The Usage of System Dynamics in Organizational Interventions

Birgitte Snabe
The Usage of System Dynamics

in Organizational Interventions
WIRTSCHAFTSWISSENSCHAFT
Birgitte Snabe
The Usage of
System Dynamics
in Organizational
Interventions
A Participative Modeling Approach
Supporting Change Management Efforts
With a foreword by Prof. Dr. Peter Milling
Deutscher Universitäts-Verlag
Bibliografische Information Der Deutschen Nationalbibliothek
Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der
Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über
<http://dnb.d-nb.de> abrufbar.
Dissertation Universität Mannheim, 2006
D 17
flage Dezember 1997

1. Auflage März 2007
Alle Rechte vorbehalten
© Deutscher Universitäts-Verlag | GWV Fachverlage GmbH, Wiesbaden 2007
Lektorat: Brigitte Siegel / Anita Wilke
Der Deutsche Universitäts-Verlag ist ein Unternehmen von Springer Science+Business Media.
www.duv.de
Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt.

Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes
ist ohne Zustimmung des Verlags unzulässig und strafbar. Das gilt insbe-
sondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die
Einspeicherung und Verarbeitung in elektronischen Systemen.
Die Wiedergabe von Gebrauchsnamen, Handelsnamen, Warenbezeichnungen usw. in diesem

Werk berechtigt auch ohne besondere Kennzeichnung nicht zu der Annahme, dass solche
Namen im Sinne der Warenzeichen- und Markenschutz-Gesetzgebung als frei zu betrachten
wären und daher von jedermann benutzt werden dürften.
Umschlaggestaltung: Regine Zimmer, Dipl.-Designerin, Frankfurt/Main
Gedruckt auf säurefreiem und chlorfrei gebleichtem Papier
Printed in Germany
ISBN 978-3-8350-0711-6
Kathrine
and
Nicolaj
Foreword
Internationalization and globalization are major forces for companies to
change their organizational structures and processes fundamentally. To
master the associated problems, profound and well planned procedures are
indispensable, a task which is referred to as change management, and which
has to take into account both structural and dynamic aspects. Especially
interventions in the area of human resource management lead to manifold
repercussions—intended and unintended, enhancing or interfering with the
original intentions. Birgitte Snabe investigates in her dissertation if and to
what extent System Dynamics can be helpful to design organizational
interventions and to examine and evaluate in a next step which particular
actions offer adequate problem solutions.
In a distinction from the permanent organizational adaptation to a
changing environment, the author understands organizational interventions
as discrete and fundamental changes to the company’s structures and
processes. The management of organizational interventions consists of two
interdependent cycles: Problem formulation, analysis and solution (what to
do) on the one hand, and the resulting actions to make the change happen
(how to do it) on the other. It is a central hypothesis of the investigation that
the implementation of solutions to strategic problems often presents larger
challenges than the development of the solution itself. The methodological
support for effective implementation processes is the core topic of the
dissertation.
Following an ‘action research’ approach, a delicate and far reaching
personnel decision in a large corporation was investigated and is discussed.
System Dynamics uses participative model building since about 1990 and
offers the prerequisites for mapping contexts which are difficult to quantify.
The author presents the development of a complex system model and the
implementation of its recommendations both in the practical steps of the
concrete case under investigation and in the abstract form necessary for
scientific analysis. She demonstrates the power of the selected modelling
approach and points out how group dynamics lead to the integration of
VIII Foreword
initially individual objectives and strategies into a generally accepted

process model – i.e. into a “shared mental model”.
Mrs. Snabe’s work skillfully combines theoretical considerations and
aspects of the practical implementation. The propositions about organi-
zational interventions developed in the conceptual parts of the study are
tested in a rigorous – even though not extensive and representative –
practical setting. Their viability is shown in the context of top-management
decision making.
Peter Milling
Preface
The managers below the top executives in large corporate organizations are
often placed in the challenging situation of implementing other people’s
ideas. Top executives will frequently launch strategic initiatives, and expect
the managers at lower levels to act as change leaders even though they have
often played only a small or no part at all in the groping strategy forming
process where the strategic initiative has its origin. Consequently, there is a
need for learning processes that focus on the transfer of the insights and
reasoning behind the decision, as well as supporting the refining of
implementation plans. Furthermore, the processes should allow iterations
with top executives, with the dual objective of adjusting the strategic
initiative according to implementation issues and giving the managers
responsible for implementation true influence on the entire process. To a
great extent, this dissertation addresses the process of transfer of insights
and ownership as well as the operationalization of strategic initiatives and
other change projects. The main topic is the usage of system dynamics
modeling in organizational interventions in general, and specifically the use
of system dynamics modeling for the purposes of change management. The
first two chapters mainly discuss organizational interventions and the use of
modeling in decision-making and policy forming processes, which is the
predominant application of system dynamics. The last three chapters
concentrate on a rather specific application of system dynamics: modeling in
a change management context. Change management dedicated application of
system dynamics builds upon the theories and methods of system dynamics
in a decision-making and policy-forming context, but aims at the transfer of
insights and ownership from decision-makers to implementers, as well as
refining and aligning cross-organizational implementation plans.
Writing this doctoral dissertation has been an interesting journey for
me. I have enjoyed the opportunity to take the time to go into depth with the
literature especially from the disciplines of system dynamics and
organizational psychology. Coming directly from 10 years of management
consulting, the academic experience has been of great personal and
X Preface
educational value to me. I would therefore like to sincerely thank first of all
Professor Dr. Peter Milling for supporting me on my journey. He and
assistant Professor Dr. Andreas Größler have guided me with great patience
through the learning process – helping me to adjust my normative and
solution-oriented worldview from the consulting world to also embrace
scientific and academic viewpoints. I would also like to thank all my
doctoral colleagues at ‘Industrieseminars’ for the weekly discussions at the
doctoral seminars, and Markus Salge and Dr. Nadine Schieritz especially for
always volunteering to finding literature and discussing modeling issues.
Furthermore, I want to thank my good friend since early childhood, Kirstine
Munk, who has been struggling with her own dissertation at the same time as
me. Although our subjects are very different, we have had many and
interesting discussions on a wide variety of issues including theories of
science, social constructivism, aesthetic in consultations and workshops, and
using cognitive frameworks (being both models and horoscopes!) to reduce
personal barriers for involvement and honesty in discussions. Last but not
least, I want to thank my husband, Jim Hagemann Snabe. As well as
receiving personal support in many ways, I have also been very privileged to
be able to draw on his extensive business experience and conceptual
capabilities.
Birgitte Snabe
List of Contents
Foreword .............................................................................................. VII
Preface .................................................................................................. IX
List of Figures ...................................................................................... XV
List of Tables..................................................................................... XVII
A. Organizational Intervention Skills as Corporate Competence........ 1

I. The Need for and the Challenges of
Organizational Interventions ......................................................... 1
II. Foundation and Strategies for Planned Change Interventions....... 10
III. The Usage of System Dynamics Modeling

in Organizational Interventions ................................................... 15
B. Conceptual Foundation for the Usage of System Dynamics ......... 27

I. The Usage and Utility of Modeling in Decision-Making .............. 27
II. Cognitive and Behavioral Rationale for the
Usage of System Dynamics ......................................................... 35
1. Individual Learning and Change of Behavior in a
Complex and Dynamic Environment ...................................... 35
2. Establishing Group Consensus by Sharing Mental Models...... 46
3. Enhancing Organizational Learning through System
Thinking Experience and Double-Loop Learning ................... 51
XII List of Contents
III. The Development Process of System Dynamics Models

in Corporations ........................................................................... 56
1. Decompositions and Iterations in Model Development ........... 58
a. Problem Definition and System Conceptualization............. 58
b. Model Formulation and Testing......................................... 62
c. Policy Formulation and Implementation ............................ 66
2. Designing System Dynamics Modeling-Based Interventions... 68
a. Experimentation-Based Learning Cycles ........................... 68
b. Knowledge Acquisition in Modeling Projects .................... 71
c. Designing Participative Modeling Interventions ................ 73
d. Facilitation of Participative Modeling Interventions .......... 77
C. A Case Study Using Participative System Dynamics Modeling

in the Implementation of a Sensitive Change Project .................... 83
I. Research Considerations for the Case Study Application ............. 83
II. Case Study: Refining and Implementing a Location Strategy ....... 89

1. The Problem and its Context .................................................. 89
2. Intervention Process............................................................... 93
3. The Model and Selected Simulations...................................... 98
III. Evaluation of the Case Study .....................................................108

1. A Framework for Evaluating the Effectiveness and
Efficiency of the Case Study .................................................108
2. Conclusions on Case Study Effectiveness and Efficiency.......116
List of Contents XIII
D. The Usage and Utility of Participative Modeling

in Change Management .................................................................119
I. Context Factors Relevant for Deciding on

Usage of Modeling in Change Management ...............................120
II. Process Considerations ..............................................................128

1. Business Objectives and Targets Directing and
Framing the Intervention .......................................................131
2. Structured Development of Change Leaders ..........................135
3. Designing the Change Process...............................................140
4. Facilitation of modeling and simulation sessions ...................148
III. Outcomes of Participative Modeling Efforts in the

Implementation of Change Programs..........................................155
1. Modeling and simulation as a tool for transfering insights
and ownership from decision-makers to implementers ...........155
2. Refining and Aligning Implementation Plans
Through Scenario Simulation ................................................157
3. Organizational Learning in Change Management
Oriented Modeling ................................................................158
E. Targeted Participative Modeling in Change Management...........161
Appendices ...........................................................................................165
Bibliography.........................................................................................201
List of Figures
Figure A-1: The Basic Model of Corporations ...................................... 2
Figure A-2: Model of stages of problem solving ................................... 5
Figure A-3a: Diagnostics and decision-making (cycle I) ......................... 6
Figure A-3b: Change management (cycle II) ........................................... 6
Figure A-4: Goal-seeking system ........................................................ 16
Figure B-1: Accumulative levels of models in the usage of system

dynamics ......................................................................... 30
Figure B-2: Limited linear perception of system ................................. 36
Figure B-3: Theory of planned behavior ............................................. 43
Figure B-4: Different limited linear perceptions of a system ............... 47
Figure B-5: Mental models as instruments between actual systems

and formal models ........................................................... 51
Figure B-6: The basic structure of organizational learning .................. 54
Figure B-7: Formal models supporting organizational learning ........... 56
Figure B-8: The learning cycle for learning labs ................................. 70
Figure B-9: Mental database and decreasing content

of written and numerical databases................................... 71
Figure B-10: Maps, frameworks and microworlds................................. 74
Figure B-11: A model of communication .............................................. 79
Figure B-12: The Wallow Curve at work .............................................. 81

XVI List of Figures
Figure C-1: The reinforcing growth loop underlying the

intervention ..................................................................... 91
Figure C-2: Intervention process as communicated in the project ........ 94
Figure C-3: The location strategy model ............................................. 99
Figure C-4: Fraction of employees in low-cost countries compared

to total number of employees in the division ................. 104
Figure C-5: Development in cost-index for an average productive

unit (e.g. cost for one employee for a fixed period) ....... 105
Figure C-6: Development of productivity index for an average unit

(e.g. output/month/unit) ................................................. 105
Figure D-1: Conceptual modeling steps in the case study .................. 128
Figure D-2: Four antecedent processes in organizational

interventions.................................................................. 144
Appendix D, Figure 1: The preliminary model in the case study....……. 185
Appendix E, Figure 1: Simulation run of adjusted model

(avoiding rate-on-rate modeling)………….……..187
Appendix E, Figure 2: Model without rate-on-rate modeling……...…… 188

List of Tables
Table C-1: Roles and responsibilities as defined in the project........…. 95
Table C-2a: Main sources for evaluation of outcomes on

Individual level…………………………………….………... 111
Table C-2b: Main sources for evaluation of outcomes on group level…… 112
Table C-2c: Main sources for evaluation of outcomes

on organization level………………………………………… 113
Table C-2d: Main sources for evaluation of system dynamics

compared to other approaches……...……………………….. 114
Table C-2e: Main sources for evaluation of the usage of system dynamics
in a change management context……………………………. 115
Table C-3: Questionnaire results……..………………………….……. 118
Table D-1: Political characteristics of situations in terms of the

issues of interest, conflict, and power………………..…... 123
Table D-2: Generic symptoms of change resistance..………………… 142
Appendix A, Table 1: Parameters relevant to high-cost locations……… 167
Appendix A, Table 2: Parameters relevant to low-cost locations………. 168
Appendix A, Table 3: Parameters mainly relevant for transfer of tasks

and build-up of employees in low-cost locations………… 169
.
Appendix B, Table 1: Main equations influencing stock levels………… 171
Appendix B, Table 2: Main equations influencing production/month….. 172

A. Organizational Intervention Skills as Corporate
Competence
I. The Need for and the Challenges of Organizational

Interventions
Finding ways to lead and develop organizations is a constant quest seeking to

ensure competitiveness in a changing and dynamic world, which is well
illustrated by Forrester’s words calling change “the essence of the manager’s
environment.”1 Furthermore, industries are typically facing shorter changes
cycles in new technologies, competition, value chain, environmental factors, and
customer demands, resulting in an increased need for effectiveness and efficiency
in organizational change processes.2 The changes constitute challenges
representing both threats and new opportunities for the individual business
organization, putting pressure on its ability to learn and transform.3 Organizations
change in various ways: in organic, incremental processes of adapting to
changing environments or in more abrupt organizational interventions.4 The latter
way is the focus of this dissertation.
1
Forrester, Jay W.: Industrial Dynamics, Cambridge, 1961, p. 1.
2
See Kotter, John P: Leading Change, Boston, 1996, p. 18; Fine, Charles H.:
“Clockspeed-based strategies for Supply Chain Design”, Production and
Operation Management, Vol. 9, No. 3, 2000, p. 213; Brown, John Seely:
“Research That Reinvents the Corporation”, Harvard Business Review, August
2002, p. 105.
3
In de Geus, Arie P.: “Planning as Learning”, Harvard Business Review, Vol. 66,
No. 2, March-April 1988, p. 71, it is proposed that ”the ability to learn faster
than competitors may be the only sustainable competitive advantage.“
4
The Japanese concept of Kaizen is an example of a continued process
improvement focus, whereas the western world typically is more oriented
towards innovation- and result-oriented thinking, see Imai, Masaaki: Kaizen:
Der Schlüssel zum Erfolg der Japaner im Wettbewerb, German translation,
München, 1992, p. 15.
2 A. Organizational Intervention Skills as Corporate Competence
Discussions of the concepts of ‘organizations’ and ‘interventions’ cover a

wide range of literature from the sciences of management, the social sciences and
system sciences. Taking a departure from the field of cybernetics, Takahara
offers a rather generic definition of an organization as “a complex system of
interconnected human and nonliving machines; and it is formed for a purpose, to
achieve a certain goal.”5 In line with this definition, Milling has described the
basic model of corporations as a goal-seeking input-output feedback system:
Managing Stratum
Goals
Interventions Feedback
Resource Achieved
Causal Stratum output
input
Figure A-1: The Basic Model of Corporations 6
In this basic model of corporations, two different, but interrelated, conceptual

