, leaving
only implementation activities at the Plant level. Some double-loop learning occurs as a result of
the quality monitoring.
Action norms
The Plant and company emphasize the importance of work satisfaction created through a 'we-
feeling' and the use of internal awards to individuals and groups. In this West European case this
system is possibly not effective. Employees stated that they were very committed to the company's
success because this is essential for maintaining the relatively well-paid jobs. The intrinsic
motivation to help the company succeed and excel is therefore closely related to extrinsic
motivations. People prefer financial rewards.
No indications of defensiveness were found. According to a tester interviewed, people on the shop
floor were eager to improve their work when they received comments about failures. Additionally,
the quality assurance group communicates clearly about what they will audit and when they will do
so. Findings lead to improvements, some through simple changes of work, sometimes through
training, sometimes through projects.
New findings and technologies are easily introduced and adopted. At the same time it is likely that
people on the shop floor will resist these changes, because many involve labor-displacing
automation. For some workers several years of training effort also can become worthless because
the new technologies do not require the new skills.
Conclusion: Hitec has action norms that are according to the 'team-fast' extreme. At Hitec we also
see the close relation between theory (the managerial philosophy of excellence) and action.
Implementation of insights is carried out persistently and carefully during the course of many
years.
33
SMD is a new soldering technique, completely automated and extremely quick.
248 Organizational Learning and Information Systems
Procedural norms
The feedback frequencies are much higher than expected from a classic manufacturer but still are
not on-line (most are weekly or monthly reports). The high frequencies result from the Plant's
desire for excellence, through detecting errors and the source of the mistake as soon as possible, so
that people will connect it more easily with what they have done. For these purposes testers give
feedback to the shop floor, and the quality assurance group analyzes outgoing quality data, and
communicates problems when found.
In all cases the existing information norms seem to fit with the lean type. The only deviation is the
fact that data from the main MRP system cannot be analyzed and accessed on-line. The MRP
system now is being replaced by one that has this on-line ability.
Conclusion: Hitec's procedural norms belong to the 'free and continuous' extreme, that was
regarded as typical of lean organizations.
It is particularly interesting to mention the main issues and systems with which the organization
assesses its performance, because this gives an operational description of the management theory
used and the procedure for communicating about it. The official management theory emphasizes:
Total quality commitment, People involvement, Manufacturing resource planning and logistics,
and Just-in-time production34.
From these major issues a number of indicators are derived:
• Total quality commitment indicators: customer service level, outgoing quality, reliability of
deliveries, error registration in delivery and production, vendor analysis and performance
rating, and the Malcom Baldridge assessments35.
• People involvement with absenteeism indicators.
• Manufacturing resource planning and logistics indicators: inventory as percentage of net
sales, work-in-progress volume costs also specified by components and parts and,
manufacturing costs of sales.
• Just-in-time production, or OPT indicators are: manufacturing velocity, inventory turns, and
manufacturing service level.
Some additional reports are present as well:
• Efficiency measures: net sales per employee (NOE); administration expenses as percentage
of net sales (these measures are extracted from the accounting system).
• Several financial performance indicators.
34
In fact, Hitec has not yet achieved JIT in all its processes, but it has OPT which means just-in-time
delivery and adjusting processes to accomplish this with the least effort (cf. Aggerwal, 1985).
35
See Evans and Lindsay (1993) for a detailed treatment of the Malcom Baldridge assessment.
Case Studies 249
Single-loop Learning
Adaptation
Adaptation of knowledge at the Hitec Plant consists of testing the quality of the way people are
working, and thus could lead to adjustments of theories by critical evaluation and training. MICS
has several functions here in detecting problems. Several people are involved in analyzing these
data and suggesting improvements. Data about outgoing quality and delivery are available.
Adaptation of knowledge about planning, ways of budgeting, ways of organizing (work cell
construction etc.) is done on a permanent basis through open communications. According to some
managers the distance between management and employees is rather small, therefore leading to
effective feedback. It is not clear if this is just wishful thinking. The distance between the
management team and employees seems to be too large. The quarterly meetings are ineffective for
bottom-up communication (they are clearly top-down).
Storage
Storage of knowledge is done whenever new procedures are discovered that could improve
effectiveness. Knowledge is written down in handbooks and the modular construction of the
product concepts, makes it easy to build some variants on a product line. New product lines lead to
250 Organizational Learning and Information Systems
more severe changes. This recently happened. Training then can become obsolete, which is quite
frustrating for people who have passed difficult exams. The stored knowledge is also to a large
extent available by electronic means. It was quite easy to obtain data about performance in recent
years, so that improvements could be related to managerial practices.
Dissemination
Much of the conceptual dissemination occurs through training in the basic issues of manufacturing
excellence. These trainings are followed by all members (top and shop floor). This training is about
social communicative aspects of successful team membership, management and motivation, and
quality concepts. The role of MICS in the dissemination of concepts is implicit. The information
systems supply data that are interpreted via the concepts. At the same time, concepts used for data
gathering and understanding were disseminated during the information systems user training and
systems development processes. Because concepts and information are mentally and
organizationally closely connected, interpretation problems are not problems about the syntactics
of the data, but problems about the structure of reality. The data can support alternative points of
view (Hegelian principles according to Kirsch and Klein, 1978). Nevertheless there is one clear
organizational philosophy, which decreases the chance of conflict about interpretations.
(Re-)use
(Re-)use is not problematic, because knowledge and data are closely connected and knowledge and
action are closely connected via training and the fact that learning processes are an inherent part of
everyday life in the Plant (work cell meeting, Tuesday's quality meetings, and Thursday's
production meetings).
Conclusion: MICS is absolutely crucial for the adaptation process. It only has a limited role in
storage. Of course set targets are part of the monitoring process, but even more essential is how
people understand these targets and norms. For this purpose extensive training and many meetings
and handbooks are essential. The dissemination process is supported by the quick way reports can
be generated and distributed by hard copy and by data on the terminal. Dissemination of concepts
(targets, norms and theory) goes through trainings and meetings. (Re-)use happens through the
interpretation of data that are the output of the various information systems. The MRP-system
seems to be particularly effective here because no additional informal information circuit exits in
the organization that could lead to conflicting interpretations. This means that the system must be
always right.
Hitec's SLL-learning effort therefore concerns all four activities and two fields (transformation and
human resources). All the SLL-activities for the two fields are performed, thus its SLL-effort score
is 8. MICS contributes to all these values and no negative impacts were found, so MICS' value for
SLL-effort is +8. MICS' role is in problem anticipation and critical evaluation.
Double-loop Learning
At Hitec, a production Plant, the opportunities for market and product developed are almost
restricted to zero. This type of double-loop learning is the task of the marketing department located
at Headquarters in the USA. The Plant has no opportunities to seriously influence what happens
there. One could describe the larger division best as a classic divisionalized form, with a command
and report relation between Headquarters and divisional units (Plants). This could seriously
threaten the Plant's longer term survival despite its excellence in quality.
Much effort however is placed in improvements of transformation. Some of the major innovations
in this area were not developed and chosen by the Plant itself. Again divisional Headquarters
decides what investments are made in transformations. Project management is an important
organizational information system to support the introduction of these innovations. MICS is often
used to discover required innovations.
Conclusion: Hitec has a very restricted double-loop learning process. Most double-loop learning
Case Studies 251
occurs in the USA with no participation of Hitec. Hitec is a successful implementor of ideas
developed by Headquarters. Here we touch the delicate question of the effectiveness of excellence
for longer term survival (cf. replications of Peters and Waterman, as described in Lammers, 1986).
Because Hitec is a successful implementor, its unlearning capacity is high. It has done this in the
past through innovation in transformation and products, which were externally developed. This
means a learning score of 2 (1 activity and 2 fields). MICS' value is very limited for double-loop
learning. Market data and strategic product data are not processed at Hitec. This means that MICS'
value for DLL is 0.
MICS could accelerate the single-loop learning process when technical features are improved
(compatible data structures, integration of systems, on-line high quality user interface). The social
system is ready to profit from these investments. The double-loop learning process is inhibited by
responsibility norms that cannot be changed by information technology. When changes occur,
MICS could be an important support instrument when market and commercial data are made
accessible to Hitec's management, and when the design and product engineering group of Hitec is
given more responsibilities.
Conclusion evaluation
Hypothesis Case 5
Con 4: Learning needs determine the learning norms required for survival. True and false
Statement 14: Lean norms emphasize problem anticipation and the critical True
evaluation roles of MICS, whereas classic norms emphasize the problem anticipation
and accounting role of MICS.
Con 6: MICS improves single-loop learning effort and inhibits double-loop learning False
effort.
Con 7 Depending on the Learning Norms, MICS contributes to or decreases True
complexity and dynamics.
Table 8.18: Evaluation Table for Cross-Comparative Assessment.
Comments
Concerning Con 4: learning needs are important, but the way an organization reacts
to learning needs also depends on issues like internal power relations (compare The
Bank), and its being part of a larger consortium (to have shared learning resources in
a centralized divisional R&D department). After case 1, I stated that Con 4 is a
truism and requires the definition of learning norm profiles in order to become
informative. Table 8.19 presents the learning profile for Hitec.
having implementors involved in the theory development and adaptation. Classic MBs require
responsibility norms that have implementors involved in the development and adaptation
process.
Conclusions and Discussion 255
The objective of this study was to understand and explain under what conditions computer-based
MICS contribute to organizational learning in lean and classic machine bureaucracies.
Management information systems were studied because of their pretention of creating smarter
organizations. Because of their particular problems with organizational learning, the study
focussed on machine bureaucracies (MBs) and monitoring information and control systems
(MICS). In seeking to answer this question, we were first faced with the lack of clear concepts and
theory in the joint area of organizational learning and information systems. Here we set ourselves
the following tasks:
1. Review the theory of organizational learning. This resulted in definitions and
operationalizations of organizational learning, and a basic assumption that states that
learning effort should be based on the learning need, and that learning can be facilitated by
learning norms and information systems.
2. Study the specific organizational and technical features of monitoring information and
control systems, and especially the way these features are combined with socio-technical
learning systems in machine bureaucracies.
3. Generate a theory to explain MICS' impact on organizational learning.
Tasks 1 and 2 should result in concrete hypotheses and variables for the construction of a theory.
On the basis of these tasks the following questions were formulated:
1. What are the basic dimensions of organizational learning?
2. How do lean and classic machine bureaucratic organizations learn?
3. Do lean and classic machine bureaucracies differ significantly in their way of using MICS
for organizational learning?
4. What is the influence of MICS on organizational learning in machine bureaucratic contexts?
5. How can one observe the impacts of MICS in machine bureaucratic environments?
These five questions are evaluated in this chapter, some comments are made about the results, and
issues for further research are mentioned.
Organizational learning consists of three processes, each with several activities (see table 9.1),
which are governed by organizational learning norms. This preliminary answer was based on
insights from four approaches to organizational analysis: cybernetics, organization development,
scientific management, and soft systems analysis.
Unlearning
Deutero Learning Learning identity and policy norms creation
(learning about learning)
Learning responsibility norms creation
Learning action norms creation
Creation of procedural learning norms (included MICS)
MBs, simple structures, professional bureaucracies, divisionalized forms and adhocracies have
different learning needs and learning norms. As the complexity and dynamics of the business
environment give rise to the learning needs, complexity is higher in MBs than in simple structures
but lower than in the other types (Mintzberg, 1983). Machine bureaucracies have a less dynamics
environment than all other types of organizations (Mintzberg, 1983). Nevertheless, environmental
pressures show that in all MB-cases studied, the learning needs score was increasing. As a result,
four profiles of effective (ideal) MB-learning combinations of norms and needs are described.
These profiles are summarized in table 9.2.
Learning needs 1: low complexity 2 high complexity and 3: High dynamics and 4: High dynamics and
score: and low dynamics low dynamics low complexity high complexity
Ideal learning
norms
Identity & policy None, or 'work Emphasizing core Learning policy and Learning policy and
norms harder' competencies and identity norms identity norms
managing them to 'work described as 'work described as 'work
harder' smarter' smarter'. Learning
infra-structures are
designed and core
competencies are
developed.