processes form the basic structure: the causal stratum, which is the operative
5
Takahara, Yasuhiko: “A Formal Model of Organization”, in Takahashi, Singo,
Kyoichi Kijima and Ryo Sato (eds.): Applied General Systems Research on
Organizations, Tokyo, 2004, p. 3.
6
Own translation of figure in Milling, Peter: Systemtheoretische Grundlagen zur
Plannung der Unternehmenspolitik, Berlin, 1981, p. 17. It should be noted that
the original figure uses the German term “Führungsstratum” (translated to
Managing Stratum) which is a broader term also encompassing the meaning of
leading, steering, controlling. This model is chosen due to its abstraction level
suitable to illustrate the concept of interventions. Takahara offers a more
detailed basic model of organizations decomposing the operational level (the
causal stratum), which inherent has a stronger focus on internal structures and
coordination challenges; see Takahara, Yasuhiko: A Formal Model of
Organization, in Takahashi, Singo, Kyoichi Kijima and Ryo Sato (eds.): Applied
General Systems Research on Organizations, Tokyo, 2004, pp. 10—13.
A. Organizational Intervention Skills as Corporate Competence 3
domain producing the output, and the higher-level managing stratum, which
includes information processing to goal compliant forming and controlling of the
causal stratum.7
In the social-psychological field of science, Argyris describes
organizational interventions from a task point of view stating: “the
interventionist’s primary tasks are to generate valid information, to help the client
system make informed and responsible choices, and to develop internal
commitment to these choices”.8 The terms ‘the interventionist’ and ‘the client’
are often used in intervention literature.9 Although disagreement exists with
regards to the importance of independence of the system and the intervener, the
contemporary literature focusing on organizational development typically sees
organizational interventions as embracing both change processes with and
without the use of external interventionists.10 The client system can, in the social-
psychological field of science, be an individual, a group of people or an
organization. At these three levels, behavioral changes are largely explained with
7
Milling, Peter: Systemtheoretische Grundlagen zur Plannung der Unternehmens-
politik, Berlin, 1981, p. 18. Milling later decomposes the managing stratum into
four levels: the normative level (formulation of long-term goals), the structuring
level (determination of the basic structures), the adaptive level (specification of
change programs) and the operative level (selection of actions), p. 20.
8
Argyris, Chris: Interventions Theory and Method – A Behavioural Science
View, Reading, Massachusetts, 1970, p. 21.
9
Cummings, Thomas G. and Christopher G. Worley: Organizational Development
and Change, Ohio, 2001, p. 142, describe the term intervention as “sequenced
planned actions or events intended to help an organization increase its
effectiveness. Interventions purposely disrupt the status quo; they are
deliberately attempts to change an organization or sub-unit towards a different
and more effective state.” Linguistic, the term ‘intervention’ indicates, that a
party is proactively doing something to change the system. This is also seen in
fields like economy (state interventions) and foreign affairs (armed conflicts).
10
In Argyris, Chris: Interventions Theory and Method – A behavioural Science
View, Reading, Massachusetts, 1970, p. 15, the importance of independency
between the client and the interventionist is stressed, whereas in Schein, Edgar
H.: Process Consultation, Boston, 2000, part II, p. 35, (collection of work first
published in the 1960’s), it is argued that both external consultants as well as
managers from within the company can serve the role of the interventionist.
Recent textbooks generally support the latter view.
the same mechanisms, having the change of attitudes and intentions of

individuals as a central element.11
Offering an additional perspective from the social-psychological field of
science, Schein identifies three basic models of organizational interventions as
being (1) The Purchase of Expertise Model, (2) The Doctor-Patient Model, and
(3) The Process Consultation Model.12 In the Purchase of Expertise Model, the
role of the interventionist is to provide recommendations based on expert
information and services, whereas the Doctor-Patient Model starts with an
investigation of ‘symptoms’ followed by analyses and recommendations made by
the interventionist. In both The Purchase of Expertise Model and The Doctor-
Patient Model, the primary objective is the identification of ‘the solution.’ The
third intervention model, The Process Consultation Model, focuses rather on
strengthening the organization’s own ability to identify the core problem in
general, as well as finding a suitable solution in the specific situation.
Furthermore, the Process Consultation Model focuses strongly on stakeholder
involvement in the search for sustainable change, as was also seen in Argyris’
discussion of establishment of internal commitment.
Theories of organizational interventions are closely linked with those of
decision-making and problem solving. The processes of decision-making and
problem solving from individual, group and organizational perspectives are
extensively discussed in the literature.13
11
See Chin, Robert and Kenneth D. Benne: “General Strategies for Effecting
Changes in Human Systems”, in Bennis, Warren G., Kenneth D. Benne and
Robert Chin: The Planning of Change, 4th edition, New York, 1985, p. 24; and
Bungard, Walter and Catrin Niethammer: “Psychologische Aspekte des Change
Management im interorganisationalen Kontext”, in Walter Bungard, Jürgen
Fleischer, Holger Nohr, Dieter Spath and Erich Zahn (eds.), Customer
Knowledge Management, Stuttgart, 2003, p. 109.
12
Schein, Edgar H.: Process Consultation, Boston, 2000, part I, pp. 9—12, and
part II, pp. 29—35. It should be noted, that this book mainly consists of reprints
from his work in the late 1960’s. Schein’s work in general addresses the last
intervention model type, the Process Consultation Model.
13
In Akkermanns, Henk: Modelling With Managers, Breda, The Netherlands,
1995, pp. 7—12, a literature overview of decision-making and problem solving
is found, covering Operation Management, System Dynamics, Strategic
Management, Operations Research/“Soft OR”, Group Decision Support Systems
and Organizational Psychology.
4
Action planning
1
Problem
formulation
3 Felt 2
need
Forecasting Producing
consequences, proposals
6 testing proposals for solutions 5
Evaluating Taking
outcomes action steps
Figure A-2: Model of stages of problem solving14
Figure A-2 illustrates a problem-solving process of iterative stages with

two conceptual cycles succeeding the problem identification (the “felt need” in
the center of the figure). Cycle I (figure A-3a) includes the problem formulation,
producing proposals for solutions and forecasting consequences and testing
proposals. Cycle II (figure A-3b) includes action planning, taking action steps
and evaluating outcomes.
14
See Schein, Edgar H.: Process Consultation, Boston, 2000, part I, p. 61. The
model is an elaboration of a model originally developed by Richard Wallen for
use in sensitivity training programs. The model has strong similarities with
Dörner’s “Steps in Planning and Action”, although Dörners model less sharp
separate in a planning and an implementation part, see Dörner, Dietrich: The
Logic of Failure, New York, 1996, p. 43. Also, the model has similarities with
the PDCA-cycle (Plan-Do-Check-Act) as seen in TQM. The PDCA-cycle is a
generic version of the Deming Cycle, see Imai, Masaaki: Kaizen: Der Schlüssel
zum Erfolg der Japaner im Wettbewerb, german translation, München, 1992,
p. 87.
4
Action planning
1
Problem
formulation
3 Felt 2
Forecasting need Producing
6 testing for solutions 5
Evaluating proposals Taking
Figure A-3a: Diagnostics and decision-making (cycle I)
4
Action planning
1
Problem
formulation
3 Felt 2
Forecasting need Producing
6 testing for solutions 5
Evaluating proposals Taking
Figure A-3b: Change management (cycle II)
For organizational interventions addressing the strategic problems of

larger organizations, the two cycles of problem solving are typically discussed in
two distinct areas: the area of strategy forming and the area of strategy
implementation.15 The latter is described in the change management focused
literature from the disciplines of organizational psychology and organizational
development (OD). Cycle I of interventions have their primary focus on
diagnostics and decision-making, which for strategic organizational interventions
can be understood as strategy forming. In the field of organizational psychology,
cycle I activities are normally labeled ‘organizational diagnostics’ which is
leading to the initiation of the planned change process.16 Cycle II interventions,
focusing on planning, carrying out and following up on implementation, comprise
what in the fields of organizational psychology and OD are typically categorized
as ‘planned change’ interventions.
The iterative and recursive nature of the entire problem solving process
(both cycle I and cycle II) should not be underestimated, as also discussed in the
problem solving system described by Flood, “Total System Intervention”.17 Total
System Intervention focuses on creative problem investigation and deliberate
selection of methods to solve problems, through an iterative and recursive
process of three phases, (1) creativity, (2) choice, (3) implementation. In this
context, strategy forming is influenced by implementation considerations and
experiences, and the strategy implementation constitutes in itself a new cycle
with the need for creative ideas on how best to implement the strategy and the
choice of the best methods to achieve the implementation.
Research within the area of strategy forming proposes that the way
corporations address strategic problems should be considered as groping,
interactive processes emphasizing learning, creativity, synthesis, and sharing of
mental models among decision-makers.18 In this regard, the understanding of
15
Huff, Anne S. and Rhonda Kay Reger: “A Review of Strategic Process
Research”, Journal of Management, Vol. 13, No. 2, 1987, p. 212. It should be
noted, that Huff and Reger use the term strategy formulation rather than strategy
forming.
16
The term ‘diagnostics’ is often used as heading for activities leading to the
planned change interventions, e.g. see the list of contents in Bennis, Warren G.,
Kenneth D. Benne and Robert Chin: The Planning of Change, 4 th edition, New
York, 1985 as well as in Cummings, Thomas G. and Christopher G. Worley:
Organizational Development and Change, Ohio, 2001.
17
Flood, Robert L.: Solving Problem Solving, Chichester, 1995, p. 32.
18
See Mintzberg, Henry: The Rise and Fall of Strategic Planning, New York,
1994, p. 77; de Geus, Arie P.: “Planning as Learning”, Harvard Business
mental models offers some interesting perspectives on the challenges in the

process. The interpretation of mental models varies significantly in the literature
from understanding mental models as “pre-compiled” limited conceptual
representations to seeing them as implicit, foggy, intuitive system perceptions
also involving the subconscious.19 Despite difficulties in the literature to agree on
a definition on mental models, it seems that general agreement exists in the
understanding, that mental models influence behavior and decisions significantly,
and that in information selection and interpretation, people subconsciously seek
confirmation of their existing mental models, which can also result in them
rejecting or ignoring information that contradicts their beliefs.20 Due to the
complexity of social systems, the strategy forming processes must encourage
strategies and assumptions to be challenged from inside or outside the problem-
solving environment to challenge improper beliefs people may have about causal
relations in their mental models.21
Review, March-April 1988, p. 71; Davenport, Thomas H.: Process Innovation,

Boston, 1993, pp. 278—279. In Eisenhardt, Kathleen M.: “Strategy as Strategic
Decision Making”, Sloan Management Review, Spring 1999, p. 66, this process
is described as “building collective intuition.”
19
For a discussion on mental models and literature on mental models, see among
others Senge, Peter M.: The Fifth Discipline, New York, 1994, pp. 174—204;
Doyle, James K. and David N. Ford: “Mental models concepts for system
dynamics research”, System Dynamics Review, Vol. 14, No. 1, Spring 1998,
pp. 3—29, Doyle, James K. and David N. Ford: “Mental models concepts
revisited: some clarifications and a reply to Lane”, System Dynamics Review,
Vol. 14, No. 1, Spring 1998, pp. 3—12. The word ‘subconscious’ is used in this
dissertation as “existing or operating in the mind beneath or beyond
consciousness”, see Webster’s Encyclopedic Unabridged Dictionary of the
English Language, New York, 1989, p. 1414.
20
See Vennix, Jac A. M.: Group Model Building, Chichester, 1996, p. 21; Senge,
Peter M.: The Fifth Discipline, New York, 1994, p. 175; Kampmann, Christian
P. E.: Feedback complexity and market adjustment – An experimental approach,
Boston, 1992, p. 29; Bakken, Bent E.: Learning and Transfer of Understanding
in Dynamics Decision Environments, Boston, 1993, p. 30; Hogarth, Robin:
Judgment and Choice – The Psychology of Decision, 2 nd edition, Chicago, 1987,
p. 130.
21
See Bakken, Bent E.: Learning and Transfer of Understanding in Dynamics
Decision Environments, Boston, 1993, p. 30; Argyris, Chris: Reasoning,
Learning, and Action – Individual and Organizational, San Francisco, 1982,
p. 39.
The challenge of improper beliefs and the improvement and realignment

of mental models is not only relevant among the executive decision-makers, but
is often critical in larger circles. In many organizations, decentralization and
empowerment have resulted in business decisions being made also at the lower
levels of the organizational hierarchy. Therefore, new strategies can seldom be
implemented only by introducing new guidelines or policies, as it is required that
a larger number of employees understand why the organization must change as
well as understand the reasoning behind the new strategy. Consequently, the
transfer of insights gained by the decision-makers in the strategy forming process
is of great importance. In terms of the previously introduced problem solving
cycles, this means that the insights gained in Cycle I need to be transferred to the
people responsible for the implementation (cycle II). A further argument for this
transfer of insights to take place in the intervention process is the trend among
business organizations to motivate employees using non-monetary instruments
such as involvement and influence, which requires the employees to have in-
depth understanding of the relevant strategic issues.22 Some researchers, however,
are questioning the importance of this value substitution at times of high
unemployment rates and strong focus on cost rationalizations.23
A number of major researchers within the field of strategic planning
devote much attention to the discussion of problems in the implementation of
strategies and policies, as implementations far too often remain unsuccessful.24 A
parallel can also be drawn with Repenning and Sterman’s view on improvement
programs, which argue that successful implementation of new methods represents
a bigger challenge than identifying or learning new improvement methods.25
22
See Schein, Edgar H.: Organisationspsykologi, Danish translation, Herning,
1990, p. 53; Kieser, Alfred: “Human Relations-Bewegung und Organisations-
psychologie”, in Kieser, Alfred (ed.): Organisationstheorien, 3 rd edition,
Stuttgart, 1999, pp. 101—131.
23
This viewpoint is discussed in Jöns, Ingela: Managementstrategien und
Organisationswandel, Mannheim, 1995, p. 156.
24
See Preface, Warren, Kim: Competitive Strategy Dynamics, Chichester, 2002.
For a theoretical discussion of the implementation problem, see also McPherson
III, L. Fillmore: “Organizational Change: An Industrial Dynamics Approach”, in
Edward B. Roberts (ed.): Managerial Applications of System Dynamics,
Cambridge, Massachusetts, 1978, pp. 447—449.
25
See Repenning, Nelson P. and John D. Sterman: “Nobody Ever Gets Credit for
Fixing Problems that Never Happened: Creating and Sustaining Process
Senge furthermore lists a number of studies documenting significant failure rates

in achieving sustainable change in top management-driven change initiatives.26 In
other words, the implementation of a solution to a strategic problem can even
constitute a bigger challenge than finding the solution.27
II. Foundation and Strategies for Planned Change Interventions

Lewin is often accredited as being the intellectual founder of ‘planned change’
organizational interventions, and he is still one of the most frequently quoted
authors in the social sciences.28 Planned change refers to attempts where change
is “conscious, deliberate, and intended, at least on the part of one or more agents
related to the change attempt”.29 The theories of planned change are concerned
with the complex processes of learning and change necessary to overcome the
normal resistance most humans have towards change, even when the goals are
apparently highly beneficial.30 Planned change in Lewin’s perspective is also
known as the “unfreezing-movement-freezing” process. Social systems seem to
have some sort of “inner resistance” to change, indicating that in spite of the
Improvement”, California Management Review, Vol. 43, No. 4, Summer 2001,

p. 65.
26
Senge, Peter M.: The Dance of Change, New York, 1999, pp. 5—6.
27
This does not imply that strategy making is easy. Rethinking strategies and entire
business concepts is both a difficult and crucial task in given situations; see
Hamel, Gary: Leading the Revolution, Boston, 2000, p. 28; Fine, Charles H.:
“Clockspeed-based strategies for Supply Chain Design”, Production and
Operation Management, Vol. 9, No. 3, 2000, pp. 213—221.
28
See preface in Gold, Martin: The Complete Social Scientist: A Kurt Lewin
Reader, Washington, 1999; Schein, Edgar H.: Organisationspsykologi, Danish
translation, Herning, 1990, p. 249.
29
See Chin, Robert and Kenneth D. Benne: “General Strategies for Effecting
Changes in Human Systems”, in Bennis, Warren G., Kenneth D. Benne and
Robert Chin: The Planning of Change, 4 th edition, New York, 1985, p. 22.
30
The intervention process described by Argyris concerning moving an individual,
a group or an organization towards Model II theory-in-use and double-loop
learning is an example of an extensive and detailed planned change intervention
in action research tradition, see Argyris, Chris: Reasoning, Learning, and Action
– Individual and Organizational, San Francisco, 1982, especially pp. 162—162
and pp. 468—474.
application of a force, the social process will not change without an additional
force to break the habit, to unfreeze existing customs.31 Unfreezing existing
behavior or attitudes can take place through the mechanisms of weakening
existing behavior or attitudes, then the establishment of feelings of dislike
regarding the present situation, and also establishment of psychological feeling of
safeness in the change process.32 The second phase in the planned change process
is the actual change part with development of new attitudes and behavior based
on new information and cognitive and affective redefinitions. The last phase in
the planned change process (freeze) is concerned with how to change behavior in
a sustainable way, avoiding it sliding back to its old level in a short time.33 To
strive for sustainable and continuous benefit of new attitudes and behavior, useful
mechanisms could be testing the congruence between the change and the
individual’s own situation, team-building efforts and continuous support, or
recognition from both formal and informal leaders in the organization.34 Lewin’s
theories of planned change originally focus on cognitive and behavioral change
aspects related to a specific change situation, but researchers often also
emphasize the broader term organizational learning. Argyris and Schein both
place interventions as part of the continuous learning and forming of the
organization and its change readiness, Senge emphasizes the importance of
improving system thinking skills, and Sterman emphasizes the importance of
helping organizations to improve the critical thinking skills necessary to
challenge future mental models and biases, opposed to only helping to solve a
specific problem.35
31
Lewin, Kurt: “Group Decision and Social Change” (first published in Newcomb
and Hartley’s Readings in social psychology, 1948, pp. 330—341), in Gold,
Martin: The Complete Social Scientist – A Kurt Lewin Reader, Washington,
1999, p. 281.
32
See Schein, Edgar H.: Organisations Psykologi, Herning, 1990, pp. 254—255.
33
Lewin, Kurt: “Group Decision and Social Change”, in Gold, Martin: The
Complete Social Scientist – A Kurt Lewin Reader, Washington, 1999, p. 265.
The arguments include, that behavior observed in a training program is often not
continued when the person goes back to his normal routines.
34
See Schein, Edgar H.: Organisationspsykologi, Herning, 1990, pp. 256—257;
and Change, Ohio, 2001, pp. 22—30.
35
View, Reading, Ma., 1970, chapters 1 & 2; Schein, Edgar H.: Organisations-
The planning of organizational interventions always, more or less