Responsi-bility Learning in Learning in specialist Committed learning: decentralized learning
norms functional groups, groups (mostly where possible, and competence-based.
and power-based technostructure), which
have power
Action norms Money and pain Money and pain Quick implementation of insights is required, and
Conclusions and Discussion 257
avoidance avoidance. Quick enabled via effective and fast teamwork. Workers
motivation. Quick implementation because are creative members of the team.
implementation of of expert power and
commands hierarchy.
Procedural norms Discrete and Discrete and Dissemination of data Dissemination of data
constrained constrained is free and continuous, and knowledge is free
dissemination of dissemination of improves and continuous,
performance data. performance data communication improves
(among specialists and between management communication
managers only). and shifts, and supports between
shifts' self-management management, experts
and employees
Example Low tech The Roman Catholic Classic MBs that are High tech companies,
manufacturers, like Church. No case like moving to lean, like like Hitec
Cardboard Co. this was found in our Health Co. and
study. Chemical Plant
8.9 Do Lean and Classic Machine Bureaucracies Differ in How They Learn?
This research question was valuable for detecting major insights into MICS' roles in organizational
learning. We compared several cases, some clearly classic and some clearly lean, and some that
were moderately lean or moving from classic to lean (table 9.3).
1 4 3 3 4
Learning need
(var. 1)
work harder work harder work harder work harder work harder and work
Policy norms smarter
(var. 3.1)
Power-based and functional Power-based and Power-based and Competence-based
Responsibility functional, but also R functional but
norms & D department and moving to
(ineffective) project competence-based
(var. 3.2) groups
Money (financially motivated) and slow to Money and slow Team and fast. Team and fast.
Action norms adopt new theories, but quick implementation of
(var. 3.3) operational insights.
Discrete but less Discrete and Continuous and free Discrete and Free and continuous
Procedural constrained because constrained. Free constrained via
norms shifts have access within branch, hierarchy
to mutual constrained via
(var. 3.4) performance data. hierarchical lines
Classic Classic, but very large Technically lean, but Classic (however Lean. Integration via
Description of socially separated large) but not MRP, no common
MICS problem anticipation computer-based. model
and critical evaluation
(var. 4)
8, four points 16, extremely high for 4 8 8
SLL-effort higher than classic service
(0..16) expected
(var. 5)
0 2 0 4 2
DLL-effort
(0..8)
(var. 6)
Problem Accounting, problem Problem anticipation Problem anticipation Problem anticipation
MICS' role anticipation and anticipation and and Critical and Critical and critical evaluation
(var. 7) Critical evaluation critical evaluation evaluation evaluation
(16..+16)
(var. 8.1)
0, not negative +2 0, not negative 0, not negative 0
MICS' DLL-
value (8..+8)
(var. 8.2)
Unexpected values are italicized in the table.
Table 9.3: Score Card for all Cases.
For all the other organizational learning variables, no systematic patterns along the
lean-classic line were found. This was probably not the result of research artefacts,
because the rating of lean and classic characteristics was done with care.
We expected MICS in classic cases to only be an appendix to accounting, problem
anticipation or critical evaluation systems. This was not true, because in our
observations no differences existed about MICS' roles. More interesting was the
question whether critical evaluation and problem anticipation communications were
done by separate groups or among groups that formed a closely knit network. The
lean case (Hitec) indeed showed that planners (performing problem anticipation)
and performance evaluators (performing critical evaluation) exchanged insights and
this demonstrated a closed learning loop at the organizational level. This type of
loop was absent in all other (classic) MB-cases, where problem anticipation and
critical evaluation were performed by strongly separated groups.
Organizational learning includes three very different processes with different impacts of MICS:
single-loop, double-loop and deutero learning, the third not the subject of this but of a new study36.
To assess the value of MICS for organizational learning, we need to understand these two learning
processes (Con 2, 3 and 4). Finally, we assumed that the effectiveness of learning can be not only
the result of MICS, but also of the learning norms (Con 7).
36
This study is titled 'Information Systems to Enable Organizational Learning' and is being conducted by
Marc C.P. Hafkamp.
Conclusions and Discussion 261
Our opinion about Con 4 evolved during the project. After case 1, we felt that the
conclusion was not concrete enough, because no answer could be given about the
question which learning norms are required given a certain extent of learning needs.
Therefore we started by describing learning profiles for each case. This resulted in
table 9.2, which prescribes certain learning norms for each learning need level. Table
9.2 only offers hypotheses. We were however not able to test the quality of these
ideas yet, because no scores for 'survival chance' were defined. As a next step in this
subject, the researcher should also check the influence of interpersonal relations,
abilities of sharing learning resources, and the size of the organization on the
required learning norms profile.
Statement 14 is correct in the two lean cases, but in the classic cases MICS also had
critical evaluation and problem anticipation roles. The most important difference
between lean and classic MICS is therefore not so much in their roles, but in how
the MICS (as formalized procedural norms) are linked with the other organizational
learning norms. Especially important here are the policy norms (that state the
importance of MICS explicitly and create an infrastructure), the action norms (how is
behavior changed as a consequence of conclusions drawn from data-analysis), the
responsibility norms (who takes responsibility for the quality of data and its analysis)
and other procedural norms (the way the insights are communicated and discussed).
More specifically we found that lean organizations have a more developed social
aspect of MICS. In particular we found that these organizations have an intergrated
network of problem anticipation and critical evaluation, which is not available in the
classic organizations.
262 Organizational Learning and Information Systems
In the first two cases we found that Con 6 was correct with respect to SLL. The
impact of MICS on DLL was not clear (in one case positive, and in another no
impact was found at all). I concluded that MICS does not have much impact when
the learning norms are not appropriate (cf. Markus' interaction paradigm, 1983).
MICS facilitates learning by providing data, but these can only be interpreted via
mental models. When organization members have incompatible mental models, the
learning situation can become political and disruptive. Information systems can then
only be supportive when they aid in discussion and negotiation. This type of system is
called semi-confusing (Hedberg and Jönsson, 1978). These insights considerably
influence our understanding of the relation between MICS and single-loop and
double-loop learning, and are summarized in the following paragraph.
MICS is a provider of data. The better MICS is, the more relevant are data at any
time and place (S17). The type of data that are available from a MICS depend on a
mental model. This model has semantic implications which are incorporated into the
structure of the information system (S18). The availability of data and (a) mental
model(s) are a prerequisite for an interpretation of reality (S19). In organizations
people or groups of people can share models, but also can have unconnected and
even incompatible mental models (S20). When, in interpretation processes (the
cognitive part of the learning processes), incompatible mental models are applied (as
in the Cardboard case), the result is an increased double-loop learning effort, if
people are not trying to avoid the problems involved (S21). This leads to the folowing
new statement: When shared mental models are used, this leads to more effective single-loop
learning processes (even with the same amount of SLL-effort) (S28). When, in interpretation
processes, shared (compatible) mental models are applied, the interpretation process
will be followed by actions (the behavioral part of organizational learning), if people
are motivated (action norms) to take these actions (S22). The greater the model
incompatibility, the more action norms exist inhibiting behavioral learning. The
more compatible the mental models, the better the action norms lead to quick
implementation of new theories and insights (S23).
Conclusion 8 now can be drawn, based on S24, S25, S26 and S27,:
Lean organizations have very effective theory implementation processes without having
implementors involved in the theory development and adaptation. Classic MBs require
responsibility norms that have implementors involved in the development and adaptation
process.
The insights of model 9.1 enable us to predict MICS-impact in different settings, and
use this knowledge to design an effective MICS as a socio-technical accomplishment.
The model is formed on the basis of insights gained in MBs, but is applicable to all
organizations, because the variables involved can be observed in all other
organizations. This means, in Glaser and Strauss' terms, that the 'substantive' theory
has become 'formal' (cf. chapter 3.2.2 and 3.5). Subsequent research should find out
the validity of this formal theory, by testing it with evidence from other organization
types. The theory is however limited to MICS systems.
The diagnosticum in the cases, if effective, may in the future be applied to practical business
problems. The scales used thus far could be improved by evaluating the construct validity and
reliability. The ratings as used thusfar are briefly evaluated here.
MB-type (var. 2)
The score of this variable was based on an index of leanness and a transformation. The ratings of
leanness and transformation were in some cases ambiguous, because most cases did not score
neatly along the extremes of the scales, and in some cases the organizations scored on both
extremes for the same item. For instance Health Co., although scoring high on the service extremes
of the transformation scale, also scored high on the manufacturing extremes of this same scale,
264 Organizational Learning and Information Systems
because it was an industrialized service company. A solution to this 'problem' might be to regard
these scales as multi-dimensional. It would be useful to find out if a factor analysis also finds
evidence for correlations among the dimensions. This is however useless without a rather large
statistical sample. If these correlations are strong, the items could also be reformulated.
were only of limited learning value because of the lack of well-designed procedural and action
norms. Learning in that case was almost completely based on the informal system. The Hitec case
showed that the advanced information systems for quality control were successful because this
system was well connected to the company's philosophy (identity and policy norms). The analysis
of informal systems is however still at an early stage. Some consultancy companies are now trying
to develop this as a service under the headings of communication audit. This is a valuable
approach. Stamper's MEASUR methods might be particularly interesting to apply here (Stamper et
al., 1994).
• Data dissemination must be solved less as a technical problem (but see Health Co.), and
more as a problem of procedural and responsibility norms.
For these arguments, the two-value variable we applied is an invalid operationalization of MICS'
roles. MICS' roles are better operationalized in three values: problem anticipation, critical
evaluation and procedural links between problem anticipation and critical evaluation. In MICS
design processes, it might be more useful to look at all SLL- and DLL-activities separately,
because MICS then is expected to support different learning activities.
A learning audit requires a frame of reference for selecting issues to investigate and instruments
for observation. Additionally, the auditor might adopt a prescribed method. This study's theory can
be used as a frame of reference with the operationalized variables. This section deals with a
possible audit method.
Learning audits should reveal significant problems and make recommendations about
improvements. The theory is important as a guide to observation. It also gives criteria for what is
going right or wrong.
This study emphasized contingency factors (called learning needs) for explaining effective
organizational learning and information systems' effectiveness. Deutero learning is the process by
which the information about these contingencies is used in developing skills, structures, policies
and instruments for SLL and DLL. So by defining the organizational learning needs we can detect
a mismatch between needs and learning abilities (norms, efforts and MICS) and suggest
improvements. If this discrepancy is very large, an organization must go through several stages to
evolve the required abilities. Additionally, if stages of learning can be defined, organizations can
also be prescribed how to develop learning abilities.
In this study we have found four learning profiles, consisting of sets of learning norms that match
with certain levels of learning needs (defined as an index of complexity and dynamics). In this
section we will describe these profiles, and add the learning efforts to enable a complete evaluation
of SLL, DLL and their learning norms. This leads to four stages of organizational learning.
Consistent with organizational growth theories (e.g. Greiner, 1972; Quinn and Cameron, 1983) an
additional stage is added in which learning is still at a beginning (birth) stage.
Stage 1: Fighting the moras. This first stage as yet has no procedures or policy concern for
organization learning, and we find no allocation of organizational resources to learning. This can
happen in at least two types of organizations:
• Organizations in their infancy, when it is too early to have developed procedures for
organizational learning. They are preoccupied mainly with survival. If this period lasts too
long, they will be beaten by organizations that are learning to acquire more efficiency and
competitiveness.
• Organizations in a hyper-dynamic and complex environment. In this case anything learned
Conclusions and Discussion 267
will be obsolete before it can be applied (Hedberg, 1981). At this stage an organization
cannot even define the knowledge needed, and when it is acquired, applying it would
endanger service and products. This is typically the case in organizations which must deliver
unique services rapidly (e.g. police and hospitals).
Stage 2: Bureaucratic learning. At this stage learning needs are recognized and met by learning
norms consisting of procedures and rules. This may be done provided that the environment is
analyzable and not too dynamic, the typical environment of the classic machine bureaucracy. This
situation can lead to efficient learning, but not without danger:
• The shop floor can easily feel alienated from a formal and centralized learning process
(organized in Research & Development, Management, Technostructure etc.) as they are
trained uncritically to obey commands.