deliberately, reflects underlying change strategies.36 Chin and Benne have
developed a taxonomy of strategies for effecting changes in human systems,
consisting of three types of general strategies:37
• Empirical-Rational Strategies assume changes to be adopted

if they are rationally justified. Examples in business
organizations include allocation of funding, personnel
replacement and scientific management projects.38
• Normative-Re-educative Strategies focus the change process

on attitudes, values and skills, inspired from the fields of
sociology and psychology. These strategies are often less
concerned with solving specific problems but rather focus on
organizational development and improving the problem-
solving capabilities of the organization.
• Power-Coercive Strategies emphasize political and

economical sanctions in the exercise of power, or even
playing upon sentiments of guilt and shame. Top-down
implementation of new strategies or policies often has
implicit elements of use of power.
psykologi, 1990, p. 40; Senge, Peter M.: The Fifth Discipline, New York, 1994,
pp. 57—67; Sterman, John D.: “All models are wrong: reflections on becoming a
systems scientist”, System Dynamics Review, Vol. 18, No. 4, Winther 2002,
p. 526.
36
Borum, Finn: Strategier for organisationsændringer, Copenhagen, 1995, p. 15.
37
Chin, Robert and Kenneth D. Benne: “General Strategies for Effecting Changes
in Human Systems”, in Bennis, Warren G., Kenneth D. Benne and Robert Chin:
The Planning of Change, 4 th edition, New York, 1985, pp. 22—45.
38
The school of scientific management dates back to the early 20th century with
Taylor’s work on rational optimization of work processes. The most well known
mechanism element might be the detailed time studies of work procedures,
although this should be seen in context with the underlying principles, including
focus on science development, scientific basis for selection and development of
workmen, and friendly cooperation between the management and the men, see
Taylor, Frederick W.: The Principles of Scientific Management, New York,
1967, (first published in 1911), pp. 129—130.
Another taxonomy proposed by Borum includes The Technical-Rational

Change Strategy, The Humanistic Change Strategy and The Political Change
Strategy.39 The first two types in each of the taxonomies are rather similar,
whereas Borum’s political change strategy has less focus on coercive elements
and more focus on interpersonal negotiations, handling of personal interests,
personal power-bases and interpersonal conflicts. Although using the terminology
‘change mindsets’ rather than ‘change strategies’, Anderson and Anderson
compare the ‘industrial mindset’, including power and control, predictability,
discrete events, with the ‘emerging mindset’, including participation, uncertainty
and self-organization.40 The study of change strategy taxonomies contributes to
enabling change agents to deliberately construct intervention strategies based on
scientific research, and intervention strategies of large and complex corporate
change programs are likely to combine elements from two or more of the general
change strategies. A central element in such intervention strategies is the
overcoming of human resistance to change.41 Literature discussions on change
resistance mostly center on attitudes and intentions towards change, with attitudes
being influenced by cognitive, affective and conative elements. Ajzen offers a
widely used framework for the study of change of behavior based on the change
39
In Borum, Finn: Strategier for organisationsændringer, Copenhagen, 1995, the
change strategies are discussed thoroughly, and at p. 117, a schematic overview
can be found. Furthermore a fourth change strategy regarding network
organizations/communities is discussed.
40
Anderson, Linda A. and Dean Anderson: “Awake at the Wheel: Moving beyond
Change Management to Conscious Change Leadership”, OD Practitioner,
Vol. 33, No. 3, 2001, p. 44. The two mindsets seem to a high extend to
correspond with the traditional two views on Man-in-Organization: The Human
View vs. The Resource View, see Leavitt, Harold J., William R. Dill, and Henry
B. Eyring: The Organizational World – A systematic view of managers and
management, New York, 1973, pp. 122—123.
41
Overcoming human resistance to change is among the most discussed topics the
literature of strategy implementation and change management, see Cummings,
Thomas G. and Christopher G. Worley: Organizational Development and
Change, Ohio, 2001, pp. 154—173; Argyris, Chris: Interventions Theory and
Method – A behavioral Science View, Reading, Ma., 1970, p. 70.
of intentions leveraging on attitudes, norms and perceived behavioral control.42

This framework will be discussed further in chapter B.
The intervention context, primarily understood as organizational and
problem characteristics, is the main factor influencing the intervention strategy
and the decisions on intervention mechanisms.43 Context elements include the
organization’s management traditions such as authoritarian or democratic
decision processes.44 According to Jöns, when strategy implementation also
encompasses changes in the organization’s philosophies, it is insufficient that the
implementation process addresses qualifications and acceptance; the
implementation process must also address the underlying values of the
organization and to a higher degree includes employee development, information
and participation.45
Different science schools exist in regards to using theory or the real world
as the starting point for an academic research approach.46 In the field of planned
change, action research belongs to the most recognized approaches. Action
research takes the theoretical point of departure, that dynamic systems such as
organizations can be examined through carefully planned, theory-based
interventions.47 Action research is formative as well as summative, as the
42
Ajzen, Icek: Attitudes, Personality and Behavior, Chicago, 1988, pp. 20—131.
See also Rouwette, Etiënne: Group model building as mutual persuasion,
Nijmegen, 2003, pp. 104—111, for a discussion on Ajzen’s framework.
43
See Rouwette, Etiënne: Group model building as mutual persuasion, Nijmegen,
2003, p. 103.
44
See Schein, Edgar H.: Organisationspsykologi, Herning, 1990, p. 142 for a
literature overview of management traditions in regards to involvement of
subordinates in decisions.
45
Jöns, Ingela: Managementstrategien und Organisationswandel, Mannheim, 1995,
p. 157.
46
An extensive literature-based discussion of the relationship between practice and
theory in organizations theories in general (not specific related to action
research) can be found in Scherer, Andreas G.: “Kritik der Organisation oder
Organisation der Kritik? Wissenschaftstheoretische Bemerkungen zum Umgang
mit Organisationstheorien”, in Kieser, Alfred (ed.): Organisationstheorien,
3 rd edition, Stuttgart, 1999, pp. 1—37.
47
See Schein, Edgar H.: Organisationspsykologi, Herning, 1990, pp. 249—259;
and Change, Ohio, 2001, pp. 22—30. Action research has some parallels to the
field of cybernetics, where focus is on behavior of systems (what does it do)
interventions are altered if resultant data or changing conditions suggest the

appropriateness.48 Action research projects do not involve traditional science
evaluation models, e.g. the use of control groups, as the complex social
environment is not controllable to a degree that allows isolation of the true
variables. Furthermore, action research projects and practical use is supposed to
coexist with mutual benefit, also in regards to evaluation. Evaluation of human
systems will influence the system, for example as a Hawthorne effect or as
expectation settings.49 Consequently, interventions in action research tradition
must be designed in a way where the academic evaluation is not counter-
productive with regards to the desired results of the intervention.
III. The Usage of System Dynamics Modeling in Organizational

Interventions
On an abstract, conceptual level, modeling takes place in organizations all the
time. Every time a decision is made, the decision maker’s cognitive model of the
situation will influence the decision.
Figure A-4 is a model of decision maker (D) and the process (P), which is
the target for goal-seeking decision making. Ideally, rational decision-making
should be a function of P, input and output of P, as well as system goals (G) and
decision principles (DP).50
rather than on a detailed understanding of the system elements (what is this

thing), see Ashby, W. Ross: An Introduction to Cybernetics, paperback version,
London, 1964, p. 1.
48
See Gold, Martin: The Complete Social Scientist – A Kurt Lewin Reader,
Washington, 1999, p. 253.
49
The Hawthorne effect is widely discussed in various literatures and typically
refers to performance improvements among workers participating in
experiments, although no theoretical basis exists for the effects. According to
Wikipedia, accessed April 2006, Mayo Elton has interpreted the performance
improvement among workers (the Hawthorne effect) as: “it was the feeling they
were being closely attended to that caused the improvement in performance.”
50
Only external information input (u e) is made explicit, as resource input is viewed
as controllable in P. M, Ue, Y, represent the sets of values of the manipulating
value, the external input and the outcomes. The decision problem is to find the
decision variable (m) in M such that G(m,ue,P(m,u e)) is maximized. For further
G, DP
m
D, P
ue P y
Figure A-4: Goal-seeking system51
However, in social-economic systems, due to uncertainty and complexity, only a

limited cognitive model of P will be available (Pm) for the decision-making.
Furthermore, D will be characterized by perceptions, motivation, and personal
values.
Figure A-4 illustrates the often-implicit use of modeling in decision-
making, whereas system dynamics offers an explicit, deliberate use of modeling.52
description see Takahara, Yasuhiko: “A Formal Model of Organization”, in

Takahashi, Singo, Kyoichi Kijima and Ryo Sato (eds.): Applied General Systems
Research on Organizations, Tokyo, 2004, pp. 15—21.
51
Takahara, Yasuhiko: “A Formal Model of Organization”, Tokyo, 2004, p. 16.
52
It should be noted, that a model is always only a limited reflection of a real
system, representing a given viewpoint on a problem or a system, based on
human decisions on parameters and structures to be included in the model. In
Sterman, John D.: “All models are wrong: reflections on becoming a systems
scientist”, System Dynamics Review, Vol. 18, No. 4, Winter 2002, p. 525, a
model is called “a simplification, an abstraction, a selection“ and “inevitably
incomplete, incorrect – wrong.” Nevertheless, modeling offers an opportunity to
overcome a number of the problems in unsupported decision-making, as
discussed later in this chapter.
The system dynamics field has its origin as a primarily analytical and rational
oriented problem-investigating and policy-forming discipline.53 Forrester states
that the purpose of system dynamics in corporate environments is to aid in the
design of improved industrial and economical systems, and system dynamics has
over the years contributed significantly to create insight being used in strategic
planning and policy design.54 System dynamics offers a complementary
opportunity to analyze complex and dynamic problems, as most of the traditional
tools offered by the strategic planning field, are largely static, and thereby often
insufficient in our present-day environment of complexity and dynamics,
consequently resulting in actions often being made based on intuition and
experience.55
System dynamics addresses the need for decision makers to learn and
understand complex problems and situations. Since Descartes, cognitive science
has been interested in how humans learn.56 Human brain processes are event-
orientated, which – without long experience or effective learning - makes it
53
In Forrester, Jay W.: Industrial Dynamics, Cambridge, 1961, p. 56, objectives in
using mathematical models are described as follows: “A mathematical model of
an industrial enterprise should aid in understanding that enterprise. It should be
a useful guide to judgment and intuitive decisions. It should help establish
desirable policies.” Milling, Peter: “Leitmotive des System-Dynamics-
Ansatzes”, Wirtshaftswissenschaftliches Studium, Vol. 10, 1984, p. 508, also
supports this understanding of system dynamics: “System Dynamics verwendet
formale Modelle, um zu einem verbesserten Verständnis des zu studierende
Phänomens zu gelangen und um Eingriffe in das System auf ihre Konsequenzen
hin zu untersuchen.”
54
See Forrester, Jay W.: Industrial Dynamics, Cambridge, 1961, p. 115. For
examples of the practical usage of system dynamics, see the numerous cases
published in System Dynamics Review over the years.
55
See Lyneis, James M.: Corporate Planning and Policy Design: A System
Dynamics Approach, Massachusetts, 1980, p. 3; and Warren, Kim: Competitive
Strategy Dynamics, Chichester, 2002, preface; Mintzberg, Henry: The Rise and
Fall of Strategic Planning, New York, 1994, p. 319.
56
In René Descartes first major contribution, 1628, “Regulae ad directionem
ingenii,” regarding rules for the use of the human’s cognitive means, a method
for acquiring scientific or any other type of rational founded insight is described,
see Lübcke, Poul (ed.): Politikens filosofi leksikon, Copenhagen, 2001,
pp. 82—87. Wikipedia (accessed April 2006) describes the work as a method for
scientific and philosophical thinking and translates the title of the book into
“Rules for the Direction of the Mind.”
difficult to comprehend feedback-loops of even relatively simple and small

dynamic structures, resulting in problem-solving in the area of strategic, complex,
and dynamic problems often not taking unwanted side-effects, delayed reactions
and policy resistance into consideration.57 Although companies deal with these
types of complex problems every day, they have often not been solved using
analytical tools, but have in many respects been managed based on past
experiences on what works and what does not – with respect to the existing
production facilities, portfolios of customers, products, etc. This experience-
based decision-making is regularly implemented as heuristics, rules of thumb,
organizational routines or the use of simplifications and traditions.58 No explicit,
formal models underlie this type of decision-making, but the experience-based
decision-making is building on mental models of individuals.
While intuitive, implicit knowledge such as simple heuristics and
experience proves to be helpful in many situations (particularly when dynamics
are rather low), it is insufficient in innovative and rapidly changing situations.59
This dependence on erroneous intuitive solutions is, in the view of Forrester, the
cause of most misbehavior in corporate systems.60
System dynamics can help decision-makers cope with and understand
situations and problems, that would have taken years to understand based on
empirical experiences. The accelerated learning is partly due to the structured
emphasis on understanding and exploring how behavior is influenced by
corporate structures and policies, partly due to the aid of computer modeling to
design improved policies and resource allocation and the utilization of the
computers’ ability to calculate thousands of iterations, and partly due to the
57
See Sterman, John D.: Business Dynamics – Systems Thinking and Modeling for
a Complex World, Boston, 2000, pp. 10—11; Kampmann, Christian P. E.:
Feedback complexity and market adjustment – An experimental approach,
Boston, 1992, p. 31; Bakken, Bent E.: Learning and Transfer of Understanding
in Dynamics Decision Environments, Boston, 1993, pp. 29—30; Dörner,
Dietrich: The Logic of Failure, New York, 1996, pp. 38—42.
58
Größler, Andreas: “A Content and Process View on Bounded Rationality in
System Dynamics”, Systems Research and Behavioral Science, Vol. 21, No. 4,
July/August, 2004, p. 320.
59
Bonabeau, Eric: “Don’t Trust Your Guts”, Harvard Business Review, May 2003,
pp. 118—119.
60
Forrester, Jay W.: “System Dynamics, System Thinking, and Soft OR”, System
Dynamics Review, Vol. 10, No. 2, Summer 1994, p. 249.
creation of improved and shared mental models among the decision-makers

across multiple organizational units.61
De Geus gives three main reasons for going through the trouble of making
and simulating computer models.62 These main reasons are: (1) that most people
only are able to deal with a few variables at a time, and this only in one or two
time iterations, (2) the need for separation of cause and effect in time and space,
and (3) computer models help to identify what information is most relevant. The
first reason in particular, is also supported in Miller’s work on limitations on the
amount of information humans are able to receive, process and remember.63 De
Geus’ arguments further include statements on computer models, often revealing
counter-intuitive behavior, which is a view also supported by Lane.64 Milling
furthermore emphasizes the synergy of combining human creativity with the
capabilities and the power of high-speed computing.65 The system dynamics
literature often points out promising results from the transfer research regarding
computer simulations, which according to Bakken “may be attributed to
motivational side-effects of the interactive pedagogy”.66 Lastly, the data acquiring
process in quantitative modeling processes is in itself valuable for understanding
61
Lyneis, James M.: Corporate Planning and Policy Design – A System Dynamics
Approach, Massachusetts, 1980, p. 9 and p. 15; Lane, David C.: “Should System
Dynamics be Described as a ‘Hard’ or ‘Deterministic’ System Approach?”
Systems Research and Behavioral Science, Vol. 17, 2000, p. 4.
62
de Geus, Arie P.: “Planning as Learning”, Harvard Business Review, Vol. 66,
No. 2, March-April 1988, pp. 70—74.
63
Miller, George A.: “The Magical Number Seven, Plus Minus Two: Some Limits
on Our Capacity for Processing Information”, The Psychological Review,
Vol. 63, No. 2, March 1956, p. 95.
64
Lane, David C.: “Should System Dynamics be Described as a ‘Hard’ or
‘Deterministic’ System Approach?”, Systems Research and Behavioral Science,
Vol. 17, 2000, p. 4.
65
Milling, Peter: “Modeling Innovation Processes for Decision Support and
Management Simulation”, System Dynamics Review, Vol. 12, No. 3, 1996,
p. 227.
66
Bakken, Bent E.: Learning and Transfer of Understanding in Dynamics Decision
Environments, Boston, 1993, p. 31.
the problem parameters, and stimulates best practice discussions and brings to
surface misperceptions among key individuals.67
In the 1990’s a school within the field of system dynamics, oriented
towards participative modeling approaches, emerged with increased embracing of
softer aspects such as organizational learning, group processes, and the
importance of consensus and commitment.68 This development might follow from
the change in organizational structures in many organizations. Modern
organizations with a high degree of employee empowerment typically have a
need for a large number of people to have an understanding for the whole of the
organization and its strategy, including the dynamics and the interdependencies,
to be able to make the right decisions in their daily work as well as for
motivational factors. This to some degree substitutes the “old way” with a few
executives directing strategies and policies to be implemented downwards in the
organization. The challenge of interventions nowadays is therefore not only to
find good solutions to problems or new situations. The solution must also be
understood and find acceptance among the many stakeholders, and efforts of
establishing internal commitment in organizational interventions are often
centered on the creation of awareness, consensus, and confidence regarding the
goals and the change process.69
Although system dynamics projects (participative modeling, in particular)
are concerned with both cognitive and behavioral aspects relevant for
implementation, system dynamics modeling efforts are typically elements in
organizations’ strategy forming, with less focus on strategy implementation.
However, the value creation of a corporate modeling study is seldom “a new
67
See Warren, Kim: Competitive Strategy Dynamics, Chichester, 2002, p. 30;
Snabe, Birgitte and Andreas Größler: “Targeted Participative Modelling as
Organisational Intervention: Concept and Case Study”, Journal of Systems
Research and Behavioral Science, Vol. 23, No. 4, in print, 2006, p. 20.
68
Through the introduction of participative model-building methodologies and
“planning as learning”, focus has been put on creating conceptual insights,
changing mental models of decision-makers and creating consensus and
commitment; see Vennix, Jac A. M.: Group Model Building, Chichester, 1996,
p. 97; de Geus, Arie P.: “Planning as Learning”, Harvard Business Review,
March-April 1988, p. 70; Lane, David C.: “Modelling as Learning: A
consultancy methodology for enhancing learning in management teams”,
European Journal of Operational Research, Vol. 59, No.1, 1992, pp. 64—84.
69
See Akkermans, Henk: Modelling With Managers, Breda, p. 20.
policy” to be implemented subsequently, but rather the learning that appears

among modeling participants through the exploratory process examining the
problem and the system behavior.70 New improved mental models among
decision-makers will constitute a part of the solution, as the new insight and new
decided policies will influence future operational decisions. In many cases, no
formal implementation is needed, as the learning among decision-makers is all
that was required to solve the problem.71 Some projects, nevertheless, are taking
place in a context, where more formal implementation is needed, and for this
reason most of the system dynamics literature and textbooks calls for the
implementation challenge to be considered throughout the intervention process,
and also considers the iterative nature of the process.72
The purpose of a system dynamics modeling study is likely be articulated
as an exploratory exercise addressing a problem and possible solutions, and only
seldom to be stated as strategy forming or strategy implementation. Although
seen from a strategic organizational intervention viewpoint, addressing the
context of the modeling studies, such a modeling study has conceptual orientation
towards strategy forming due to the strong Cycle I focus. As explicit support of
the change management phase (Cycle II), in the implementation efforts when a
strategic decision is already made, almost no tradition exists for using system
dynamics in a modeling-oriented way, as most implementation-oriented SD
70
See Lane, David C.: “Modelling as Learning: A consultancy methodology for
enhancing learning in management teams”, European Journal of Operational
Research, Vol. 59, No.1, 1992, p. 64; Vennix, Jac A. M.: Group Model
Building, Chichester, 1996, pp. 98—99.
71
See Akkermanns, Henk: Modelling With Managers, Breda, 1995, p. 17.
72
Sterman, John D.: Business Dynamics, Boston, 2000, p. 80 and p. 88; Roberts,
Edward B.: “Strategies for Effective Implementation of Complex Corporate
Models”, in Edward B. Roberts (ed.): Managerial Applications of System
Dynamics, Cambridge, 1978, pp. 79—84. A more critical view on system
dynamics efforts in organizational interventions can be found in Zock,
Alexander: “A critical review of the use of System Dynamics for organizational
consulting projects”, at CD-ROM of Proceedings, System Dynamics Conference,
System Dynamics Society, 2004, p. 7, where it is argued that not even the
participative modeling approaches are sufficient attentive to the overall
challenges of change processes.
studies and approaches are concerned with gaming-oriented simulations such as

flight simulators and planning games.73
Taking the view of change management versus decision-making or
strategy forming, two conceptually different usages of system dynamics modeling
in organizational interventions appear: 74
• Exploratory modeling supporting diagnosing, learning and decision-

making75
• Transfer-oriented modeling usage supporting change management
The focal point in this differentiation is the purpose of the usage of system
dynamics: exploring a problem versus to transferring existing insights. Modeling
used in supporting diagnosing, learning and decision-making is driven by the
desire to explore and understand system behavior and to identify and simulate
possible new policies addressing a complex problem. Such interventions do not
have the same degree of control characteristics as typical implementation
projects, for example detailed project plans or thorough stakeholder analyses and
73
The use of modeling-oriented simulations vs. gaming-oriented simulations
comes from the taxonomy proposed by Maier, Frank und Andreas Größler:
“What are we talking about? A taxonomy of Computer Simulations to Support
Learning”, System Dynamics Review, Vol. 16, No 2, 2000, p. 143. The term
‘modeling-oriented simulations’ does not refer to the context of the project in
regards to decision-making or implementation.
74
In this dissertation, strategic formulation is understood to include both strategic
planning and policy formulation. In the system dynamics society, the term
“policy formulation“ is often used as the aim of modeling projects with policies
being rules stating how the day-by-day operating decisions are made, see
Forrester, Jay W.: Industrial Dynamics, Cambridge, 1961, p. 93. Strategies are
constituted by both corporate goals and corporate policies, and strategic
planning is defined as the process of transforming corporate goals into policies,
see Lyneis, James M.: Corporate Planning and Policy Design: A System
Dynamics Approach, Boston, 1980, p. 19 and p. 3.
75
‘Exploratory modeling’ should not be mistaken with ‘exploratory models’ as
described in Homer, Jack B.: “Why we iterate: scientific modeling in theory and
practice”, System Dynamics Review, Vol. 12, No. 1, Spring 1996, p. 1. Homer
defines exploratory models as a kind of easy-made, draft models less occupied
with validation. In this dissertation ‘exploratory modeling’ refers to the purpose
of the project: to explore and understand a given problem, no matter if the model
used is less detailed or if it is highly developed and refined with scientific rigor.
communication plans. This is due to the fact, that for exploratory modeling
interventions, the organizational change process cannot be defined before the
outcome of the modeling process is (at least to some extent) clear.76 Often, a
strategy forming modeling intervention will result in changed mental models
among decision-makers; frequently, implementation will not take place in an
explicit, planned change manner.77 Exploratory modeling can take place as
‘participative modeling’ or as ‘expert modeling’, the difference primarily being
the way people are involved. In expert modeling, people – apart from the main
decision-makers and a few modelers - are primarily involved for information
collection purposes.78 In participative modeling, such as Group Model Building
and Modeling for Learning, people representing an extensive array of viewpoints
are involved in the modeling process itself, with strong focus on mental model
alignment and refinement.79
Transfer-oriented usage of system dynamics modeling belongs to the
planned change type of organizational interventions and has some common
characteristics with the field of action research, with its parallel focus on the
implementation of planned change as well as on continued knowledge
80
development. The modeling project supports implementation processes where a
76
In Vennix, Jac A. M.: Group Model Building, Chichester, 1996, p. 99, it is
argued that learning cannot be predicted in the outset of a project.
77
In Vennix, Jac A. M.: Group Model Building, Chichester, 1996, p. 97, it is
argued that insights are conceptual rather than instrumental, and although
stressing that the purpose of system dynamics is performance improvement, he
also states (p. 99) that “implementation becomes evasive.” Richardson, George
P. and Alexander L. Pugh: Introduction to System Dynamics Modeling with
DYNAMO, Cambridge, 1981, p. 355, write that “a modeling study usually
focuses on what policies will help, not on how those policies ought to be
introduced into the system.”
78
See Forrester, Jay W.: “Policies, decisions and information sources for
modeling”, European Journal of Operational Research, Vol. 59, No. 1, 1992,
pp. 59—60, and Forrester, Jay W.: Industrial Dynamics, Cambridge, 1961,
p. 364, where it is recommended to use industrial dynamics in a business
company by initiating in a small, exclusive group of people with the right
qualities to go in-depth with the dynamics of the company including the
“innermost secrets and hopes of the organization.”
79
Vennix, Jac A. M.: Group Model Building, Chichester, 1996, p. 112.
80
For definitions of action research, see Cummings, Thomas G. and Christopher G.
Worley: Organizational Development and Change, Ohio, 2001, p. 23.
strategy forming process has outlined business objectives and targets, but where
the optimization of the strategy or policy is left as a part of the implementation.
Therefore a balance is needed between the initial detail level of the objectives
81
and the degree of freedom to make decisions in the implementation process.
This could be called “framing the intervention”, giving participants
empowerment to explore, decide and act within a given ‘frame’ (how to do), but
not to explore, decide and act outside the given ‘frame’ (what to do). Transfer-
oriented usage of modeling could be called instrumental usage of modeling for
change management purposes. In respect to transfer of existing learning from one
group of people to another group of people, this type of modeling has similarities
with gaming-oriented simulations, such as flight simulators or educational games.
Gaming-oriented simulations make use of fixed models, with the purpose being
to transfer the understanding of the causal relations and the behavior of the
system. Compared with gaming-oriented simulations, transfer-oriented modeling
to a higher extends aims to transfer commitment in the search for sustainable
change. Through involvement and participation in modeling sessions,
implementers take part in the refining of the change program and the
operationalization of the implementation. In his description of the system
dynamics process, Forrester proposes a phase called “Educate and Debate”, in
which consensus for implementation is aimed for.82 The phase is placed after the
actual exploratory modeling, but is expected to raise questions resulting in
repeated analysis in the previous phases. If modeling were to be used in the
Educate and Debate phase, it would be an example of change management
oriented modeling. Although for transfer-oriented usage of modeling, it may or
may not be the case that a model has been developed in an earlier strategy
formulation phase.83 The modeling can also be based on a model especially
81
Borum, Finn: Strategier for organisationsændringer, Copenhagen, 1995, p. 58,
discuss the problem of “a free, informed choice as a condition for establishment
of commitment” in a change process planned and controlled by consultants.
82
Dynamics Review, Vol. 10, No. 2, 1994, p. 247.
83
An interesting case, where a modeling project was continued into instrumental
implementation activities is the well-known “Maintenance Game” described in
Repenning, Nelson P. and John D. Sterman: “Nobody Ever Gets Credit for
Improvement”, California Management Review, Vol. 43, No. 4, Summer 2001,
pp. 64—88.
drafted for the project. Using such a preliminary model of the problem-system,
the process should allow for interactive refining and evaluation of the model
itself, and through both model adjustments, model enhancements and model
simulations, the modeling approach has the threefold aim of change program
refining, transfer of system understanding, and establishment of commitment.
The design of participative modeling interventions supporting change
management can draw from the normative, prescriptive management literature of
planned change with regards to activities such as intervention planning,
stakeholder management and implementation planning and review. Intervention
planning includes the definition of business objectives and targets, the framing of
the intervention, the identification of consultation relationships, roles and
responsibilities in the project organization, and time and budget planning.
Stakeholder management involves a thorough analysis of all the major interest
groups and individuals who have significant influence - directly or indirectly - on
the success of the intervention. Focus is on interests and power, importance for
solution design and implementation, and relevant means of involvement and
84
communication. Stakeholder analysis is a major input to intervention planning,
both to secure relevant parameters to be included in the process, and to secure
appropriate involvement and communication with stakeholders and other
85
employees. Implementation planning and review deals with the planning of the
implementation, including a communication plan and a clear assignment of
responsibilities. The communication plan develops over the course of the
intervention and includes elements such as motivating change and the
communication of visions, results, implementation plan and successes. Planning
activities for implementation should to be understood as something to be done
84
See Flood, Robert L. and Michael C. Jackson: Creative Problem Solving – Total
Systems Intervention, Chichester, 1991, p. 12; Argyris, Chris: Interventions
Theory and Method – A Behavioural Science View, Reading, Massachusetts,
1970, p. 81; and Borum, Finn: Strategier for organisationsændringer,
Copenhagen, 1995, pp. 77—89.
85
For discussions on “Employee Involvement”, see both Cummings, Thomas G.
and Christopher G. Worley: Organizational Development and Change, Ohio,
2001, p. 317; and Thun, Jörn-Henrik, Peter M. Milling, and Uwe Schwellbach:
“The Impact of Total Employee Involvement on Time-based Manufacturing”, in
Blackmon, Kate, Steve Brown, Paul Cousins, Andrew Graves, Christine
Harland, Richard Lamming, and Harvey Maylor (eds.): “What Really Matters in
Operations Management”, Bath, 2001, pp. 133—135.
only after a strategy or policy has been designed. Involving the right people in the
right way early in the process might be one of the most important criteria for
successful implementation later, together with communication strategies
including timely information and dialogues. Also, the iterative process continues
after implementation activities, as follow-up activities will create learning to be
used for further corrective actions.
The main differentiator of modeling supporting change management
compared to other approaches within the field of system dynamics, is the usage
and utility of participative modeling in a new context, namely in strategy
implementation rather than strategy formation. The main research objective is to
investigate whether change management-oriented participative modeling seems
to be an effective approach seeking sustainable change through:
• Transfer of insight from decision-makers to implementers in such a way
that not only the decisions but also the underlying arguments are
effectively transferred,
• Allow true involvement of implementers through participative strategy
refinement within a given decision ‘frame’ and strategic direction.
The rest of the dissertation is structured as follows: chapter B is a

theoretical discussion of the conceptual foundation for the usage of system
dynamics in organizational interventions. The discussion is not limited to usage
of system dynamics in change management, but rather it aim to investigate the
general purposes and methods for using system dynamics, which is traditionally
placed in decision-making and strategy and policy forming contexts. Focus on
usage of system dynamics in a change management context, begins in chapter C,
which describes a field study and an underlying research approach. In Chapter D
the insights from the field study are discussed in terms of both the theory basis
from chapter B and the normative, prescriptive management literature from the
field of organization development and planned change. Chapter E concludes with
the discussions from the previous chapters and proposes areas for further
research.
B. Conceptual Foundation for the Usage of
System Dynamics
The description of the conceptual foundation for the usage of system dynamics
involves a journey through a variety of disciplines: although it is based in
mathematics, physics and engineering, system dynamics also draws on cognitive
and social psychology, economics and other social sciences.86 According to
Martinez and Richardson, conceptual differences in research designs can be
discussed in terms of theory, method and procedure elements.87 The theories
underlying the usage of system dynamics (why use system dynamics) are
discussed in chapter B.II: Cognitive and Behavioral Rationale for the Usage of
System Dynamics. The methods and procedures for the usage of system
dynamics in organizational interventions (how system dynamics is used) are
described in chapter B.III: The Development Process of System Dynamics
Models in Corporations. Before the theory, method and procedure discussions,
however, chapter B.I seeks to place the usage of system dynamics within the
overall context of decision-making.
I. The Usage and Utility of Modeling in Decision-Making