• This easily leads to an under-utilization of human potential, which after a while is not able to
do more than obey.
• Although a lot of learning can best be done on the shop floor, where a close connection
between thinking and doing can be established, it will however not happen because these
organizations have no learning responsibility for shop floor people.
• Creating major changes is extremely difficult because of a lack of understanding on the shop
floor. At the same time, problems are incompletely comprehended at the top because no
knowledge goes bottom-up. A lot of senior management courage and effort are required to
achieve changes, a process which may take many years.
The organization in this stage corresponds with the profile mentioned in the simple-stable situation
and is typical of the classic MB. It only learns in the transformation field. When looking at the
priorities among learning activities, most priority will be given to knowledge storage (archiving).
The second priority is the adaptation of existing rules and the dissemination of these adaptations
via formal letters. Hopefully this knowledge is also (re-)used. The lowest priorities are given to
theory development and unlearning, thus resulting in very inert organizations.
Stage 3: Expert learning. At this stage management regards knowledge as a strategic asset. It
allocates budgets to specialist groups to develop the very complex new knowledge required.
Organizational learning therefore is institutionalized learning. Research & Development
departments and project groups are created. After the development activities most emphasis is
placed on storing and adapting the findings, considered to be of strategic value and kept secret or
patented. This is appropriate in environments of high complexity and low dynamics, but also can
lead to serious problems. The dissemination of the findings is often a neglected topic, leading to
under-utilizations of knowledge (Hamel and Prahalad, 1990). A critical success factor in these
cases is not only the creation of knowledge and skills, but especially the communication of
findings and the speed at which these findings can be implemented as new products, production
processes and services. Almost no emphasis is placed on the implementation of the new ideas in
the area of human resources and marketing, which is the major reason for the failures of the great
new ideas (Hamel and Prahalad, 1990).
Stage 4: Dispersed learning. To solve the many communication problems in organizational
learning typical of stage 3, some organizations allow strong decentralization (vertical and
horizontal) to encourage learning in all corners of the organization. The guiding principles for
learning are in this case:
• Support creativity and critical thinking.
• Allow experimentation.
• Encourage democratic relations among organization members.
• Create strong commitments to the organization via indoctrination and reward systems.
• Give priority to innovation in relation to costs.
An excellent example of such an organization is documented by Leonard-Barton (1992) in her
study of Chaparral Steel. The strong point of this learning configuration is the enormous amount of
energy that is freed from organization members, encouraging learning everywhere in the
organization. The most detailed problems are taken up by someone for analysis, improvement,
268 Organizational Learning and Information Systems
and/or innovation (people are allowed to create a group or project to think about possible new
products, even when in the initial stage the ideas might seem stupid).
Dispersed organizational learning is very strong in the adaptation of existing work procedures,
especially because the improvements are created by the people who must use them. The findings
are disseminated well in the organization, and picked up by people who have the knowledge and
skills to develop more profound innovations. A relatively weak point in this organization type is
the way how improvements and insights are stored. The human factor (individuals' memories,
culture, stories, myths and beliefs) are vital organizational memories here. On the other hand,
however, unlearning is accepted more than sticking to old experiences and insights. This type of
organization learns on the fields of transformation (especially quality), and human resources (by
appealing to people's interest in growth and personal curiosity).
The organization corresponding to this stage has high dynamics and low complexity, and is
sometimes called a 'learning lab' (Leonard-Barton, 1992). The basic problem with this type of
organizational learning is the allocation of limited learning resources and stopping projects. The
learning activities can become so popular that they could go out of control. Another problem is that
the amount of communication required, though possibly less than in the case where learning is
organized via hierarchical procedures, will generate problems of too much decentralization (high
coordination costs and agency problems. (Galbraith, 1973; Douma and Schreuder 1991).
Stage 5: Middle-Up-Down Management. This is a stage in which dispersed learning is guided by
business ideas, which are made more concrete by middle management functions that facilitate the
learning process. The following statements are typical of this stage of learning:
• Management gives direction about learning priorities via its strategic view of core
competencies.
• Management becomes receptive to ideas flowing upward, connecting them with their own
priorities and insights.
• Middle management makes general ideas about learning concrete by formulating budgets,
organizing work (especially the relation between on-going work and change work) and other
resources such as learning infrastructures (computers, archives, communication networks
etc.).
• Top management lays down a policy about the importance of learning and how learning
should be supported.
• A major issue is that learning does not occur within the confines of a separate organizational
unit, but is close to the everyday work process and includes awareness of its effects on other
organizational departments and work groups.
In this organization type, top managers that have brilliant and appealing visions (theory
development) are important, but even more important are the middle managers that can bring these
visions to ground level (adaptation). This also requires strong communication and indoctrination
processes as a coordination instrument (Mintzberg's 'missionary form'), so that learning is better
organized than in the dispersed type. As a result of this communication process people must feel
committed to using and re-using the gained insights, although more emphasis is given to the first
than the second because of the high demands of innovation. This learning type (originating from
Ikujiro Nonaka Honda Company experiences, 1988) serves an environment of high learning needs
(high dynamics and high complexity).
The stages mentioned describe levels of learning norms, and must themselves be learned. It is not
likely that an organization that has not yet reached stage 2 (the bureaucratic procedures that are
minimally required to stay in business) could be ready for stage 3 (expert learning), because expert
learning requires an administration that can be counted on. Stage 4 is mostly a reaction to the
limitation of the expert learning experiences, when organizational dynamics increase. At stage 5,
management realizes the essential importance of managing the learning process, not only for
keeping the process within the confines of budgets, but also for linking the efforts with strategic
demands.
The single-loop and double-loop learning efforts have costs involved that are valued differently in
Conclusions and Discussion 269
the different learning stages. The first stage sees learning as not relevant yet and therefore
evaluates learning efforts as costs that should be omitted. The second stage sees learning as an
improvement of administration and operational management. At the third stage, learning is
initiated by the management, and thus management costs. The fourth stage has learning
implemented in its on-going operational processes. The costs thus are embedded in operations, and
impossible to administer separately from these. The fifth stage regards learning as vital for the
organization's mission, because it is initiated by top management and requires the involvement of
all organization members.
When a limited budget exists for learning, priorities will be shared out differently at the different
stages. These priorities are summarized in table 9.3.
Table 9.5: Stages of Organizational Learning and Ranking of Priorities for Learning
Activities and Learning Fields
The audit could follow the steps used in the cases, starting with a general focus and then narrowing
to SLL, DLL and finally MICS' role and value.
The aim however should not only be to gain quick and valid scores but also reliable insights to
help the client understand his problems, and find a course of action. Well-designed presentation
tools can help support discussions and the development of an opinion. Our empirical work suggests
that the existing tables about the learning norms (table 9.2 and the tables in chapter 7) are well
designed for presenting major distinctions. The tables about learning effort and the learning value
of MICS could however better be split into a table for SLL and a table for DLL, to make clear that
they are very different types of learning (see tables 9.6 and 7.7).
Dissemination
(Re-)use
Each cell can be discussed separately to aid the systematic discovery of opportunities
to improve organizational learning activities and to detect and solve problems with
MICS, or other information systems. Finding which combinations of activities and
fields appear in the organization can reveal learning limitations. The application of
these tables however must be done in the broader context of the organization's
learning needs and the related learning norms and learning efforts.
This study has a number of limitations. As with every project with a deadline, choices had to be
made. For instance I decided to operationalize single-loop and double-loop learning processes but
not deutero learning processes. Statements were formulated about the link between learning needs
and learning norms, but no description is made of how organizations can organize the deutero
learning process themselves, and thus assess learning needs and norms.
Another limitation is related to the implicit rationalist approach to reality that is adopted here,
which implies that people, by understanding organizations and their problems, will find better
(transformation) technologies. The technology involved should imply a scientifically correct
relation (management theory) between the goals and the most effective and efficient means. This
optimistic view has been challenged by writers such as H.A. Simon (1976) emphasizing people's
bounded rationality. Organizations have many ways of coping with these human limitations, and
accomplish very complex tasks by such devices as division of labor, use of standard operating
procedures, use of decision premisses (task goals and constraints), which can be effective when the
organizational environment can be analyzed. When the environment is difficult to analyze, it is
difficult to find an optimal organization or information systems design. In that case the rationalist
paradigm reveals its limitations. Consequently, Cohen, March and Olsen (1972) developed a so-
called 'garbage-can' model, that states that decisions are not made via a rational selection from
several alternatives, but emerge from the interactions among more or less independent streams: a
stream of problems (with dates of first appearance of the problems, energy required for finding
solutions, and a list of decisions), a stream of choices (containing decision moments, agendas, and
participants), a stream of decisions, and a stream of participation energy per participant. This
model explains that solutions for problems in organizations depend on the moment the problems
were recognized, the energy participants put into solving them, and group structure and task
division. In this unanalyzable situation knowledge about the right solutions is not a critical issue. In
principle, someone with the least knowledge about a subject could become the most important
Conclusions and Discussion 271
decision-maker, when he has the power gained via his position, or by putting more energy into the
decision-making process than the others. Information systems could help to identify problems
more quickly, after which the priority on the decision-making agenda becomes a problem and a
part of the garbage-can. This model gives an insight into the limitations of organizational
rationality and learning. It is however only valid in cases of unanalyzable environments. When the
environments are analyzable, the best learning organization will be most effective. It seems as if
lean production organizations have solved many limitations of learning. Competitors then have no
other choice than to improve their learning ability as well.
A third limitation of this study is its restriction to the organizational level of analysis. Many
improvements in organizational rationality imply the adoption of new production technologies,
especially automation of manufacturing and services. This process of impovement is not limited to
within the gates of the factory, but increasingly involves improvement along the external value
chains as well. Some examples are the demands for better supplies of materials by applying ISO
9000 standards (cf. case of Cardboard Co. and its adhesive paper supply). Another example is the
trend to outsource the activities of a company (e.g. its administration) that are considered not to
belong to its core business. This outsourcing process is not limited to national borders, but often
involves the search for optimal physical and socio-political environments for production and
service anywhere in the world. Famous examples are the outsourcing of computer programming by
European software houses to India, where many skillful programmers are available for low wages.
Other examples are the establishment of West European manufacturing plants in developing
countries. The problems of coordination that limited these opportunities in the past are nowadays
solved via telecommunications means. The consequences of these trends are large, because
(western) countries must now rethink their economic function in the world.
A fourth limitation concerns about the conceptualization of learning needs. It was decided to score
learning needs in terms of organizational uncertainty (complexity and dynamics) following Duncan
and Weiss (1979). This scale does not rate individual learning needs nor did it rate specific
problems or problem classes and their urgency of solution. Individual learning needs were
excluded because this study is about organizational learning. Organizational learning can also be
approached by studying problem classes and how organizations could manage them (Etheredge,
1980 and Mason and Mitroff, 1973). The only reason why this was not done is that learning about
problem classes, for an unknown reason, is not a main issue in the tradition of organizational
learning and information systems.
This study also clarifies the need for further research in many areas. Some of these needs are listed
below.
• Chapter 6 proposed that dynamics and complexity have a different impact on the value of
knowledge. This proposition needs further empirical research, because it could shed light on
the value of organizational knowledge and organizational memory stores. In such a study
also criteria for investments in different organizational memories could be developed.
• The scales and indexes we developed are only a beginning for further research, which
should be more codified, to support surveys on the issues of organizational learning. It
would be particularly interesting to develop a measure for organizational learning
effectiveness. This requires the development of a scale that measures learning effort and
learning needs. The match between needs and effort could be an indicator of learning
effectiveness. In a survey, measures could be tested for their validity and reliability, and it
could clarify the quality of organizational learning in the research population.
• Deutero learning was not studied. Rather then describing learning processes (many
researchers have done this already under the headings of quality research, organizational
change, and innovation) one could study the effectiveness of techniques being marketed at
this moment (such as computer-aided systems engineering, micro-worlds, executive
information systems, knowledge-based systems, case-based reasoning, learning labs).
• Research also might be fruitful in the area of novel types of information systems such as
group decision support systems, electronic highways, knowledge-based systems. What are
272 Organizational Learning and Information Systems
the opportunities and limitations of these systems for organizational learning? What kinds of
changes in organizations and people are required to make them profitable from an
organizational learning perspective?