In the field of decision-making Baron discusses utility theory as a normative
model concerned with elements of (1) the trade-off between the probability of an
outcome and its utility, (2) the trade-offs among different goals, (3) maximizing
utility over all relevant people as a normative model for moral decisions, and (4)
86
Sterman, John D.: Business Dynamics – Systems Thinking and Modeling for a
Complex World, Boston, 2000, pp. 4—5.
87
Martinez, Ignacio J and George P. Richardson: “An Expert View on the System
Dynamics Modeling Process: Concurrences and Divergences Searching for Best
Practices in System Dynamics Modeling”, at CD-ROM of Proceedings, System
Dynamics Conference, System Dynamics Society, 2002, p. 25.
28 B. Conceptual Foundation for the Usage of System Dynamics
handling conflicts among outcomes that occur at different times.88 The rationality
based expected-utility decision-making is often seen analyzed in the traditions of
operations research, decision trees, game theories, etc. 89
Whereas principles of rational choice are considered as reasonable in
abstract form, their implications are often violated in actual choices. In socio-
economic systems it is an illusion to assume perfect rational decision-making due
to complexity, uncertainty and human factors. The topic of complexity and
complex systems has been of great interest to scientists using terms such as
theories of holism, cybernetics, general system theory, chaos theories etc. since
World War I.90 Although not undertaking a formal definition of complexity,
Simon explains a complex system as “one made up of a large number of parts
that have many interactions.”91 In system theory traditions, Senge and Sterman
describe complexity as consisting of detail complexity and dynamic complexity.92
Milling further divides detail complexity into three sub-dimensions: number of
relevant elements (variety), number of connections between elements
(connectivity), and functional relationship between elements (functionality).93
88
Baron, Jonathan: Thinking and Deciding, 3 rd edition, Cambridge, UK, 2000,
pp. 223—243.
89
See Baron, Jonathan: Thinking and Deciding, Cambridge, UK, 2000, p. 227, for
a discussion on using game theory to examine expected-utility decision-making.
Expected-utility is also a cornerstone in the expected-monetary-value method;
see e.g. Tversky, Amos: “Additivity, utility and subjective probability”,
in Edwards, Ward and Amos Tversky (eds.): Decision Making, 1967,
pp. 208—238.
90
In Simon, Herbert A.: The Science of the Artificial, 3 rd edition, Cambridge,
1996, pp. 169—181, an overall discussion is offered on the major scientific
trends in this field.
91
Simon, Herbert A.: The Science of the Artificial, Cambridge, 1996,
pp. 183—184.
92
Senge, Peter M.: The Fifth Discipline, New York, 1994, p. 71; Sterman, John
D.: Business Dynamics – Systems Thinking and Modeling for a Complex World,
Boston, 2000, p. 21.
93
Milling, Peter: “Kybernetische Überlegungen beim Entscheiden in komplexen
Systemen”, in Entscheiden in komplexen Systemen, Wirtschaftskybernetik und
Systemanalyse, Band 20, Berlin, 2002, p. 12.
B. Conceptual Foundation for the Usage of System Dynamics 29
Simon’s view on rationality is, that whereas human decision-making is not

rational from an economic standpoint, it is still purposeful.94 Simon argues that in
real-world context even rational expectationists are retreating from rational utility
maximizing to a more realistic scheme of adaptive expectations.95 Decisions are
constrained not only by human process capabilities, but also by an incomplete
search for information. This only continues until a satisfactory solution is found
(contrary to seeking an optimal solution).96 History is full of grave examples of
people seeking to solve a problem and actually managing to worsen the situation
despite the best intentions. This is often due to what Forrester calls
“counterintuitive behavior of social systems”, or “policy resistance” in Sterman’s
terminology, where unintended side effects and neglected feedback loops make a
system behave differently from the intentions of the intervener.97 The literature
offers extensive discussion on these phenomena: the descriptive literature
identifies the deficiencies of traditional, unsupported decision making, whereas
the prescriptive tradition offers a number of methods and techniques to overcome
these limitations.98
When seeking to improve the organizational decision-making,
organizations make different types of analyses and models. For problems
characterized by feedback loops and delays, organizations can make use of
94
Herbert Simons view was expressed in the 1950’s, and discussed in Hogarth,
Robin: Judgement and Choice – The Psychology of Decision, 2nd edition,
Chicago, 1987, p. 63.
95
Simon, Herbert A.: The Science of the Artificial, 3 rd edition, Cambridge, 1996,
p. 39.
96
Vol. 63, No. 2, March 1956, p. 95; Vennix, Jac A. M.: Group Model Building,
Chichester, 1996, p. 27.
97
Forrester, Jay W.: “Counterintuitive Behavior of Social Systems”, in Collected
Papers of Jay W. Forrester, Cambridge, 1975, p. 216. In Sterman, John D.:
Business Dynamics – Systems Thinking and Modeling for a Complex World,
Boston, 2000, pp. 5—9, a larger number of examples of policy resistance are
described. In Dörner, Dietrich: The Logic of Failure, New York, 1996, a few,
but more detailed examples are discussed throughout the book; including
failures in Third World efforts and the Chernobyl disaster.
98
See Rouweette, Etiënne: Group model building as mutual persuasion, Nijmegen,
2003, pp. 19—29, for a description of descriptive and prescriptive view-points
in decision-making.
explicit conceptual feedback models (system thinking). Examples of such system

thinking efforts are soft operations research, cognitive mapping, structured
system discussions etc. A particular category of explicit, conceptual feedback
models is taking advantage of mathematical computer models and simulation.
These are the system dynamics models.
Figure B-1 tries to graphically depict the accumulative use of models in system
dynamics.
System
Dynamics
System Use of Formal,
Thinking Mathematical Models
(Simulation Models)
Intuition & Use of Explicit, Conceptual Feed-back Models

Experience (e.g. CLD or Soft OR)
Use of Implicit, Experience-based Models (Mental Models)
Figure B-1: Accumulative levels of models in the usage of system dynamics
The three levels in figure B-1 are accumulative; i.e. in addition to the use
of formal mathematical models, system dynamics also comprises explicit,
conceptual feedback models as well as intuitive and experience-based models.
This is in accordance with Kampmann, who stresses that intuitive assumptions
underlie any type of model.99 The usage of qualitative and quantitative models
serve the purpose of changing the mental models of the decision makers, as
mental models are seen as a vehicle to change decisions and organizational
action.100 The difference between system thinking and system dynamics cannot be
99
Kampmann, Christian P. E.: Feedback complexity and market adjustment,
Boston, 1992, p. 28.
100
Keough, Mark and Andrew Doman: “The CEO as organization designer – An
interview with Professor Jay W. Forrester, the founder of system dynamics”, The
seen as reflecting the differences between hard and soft modeling approaches, as
system dynamics is located somewhere between the two extremes.101 Hard
modeling is a term used for single objective optimization, typically without
taking people and organization into account.102 Although system dynamics uses
mathematical formulas and relatively rigid model structures, it also encompasses
soft modeling fundamentals like focus on generating debate and new insights
about the problem at hand.103
The field of system dynamics often has vital debates concerning the
advantages of qualitative modeling as seen in the system thinking area vs. the
advantages of quantitative modeling as seen in system dynamics. The opinions
differ from the one extreme, that only if a model is quantified and simulated, a
study can be said to be complete, to the other extreme, that for a complex system
with many soft relationships, quantification itself can be damaging.104
Some of the better known qualitative modeling approaches include
Checkland’s Soft System Methodology (SSM), which is a “process of enquiry”,
Eden’s Strategic Option Development and Analysis (SODA), using cognitive
mapping for strategic options development and Senge’s use of Cause-Loop-
Diagrams (CLD) in building learning organizations.105 The goals of qualitative
McKinsey Quarterly, No. 2, 1992, p. 5; Kim, Daniel H. and Peter M. Senge:

“Putting systems thinking into practice”, System Dynamics Review, Vol. 10,
Nos. 2-4, Summer-Fall 1997, p. 280.
101
In Sterman, John D.: Business Dynamics – Systems Thinking and Modeling for a
102
Maani, Kambiz E. and Robert Y. Cavana: Systems Thinking and Modelling –
Understanding Change and Complexity, Auckland, 2000, p. 21.
103
In Forrester, Jay W.: “System dynamics, system thinking, and soft OR”, System
Dynamics Review, Vol. 10, No. 2, 1994, p. 226, it is stated: “Understanding
comes first, but the goal is improvement;“ in Fey, Willard and John Trimble:
“The Evaluation and Development of Knowledge Acquisition in System
Dynamics Studies”, in Proceedings, System Dynamics Conference, System
Dynamics Society, 1992, p. 174, the process orientation of system dynamics is
compared to the product (being a model) orientation among hard system
developers.
104
Groessler, Andreas, Peter Milling and Graham Winch: “Perspectives on
rationality in system dynamics: a workshop report and open research questions”,
System Dynamics Review, Vol. 20, No. 1, 2004, p. 84.
105
Checkland, Peter: “Soft System Methodology”, in Rational Analysis, Jonathan
Rosenhead (ed.): Chichester, 1989, pp. 71—100; Eden, Colin: “Using cognitive
modeling include individual learning, challenging and alignment of mental

models, establishment of consensus and the search for ways to improve the
system.
Although quantitative modeling in organizational interventions also
includes simulations, the overall goals seldom differ radically from the goals of
qualitative modeling, but differ rather in the means of reaching those goals.
Devotees of quantitative models argue that softer models like Cause-Loop-
Diagrams leave open the risk of different interpretations of the same model by
different individuals. This is mainly due to the fact that the qualitative models do
not incorporate any test of logic. In mathematical formal models, built-in logical
constraints force model builders to have a more precise description and
understanding of the model. Forrester states that qualitative studies to a higher
degree depend on intuition compared to quantitative studies, where level and rate
diagrams discipline the thinking process in model formulation and simulation.106
Forrester gives examples where Harvard Business School graduates arrive to
wrong policy recommendations, inconsistent with their own quantitative system
description.107 This is in accordance with observations by other researches
observing students revealing significant differing interpretation of relatively
simple Cause-Loop-Diagrams.108 Whereas the opinion differs with respect to the
general applicability of quantitative vs. qualitative studies, there is a broad
agreement that determinants for selection of methods include: (1) problem
characteristics, (2) how the methods fit with the decision-making context, and (3)
purposes and goals of the decision situation.109
mapping for strategic options development and analysis”, also in Rosenhead

(ed.): Rational Analysis, Chichester, 1989, pp. 21—42; Senge, Peter M.: The
Fifth Discipline, New York, 1994. Though it should be noted, that in Forrester,
Jay W.: “System Dynamics, System Thinking, and Soft OR”, System Dynamics
Review, Vol. 10, No. 2, 1994, p. 253 it is argued that Senge’s system archetypes
and behavioral descriptions are based upon extensively explored system
dynamics models.
106
Dynamics Review, Vol. 10, No. 2, Summer 1994, p. 252.
107
Forrester, Jay W.: “System Dynamics, System Thinking, and Soft OR”, p. 240.
108
Observations by faculty members at Mannheim University.
109
Systemen”, in Entscheiden in komplexen Systemen, Wirtschaftskybernetik und
Systemanalyse, Band 20, Berlin, 2002, pp. 12—16.
The discussion has until now focused on the usage of modeling in

decision-making. As a side-remark, it is interesting to note, that system dynamics
has also been used to study the concept of decision-making. Based on a study
investigating a multiplier-accelerator model of capital investments, Sterman
concludes that it appears to be feasible to do experimental exploration of
dynamic decision-making strategies in aggregate systems, with the results being
directly compared to formal models of behavior.110
In this discussion, it is interesting to take a look at some of the critics of
the usage of system dynamics in strategic decision-making. Mintzberg has
criticized the utility of system dynamics, being concerned whether the
methodology allows sufficient creativity, and he states that analytical thinking
can be as wrong as intuitive thinking, especially as he finds that analysis does not
seem to encourage creativity.111 It is worth noticing that Mintzberg in his criticism
uses the argument that system dynamics focuses on analyses and aggregation and
pays little attention to comprehending and synthesizing, which is a standpoint
that most system dynamics practitioners have opposed to.112 It is also worth to
notice that his criticism was stated more than 20 years ago. The field has
developed since then, especially in the “softer” aspects with extensive research
within system thinking and participative model-building approaches, which focus
on discussions and involvement. Nevertheless, his concerns regarding creativity
should still be taken into consideration, as additional focus on how to conceive
creative potential new policies to be simulated in system dynamics models could
add value.
110
Sterman, John D.: “Misperceptions of Feedback in Dynamic Decision Making”,
in Milling, Peter M. and Erich O.K. Zahn (eds.): Computer-Based Management
of Complex Systems, Proceedings of the 1989 International Conference of the
System Dynamics Society, 1989, p. 30.
111
In Mintzberg, Henry: The Rise and Fall of Strategic Planning, New York, 1994,
pp. 298—299 and pp. 326—328, system dynamics is criticized for being shallow
in depth and not embracing creativity and intuition, although on pp. 376—378 in
the same book, credits are given to a number of system dynamics case stories.
112
The whole article of Lane, David C.: “Should System Dynamics be Described as
a ‘Hard’ or ‘Deterministic’ System Approach?”, Systems Research and
Behavioral Science, Vol. 17, 2000, pp. 3—22, is a discussion of the
misinterpretations of system dynamics, and also holds the quote “It may seem
paradoxical but the results of a quantitative system dynamics study are
qualitative insights” (p. 17).
Another interesting criticism of system dynamics, or actually of the earlier

modeling approaches, is made by de Geus, who states that most managers have
resistance to computer models to a degree where a computer-model becomes a
barrier for starting up discussions and exploration of mental models.113 For this
reason De Geus proposes a system dynamics modeling approach where soft
mapping techniques are used to start up the process capturing mental models.
Using soft modeling techniques in the beginning of the modeling activities is also
incorporated in many other modeling approaches.114 De Geus furthermore warns
against having expert modelers transform the soft models to hard models, as
subject matter knowledge is needed in this process.115 This also complies with the
arguments for participative modeling, namely, that insights are gained primarily
through participation in the modeling itself and that insights are difficult to
transfer to others, who were not involved in the modeling process.116 Akkemans
furthermore agues, that each type of diagram provides a different and useful view
of the problem situation, and that synthesis cannot be automated.117
113
De Geus, Arie P.: The Living Company, Boston, 1997, p. 71.
114
Luna-Reyes, L.F. and D. L. Andersen: “Collecting and analyzing qualitative data
for system dynamics: methods and models”, System Dynamics Review, Vol. 19,
No. 4, 2003, pp. 271—296 give an overview of many qualitative data collections
methods to be used not only in the beginning of a modeling process but also in
the later stages. Furthermore, Hodgson, A. M.: “Hexagons for system thinking”,
European Journal of Operational Research, Vol. 59, 1992, pp. 123—136,
introduces a soft modeling technique that is incorporated in many modeling
approaches; e.g. in Group Model Building.
115
De Geus, Arie P.: The Living Company, Boston, 1997, p. 72.
116
Decision Environments, Boston, 1993, p. 31; Vennix, Jac A. M.: Group Model
117
Akkermans, Henk: Modelling With Managers, Breda, 1995, p. 116.
II. Cognitive and Behavioral Rationale for the Usage of

System Dynamics
Interventions in social systems are typically discussed in terms of impact on an

individual level, group levels (formal and informal groups) and organizational
levels.118 On the individual level, learning and change of intentions and behavior
are often the focuses of interest. On the group level, alignment of mental models
and understanding group dynamics are often seen as corner stones, and on the
organizational level, interest often focuses on the creation of a learning
organization. This will be addressed in the next three subchapters.
1. Individual Learning and Change of Behavior in a Complex and