• Research on the method of information systems development could profit from an
organizational learning perspective in two ways. First, information systems development is a
learning activity, which means that the knowledge created should be stored, removed, (re-
)used, maintained and disseminated. How are these activities linked with the practice of
systems development, and how should this be done? Secondly, system development has
often suffered from a technological determinism, for example in business re-engineering
which usually fails to consider how to make business re-engineering a social activity,
involving intelligent people each with their own experience and knowledge. Business re-
engineering practice could thus profit considerably from applying the organization
development and soft systems perspectives, by which it creates a more complete and closed
learning loop, rather than by limiting itself to the scientific management and cybernetic
approach.
• A final research project could concern problem classes and the types of organizational
norms that are most suited to cope with these problem classes (Etheredge, 1980). Strangely
enough, as far as I know, it has not led to any larger research project until now.
Researchers interested in organizational learning are very much encouraged to participate in this
fascinating field.
Conclusions and Discussion 273
274 Organizational Learning and Information Systems
References
Argote, Linda, Sara L. Beckman and Dennis Epple (1990),
The Persistence and Transfer of Learning in Industrial
Acar, William and Timothy J. Heintz (1992), Toward Settings. In: Management Science, vol. 36, no.2,
Computerizing a Causal Modeling Approach to Strategic pp.140-154
Problem Framing. In: Decision Sciences, vol. 23, pp.
1220-1230. Argyris, Chris (1970), Intervention Theory and Method: A
Behavioral Science View. Addison Wesley, Reading Mass.
Ackoff, Russell L. (1992), Corporate Perestroika: The
Internal Market Economy. Paper belonging to Ackoffs Argyris, Chris (1971), Management Information Systems:
presentation submitted to the 'Systems Thinking in Action The Challenge to Rationality and Emotionality. In:
Conference: Creating Learning Organizations', Boston, Management Science, vol 17, no. 6, pp. B-275-292.
December 1992.
Argyris, Chris (1977), Organizational Learning and
Adelman, Leonard (1992), Evaluating Decision Support Management Information Systems. In: Accounting,
and Expert Systems. Wiley & Sons, New York Organizations and Society, vol. 2, no. 2, pp. 113-123.
Adler, Paul S. (1990), Shared Learning. In: Management Argyris, Chris and Donald Schön (1978), Organizational
Science, vol. 36, no. 8, pp. 938-957. Learning: a Theory of Action Perspective.
Addison-Wesley, Reading, Massachusetts.
Adler, Paul S. and Robert E. Cole (1993), Designed
Learning: A Tale of Two Auto Plants. In: Sloan Argyris, Chris (1980), Some Inner Contradictions in
Management Review, Spring, pp. 85-94. Management Information Systems. In: Lucas, H. C. jr., F.
F. Land and T. J. Lincoln (eds.), The Information Systems
Aggarwal, Sumer C. (1985), MRP, JIT, OPT, FMS? Environment. IFIP-proceedings 1980.
Making Sense of Production Operation Systems. In:
Harvard Business Review, September-October, pp. 8-16. Argyris, Chris (1982), Reasoning, Learning and Action:
Individual and Organization. Jossey-Bass Publ., San
Anderson, T. R. and S. Warkov (1961), Organizational Francisco.
Size and Functional Complexity: a study of administration
in hospitals. In: American Sociological Review, vol. 26, Ashby, W. Ross (1956/64), An Introduction to
pp. 23-28. Cybernetics. Metuen & Co Ltd., London.
Ansari, S. L. (1977), An Integrated Approach To Control Ashworth, William (1975), A Short History of the
System Design. In: Accounting, Organizations and International Economy Since 1850. Third Edition.
Society, vol. 2, pp. 101-112. Longman, Lomdon.
Ansoff, Igor (1987), Corporate Strategy, Revised Edition, Barnard, Chester I. (1938/62), The Function of the
Penquin Books, London. Executive. Cambridge (Mass), Harvard University Press,
1938, 15th print 1962
Anthony, R. N. J. Daerden and N. M. Bedford (1984),
Management Control Systems. Homewood, Bateson, Gregory (1971), Steps Toward an Ecology of
Mind. Ballentine, New York.
Aquilar, F. (1967), Scanning the Business Environment.
New York, MacMillan.
Bateson, John E. G. (1989), Managing Services
Marketing: Text and Readings. The Dryden Press, Blau, Peter. M. and Richard A. Schoenherr (1971), The
Chicago. Structure of Organizations. Basic Books, New York.
Beer, Stafford (1981), Brain of The Firm: The Managerial Blau, Peter M. and W. Richard Scott (1963), Formal
Cybernetics of Organization. John Wiley & Sons, Organizations. Routledge & Kegan Paul.
Chichester. (Companion volume to The Heart of
Enterprise, 1979) Boesjes-Hommes, R. W. (1970), The Valid
Operationalization of Concepts: a Proposal (In Dutch: De
Bell, Daniel (1979). Thinking Ahead: An Information Geldige Operationalisering van Begrippen). Boom,
Revolution Presents New Opportunities and New Meppel.
Problems for Business and Society. In Harvard Business
Review, May-June, pp. 20-42. Boisot, Max H. (1986), Markets and Hierarchies in a
Cultural Perspective. In: Organization Studies, vol. 7, pp.
Berger, Peter L. and Thomas Luckmann (1967), The 135-158
Social Construction of Reality: A Treatise in the Sociology
of Knowledge. Penguin Books, Harmondsworth. Boone, Mary E. (1991), Leadership and the Computer:
Top Executives Reveal How They Personally Use
Bernstein, Richard, J. (1976/1979), The Restructuring of Computers to Communicate, Coach, Convince and
Social and Political Theory. Methuen & Co, London. Compete. Prima Publ.,Rocklin CA
References 275
Botter, C. H. (1980), Industry and Organization (In Dutch: Burrows, Brian (1994), The Power of Information:
Industrie en Organisatie: een overzicht en uitzicht. Elfde Developing the Knowledge Based Organization. In: Long
druk) Kluwer publ., Deventer. Range Planning, vol. 27, no. 1, pp. 142-153.
Braverman, H. (1974), Labor and Monopoly Capital. Bushe, Gervase R. and A. B. Shani (1991), Parallel
Monthly Review Press, New York. Learning Structures: Increasing Innovation in
Bureaucracies. Addison-Wesley Publ., Reading Mass.
Brown, John Seely and Paul Duguid (1991),
Organizational Learning and Communitiesof-Practice: Buzzell, Robert D. and Bradley T. Gale (1987), The PIMS
Toward a Unified View of Working, Learning, and Principles: Linking Strategy to Performance. The Free
Innovation. In: Organization Science, vol. 2, no. 1, Press, New York.
February 1991, pp. 40-57.
Camp, Robert C. (1989), Benchmarking: The Search for
Buchko, Aaron A. (1994), Conceptualization and Industry Best Practices that Lead to Superior
Measurement of Environmental Uncertainty: An Performance. Quality Press, Milwaukee, Wisc.
Assessment of the Miles and Snow Perceived
Environmental Uncertainty Scale. In: Academy of Cash, James I. jr., F. Warren McFarlan, James L.
Management Journal, vol. 37, no. 2, pp. 410-425. McKenney and Lynda M. Applegate (1992), Corporate
Information Systems Management: Text and Cases. Irwin,
Burns, Tom and G.M. Stalker (1961), The Management of third edition, Burr Ridge, Illinois.
Innovation. Tavistock, London.
Checkland, Peter (1981), Systems Thinking, Systems
Burrell, and Morgan (1979), Sociological Paradigms and Practice. Wiley & Sons, Chichester.
Organizational Analysis: Elements of the Sociology of
Corporate Life. Heinemann, London.
Chew W. Bruce, Dorothy Leonard-Barton and Roger E. In: Administrative Science Quarterly, pp. 207-224.
Bohn (1991), Beating Murphy's Law. In: Sloan
Management Review, Spring, pp.5-16. Daft, Richard L. and Karl E. Weick (1984), Toward a
Model of Organizations as Interpretation Systems. In:
Child, John (1972), Organizational Structure, Environment Academy of Management Review, vol. 9, pp. 284-295.
and Performance: The Role of Strategic Choice. In:
Sociology, vol. 6, pp. 1-22. Daft, Richard L. and Robert H. Lengel (1986),
Organizational Information Requirements, Media
Child, John (1987), Information Technology, Richness and Structural Design. In: Management Science,
Organization, and the Response To Strategic Challenges. vol. 32, no. 5, pp. 554-571.
In: California Management Review, (Fall), pp. 33-50.
Date, C.J. (1986), An Introduction to Database Systems.
Coats, Pamela K. (1991), A Critical Look at Expert Volume 1, Fourth Edition. Addison-Wesley Publishing
Systems for Business Information Applications. In: Company.
Journal of Information Technology, pp. 208-216.
Davenport, Thomas (1993), Process Innovation:
Cohen, Michael D., James G. March and Johan P. Olsen Reengineering Work through Information Technology.
(1972), A Garbadge Can Model of Organizational Choice. Harvard Business School Press.
In: Administrative Science Quarterly, vol. 17, pp. 1-25.
Davenport, Thomas H. and James E. Short (1990), The
Coleman, James (1973), The Mathematics of Collective New Industrial Engineering: Information Technology and
Action. Heineman, London. Business Process Redesign. In: Sloan Management
Review, Summer, pp. 11-27.
Cronbach, L. J. and P. E. Meehl (1955), Construct Validity
in Psychological Tests. In: Psychological Bulletin, no 52, Davis, Gordon B. and Margrethe H. Olson (1984),
pp. 281-301. Reprinted in H. Fiegl and M. Sriven (eds.) Management Information Systems; Conceptual
(1956), Minnesota Studies in the Philosophy of Science, Foundations, Structure and Development. Second Edition,
pp. 174-204. McGraw-Hill.
Crozier, M. (1964), The Bureaucratic Phenomenon. De Geus, Arie P. (1988), Planning as Learning. In:
London, Tavistock. Harvard Business Review, March-April, pp. 70-74.
Cyert, R. M. and J. G. March (1963), A Behavioral Theory De Raadt, J. D. R. (1991), A Cybernetic Approach to
of The Firm. New Jersey, Englewood Cliffs, 1963. Information Systems and Organisational Learning. In:
Kybernetes, vol. 20, no. 1, pp. 29-48.
Daft, Richard L. (1991), Management. Second Edition.,
The Dryden Press, Chicago etc. Deal, Terrence E. and Allen A. Kennedy (1982),
Corporate Cultures: The Rites and Rituals of Corporate
Daft, Richard L. and Norman B. MacIntosh (1981), A Life. Addison Wesley Publ, Reading Mass.
Tentative Exploration into the Amount and Equivocality
of Information Processing in Organizational Work Units. Dobson, J.E., A.J.C. Blyth, J. Chudge and R. Strens
276 Organizational Learning and Information Systems
(1994), The ORDIT Approach to Organisational (1993), Fit, Equifinality, and Organizational Effectiveness:
Requirements. In: M. Jirotka and J. Goguen: Requirements A Test of Two Configurational Theories. In: Academy of
Engineering, pp. 87-106. Academic Press, London. Management Journal, vol. 36, no. 6, pp. 1196-1250.
Duncan, Robert B. (1972), Characteristics of Flamholtz, Eric G. (1983), Accounting, Budgeting and
Organizational Environments and Perceived Control Systems in Their Organizational Context:
Environmental Uncertainty. In: Administrative Science Theoretical and Empirical Perspectives. In: Accounting,
Quarterly, vol. 17, pp. 313-328. Organizations and Society, vol. 8, pp. 153-169.
Duncan, Robert and Andrew Weiss (1979), Organizational Flamholtz, Eric G., T. K. Das and Anne S. Tsui ( 1985),
Learning: Implications for Organizational Design. In: Toward An Integrative Framework of Organizational
Research in Organizational Behaviour, vol 1, pp. 75-123. Control. In: Accounting, Organizational and Society, vol.
10, pp. 35-50.
During, W. E. (1986), Innovatieproblematiek in kleine
industriele bedrijven (Innovation problems in Small Franke, A. (1987), Risk Analysis in Project Management.
Industrial Companies; in Dutch). Van Gorcum, Assen. In: Project Management, vol. 5 (1), pp. 29-35.
Earl, Michael J. (1989), Management Strategies for Galer, Graham and Kees van der Heijden (1992), The
Information Technology. Prentice Hall, New York. Learning Organization: How Planners Create
Organizational Learning. In: Marketing Intelligence &
Earl, Michael J. (1994), Knowledge as Strategy: Planning, vol. 10, no. 6, pp. 5-12.
Reflections on Skandia International and Shorko Films. In:
Ciborra, Claudio and Tawfik Jelassi (eds.), Strategic Galbraith, Jay ( 1973), Designing Complex Organizations.
Information Systems: A European Perspective. J. Wiley & Reading, Mass: Addison Wesley.
Sons, Chichester.
Galbraith, Jay (1977), Organization Design. Reading,
Espejo, Raul (1989), A Cyberbetic Method to Study Mass, Addison Wesley.
Organizations. In: Espejo, R. and R. Harnden (eds.), The
Viable System Model: Interpretations and Applications of Garratt, Bob (1987), The Learning Organization and The
Stafford Beer's VSM, pp. 361-382. J. Wiley & Sons, New Need for Directors Who Think. London, Fontana/Collins.
York.
Garvin, David A. (1987), Competing On the Eight
Etheredge, Lloyd S. (1980), Government Learning: An Dimensions of Quality. In: Harvard Business Review,
Overview. In: The Handbook of Political Behavior, vol. 2, (November-December), 101-109.
pp. 73-161, Plenum Press, New York.
Garvin, David A. (1993), Building a Learning
Etzioni-Halevy, Eva (1985), The Knowledge Elite and the Organization. In: Harvard Business Review, July-August,
Failure of Prophecy. George Allen & Unwin, Boston, pp. 78-91.
Sydney.
George, C. S. jr. (1972), The History of Management
Evans, James R. and William M. Lindsay (1993), The Thought. Englewood Cliffs, N.J.
Management and Control of Quality. Second Edition,
Gilad, Tamar and Benjamin Gilad (1986), Business Organizations Grow. In: Harvard Business Review, July-
Intelligence: the Quiet Revolution. In: Sloan Management August, 1972, pp. 37-46.
Review, pp. 53-61.
Grönroos, Christian (1990), Service Management and
Ginzberg, Eli and George J. Vojta (1981), The Service Marketing: Managing the Moments of Truth in Service
Sector of the US Economy. In: Scientific American, March Competition. Lexington, D.C. Heath.
1981, vol. 244, no. 3, pp. 31-39. Reprinted In Bateson: pp.
23-37. Guetzkow, Harold (1965), Communications in
Organizations. In: March, J.G. (ed.), Handbook of
Glaser, Barney G. and Anselm L. Strauss (1967), The Organizations. Rand McNally, Chicago.
Discovery of Grounded Theory: Strategies For Qualitative
Research. Aldine Publ., New York. Gurbuxani, Viyay and Seungjin Whang (1991), The
Impact of Information Systems on Organizations and
Gouldner, Alvin W. (ed.) (1950), Studies in Leadership: Markets. In: Communications of the ACM, January, 1991,
Leadership and Democratic Action. Harper, New York. vol. 34, no. 1, pp. 59-73.
Greiner, Larry E. (1972), Evolution and Revolution as Hage, J. (1965), An Axiomatic Theory of Organizations.
References 277
In: Administrative Science Quarterly, vol. 10, pp. 289-320. Changing Environments. In: Accounting, Organizations
and Society, vol. 3, pp. 47-64.
Hage, J. and M. Aiken (1967), Program Change and
Organizatonal Properties: A Comparative Analysis. In: Hedberg, Bo (1981), How Organizations Learn and
American Journal of Sociology, 1967, vol. 72, no. 5, pp. Unlearn. In: Nystrom, Paul C. and William H. Starbuck
503-519. (eds.), Handbook of Organizational Design, vol. 1,
Adapting Organizations to Their Environments. Oxford
Hamel, Gary and C.K. Prahalad (1990), The Core University Press, 1981, pp. 3-27.
Competence of the Corporation. In: Harvard Business
Review, May-June, pp. 79-91. Hedberg, Bo (1991), The Role of Information Systems in
Imaginary Organizations. In: Stamper et al. (eds.):
Hamel, Gary and C. K. Prahalad (1991), Corporate Collaborative Work, Social Communications and
Imagination and Expeditionary Marketing, Harvard Information Systems, pp. 1-8. North-Holland, Amsterdam.
Business Review, 1991, July-August, pp. 81-93.
Hedlund, Gunnar (1994), A Model of Knowledge
Hammer, Michael (1990), Rengineering Work: Don't Management and the N-Form Corporation. In: Strategic
Automate, Obliterate. In: Harvard Business Review, Management Journal, vol. 15, pp. 73-90.
July-August, pp. 104-112.
Hempel, Carl G. (1965), Aspects of Scientific Explanation
Hannan, Michael T. and John Freeman (1977), The and Other Essays in the Philosophy of Science. The Free
Population Ecology of Organizations. In: American Press, New York.
Journal of Sociology, vol. 82, pp. 929-965.
Hersey, Paul, and Ken Blanchard (1982), Management of
Hannan, Michael T. and John Freeman (1984), Structural Organizational Behavior: Utilizing Human Resources.
Inertia and Organizational Change. In: American Fourth edition. Prentice-Hall, London.
Sociological Review, 49 (April), 149-164.
Herzberg, F. (19..), Work and the Nature of Man.
Hedberg, Bo and Sten Jönsson (1978), Designing Harcourt, Brace and World, New York.
Semi-confusing Information Systems for Organizations in
Hill, Terry (1983), Production/Operations Management. Jayaratna, N. (1990), Systems Analysis: The Need for a
Englewood Cliffs, N.J., Prentice Hall. Better Understanding. In: International Journal of
Information Management, pp. 228-234.
Hofstede, Geert (1981), Management Control of Public
and Not-for-profit Activities. In: Accounting, Jelassi, M. Tawfik and Renee A. Beauclair (1987), An
Organizations and Society, vol. 6, pp.193-211. Integrated Framework for Group Decision Support
Systems. In: Information & Management, vol. 13, pp.
Hoogerwerf, A. (1984), Public Policy is Based on 143-153.
Presumptions. (In Dutch: Beleid berust op
veronderstellingen: de beleidstheorie.) In: Acta Politica, Jelinek, Mariannn (1977), Institutionalizing Innovation.
vol. XIX, pp. 493-531. D.B.A.-thesis Harvard University, 1977. Xerox University
Microfilms, Ann Harbor, Michigan.
Huber, George P. (1983), Cognitive Style as a Basis for
MIS and DSS Designs: Much Ado About Nothing? In: Jelinek, Mariann, Joseph A. Litterer and Raymond E.
Management Science, vol. 29, no. 5, pp. 567-579. Miles (1981), Control in Organizations. In: Organizations
by Design Theory and Practice, 1981, Business
Huber, George P. (1990), A Theory of the Effects of Publications, Plano, Texas.
Advance Information Technologies on Organizational
Design, Intelligence, and Decision Making. In: Academy Juran, J.M. (1964), Managerial Breakthrough: a New
of Management Review, vol. 15, pp. 47-71. Concept of the Managers Job. New York, McGraw-Hill.
Huysman, Marleen (1992), Different Perspectives on Kaydos, Will (1991), Measuring, Managing and
Organizational Learning: a Literature Survey. Maximizing Peformance: What Every Manager Needs to
Research-Memorandum 1992-1993, Faculty of Economics know about Quality and Productivity to Make Real
and Econometrics, Free University, Amsterdam. Improvements in Performance. Productivity Press,
Cambridge, Mass.
Idenburg, P.J. (1993), Four Styles of Strategy
Development. In: Long Range Planning, vol. 26, pp.132- Keen Peter G. (1991), Shaping the Future: Business
137. Design through Information Technology. Havard Business
School Press, USA.
Indik, B. P. (1964), The Relation Between Organization
Size and Supervision Ratio. In: Administrative Science Kennedy, Carol (1994), Re-Engineering: The Human
Quarterly, vol. 9. pp. 301-312. Costs and Benefits. In: Long Range Planning, vol. 27, no.
5, pp. 64-72.
Ives, Blake and Margrethe H. Olson (1984), User
Involvement and MIS Success: A Review of Research. In: Kerlinger, Fred N. (1986), Foundations of Behavioral
Management Science, vol. 30, no. 5, pp. 586-603. Research. Third edition. Holt, Rinehart and Winsten, New
York.
278 Organizational Learning and Information Systems
Theories. (In German: Organisations Theorien).
Kerr, Roger (1991), Knowledge-Based Manufacturing Kohlhammer, Stuttgart, BRD (two volumes).
Management: Applications of Artificial Intelligence to the
Effective Management of Manufacturing Companies. Kim, Daniel H. (1993), The Link between Individual and
Sydney, Addison-Wesley. Organizational Learning. In: Sloan Management Review,
Fall 1993, pp. 37-50.
Kieser, Alfred and Herbert Kubicek (1978), Organization
Kirsch, Werner and Heinz K. Klein (1977), Management Lammers, C. J. (1986), Excellent Companies as
Information Systems. Part 1: Ways of Rationalizing Organization Model. (In Dutch: De Excellente
Leadership. Part 2: To a New Taylorism. (In German: Onderneming als Organisatiemodel). In: Harvard Holland
Managementinformationssysteme. Part 1, Wege zur Review, (Herfst), 18-27.
Rationalisierung der Fuhrung; Part 2: Auf dem Weg zu
einem neuen Taylorismus). Verlag W. Kohlhammer, Land, Frank (1982), Adapting to Changing User
Stuttgart. Requirements. In: Information & Management, pp. 59-75.
Kling, Rob (1980), Social Aspects of Computing: Lawler, Edward E. and John Grant Rhode (1976),
Theoretical Perspectives in Recent Empirical Research. In: Information and Control in Organizations. Goodyear,
Computing Surveys, vol. 12, (March), pp. 61-110. Beverly Hills, C.A.
Kolb, David A. (1974), Learning and Problem Solving: On Lawrence, Paul R. and Jay W. Lorsch (1969),
Management and the Learning Process. In: David A. Kolb, Organization and Environment: Managing Differentiation
Irwin M. Rubin and James M. McIntyre, Organizational and Integration. Richard D. Irwin inc.
Psychology: a Book of Readings, second edition.
Englewood Cliffs, Prentice-Hall. Leavitt, Harold J. and Thomas L. Whisler (1958),
Management in the 1980's. In: Harvard Business Review,
Kolb, David A. (1984), Experiential Learning. Experience November-December. Reprinted in Gordon B. Davis and
as the Source of Learning and Development. Gordon C. Everest (eds.) (1985), pp. 267-274.
Prentice-Hall, Englewood Cliffs, N.J.
Lee, Soonchul and P.J. Guinan (1991), The Impact of
Kolkman, Martin (1993), Problem Articulation information Technology on Work Group Innovation and
Methodology. Thesis University of Twente. Control. In: Edward Szewczak, Coral Snodgrass and
Mehdi Khosrowpour (eds.), Management Impacts of
Kotler, Philip (1988), Marketing Management: Analyses, Information Technology: Perspectives on Organizational
Planning, Implementation and Control. Prentice Hall, Change and Growth, pp. 241-272, Idea Group Publ,
Englewood Cliffs, N.J. Harrisburg, Pennsylvania.
Kraemer, Kenneth L. and John Leslie King (1988), Lee, Allen S. (1989), A Scientific Methodology for MIS
Computer-Based Systems for Coorperative Work and Case Studies. In: MIS-Quarterly, March, 1989, pp. 33-50.
Group Decision Making. In: ACM Computing Surveys,
vol. 20. no. 2, pp.115-146. Lee, S., F.F. Courtney Jr, and R.M. O'Keefe (1992), A
System for Organizational Learning Using Cognitive
Kumar, Kuldeep (1990), Post Implementation Evaluation Maps. In: Omega International Journal of Management
of Computer-Based Information Systems: Current Science, vol. 20, no. 1, pp. 23-36.