Dynamic Environment
Human beings have a tendency to think in events or limited linear causal
structures and more often than not, they underestimate or ignore complex
dynamic processes (illustrated in figure B-2).119 As a consequence, human
decision-makers leave out concerns for side effects and self-reinforcing dynamics
and fail to adjust their decision strategies to account for delays in the system and
expect feedback to arrive before the system can provide such information.120
118
These three levels for impacts of interventions are widely used in the system
dynamics literature, see Rouwette, Etiënne: Group model building as mutual
persuasion, Nijmegen, 2003, pp. 21—27 for a discussion on the three levels in
the literature of decision-making. In Argyris, Chris: Interventions Theory and
Method – A Behavioural Science View, Reading, Massachusetts, 1970, p. 38,
the three levels are listed together with an additional level; called intergroups
(formal and informal).
119
In Dörner, Dietrich: The Logic of Failure, New York, 1996, p. 6 cognitive
limitations in analytical, serial and visualized thinking are mentioned (as
opposed to female, “parallel” or non-western thinking); In De Bono, Edward:
Lateral Thinking for Management, England, 1971, pp. 4—9, it is argued that
linear vertical thinking being overly dominant in our education system. In
Vol. 63, No. 2, March 1956, p. 95, limitations on the amount of information
humans are able to receive, process and remember are discussed.
120
Environments, Boston, 1993, pp. 29—30; Kampmann, Christian P. E.: Feedback
Several studies within the field of system dynamics confirm these theories: based
on a study investigating a multiplier-accelerator model of capital investments,
Sterman concludes that many subjects fail to adequately account for the effects of
delays, and fail to understand the feedback between their own decisions and the
environment – even when provided with perfect information and knowledge of
the system structure.121 In gaming environments, Dörner has observed students
and professionals carrying out simulations of a variety of systems (small town,
eco-system etc.), with the majority of subjects failing to achieve “good results”;
failures he believes are primarily due to lack of system understanding and
tendencies to focus on short term, immediate effects rather than more long-term
and fundamental processes.122
E C
A => E => C => F
F
B
A
D -
System Perception
Figure B-2: Limited linear perception of system
“Mental models” is a repeatedly used term for the cognitive structures of

individuals; introduced by Johnson-Laird in the late 1970’s as internal, deducted
representations complementing the use of logic, affecting the way humans make
complexity and market adjustment – An experimental approach, Boston, 1992,

p. 31.
121
Sterman, John D.: “Misperceptions of Feedback in Dynamic Decision Making”,
in Peter M. Milling and Erich O.K. Zahn (eds.): Computer-Based Management
of Complex Systems, Proceedings of the 1989 International Conference of the
System Dynamics Society, Heidelberg, 1989, p. 30.
122
Dörner, Dietrich: The Logic of Failure, New York, 1996, p. 18.
inferences.123 The underlying ideas and concepts can be traced further back in
psychology literature, including the work of Craik’s in the mid 1940’s proposing
that people construct internal symbolic representations or models of external
events.124 Definitions in the fields of psychology, communication and system
dynamics vary from seeing mental models as being stable internal models to
being reconstructed on an ad hoc basis; from being simple picture-like images to
being complex and intuitive theories; from considering people being aware of
their own mental models to being intuitive and inaccessible models.125
Doyle and Ford propose a definition of mental models suitable to the field
of system dynamics:
“A mental model of a dynamic system is a relatively enduring

and accessible but limited internal, conceptual representation
of an external system (historical, existing or projected)
whose structure is analogous to the perceived structure
of that system.”
- Doyle and Ford126
123
Johnson-Laird, P. N.: Mental Models, Cambridge, 1983, pp. 144—145. See also
p. 74.
124
In Doyle, James K. and David N. Ford: “Mental models concepts for system
p. 8, reference is made to Craik’s (1943) book “The Nature of Explanation”.
This reference is also made in Johnson-Laird, P. N.: Mental Models, Cambridge,
1983, p. 2.
125
Doyle, James K. and David N. Ford: “Mental models concepts for system
p. 4, p. 9 and p. 14.
126
Doyle, James K. and David N. Ford: “Mental models concepts revisited: some
clarifications and a reply to Lane”, System Dynamics Review, Vol. 15, No. 4,
Winter 1999, p. 414. The definition is a revised version from an earlier article,
based on comments in Lane, David C.: Friendly amendment: A commentary on
Doyle and Ford’s proposed re-definition of “mental model”, Systems Dynamics
Review, Vol. 15, No. 2, Summer 1999, pp. 185—194.
With Doyle and Ford’s definition the system dynamics society to some
degree has an aligned use of the term mental models. In line with this definition,
cause-loop-diagrams or stock-and-flow diagrams could be viewed as externalized
mental models.127 Lane recommends not using the term “cognitive map” for these
diagrams, as this is a well-established term in management science literature for
the representation used in the SODA methodology.128 Doyle and Ford have a
broader definition, and argue that cognitive map is a well suited term, due to its
intuitive appeal, and arguing that Eden’s use of the term is merely a particular
form of cognitive mapping.129
The conceptual application of mental models gives meaning in both
everyday life and in more complex decision-making situations. In a normal
discussion of a single topic, participants will implicitly employ different mental
models, with different underlying assumptions, and with different goals.
Underlying relationships, assumptions and goals are typically left unstated and
never brought into the open. With this in mind, Forrester argues that it is easy to
understand why compromises often takes long and often fail in their objectives or
produce new difficulties.130 A different view could be, that the “slack” in
understanding does not make compromising more difficult, but allows everyone
to have a different understanding of the compromise. For important and complex
decision-making it consequently makes sense to invest in the efforts of making
the models explicit. Forrester further argues for not only making mental models
explicit, but also using computer analyses to investigate the problem, as even
models correct in structure and assumptions due to the fact that limited
127
Lane, David C.: Friendly amendment: “A commentary on Doyle and Ford’s
proposed re-definition of ‘mental model’”, Systems Dynamics Review, Vol. 15,
No. 2, Summer 1999, p. 186.
128
Lane, David C.: “A commentary on Doyle and Ford’s proposed re-definition of
‘mental model’”, Systems Dynamics Review, Vol. 15, No. 2, Summer 1999,
p. 186.
129
Spring 1998, pp. 412—413.
130
Papers of Jay W. Forrester: Foreword by Gordon S. Brown, Cambridge, 1975,
p. 213.
human brain capacities often yield wrong conclusions (individually or as group

consensus).131
An interesting aspect of mental models is that people seek confirmation
for their theories rather than questioning them, and as a consequence they are
often stuck in severely sub-optimal decision and problem-solving strategies.132
This is an effect of prior beliefs: people tend to find in data what they expect to
find.133 A classical study, replicated several times, is the Rosenthal experiment
illustrating the self-fulfilling prophecy that is the result of people seeking
confirmation of their mental models and in their mind rejecting information
contradicting their beliefs.134 It is a double-blind experiment where a large
number of teachers, before meeting their new classes, are handed names of those
20% of the students, who based on a certain test would be expected to make
above average intellectual progress. In reality the students were picked randomly,
but at the end of the school year the randomly picked students showed a real
above-average performance increase as well as being rated by the teachers as
showing distinguished intellectual curiosity. Another classic example of self-
fulfilling prophecy is a waiter having a mental model being that well-dressed
people tip better. Subconsciously, the waiter will give well-dressed people better
service and thereby create a positive loop confirming his or her prior beliefs.
131
Papers of Jay W. Forrester, Cambridge, 1975, p. 214.
132
Environments, Boston, 1993. pp. 29—30. An often mentioned example: Newell,
Allen and Herbert A. Simon: Human Problem Solving, Englewood Cliffs, New
Jersey, 1972, pp. 90—91, describe the well known ‘Nine Dot Problem’ (with
nine dots arranged in a 3 by 3 square array), where each subject is “directed to
draw four straight lines, without raising his pencil from the paper, that pass
through all nine dots.” Most subjects subconsciously assume, that the lines may
not continue outside the boundaries of the square, which make the problem
unsolvable.
133
p. 182.
134
The Rosenthal experiment was carried out in the 1960’s, and a description can
be found in Vennix, Jac A. M.: Group Model Building, Chichester, 1996, p. 20.
Other experiments showing conservatism in mental models include gaming
environments where participants have a tendency to stick to their initial
understandings of the system even when data would suggest otherwise, see
Prior beliefs can also result in illusory correlations, e.g. if a teacher thinks a child
is intelligent, he or she will often tend to overestimate how well behaved the child
is.135
Before discussing how to bring about changes in individuals’ cognitive
structures and mental models, it is relevant to take a look at the theories of
judgment. Kahneman and Tversky argue, that human judgment relies on
heuristics rather than calculus of chance or statistical based prediction.136 Human
judgment is affected by a number of elements: contextual effects, the extent to
which cues are available, the order in which information is presented, whether
comparative judgments involve similar or dissimilar information, qualitative and
quantitative data, and other factors.137 Hogarth especially points out the
importance of availability of information in the forming of judgments.
Individuals tend to put more emphasis on clues available to them, and
furthermore, some information from the past has stuck better, and therefore
influences judgments to higher degree.138 The contextual effects influencing
perceptions include the absolute variation in size, as experiments have shown
distorted perceptions due to absolute sizes.139 Another contextual effect shows in
the study of scenario planning where scenarios that include both a possible cause
and an outcome seems to appear more probable than scenarios merely involving
the outcome.140
An interesting hypothesis stated by Enhager, is that the intensity of
emotions activated (being ether positive or negative) strongly influences how
well humans remember a given situation.141 He argues, that people typically
135
p. 183.
136
Kahnemann, Daniel and Amos Tversky: “On the Psychology of Prediction”,
Psychological Review, Vol. 80, No. 4, July 1973, p. 237.
137
Hogarth, Robin: Judgment and Choice: The Psychology of Decision, 2 nd edition,
Chicago, 1987, p. 55.
138
Hogarth: Judgment and Choice, Chicago, 1987, p. 53. Hogarth mentions work
by Tversky and Kahnemann in much of his argumentation.
139
Hogarth: Judgment and Choice, 1987, p. 52.
140
141
Kjell Enhager is a known speaker on the psychology of sports, and he has used
his theories in the training of the female national golf team in Sweden, e.g. by
having the women forcing a positive feeling with a smile and a happy
exclamation for each good swing and trying to ignore the bad ones.
remember especially positive or especially negative situations from their pasts,

and by forcing positive feelings by different means, effectiveness in learning
situations can dramatically improve. This is in accordance with the line of
reasoning for storytelling: that activating feelings make a message memorable.142
Baron discusses dedicated scientific research on how human learning “sticks”
from experiences, and is biased by attention factors, such as giving more
attention to easily available information or being more or less open-minded.143
With regards to learning from experiences, it seems that positive feedback is
weighted more heavily in memory than negative feedback, and furthermore
indications exist, that explicit focus on identifying underlying rules lower the
attentional bias.144
In handling judgment, anchoring is a commonly used strategy. The
judgment process is based on a starting point (the anchor) from where predicted
adjustments are made – and often this type of judgment is believed to be the
foundation for much intuitive anticipation.145 Even randomly selected anchors
that have nothing to do with the judgment situation, or the information available,
have shown to influence judgment processes.146 For decision-making under
complexity and uncertainty this means anchoring outcome of choices as
deviations (losses and gains) from reference points, although often weighting
losses larger that gain reflecting a negative attitude towards risk.147
Cognitive learning is often a necessary prerequisite in organizational
interventions; although not adequate for creating changes in behavior. Thinking
is rooted in the total process of psychic activity, and is linked with emotions,
142
McKee, Robert: “Storytelling That Moves People”, Harvard Business Review,
June 2003, p. 52.
143
pp. 176—181.
144
See Hogarth, Robin: Judgment and Choice: The Psychology of Decision,
Chicago, 1987, p. 130; Baron, Jonathan: Thinking and Deciding, 3 rd edition,
Cambridge, UK, 2000, p. 182.
145
See discussions in Hogarth, Robin: Judgment and Choice, Chicago, p. 54;
146
Vennix: Group Model Building, 1996, p. 28, offers a number of examples from
psychology literature.
147
values and motivations.148 Ajzen and Fishbein have developed a theory of

planned behavior building upon research within attitudes and personality, as well
as behavioral, normative and control beliefs (see figure B-3). The framework is
based upon their early research implying that changes in attitudes due to
persuasive communication is insufficient to produce behavioral changes.149
Consequently, the model includes elements such as group dynamics, social
norms, and control beliefs.
Attitudes towards behavior include cognitive, affective and conative
dimensions, and are determined by behavioral beliefs reflecting the person’s
understanding of the likely outcomes given certain behavior, weighted with the
person’s individual evaluation of the possible outcomes.150 This tripartite model
is also used in the field of consumer behavior, where Kotler discusses a number
of different response hierarchy models with regards to not only the elements
included, but also how different buying situations call for different order
sequences.151
148
Dörner, Dietrich: The Logic of Failure, New York, 196, p. 8. An example of
personal motives and perceived behavioral control influencing behavior is given
in Berger, Ulrike and Isolde Bernhard-Mehlich: “Die Verhaltenswissen-
schaftliche Entscheidungstheorie”, in Kieser, Alfred (ed.): Organisations-
theorien, 3 rd edition, Stuttgart, 1999, p. 155. A practical example of how to
involve emotions in change processes is the use of storytelling, see McKee,
Robert: “Storytelling That Moves People”, Harvard Business Review, June
2003, p. 52.
149
Ajzen, Icek and Martin Fishbein: “The Prediction of Behavior from Attitudinal
and Normative Variables”, Journal of Experimental Social Psychology, No. 6,
1970, p. 483.
150
See Ajzen, Icek: Attitudes, Personality and Behavior, 1988, Chicago,
pp. 20—23, and pp. 120—121. Ajzen is using the three components of attitude
(cognitive, affective, and conative), arguing that this tripartite model of attitude
has served as the starting point of most behavioral analyses since the 1960’s.
151
See Kotler, Philip: Marketing Management, 7th edition, New Jersey, 1991,
pp. 573—575 where different buying situations are described in terms of models
embracing cognitive (learning), affective (feeling) and behavior (doing) stages.
The AIDA-model (awareness-interest-desire-action) is probably the most known
model. Kotler also argues, that high-involvement purchases with perceived high
differentiation calls for the “learn-feel-do“ sequence of marketing.
Attitude
towards the
behavior
Subjective
Intention Behavior
norm
Perceived
behavioral
control
Figure B-3: Theory of planned behavior 152
For research within organizational interventions, understanding the three

attitude components (the cognitive, affective, and conative elements) and
personality traits are important in establishing the “general laws” of human
action, as well as concepts of the dispositional nature of human behavior.153
Although many researchers argue that the evaluations between the three attitude
components can differ, Ajzen also discuss that many researchers find it
152
Taken from the figure in Ajzen, Icek: Attitudes, Personality and Behavior,
Chicago, 1988, p. 133.
153
Ajzen, Icek: Attitudes, Personality and Behavior, Chicago, 1988, p. 46. See also
Rao, Abhijit: “Recognition of Conative and Affective Behavior in Web Learning
using Digital Gestures”, North America Web-Based Learning Conference,
Online Proceedings, New Brunswick, 2001, p. 1, on conative and affective
elements improving learning experiences.
empirically difficult to distinguish between cognition and affect, as they often are
highly correlated.154
The subjective norms are assumed to be a function of the person’s beliefs
that specific individuals or groups approve or disapprove of performing this
behavior, and deal with both social norms and group dynamics aspects.155 The
importance of subjective norms is often stressed since Elton Mayo in 1945,
published a set of social assumptions regarding human nature. This focuses on
the social need of employees and recognizes the strong influence high status
individuals have on peer employees independently of formal power.156 Schein
states that norms within a group are maybe the most important influential factors
for the everyday behavior of the employees in a company.157 A related view on
social norms is what Gladwell calls the Power of Context: how seemingly small
changes in context can influence behavior. A known example is the Broken
Window theory, proposing that if small context flaws are tolerated (e.g. broken
windows), soon larger context flaws will occur (e.g. more windows will be
broken, which will start a process towards more serous crimes).158
Together the attitudes towards behavior and the subjective norms
influence intended behavior in a way that can be considered as goals to be
pursued.159 The intention to pursue these goals are affected by the person’s
perceived control over a given behavior determined by internal factors including
information, skills, abilities and control of emotions and compulsion, and
external factors like opportunity and dependence on others.160 The broken arrow
154
See Ajzen, Icek: Attitudes, Personality and Behavior, 1988, p. 21.
155
Ajzen, Icek: Attitudes, Personality and Behavior, 1988, p. 121; Ajzen, Icek and
Martin Fishbein: “The Prediction of Behavior from Attitudinal and Normative
Variables”, Journal of Experimental Social Psychology, No. 6, 1970, p. 483.
156
See Schein, Edgar H.: Organisationspsykology, Herning, 1990, p. 67.
157
Schein, Edgar H.: Organisationspsykology, 1990, p. 37.
158
See Gladwell, Malcolm: The Tipping Point, paperback edition, New York, 2002,
p. 141, giving credit of the Broken Window theory to the criminologists James
Q. Wilson and George Kelling. At pp. 142—145 an example of interventions
partly based on this theory is described: the fight against crime in New York in
the 1990’s, where fight against graffiti and fare-beating in the subway was the
starting point
159
Ajzen: Attitudes, Personality and Behavior, Chicago, 1988, p. 128.
160
See Ajzen: Attitudes, Personality and Behavior, 1988, pp. 128—131.
in figure B-3 indicates that the influence will only emerge when there is some
agreement between a person’s actual control and the person’s perceived control
of the behavior.161 Furthermore, perceived control does not only influence a
person’s current behavior. It also has long-term perspectives, as there is a general
agreement that perceived control results in positive feelings of one’s own
competence and worth, and likewise that lack of or loss of perceived control
influences both the physical and psychological well-being negatively.162
Understanding mental models, judgment processes, and behavior theories
is essential in understanding how people learn and react. People learn primarily
on the basis of what they can observe.163 A fundamental attribute of system
dynamics offers the opportunity to model and simulate the behavior of dynamic
structures and through observing the behavior reaching an understanding of a
complex problems in a few days or weeks that would have taken years to
understand by “normal” experience. Simulations have shown to result in better
transfer than case studies and lectures and this may also be attributed to
motivational side-effects of the interactive pedagogy.164 It is often difficult to
change the truly fundamental assumptions and beliefs underlying one’s thoughts
and actions, and especially when they are vague and poorly understood are they
difficult to escape from.165 People are typically unaware of their own
inconsistencies leading to self-fulfilling prophecies, and self-sealing processes.166
Modeling and simulating system structures is a way to bring forward underlying
161
Ajzen, Icek: Attitudes, Personality and Behavior, 1988, p. 134.
162
See Bungard, Walter: “Zur Implementierungsproblematik bei Business-
Reengineering Projekten”, in Perlitz, Manfred, Andreas Offinger, Michael
Reinhardt and Klaus Schug (eds.): Reengineering zwischen Anspruch und
Wirklichkeit, Wiesbaden, 1996, pp. 260—261. Also Kieser, Alfred:
Organisationstheorien, 3 rd edition, Stuttgart, 1999, pp. 129—131, describes
“Humanisierung der Arbeit” and discusses the benefits of the individual’s
influencing own work situation.
163
Hogarth, Robin: Judgment and Choice – The Psychology of Decision, Chicago,
1987, p. 130.
164
Environments, Boston, 1993, p. 31.
165
Morgan, Gareth: Creative Organization Theory, Newbury Park, California, 1989,
p. 28.
166
See Argyris, Chris: Reasoning, Learning, and Action – Individual and
Organizational, San Francisco, 1982, p. 39.
assumptions and inconsistencies, being the first step in a process to changing

them.167 Mental models are the “product” that “modelers take from students and
clients, disassemble, reconfigure, add to, subtract from, and return with value
added”.168
2. Establishing Group Consensus by Sharing Mental Models

Development of group consensus and individual learning are strongly interrelated
processes. Alignment between individuals in a group necessarily requires
individual learning, and likewise individual learning is affected by the context of
a group setting. A decision regarding individual goals made by an individual in a
group setting seems to have a significantly more enduring behavioral effect
compared to settings with 1-on-1 lecturing.169 Furthermore, it is easier to change
the ideology and social practice of a small group handled together than of single
167
In chapter A, learning was discussed in terms of single-loop and double-loop
learning. The discussions on double-loop learning and mental model refinement
are closely related. Both types of learning require impact of belief systems and
value systems, with beliefs being an understanding of causality and values being
a network where one value is supported by the other values. The belief system
especially impacts the internal process of “making sense” and defining the
situation, whereas the value system subsequent impacts the problem definition,
see Eden, Colin: “Cognitive mapping and problem structuring for system
dynamics model building”, System Dynamics Review, Vol. 10, Nos. 2-3,
Summer-Fall 1994, p. 263.
168
Spring 1998, p. 4.
169
Lewin, Kurt: “Group Decision and Social Change” (first published in Newcomb
and Hartley’s Readings in social psychology, 1948, pp. 330—341), in Gold,
Martin: The Complete Social Scientist – A Kurt Lewin Reader, Washington,
1999, pp. 276—279. Note that Lewin has a different terminology use compared
to most SD research, as he calls individual decision made in a group setting for
“group decisions“ (see p. 274).
individuals partly due to unwillingness of the individual to depart too much from
group standard of the group they belong to or wish to belong to.170
In organizational settings, where teams work together to make the business
run, and where decisions often are made in teams involving the main
stakeholders, the concept of mental model offers insight into many of the typical
problem solving challenges. Each individual has a limited linear understanding of
the main problems, and these will differ with regards to which causal
relationships are perceived as the most influential ones (see figure B-4).
A => E => C => F
E C
F
B
A A => E => C => -D A => B => D
-
D
System Perception
Figure B-4: Different limited linear perceptions of a system
The differing mental models among individuals in team settings often

result in reduced communication effectiveness, as the mental models reflect
different perceptions of the topics being discussed, typically without these
differences are being investigated. The diverse mental models or system
perceptions partly explain some of the classic conflicts in organizations. An
example being conflicts between the production and the marketing department,
170
Complete Social Scientist – A Kurt Lewin Reader, Washington, 1999, p. 273
and p. 281.
although not only differing mental models, but also conflicting interests based on
conflicting goal-settings play a role.
Baron discusses the concept of cognitive dissonance resolution, as a way
humans eliminate conflicts among beliefs.171 An example is, that after a decision
is made, it seems that increased weight is put on the reasons in favor of the
decision, and less value is given to the arguments in favor for paths that were not
selected. In organizational settings, where decision-makers typically have had a
significant role in the forming of the present situation, and have often had
different viewpoints in earlier decision-making processes, cognitive dissonance
resolution must be expected to contribute to the creation of biases in their
individual mental models.
Janis has put forward the concept of Groupthink, which is partly related to
cognitive dissonance resolution on a group level, and the concept is based on
extensive case studies on group decision situations, including major US failures
such as the Bay of Pigs decision, the escalation of the Vietnam War and the
failure to predict the attack on Pearl Harbor.172 The Groupthink phenomenon
refers to the lack of critical thinking that can occur in groups characterized by
group cohesiveness and consensus. It is in line with social-psychological research
findings, showing powerful social pressures if a dissident in a cohesive group
reveals objections to group norms or consensus. Although the term Groupthink
primarily refers to self-censorship in the critical thought process to avoid group
disunity.
Dörner discuss the Chernobyl explosion in terms of Groupthink and
overconfidence, arguing that the team responsible for the accident felt they were
so experienced that they did not have to follow safety rules.173 Although the team
was in fact very experienced they made basic errors of interaction with dynamic
171
p. 208. Leavitt, Harold J.: Top Down – Why Hierarchies Are Here to Stay and
How to Manage Them More Effectively, Boston, 2005, p. 130, discuss the same
concept, stating that “we humans don’t just do what we believe. We also believe
what we do.”
172
Janis, Irving L.: “Groupthink: The Problems of Conformity” (original printed in
Psychology Today, Nov. 1971, pp. 271—279), in Morgan, Gareth: Creative
Organization Theory, Newbury Park, California, 1989, pp. 224—228.
173
systems; they over-steered (regulation of the situation rather than the process),
they ignored side effects, and they interpreted system indicators erroneously.174
Groupthink is normally discussed is the context of smaller groups, often
being management teams, but some of the same mechanisms also apply to a
people’s philosophy of life as expressed in the below quote of Nietzsche:
“What is truth? A moving army of metaphors,

metonymies and anthropomorphisms,
in short a summa of human relationships
that are being poetically and rhetorically
sublimated, transposed, and beautified
until after long and repeated use,
a people considers them as
solid, canonical, and unavoidable.”
- Nietzsche175
Methods for overcoming Groupthink include establishing teams with

cognitive diversity, ensuring that teams to embrace different interests and
viewpoints in the organization, as well as applying different facilitation
approaches in decision-making processes.176 Examples of facilitation approaches
include the concept of “Think-hats” (where group members play different roles
during the meetings, including playing roles like out-of-the-box-thinking and
being the-devils-advocate), and Beer’s concept of Team Syntegrity. Team
174
Dörner, Dietrich: The Logic of Failure, New York, 1996, pp. 30—33; Salge,
Markus and Peter Milling: “Who is to blame, the operator or the designer? Two
stages of human failure in the Chernobyl accident”, System Dynamics Review,
Vol. 22, in print, 2006.
175
Quoted in Morgan, Gareth: Creative Organization Theory, Newbury Park,
California, 1989, p. 22.
176
For a definition on cognitive diversity, see Tilebein, Meike: “Eine struktur-
wissenschaftliche Betrachtung von Diversity Management”, Tagungsband,
GWS-Tagung, Greifswald, in print, 2006, p. 1.
Syntegrity is a process to establish shared meaning of a problem universe as well

as to create shared action plans among a larger group of people.177 This process
includes giving the participants the roles of critics and observers in different team
settings.178
Team effectiveness problems, which might be more frequently seen than
Groupthink, are the problems originating from teams characterized by politics
and differing perceptions. Especially if the problem reflects an unpleasant
situation, where the company performs unsatisfying, a normal human reaction is
to mistrust and blame others.179
Participative model building can be an approach for overcoming
Groupthink as well as for handling politics or differing perceptions. Through a
group model building process, mental models can be challenged and the process
can be seen as a vehicle for facilitating negotiation or mutual persuasion.180
Figure B-5 is a simplified illustration of how mental models influence formal
models, and how the formal models in turn also influence the mental models.
This is the loop where a shared formal model through an iterative model-building
process will facilitate an alignment between individual mental models. The new
mental model will result in new behavior and therefore indirectly affect the
system, and more directly the new mental models might result in new policies
and in this way result in system changes. Meyer offers a somewhat similar, but
more elaborated model, where the outcome of the process of formulating and
177
Pearson, Alan: “You Drive for the Show but you Putt for the Dough”, appendix
in Beer, Stafford: Beyond Dispute – The Invention of Team Syntegrity, 1994,
p. 321.
178
See Beer, Stafford: Beyond Dispute – The Invention of Team Syntegrity, 1994,
p. 59 and p. 102.
179
Kanter, Rosabeth Moss: “Leadership and the Psychology of Turnarounds“,
Harvard Business Review, June 2003, p. 61.
180
Eden, Colin: “Cognitive mapping and problem structuring for system dynamics
model building”, System Dynamics Review, Vol. 10, Nos. 2-3, Summer-Fall
1994, p. 259; Rouwette, Etiënne: Group model building as mutual persuasion,
Nijmegen, 2003, p. 251. Also the alignment of mental models has some
similarities with building collective intuition, see Eisenhardt, Kathleen M.:
“Strategy as Strategic Decision Making”, Sloan Management Review, Spring
1999, pp. 66—67.
simulating formal models results in selection and implementation of new policies,

which is how the feed-back is primarily given back to the actual system.181
Actual Mental Formal

System Model Model
Figure B-5: Mental models as instruments between actual systems and formal models
3. Enhancing Organizational Learning through System Thinking

Experience and Double-Loop Learning
System modeling and simulation projects in organizations not only address
specific problems, but also influence ongoing organizational learning efforts,
since the individual learning and improvement and alignment of mental models
described in the two previous subchapters are important instruments in
organizational learning.182 The term organizational learning has been subject to a
large variety of definitions in the literature. A few examples from these
definitions include the understanding of organizational learning as the sum of
individual learning, the understanding that it takes place, when learning is
accumulated into rules or routines that guide decisions and behavior, and also
181
Maier, Frank: Die Integration wissens- und modellbasierter Konzepte zur
Entscheidungsunterstützung im Innovationsmanagement, Berlin, 1995, p. 217.
182
In Kim, Daniel H. and Peter M. Senge: “Putting systems thinking into practice”,
System Dynamics Review, Vol. 10, Nos. 2-4, Summer-Fall 1997, p. 279, the
concept of mental models is called the transfer mechanism between individual
learning and organizational learning. In Senge, Peter M.: The Fifth Discipline,
New York, 1994, mental models and system thinking are together with team
learning, shared vision and personal mastery, described as the main elements in
organizational learning.
include the understanding that organizational learning is a fundamentally

different process than individual learning.183
In the context of this dissertation, and in accordance with Senge and
Sterman, organizational learning is understood as the improvement of system
thinking skills necessary to challenge future mental models and biases, opposed
to only learning to solve a specific problem.184 This understanding embrace the
concept of Double-Loop Learning, proposed by Argyris and Schön, focusing on
the difficult learning issues that cannot be solved unless underlying individual
and organizational values and assumptions are reexamined.185 Organizational
learning deals with the ability to learn and on a continuing basis to become better
at double-loop learning.186 For organizational interventions solving business
problems or improving business processes, organizational learning only takes
place if the underlying problem solving capability of the organization is not
ignored.187 This is also in agreement with the process consultation view of Schein,
focusing on the organization’s general ability to solve problems, rather than on
just solving a present problem.188 Taking this view on organizational learning in
SD projects means that not only should a modeling study focus on exploring the
problem and suitable solutions, it should also increase the participants ability to
think in terms of systems and causal loops in later problem solving situations.
183
For literature overview and discussions of organizational learning, see Argyris,
Chris: On Organisational Learning, 2 nd edition, Oxford, 1999, pp. 7—14; and
Kieser, Alfred and Ulrich Koch: Organizational Learning through Rule
Adaptation: From the Behavioral Theory to Transactive Organizational
Learning, Mannheim, 2000, pp. 2—26.
184
Senge, Peter M.: The Fifth Discipline, New York, 1994, p. 69; Sterman, John
D.: “All models are wrong: reflections on becoming a systems scientist”, System
Dynamics Review, Vol. 18, No. 4, Winter 2002, p. 526.
185
Argyris, Chris and Donald Schön: “Organizational Learning: A theory of Action
Perspective” (first printed as part of book with the same title in 1978), reprint in
Morgan, Gareth: Creative Organization Theory, Newbury Park, California, 1989,
p. 140; Argyris, Chris: Reasoning, Learning, and Action – Individual and
Organizational, San Francisco, 1982, p. 160.
186
Argyris, Chris and Donald Schön: “Organizational Learning: A theory of Action
Perspective”, reprint in Morgan, Gareth: Creative Organization Theory,
Newbury Park, Ca., 1989, p. 142.
187
Argyris, Chris: On Organisational Learning, 2nd edition, Oxford, 1999,
pp. 230—238.
188
Schein, Edgar H.: Process Consultation, Boston, 2000, part I, p. 194.
Organizational learning can also be discussed in terms of company

business planning. De Geus is known for the quote: “At Shell, planning means
changing minds, not making plans.” 189 He describes how the normal decision-
making processes in organizations can be viewed as learning processes, where
mental models are improved and aligned through the ongoing dialogue, and
where he perceives one of the main problems to be the speed of learning of such
groping processes. De Geus furthermore recommends striving for acceleration of
the institutional learning through using business planning as a tool to change the
mental models of the decision makers, recognizing the mutual influence between
planning and mental models.
Based on cybernetic tradition, Milling offers a conceptual illustration of
organizational learning (figure B-6) as a number of interrelated second-order
individual learning processes, combining a feedback-loop based on observations
of actual system behavior (adaptive learning), and a feedback-loop representing
individuals modification of mental models (double-loop-learning).190 Due to the
interaction between the individual learning cycles, the organizational learning is
more than a mere summation of individual learning processes. Kim and Senge
have proposed a somewhat similar illustration, of what they call the
Organizational Learning Cycle, where changes of mental models also take place
as second-order learning (double-loop learning). 191
189
De Geus Aire P: “Planning as Learning”, Harvard Business Review, March-
April 1988, p. 70.
190
Milling, Peter: “Organisationales Lernen und seine Unterstützung durch
Managementsimulatoren”, in: Zeitschrift fåür Bestriebswirtschaft, 65 Jg.
Lernende Unternehmen (Sonderausgabe), 1995, pp. 98—100.
191
Kim, Daniel H. and Peter M. Senge: “Putting systems thinking into practice”,
System Dynamics Review, Vol. 10, Nos. 2-4, Summer-Fall 1997, pp. 280—281.
This framework is more comprehensive, and includes a number of different types
of learning, but is as a consequence less intuitively understandable compared to
the framework of Milling.
Observed
Behavior
Comparison
Comparison
Statements
Comparison
Expectations
Comparison
Comparison Comparison
Comparison
Comparison Mental
Comparison Real World
Comparison
Models
Comparison
Comparison
Comparison
Reflection
Decision
Action
Figure B-6: The basic structure of organizational learning 192
This framework more explicitly operates with an organization’s shared

mental model, and when changes in individual mental models result in changes in
the organization’s shared mental models, it is called Organizational Double-Loop
Learning. Due to the significant time-delays in learning based on behavior of real
systems and the often seen breakdowns in the learning cycle involving real life
actions, Kim and Senge argue for the importance of Managerial Practice Fields.
Managerial Practice Fields are settings where teams who need to take action
together can learn together and benefit from virtual worlds allowing in a short
time to learn long-term systemic consequences of decisions (without real-life
consequences of mistakes) and as a by-product improve their learning
192
Own translation of figure in Milling, Peter: “Organisationales Lernen und seine
Unterstützung durch Managementsimulatoren”, in: Zeitschrift für
Betriebswirtschaft, Ergänzungsheft 3/95, Lernende Unternehmen (Sonder-
ausgabe), 1995, p. 100.
capabilities.193 Examples of virtual worlds or simulation games used in business

context are Senge and Sterman’s Claims Learning Lab, Repenning and Sterman’s
Maintenance Game and the often-described Beer Game.194
The opportunity to significantly speed up organizational learning through
the usage of management simulators is illustrated by a “short-cut” in the system
behavior feedback-loop in Milling’s framework (figure B-7). The arrows from
Mental Models to Formal Models reflect that a formal model should represent a
transformation of the mental models of the decision-makers. If the figure should
illustrate a participative model building process, rather than a management
simulator, one could argue that the arrows should go both ways due to the
challenge and alignment of mental models seen in the modeling processes itself.
This type of learning would be complementary to the learning achieved by
simulating the model.
193
System Dynamics Review, Vol. 10, Nos. 2-4, Summer-Fall 1997, pp. 278—279
and p. 286.
194
Senge, Peter M. and John D. Sterman: “Systems thinking and organizational
learning: Acting locally and thinking globally in the organization of the future”,
European Journal of Operational Research, Vol. 59, No. 1, 1992, pp. 146—148;
Repenning, Nelson P., Sterman, John D.: “Nobody Ever Gets Credit for Fixing
Problems that Never Happened: Creating and Sustaining Process Improvement”,
California Management Review, Vol. 43, No. 4, Summer 2001, pp. 64—88; and
Observed
Behavior
Comparison
Comparison
Statements
Comparison
Expectations
Comparison
Comparison Comparison
Comparison Formal
Comparison Mental
Comparison Real World
Comparison Model
Models
Comparison
Comparison
Comparison
Reflection
Decision
Action
Figure B-7: Formal models supporting organizational learning 195
III. The Development Process of System Dynamics Models in