Practices. In: Communications of the ACM, February, vol.
33, no. 2, pp. 203-212. Leege, David C. and Wayne L. Francis (1974), Political
Research: Design, Measurement, and Analysis. Basic
Lammers, C. J. (1983), Comparing Organizations: Book, New York.
Development and Relevance of Sociological Thinking
about Organizations. (In Dutch: Organisatie Leeuw, A.C.J. de (1982), Organizations, Management,
vergelijkenderwijs: Ontwikkeling en relevantie van het Analysis, Design, and Change: A Systems View. (In Dutch:
sociologisch denken over organisaties). Organisaties: Management, Analyse, Ontwerp en
Utrecht/Antwerpen. Verandering; een Systeemvisie). Van Gorcum.
March, James G. and Herbert A. Simon (1958), Mastenbroek, W.F.G. (1982), Confict Handling and
Organizations. John Wiley & Sons, Inc. Organization Development. (In Dutch: Conflicthantering
in Organisatieontwikkeling). Samsom, Alphen a/d Rijn,
March, James G. and Johan P. Olsen (1976), Ambiguity The Netherlands.
and Choice in Organizations. Universiteitsforlaget,
Bergen, Norway. Mastenbroek, W.F.G. (1987), Negotiating. (In Dutch:
Onderhandelen). Marka Pocket, Utrecht, 1984. Fifth print
Margerison, C.J. and R.G. Lewis (1979), How Work 1987.
Preferences Relate to Learning Styles. Bedfordshire, U.K.,
Management and Organisation Development Research Mayo, E. (1945), The Social Problems of an Industrial
Centre, Cranfield School of Management. Civilisation. Boston, Graduate Sehool of Business
Administration, Harvard University.
Markus, M. Lynne (1983), Power, Politics, and MIS
Implementation. In: Communications of the ACM, vol. 26 McAuliffe, Thomas P. and Carlyn S. Shamlin (1992),
(June), pp. 430-442. Critical Information Network: The next Generation of
Executive Information Systems. McAuliffe and Co.,
Markus, Lynne and Daniel Robey (1983), The Boston.
Organizational Validity of Management Information
Systems. In: Human Relations, vol. 36, pp. 203-226. McFadden, Fred R. and Jeffrey A. Hoffer (1991),
Database Management. Third edition.
Markus, M. Lynne (1984), Systems in Organizations; Benjamin/Cummings, Redwood, CA.
Bugs + Features. Pitman Publishing, Marshfield Mass.
McGregor, Douglas (1960), The Human Side of
Markus, M. Lynne and Daniel Robey (1988), Information Enterprise. McGraw-Hill, New York.
Technology and Organizational Change: Causal Structure
in Theory and Research. In: Management Science, vol. 34 McKeown, Patrick G. and Robert A. Leitch (1993),
(May), pp. 583-598. Management Information Systems: Managing with
Computers. The Dryden Press, Fort Worth.
Maslow, Abraham H. (1954), Motivation and Personality.
Harper & Row publ, New York.
Merton, R.K. (1957), Social Theory and Social Structure.
Glencoe Ill., Free Press, second edition. Mintzberg, Henry (1991), The Effective Organization:
Forces and Forms. In: Sloan Management Review, Winter,
Michels, Robert (1966), Political Parties: a Sociological pp. 54-67.
Study of the Oligarchic Tendencies of Modern Democracy.
The Free Press, New York. (Translated from German: Zur Mohrman, Susan Albers and Thomas G. Cummings
Soziologie Des Parteiwesens un der moderne Demokratie, (1989), Self-Designing Organizations: Learning How to
1925). Create High Performance. Addison-Wesley, Reading,
Mass.
Miles, Raymond E., Charles C. Snow, Alan D. Meyer and
Henry L. Coleman jr. (1978), Organizational Strategy, Morecroft, J.D.W. and J.D. Sterman (eds.)(1992),
Structure, and Process. In: Academy of Management Modelling for Learning. Special Issue of the European
Review, vol. 3, pp. 546-562. Journal of Operational Research, vol. 59, no. 1, May
1992.
Mills, Peter K and Dennis J. Moberg (1982), Perspectives
on the Technology of Service Operations. In: Academy of Morecroft, J.D.W. and K.A.J.M. van der Heijden (1992),
Management Review, 1982, vol. 7, no. 3, pp. 467-478. Modelling the Oil Producers: Capturing oil Industry
Reprinted in Bateson, pp. 151-165. knowledge in a behavioural simulation model. In:
Morecroft and Sterman (eds.) (1992) Modelling for
Mintzberg, H. (1973), The Nature of Managerial Work. Learning. Special Issue of the European Journal of
Harpe and Row, New York. Operational Research, vol. 59, no. 1, May 1992, pp. 102-
122.
Mintzberg, Henry (1976), Planning on the Left and
Managing on the Right. In: Harvard Business Review, Morgan, Gareth (1986), Images of Organization. Beverly
July/August, pp. 49-58. Hills, Sage Publications.
Mintzberg, Henry (1983), Structures in Fives: Designing Mumford, Enid (1983), Designing Human Systems.
Effective Organizations. Englewood Cliffs, N.J., Manchester Business School.
Prentice-Hall.
Nagel, Ernst (1961), The Structure of Science. Harcourt
Mintzberg, Henry (1989), Inside Our Strange World of Brace Jovanovich, New York.
Organizations. The Free Press, New York.
280 Organizational Learning and Information Systems
Niebel, Benjamin W. (1982), Motion and Time Study. Sloan Management Review, (Spring), pp. 9-18.
Seventh edition. Richard D. Irwin, Homewood, Illinois.
Nonaka, Ikujiro (1990), Managing Globalization as a self-
Nijssen, G.M. and T.A. Halpin (1989), Conceptual renewing process: experiences of Japanese MNC's. In:
Schema and Relational Database Design: A fact oriented Bartlett, Christopher A., Yves Doz and Gunnar Hedlund
approach. Prentice-Hall, New York. (eds.), Managing the Global Firm. Routledge, London and
New York.
Nolan, R. L. and A. H. J. B. Schotgerrits (1989),
Transformation in Organizations by Information Osgood, C.E., G.J. Suci and P.H. Tannenbaum (1957), The
Technology.( In Dutch: Transformatie in Organisatie Door Measurement of Meaning. University of Illinois Press.
Informatietechnologie). In: Informatie, vol. 31, pp.
991-1000. Parkinson, C. N. (1958), Parkinson's Law or the Pursuit of
Progress. Murray, London.
Nonaka, Ikujiro (1986), Toward Middle-up-down
Management: Accelerating Information Creation. In:
Peffers, Ken and Timo Saarinen (1993), Measuring the Model of Effectiveness Criteria: Towards a Competing
Business Value of IT Investments: Inferences from a Values Approach to Organizational Analysis. In:
Study of Senior Bank Executives. To appear in Journal of Management Science, vol. 29, no. 3, pp. 363-377.
Organizational Computing.
Rochfeld, Hubert, Arnold Tardieu and Rene Colletti
Pierce, J. L. and A. L. Delbecq (1977), Organization (1986), The Cherry Method (In French: La Methode
Structure, Individual Attitudes, and Innovation. In: Merise). Les Editions d'Organisation, Paris.
Academy of Management Review, 1977, vol. 2, no. 1, pp.
27-37. Rockart, John F. (1979), Chief Executives Define Their
Own Data Needs. In: Harvard Business Review,
Peters, T. (1992), Liberation Management: Necessary (March-April), pp. 81-93.
Disorganization for the Nanosecond Nineties. MacMillan,
London. Rogers, Everett M. (1964), Diffusion of Innovations. Free
Press, New York.
Peters, Thomas J. and Robert H. Waterman jr. (1982), In
Search of Excellence: Lessons from America's Best-run Rosenhead, Jonathan (ed.) (1989), Rational Analysis For
Companies. Harper & Row, New York. as Problematic World. Problem Structuring Methods For
Complexity, Uncertainty and Conflict. John Wiley & Sons.
Pettigrew, Andrew M. (1973), The Politics of
Organizational Decision-Making. Tavistock, London. Roszak, Theodor (1986), The Cult of Information. The
folklore of computers and the true art of thinking.
Popper, Karl (1963), Conjectures and Refutations: The Pantheon books, New York.
Growth of Scientific Knowledge. Routledge and Kegan
Paul, London. Saarinen, T. and F. Wijnhoven (1994), Organizational
Learning and Evaluation of Information Systems. Working
Porat, M. (1977), The Information Economy. Washington paper, Helsinki School of Economics and Business
D.C., U.S. Department of Commerce. Administration, Heslinki.
Porter, Michael E. (1985), Competitive Advantage: Schmenner, Roger W. (1986), How Can Service Business
Creating and Sustaining Superior Performance. Free Survice and Prosper? In: Sloan Management Review,
Press, New York. Spring 1986, pp. 21-32.
Porter, Michael E. (1988), Reports. In: Planning Review, Schrama (1991), Freedom of Choice in Organization
September/October. Forms. (In Dutch: Keuze Vrijheid in Organisatievormen:
Strategische keuzes rond organisatiestructuur en
Porter, Michael (1990), The Competitive Advantage of informatietechnologie bij het invoeren van een
Nations. MacMillan, London. personeelsinformatiesysteem bij grote gemeenten). Ph.D.-
thesis, University of Twente.
Quinn, James Brian (1992), Intelligent Enterprise: A
Knowledge and Service Based Paradigm for Industry. The Senge, Peter (1990a), The Fifth Discipline: The Art and
Free Press, New York. Practice of The Learning Organization. Doubleday
Currency, New York.
Quinn, Robert E. and K. Cameron (1983), Organizational
Life Cycles and Shifting Criteria of Effectiveness: Some Senge, Peter M. (1990b), The Leader's New Work:
Preliminary Evidence. In: Management Science, vol. 29, Building Learning Organizations. In: Sloan Management
no. 1, pp. 33-51. Review, Fall 1990, pp. 7-23.
Smith, Adam (1776/1975), The Wealth of Nations. Dent, Stinchcombe, Arthur L. (1959), Bureaucratic and Craft
London. Administration of Production: a Comparative Study. In:
Administrative Science Quarterly, vol. 4, pp. 168-187.
Sproull, Lee and Sara Kiesler (1988), Reducing Context
Cues: Eletronic Mail in Organizational Communcation. In Strassmann, Paul A. (1985), Information Payoff. The
Greiff, Irene (ed.), Computer-Supported Cooperative Transformation of Work in the Electronic Age. The Free
Work: A Book of Readings, pp. 683-712. Morgan Press, New York.
Kaufmann Publ. San Mateo, C.A.
Strassmann, Paul A. (1990), The Business Value of
Stamper (1973), Information in Business and Computers: An Executive's Guide. The Information
Administrative Systems. John Wiley & Sons, New York. Economics Press, New Canaan, Connecticut.
Stamper, Ronald, Karl Althouse and James Backhouse Sullivan, Jeremiah and Ikujiro Nonaka (1986), The
(1988), Measur Method For Eliciting, Analysing and Application of Organizational Learning Theory To
Specifying User Requirements. In: T. Olle, A. A. Japanese and American Management. In: Journal of
Verrijn-stuart, L. Bhabuta (eds), Computerized Assistance International Business Studies, vol. 17 (Fall), pp. 127-148.
During the Information Systems Life Cycle. Elsevier
Science Publishers. Swierenga, J. and A.F.M. Wierdsma (1990), .To a
Learning Organization: About Learning and Teaching in
Stamper, R.K. (1979), Legol: Modelling Legal Rules by Organizations. (In Dutch: Op Weg Naar een Lerende
Computer. In: Bryan Niblett (ed.) (1980), Advanced Organisatie. Over het Leren en Opleiden van
Workshop on Computer Science and Law, pp. 45-72. Organisaties). Wolters Noordhof.
Stamper, Ronald (1987), Semantics. In: Boland, R.J. and Tanenbaum, Andrew S. (1984), Structured Computer
R.A. Hischheim (eds.), Critical Issues in Information Organization. Second Edition. Prentice Hall.
Science, pp. 43-78. John Wiley & Sons, New York.
Tanenbaum, Andrew S. (1988), Computer Networks.
Stamper, R.K., P. Kerola, R. Lee and K. Lyytinen (eds.) Second edition. Englewood Cliffs, N.J.
(1991), Collaborative Work, Social Communications and
Information Systems. Proceedings of the IFIP TC8 Taylor, F. (1911), The Principles of Scientific
Working Conference on Collaborative Work, Social Management. New York, Harper & Sons.
Communications and Information Systems, COSCIS '91.
North-Holland, Amsterdam. Teich, J. (1991), Decision Support for Negotiations. Ph.D.
Stata, Ray (1989), Organizational Learning: The Key To dissertation, State University of New York Buffalo.
Management Innovation. In: Sloan Management Review,
Teich, J., H. Wallenius, M. Kuula and S. Zionts (1993), A Public Affairs. John Wiley & Sons, London.
Decision Support Approach for Negotiations with an
Application to Agricultural Income Policy Negotiations. Van Nievelt, M. C. A. (1992), Managing with Information
To appear in European Journal of Operational Research. Technology: a Decade of Wasted Money? In: Compact,
Summer 1992, pp. 15-24. A publication of KPMG.
Thompson, J.D. (1967), Organizations in Action. Social
Science Bases of Administrative Theory. McGraw-Hill, Van Schaik, F. D. J. (1988), Effectiveness of Decision
New York. Support Systems. Delft Unversity Press.
Toffler, Alvin (1985), The Adaptive Corporation. New Vennix, Jac A. M. (1990), Mental Models and Computer
York, McGraw-Hill. Models; Design and Evaluation of a Computer-based
Learning Environment For Policy-making. University of
Tyson, Kirk W.M. (1986), Business Intelligence: Putting it Nijmegen, Nijmegen Institute For Cognition Research and
all Together. Leading Edge Publ., Lombard, Illinois. Information Technology.
Urwick, L.F. (1937), Organization as a Technical Problem. Venugopal, V. and W.R.J. Baets (1994), An Integrated
In: L. Gulick and L.F. Urwick, Papers on the Science of Intelligent Support System for Organisational Learning: A
Administration. New York. Conceptual Framework. Nijenrode University Press,
Breukelen, The Netherlands.
Van der Vegt, Rein (1973), Educating and Evaluation: An
Field Experimental Study on Results of Company Wales, J. (1980), A Four-Culture Model of Collaborative
Trainings. (In Dutch: Opleiden en Evalueren: Een Learning and Problem Solving. In: Management
Veldexperimentele Studie naar Uitkomsten van een International Review, 1980, no. 4, pp. 100-111.
Bedrijfsopleiding). Boom, Meppel.
Walker, Rob (1992), Rank Xerox: Management
Van Gunsteren, Herman R. (1976), The Quest for Control: Revolution. In: Long Range Planning, vol.25, pp. 9-21.
A Critique of the Rational-Central-Rule Approach in
282 Organizational Learning and Information Systems
Walsh, James P. and Gerardo Rivera Ungson (1991),
Organizational Memory. In: Academy of Management Weiner, Norbert (1950/1967), The Human Use of Human
Review, vol. 16, pp. 57-91. Beings: Cybernetics and Society. Avon Books, New York.
Watts, D. L. (1990), Disorder and Contradiction; An Wijnhoven, Fons (1984), Political Apathy: A Study in the
Empirical Perspective On Self-Organization. In: Human Methodology of Concept Formation. (In Dutch: Politieke
Systems Management, pp. 239-248. Apathie). Unpublished M.A. thesis, Catholic University of
Nijmegen, The Netherlands.
Weber, Max (1964), The Theory of Social and Economic
Organization. Translation of Max Weber's 'Wirtschaft und Wijnhoven, Fons (1992a), Model Management by Means
Gesellschaft' by A. M. Henderson and Talcott Parsons. of Computer-based Information Systems in Managerial
The Free Press of Glencoe, Collier-MacMillan Limited. Contexts. In: Jac. A.M. (et al. eds.), Proceedings System
Dynamics 1992, Utrecht, The Netherlands.
Weick, Karl E. (1979), The Social Psychology of
Organizing. Second Edition, Random House. Wijnhoven, Fons (1992b), Impact of Decision Support
Systems On Decision-making Processes Within
Weick, Karl E. (1985), Cosmos Vs. Chaos: Sense and Organizations. A Framework For Assessment. In: Mark
Nonsense in Electronic Contexts. In: Organizational Silver (ed.), DSS-92 Transactions. The Institute of
Dynamics, (Autumn), pp. 51-64. Management Sciences.
Wijnhoven, F. (1993), CSFmatrix. Software for Soft Machine: The Future of Work and Power. Basic Books,
Sysems Policy Modelling. University of Twente, New York.
Department of Information Management.
Blau 39
Index Boesjes-Hommes .................................................33
Boisot 57, 62
"garbadge can"-model ........................................ 289 Boone 280
"I Think" 140 bounded rationality ..........................................288
Acar 70, 83 Braverman ...........................................................72
Ackoff 6, 81, 160 broad sense systems definition ...........................24
acquisition of knowledge................. 126, 141, 154 Brown 87, 88
acquisition of knowledge and MICS ............... 154 building shared vision ........................................67
action norms 71, 82, 178, 180, 196, 216, bureaucratic learning
232, 247, 253, 261, 273,........................................
280 283
action planning ................................................. 127 Burns 91
action planning systems.................................... 153 Burrell 17, 34, 49
action-outcome theories ..................................... 32 Burrows 25
adaptation of knowledge. . . . . 127, 200, 219, Bushe 108, 115, 167, 251
234, 248, 263 business awareness ............................................244
Adelman 280 business re-engineering .................. 8, 73, 113, 150
Adler 56, 114, 177, 274 business value of computers ...............................13
administration/production ratio ..................... 260 Buzzell 151
administrative organization ................................ 86 CAD/CAM ........................................................... 3
affiliating 139 Camp 150
Aggerwal 3, 256 canonical practice ...............................................87
Aiken 11 Cardboard Co ...................................................189
analytic induction ............................................... 39 Cash 3
Anderson95 causal structure ...................................................24
Ansari 21, 23, 24, 34, 138, 245 Centers of Service ...............................................65
Ansoff 45, 84 channels of communication ...............................55
Anthony 135 Chaparral Steel .................................................284
APMS 189, 193 Chemical Plant .................................................225
Aquilar 26 Chew 25, 62, 150, 168
Argote 56 Child 95
Argyris 17, 21, 26, 32, 49, 59, 60, 64, 82,90, 91, 117,classic machine
132, 146, 163, bureaucracy
170 ...............................94
Asean Brown Boveri ........................................... 65 closed control loop ...........................................138
Ashby 9 closed learning loop ................................. 153, 234
Ashworth 1 closed loop learning situations .........................141
assimilation ....................................................... 140 co-makership .......................................................96
attitude to quality ...................................... 95, 129 Coats 45
audience learning ............................................... 57 codification of knowledge ..................................57
Automatic Teller Machine ................................... 3 Cohen 289
Bateson 19, 59 Cole 114, 177, 274
Beckman 56 Coleman 71
Bedford 135 colonial industry ................................................... 5
Beer 90 command economy .............................................. 5
behavioral adaptation ......................................... 20 Common Market ..............................................209
Bell 1, 2, 6 communication systems......................................16
Bell Company ..................................................... 60 company administration.....................................93
Berger 58 comparative analysis .........................................150
Berliner 145 competency trap ..................................................62
Bernstein 35 complexity ................... 79, 90, 102, 166, 169, 173
Blanchard ................................... 10, 117, 120, 259 concreta 33
References 285
De afgelopen jaren is een discussie gevoerd over het belang van informatiesystemen
voor ondernemingen (Earl, 1989; Strassmann, 1990). Hierbij stond meestal de vraag
centraal wat informatietechnologische investeringen financieel opleveren. Deze vraag
veroorzaakt verwarring, ten eerste omdat informatie-technologie een zeer ruim begrip
is en ten tweede omdat voor sommige toepassingen van informatietechnologie
kosten-baten berekeningen bijna onuitvoerbaar zijn. Het tweede punt is in dit
onderzoek nader bestudeerd voor de zogenaamde management-informatiesystemen,
in het bijzonder het type management-rapportagesystemen. Dit soort systemen is
reeds veelvuldig ingevoerd in bedrijven, met vaak teleurstellende resultaten. Zij laten
zich moeilijk beoordelen op kosten en baten aangezien zij niet primair de
verandering van kostbare productieprocessen dienen, maar ten dienste staan van de
verandering van intellectuele vermogens van managers. Dit laatst is moeilijk direct te
relateren aan verbetering van kosten-baten verhoudingen van informatiesystemen,
maar is wel essentieel voor de effectiviteit van organisaties. Vanuit deze
probleemformulering is gekozen om management-informatiesystemen vanuit een
organisatorisch leerperspectief te evalueren.
Vervolgens is na gegaan hoe deze probleemstelling geconcretiseerd kan worden.
Gekozen is voor een eerste beperking van het onderzoek tot het gebied van Prestatie-
Evaluerende en Controlerende Systemen (een type management-rapportagesysteem dat in
het Engels Monitoring Information and Control System heet, en hier verder MICS
wordt genoemd). De reden hiervoor is dat met name deze systemen in het verleden
veel kritiek hebben gekregen van organisatie-deskundigen op grond van mogelijk
negatieve effecten van deze systemen op organisatie-leerprocessen. Deze argumenten
zijn echter gebaseerd op slechts summier geformuleerde theorie en empirisch
nauwelijks onderzocht. Voor zover studies op dit gebied bekend zijn (Jelinek, 1979 en
Lee and Guinan, 1991) wijzen de resultaten juist op een positieve bijdrage van MICS
op organisatorisch leren. Deze resultaten worden echter bestreden door enkele
vooraanstaande organisatiedeskundigen (o.a. Mintzberg, 1989 en Argyris, 1971).
Volgens Markus (1983) en Markus en Robey (1988) spelen organisatiecondities een
essentiële rol bij het totstandkomen van positieve of negatieve impact van
informatietechnologie. Het methodologische probleem dat zich dan voordoet is dat de
grens tussen informatiesysteem en organisatie diffuus wordt, waardoor geen
uitspraken meer mogelijk worden over de impact van MICS op organisaties en vice
versa. De gekozen oplossing is om het begrip MICS op te vatten als een set van
formele normen volgens welke informatie zou moeten worden opgeslagen, verspreid en
veredeld. Daarnaast kennen organisaties ook informele normen waarin het MICS is
Samenvatting 293
ingebed, en die behalve bepalend zijn voor de effectiviteit van het MICS, tevens de
leerprocessen in een organisatie vormgeven en sturen.
De organisatieliteratuur levert door de onderkenning van organisatietypen, logische
combinaties van organisatienormen. Door een vergelijking van organisatietypen kan
per organisatietype de invloed van MICS op leren vastgesteld worden. Bij de definitie
van deze organisatietypen is gebruik gemaakt van de configuraties zoals door Mintzberg
beschreven op basis van zijn synopsis van de organisatie-literatuur tot eind jaren
zeventig. Ter vereenvoudiging van het onderzoeksontwerp is alleen gekozen voor de
bestudering van de machine bureaucratie, een configuratie met een simpele en stabiele
omgeving. Hiermee is de tweede beperking van het onderzoeksdomein gegeven. De
keuze voor deze configuratie is ingegeven op grond van de volgende redenen. Ten
eerste wordt de machine bureaucratie door velen beschreven als het prototype van
een slecht lerende organisatie (Argyris, 1971; Senge, 1990). Bestudering van dit
organisatietype moet daarom veel informatie opleveren over leerproblemen in
organisaties. Ten tweede is dit organisatie-type gekozen omdat de omgevingsdynamiek
en -complexiteit van deze organisaties de laatste decennia aanzienlijk is toegenomen,
wat vraagt om verbetering van het organisatorische leervermogen. Hierbij kon ook
aangesluiting worden gevonden bij studies over 'lean'-productie, een machine
bureaucratie-type dat onder toenemende dynamiek en complexiteit is
getransformeerd tot een zeer efficiënte, flexibele en hoge kwaliteit genererende
organisatie (Womack e.a., 1990). De derde reden om machine bureaucratieën te
bestuderen, is gelegen in het feit dat deze organisaties te groot zijn om alleen een
informeel MICS te hebben, waardoor duidelijker de invloed van MICS kan worden
waargenomen.
Dit onderzoek concentreert zich aldus om drie hoofdvariabelen: Machine
Bureaucratieën, MICS en Organisatorisch leren. De eerste twee variabelen zijn reeds
duidelijk beschreven in de bestaande organisatiekundige en informatiekundige
literatuur. De derde variabele is echter bijzonder onduidelijk. Het was daarom
noodzakelijk een uitgebreide literatuurstudie uit te voeren naar het begrip
organisatorisch leren, alvorens criteria voor evaluatie van MICS te formuleren. De
literatuur over organisatorisch leren heeft de gedachte van Argyris en Schön dat
organisatorisch leren uit drie hoofdprocessen aanvaard. Het eerste proces noemen
Argyris en Schön 'single-loop'-leren: het creëren en evalueren van feedback-
informatie, met als doel om bestaande transformatieprocessen te verbeteren. Hierbij
worden de basisdoelen van dat proces niet ter discussie gesteld. Het tweede leerproces
wordt 'double-loop'-leren genoemd: activiteiten met als doel om de basisdoelen of
assumpties van organisatorische processen te evalueren, te vervangen en te
vernieuwen. Het derde leerproces, 'deutero leren', heeft als doel om de organisatorie
(cultuur, managementstijl, informatiesystemen etc.) dusdanig te verbeteren dat haar
leervermogen wordt vergroot. In deze studie is deze terminologie geaccepteerd. De
onderliggende cybernetische gedachte is hiermee tevens aangenomen, met de
aantekening dat organisatorisch leren plaatsvindt in een politieke en sociale context
294 Organizational Learning and Information Systems
die bepalend is voor het feitelijke leergedrag en de acceptatie van haar resultaten. Tot
slot is een nadere operationalisering van de begrippen gemaakt, door concrete
activiteiten te benoemen. Deze zijn weergegeven in de onderstaande tabel.
Leerproces Leeractiviteit
Overeenkomstig de gedachte dat leren niet alleen een proces is maar ook betrekking
heeft op onderwerpen waarover geleerd moet worden (Kolb, 1984), zijn de volgende
leervelden op basis van Quinn en Rohrbaugh (1983) geïndentificeerd:
transformatieprocessen, mensen, producten en markten.
Het Deutero leerproces is niet bestudeerd in deze studie. Wel is het belang
onderkend van de bepaling van organisatorische leernormen, die bepalend zijn voor de
wijze waarop Single-loop en Double-loop leerprocessen plaatsvinden. Deze
leernormen dienen afgestemd te zijn op organisatorische leerbehoeften, welke
beschreven worden in termen van een combinatie van organisatorische complexiteit
en dynamiek.
Het doel van het onderzoek is het leveren van een manier waarop MICS-systemen
kunnen worden beoordeeld naar hun waarde voor organisatorisch leren. Hiervoor is
het noodzakelijk te beschikken over een referentiekader waarmee bepaald kan
worden welke variabelen geobserveerd moeten worden, en hoe uit observaties
conclusies getrokken kunnen worden. Een gevalideerd referentiekader, bestaande uit
een theorie over de relatie tussen MICS, organisatorisch leren, en machine
bureaucratieën, ontbreekt echter in zijn geheel in de literatuur. De onderzoeker heeft
daarom zelf zo'n theorie geconstrueerd in twee stappen, namelijk: (1) Een
Samenvatting 295
(mensen van de bedrijfsleiding), die slankheid als een ideaal beschreven dat de
onderneming al zou hebben bereikt. Deze verwarring tussen ideaal en actuele
toestand werd helaas pas later ontdekt. Wel bleken beide gevallen achteraf toch zeer
waardevol te zijn, aangezien hierdoor stellingen te formuleren zijn over de overgang
van klassiek naar slank, een leerproces in zichzelf. De onderzoeker was aldus
genoodzaakt een extra gevalstudie (5) uit te voeren naar een bedrijf waarvan met meer
zekerheid gezegd kon worden dat het aan de eisen van slankheid voldeed.
De resultaten per geval worden hieronder kort gegeven.
Geval 1: Een kartonfabricant (Cardboard Co. genoemd) in een stabiele en simpele
omgeving (met een lage leerbehoefte).
Een plakpapierregistratiesysteem (APMS) werd in deze organisatie gebruikt in twee
van haar locaties. In de ene locatie was het succes groot en in de andere locatie
gering. De oorzaak van dit feit lag niet in het systeem (dat was in beide gevallen
identiek), maar in het vermogen van operationele managers om de gegevens uit het
systeem te interpreteren, en op grond daarvan gerichte en cumulatieve verbeteringen
voor te stellen en in te voeren. Essentieel voor het vormen van voorstellen voor
verbetering was kennelijk een goed begrip van hoe in de practijk met het plakpapier
wordt omgegaan. In locatie 1 (het succesgeval) was de operationele manager zelf
jarenlang werkzaam geweest op de werkvloer en kende de (informele)
managementtheorie daardoor van binnenuit. In locatie 2 was de operationele
manager administratief geschoold, en had geen ervaring in het productieproces.
Het succes van het single-loop leerproces in locatie 1 is daardoor meer te verklaren
uit de wijzen waarop gegevens en kennis in actie worden omgezet (actienormen) dan
uit het informatiesysteem en andere leernormen. De bijdrage van het MICS (APMS)
was zeer groot bij het single-loop leren. Locatie 1 bijvoorbeeld, verminderde in twee
jaar tijd het plakpapierverlies van 16% naar 8%, wat een kosten besparing van 1,4
miljoen US dollars per jaar met zich mee heeft gebracht.
Geval 2: Een commerciële Europese bank (The Bank).
Deze bank begeeft zich in een complexere omgeving dan geval 1 (veel producten en
veel verschillende klanten), en de dynamiek vertoont een stijgende lijn (met name
door de toename van concurrentie en verdere liberalisering van de markt). Hierdoor
heeft de bank, tegen onze aanvankelijke verwachting in, een hoge leerbehoefte (score
4 op een schaal van 1 tot en met 4). De bank heeft een zeer diverse en omvangrijke
verzameling van MICS-systemen. Deze systemen zijn echter onsamenhangend,
gebruiksonvriendelijk en hun gegevens krijgen een geringe aandacht bij het
management. Het leerproces vindt daarom minder direct via de systemen plaats, en is
sterk gecentraliseerd bij het hoofdkantoor. Leren vanuit de decentrale kantoren vindt
zelden plaats, en is een moeizame aangelegenheid vanwege de ingewikkelde
communicatiewegen die daarvoor afgelegd dienen te worden. De ineffectiviteit van
deze situatie wordt door het management onderkend, en men is nu bezig met het
veranderen van de procedurele en verantwoordelijksnormen. Tegelijkertijd is een
herautomatiseringsproject gestart dat moet leiden tot een betere samenhang in
Samenvatting 297
Onderzoeksvragen en antwoorden
onderzocht, bleek dat deze toch heel anders leert dan de klassieke machine
bureaucratie. De slanke organisatie heeft een totaal-leervermogen ontwikkeld
(middle-up-down). De drijvende kracht achter het leren in zo'n organisatie zijn
de policy en identity norms. In de klassieke organisaties wordt leren meestal
gedelegeerd aan specialisten en staat vaak los van alledaagse werkprocessen.
Enkele organisaties toonden een overgangstrend, waardoor een aantal leerstadia
konden worden beschreven. De volgende leerstadia kan een organisatie
ondergaan in het leersproces: Gevecht tegen de Chaos, Bureaucratisch Leren,
Expert Leren, Verspreid Leren en tot slot Totaal Leren. Deze indeling wordt
geschouwd als een aantal fasen in een deutero leerproces. Bij iedere fase
behoren andere leernormen.
4. Wat is de invloed van MICS op organisatorisch leren in de machine
bureaucratische context? De invloed van MICS op organisatorisch leren staat
sterk onder invloed van de managementtheorie die wordt gebruikt en de
bestaande leernomen. De managementtheorie vormt enerzijds een kader
waaruit de objecten voor gegevensverzamelingen en -verwerkingen zijn afgeleid.
Anderzijds zijn de managementtheorieën een referentiekader waardoor de
gegevens betekenis kunnen krijgen. In de meeste gevallen vormen de objecten
voor MICS slechts een beperkte afbeelding van de totale (impliciete)
managementtheorie. De niet afgebeelde elementen (tacit knowledge genoemd)
zijn evenwel net zo belangrijk voor de interpretatie van andere
informatiebronnen (Hedlund, 1994). Belangrijker is echter de bevinding dat
managementtheorieën niet altijd door iedereen gedeeld worden, en dat
managementtheorieën een incompatibele relatie tot elkaar kunnen hebben. Dit
doet zich met name voor in de klassieke machine bureaucratieën, waar
verschillende afdelingen verschillende theorieën hanteren, en elkaar soms niet
begrijpen en zodoende niet tot een synthese kunnen komen. MICS wordt dan
alleen gebruikt voor single-loop leerprocessen binnen de normen en het kader
van de afzonderlijke afdeling (suboptimalisatie), of als een wapen voor de
politieke strijd met andere organisatie-onderdelen (dialectisch gebruik). Binnen
slanke organisaties wordt minder nadruk gelegd op
afdelingsverantwoordelijkheden, en bestaat er een duidelijke bedrijfsfilosofie,
die door iedereen wordt geaccepteerd (shared mental model). Hierdoor heeft
het MICS de mogelijkheid om fundamentele organisatieproblemen te
detecteren en als communicatiemiddel te fungeren tussen afdelingen. Indien
belangentegenstellingen zich voordoen, wordt binnen de filosofie naar een
gezamenlijke oplossing gezocht. De resultaten van deze tweede benadering
komen niet alleen op het cognitieve vlak tot uitdrukking, maar ook in een
betere acceptatie en implementatie van nieuwe inzichten, waardoor de kloof
tussen theorie en practijk aanzienlijk kleiner is.
5. Hoe kan men de impact van MICS in de machine bureaucratische omgeving
bepalen? Hiervoor zijn twee meetinstrumenten ontwikkeld, één voor Single-
300 Organizational Learning and Information Systems
loop leren en één voor Double-loop leren, waarbij de scores worden bepaald
door scores op de snijpunten van de dimensies leeractiviteiten en leervelden.
Indien leren plaatsvindt op een cel wordt een 1 gescoord, anders 0. Voor het
instrument voor meting Single-loop-leer-inspanningen zijn vier leeractiviteiten
onderscheiden (aanpassing, opslag, verspreiding en (her-)gebruik) en vier
leervelden gemeten (mensen, producten, processen en markten). De waarde van
MICS op Single-loop leren wordt bepaald door per cel aan te geven of de
bijdrage van MICS positief (+1), neutraal (0), of negatief (-1) (belemmering van
leren) is. De maximumscore is aldus +16, en de minimumscore is -16. Voor
Double-loop leren zijn slechts twee activiteiten gemeten (theorie-ontwikkeling
en theorie-verwijdering). Aangezien dezelfde leervelden van toepassing zijn als
bij Single-loop leren, kan hierdoor een maximale score van +8 en een minimale
score van -8 worden gemeten. In dit onderzoek is nergens een geval van een
negatieve score gevonden, waardoor de frekwent geuitte stelling dat MICS een
belemmering is voor het Double-loop leren (stelling S16) moet worden
verworpen. Opmerkelijk was ook dat in de lean case, de scores op Single-loop
leren zeer hoog waren, maar de scores op Double-loop leren niet hoger waren
dan bij de klassieke organisaties. De verklaring hiervoor moet worden gezocht
in de beperkte leerverantwoordelijkheden die de bestudeerde 'lean' organisatie
van het moederbedrijf had gekregen (stelling S27).
Samenvatting 301
302 Organizational Learning and Information Systems