Corporations
Luna-Reyes and Andersen offer a discussion of the system dynamics modeling
process across five selected representatives of the classic system dynamics
literature.196 The described processes all divide the modeling process into a
number of iterative phases, varying from three to seven phases. No dedicated
195
Own translation of figure in Milling, Peter: “Organisationales Lernen und seine
Unterstützung durch Managementsimulatoren”, in: Zeitschrift für Betriebs-
wirtschaft, Ergänzungsheft 3/95, Lernende Unternehmen (Sonderausgabe), 1995,
p. 105.
196
In Luna-Reyes, Luis Felipe and Deborah Lines Andersen: “Collecting and
analysing qualitative data for system dynamics: methods and models”, System
Dynamics Review, Vol. 19, No. 4, 2003, pp. 274—279, primarily based on the
following mentioned sources: Randers, Richardson and Pugh, Roberts et al.,
Wolstenholme and Sterman.
participative model-building approach is represented in the Luna-Reyes and

Andersen overview, but such descriptions can be found in the work of Vennix,
Andersen, Richardson and Rohrbaugh.197 Among other additional interesting
modeling process descriptions, Forrester offers a generic process with six steps
from problem symptoms to improvement.198 The following subchapter,
Decompositions and Iterations in Model Development and Use, consists of a
general discussion of the modeling process used in the context of organizational
interventions, and is divided into three main sections:
a) Problem Definition and System Conceptualization
b) Model Formulation and Testing
c) Policy Formulation and Implementation
The second subchapter, Designing System Dynamics Modeling-Based

Interventions, primarily consists of discussions on the design of experimentation-
based learning cycles, knowledge acquisition, and intervention design
considerations with special focus on participative modeling processes.
The practical application and the special requirements for the development
process of management simulators, micro-worlds and simulation-based games
will not be discussed in this dissertation, due to its focus on participative
modeling processes.199
197
See Vennix, Jac A. M.: Group Model Building, Chichester, 1996, chapter 2 and
3; and Vennix, Jac A. M; David F. Andersen; George P. Richardson and John
Rohrbaugh: “Model-building for group decision support: Issues and alternatives
in knowledge elicitation”, European Journal of Operational Research, Vol. 59,
1992, pp. 28—41.
198
Forrester, Jay W.: “System dynamics, system thinking, and soft OR”, System
199
For a description and discussion of management simulators, see Größler,
Andreas: Entwicklungsprozess und Evaluation von Unternehmenssimulation für
lernende Unternehmen, Frankfurt am Main, 2000.
1. Decompositions and Iterations in Model Development

Modeling is an iterative process embedded in the dynamics of the system.200 The
establishment of a modeling project in itself is a new part and influencer of the
real system. Although the process naturally needs to start at a point and address
first things first, any step can yield insights making adjustments to an earlier step
necessary. Model revisions usually continue almost to the end of the project, and
even in implementing results or following up on results, insights can occur
changing the very foundation for the process with a new focus of the problem.201
The path towards a useful model is always to some degree a process of trial and
error, and the goal of effective procedures for model construction is to achieve
reasonable consistency rather than to eliminate all iterations.202 For practical
reasons, all descriptions of modeling processes group tasks of the modeling
process into modeling stages or phases.
a. Problem Definition and System Conceptualization

Defining the problem and conceptualizing the system is not only the most critical
part of a modeling process; it is often also the most difficult one.203 The word
problem is used without attachment of value; it might as well reflect a positive
opportunity as a negative problem. Problem-orientation is central in most system
dynamics literature, and the importance of modeling a problem, as opposed to a
system, is often stressed as a mean to develop focused, purposeful and relevant
200
See Sterman, John D.: Business Dynamics – Systems Thinking and Modeling for
a Complex World, Boston, 2000, pp. 87—88.
201
Homer, Jack B.: “Why we iterate: scientific modeling in theory and practice”,
System Dynamics Review, Vol. 12, No. 1, Spring 1996, p. 16.
202
Randers, Jørgen: “Guidelines for Model Conceptualization”, in Randers, Jørgen
(ed.): Elements of the System Dynamics Method, Cambridge, Connecticut, 1980,
p. 118.
203
In Forrester, Jay W.: “System dynamics, system thinking, and soft OR”, System
Dynamics Review, Vol. 10, No. 2, 1994, p. 253, it is stated that to describe the
system is “the most important and the least straightforward of the stages in
system improvement;” and Schein, Edgar H.: Process Consultation – MAOM
Capstone Course for the University of Phoenix, Boston, 2000, part I, p. 62 it
says: “In my own experience in solving problems and watching others solve
them, by far the most difficult step is the first one – defining the problem.”
models.204 Different people will have different views on the importance of

competing problems, and context and relevance are related to whom the
modeling is for.205 Problem focus is a way to think in needed organizational
changes, and it is linked to creating implementation orientation from the very
outset of the project, as well as acknowledging that the project is a part of the
ongoing organizational learning process of addressing problems in the
organization.206
Eden argues that the most common defaults to proper problem definition
arrive from not seeing or ignoring the complexity, or from the politics of “the
truth” being defined by a powerful or otherwise dominant person or group.207
To characterize the problem dynamically and to avoid a short-term event-
oriented worldview it is often recommended to establish a reference mode, with
graphs or data illustrating how the problem has developed in the past (and how it
might evolve further on).208 A reference mode will often be looked back at
204
Richardson, George P. and Alexander L. Pugh: Introduction to System Dynamics
Modeling with DYNAMO, Cambridge, 1981, p. 18, Sterman, John D.: Business
Dynamics – Systems Thinking and Modeling for a Complex World, Boston,
2000, p. 89. An exception from the problem-orientated viewpoint is the Strategy
Dynamics approach, where a company’s strategic architecture is modeled rather
than a problem, see Warren, Kim: Competitive Strategy Dynamics, Chichester,
2002, pp. 89—113.
205
1994, p. 261.
206
Roberts, Edward B.: “Strategies for Effective Implementation of Complex
Corporate Models”, in Edward B. Roberts (ed.): Managerial Applications of
System Dynamics, Cambridge, Massachusetts, 1978, pp. 83—89; Sterman, John
D.: Business Dynamics – Systems Thinking and Modeling for a Complex World,
Boston, 2000, pp. 87—88.
207
1994, pp. 261—262. Eden furthermore argues for the usage of cognitive
mapping in the problem definition phase, to overcome the problems of handling
complexity and politics (pp. 263—268).
208
Complex World, Boston, 2000, p. 90; Randers, Jørgen: “Guidelines for Model
Conceptualization”, in Randers, Jørgen (ed.): Elements of the System Dynamics
Method, Cambridge, Connecticut, 1980, pp. 121—122.
(referred to) in many stages of the project, and timescale should be appropriate to
the time frame where significant changes and feedbacks are likely to occur.209 The
literature in general furthermore recommends formulating a dynamic hypothesis
characterizing the problem in terms of underlying feedback, stock and flow
structures of the system.210 The dynamic hypothesis guides modeling efforts by
focusing on certain system structures.
In the system conceptualization, the first explicit structures of the model
will typically take form. Decisions will be made with regards to what parameters
and casual relations to include in the model, focus on feedback loops etc.211 In
general, the system dynamics literature agree on the value of softer techniques in
problem identification, whereas a vital discussion exists with regards to the usage
of softer techniques in the starting point of the conceptualization and model
formulation. The discussions especially focus on the advantages of causal-loop-
diagrams vs. stock-and flow diagrams in the search for identifying the most
important structures of the system. Sterman, Richardson, Pugh and de Geus often
recommend the combination of use of quantitative modeling and qualitative
modeling, and having the qualitative modeling serve the purpose of providing
good understanding of the problem, it’s main parameters, and it’s boundaries
before actually outlining the stock and flow structures in the quantitative
model.212 The discussion on whether to include soft modeling techniques in the
209
Warren, Kim: Competitive Strategy Dynamics, Chichester, 2002, p. 27.
210
Dynamics Review, Vol. 10, No. 2, 1994, p. 246; Sterman, John D.: Business
Dynamics – Systems Thinking and Modeling for a Complex World, Boston,
2000, p. 95; Luna-Reyes, L.F. and D. L. Andersen, “Collecting and analyzing
qualitative data for system dynamics: methods and models”, System Dynamics
Review, Vol. 19, No. 4, 2003, p. 275.
211
In Sterman, John D.: Business Dynamics, Boston, 2000, p. 97, it is
recommended to do a formal model boundary chart explicitly grouping
endogenous, exogenous and excluded parameters.
212
Sterman, John D.: Business Dynamics, Boston, 2000, p. 102; Richardson,
George P. and Alexander L. Pugh: Introduction to System Dynamics Modeling
with DYNAMO, Cambridge, 1981, pp. 25—26; de Geus, Arie P.: The Living
Company, Boston, 1997, pp. 70—73; Luna-Reyes, L.F. and D. L. Andersen:
“Collecting and analyzing qualitative data for system dynamics: methods and
models”, System Dynamics Review, Vol. 19, No. 4, 2003, pp. 271—296;
Hodgson, A. M.: “Hexagons for system thinking”, European Journal of
Operational Research, Vol. 59, 1992, pp. 123—136.
system conceptualization in system dynamics projects centers around how to

ensure the best quality of the conceptualization; if a softer discussion will allow
for more holistic and creative thinking, or if the more structured stock and flow
approach will ensure more rigidity and help to not ignore important parts.
Forrester represents the latter, recommending using stocks and flows as a starting
point.213 Warren, with his Strategy Dynamics approach, also stresses the
advantages of a direct mathematical approach, using fundamental resources of a
company as starting point for identifying the stock-flow structure of the model.214
This way, the resources and “harder” discussion around reference modes starts
off the modeling process, opposed to the use of softer techniques and models.
Although it should be mentioned that another resource-oriented approach,
Wolstenholme’s Step-wise Approach proposes first to identify the major feed-
back loops as a starting point for identification of the major resources, and
thereby the core stock-flow structure of the model.215
Vennix argued in a discussion at a course in Group Model Building, that if
a modeling project is going to include a simulation model, the soft modeling
efforts at the beginning of the projects should be of only limited character. If
significant time were invested in building a soft model, e.g. a larger Cause Loop
Diagram model, he would seldom recommend a later formulation of a
quantitative model.216 This might also be a natural consequence of the significant
time pressure most organizations experience. If an organization invests
significant time in softer modeling exercises, the decision-makers might perceive
that the 80% insights have been identified, and consequently that further time to
be spent in quantitative modeling would not being worthwhile.
213
214
Warren, Kim: Competitive Strategy Dynamics, Chichester, 2002.
215
Wolstenholme, Eric F.: “The definition and application of a stepwise approach
to model conceptualisation and analysis”, European Journal of Operational
Research, Vol. 59, 1992, p. 128. Wolstenholme describes the approach as a
combination between a feedback loop approach to model construction and a
modular approach to model construction (p. 136).
216
Jac Vennix, Professor at Nijmegen University, in a Course in Group Model
Buillding, Nijmegen, May, 2004.
The problem identification and system conceptualization phase does not

only structure and focus the intervention. Through the first analysis of the
problem and the microstructure, important insights are likely to occur.217
b. Model Formulation and Testing

The concept of stocks and flows are the basic building blocks of system
dynamics models. In 1968, Forrester formulated the basic principles for model
constructions, which are the foundation for most system dynamics modeling
work.218 Sterman offers in his comprehensive text book Business Dynamics a
thorough discussion and detailed explanations on how to construct, test, and
analyze models based on stock and flow structures, feedback-loops, delays,
decision-points, parameter-estimations etc.219
A basic understanding behind the usage of system dynamics is that system
behavior is a consequence of the system structure. Richardson and Pugh
emphasize the importance of having the model contain all significant components
relevant to the problem, in order to have the model generate the problem
behavior.220 This way the model offers endogenous explanations for system
behavior, as opposed to exogenous explanations, where external events cause the
behavior.221 Recognizing that a model is only a limited reflection of a real system,
the art is to identify the smallest possible set of truly significant components
necessary for the predictive ability of the model.222
217
Flood, Robert L. and Michael C. Jackson: Creative Problem Solving – Total
Systems Intervention, Chichester, 1991, p. 74.
218
Forrester, Jay W.: Principles of Systems, Cambridge, 1968.
219
Complex World, Boston, 2000.
220
Richardson, George P. and Alexander L. Pugh: Introduction to System Dynamics
Modeling with DYNAMO, Cambridge, 1981, p. 63.
221
Sterman, John D.: Business Dynamics, Boston, 2000, p. 95, argues: “system
dynamic models seek endogenous explanations for phenomena.”
222
Luna-Reyes, L.F. and D. L. Andersen, “Collecting and analyzing qualitative data
No. 4, 2003, p. 285.
Striving for high quality and efficiency in model formulation, the field of
system dynamics has dedicated much effort to the development of generic models
and model elements, which can serve as a starting point in a model building
process. Sterman offers a significant number of ‘example-models’ in his
extensive textbook, and the literature often discusses generic models, archetypes,
as well as microstructures to be used as building blocks in model formulation.223
In the field of group model building, Anderson and Richardson discuss scripts
supporting the entire modeling process.224
An important step in the modeling process, and closely linked to the model
formulation, is the model testing. Forrester and Senge identify three categories of
tests: tests of model structure, tests of model behavior, and tests of policy
implications.225 Tests of model structure include structure and parameter
verification (in comparison with knowledge of the real system), tests of extreme
situations, boundary-adequacy tests, and test of dimensions.226 Tests of model
behavior include behavior reproduction based on historical data, behavior
prediction tests, behavior-anomaly tests (study of anomaly features of the model
behavior), family-member test (comparisons to related models), surprise-
behavior tests (behavior which has gone unrecognized in the real system),
223
Complex World, Boston, 2000; Wolstenholme, Eric: “Using generic system
archetypes to support thinking and modelling”, System Dynamics Review,
Vol. 20, No. 4, Winter 2004, pp. 341—356.
224
Andersen, David F and George P. Richardson: “Scripts for group model
building”, System Dynamics Review, Vol. 13, No. 2, 1997, pp. 107—129.
225
Forrester, Jay W. and Peter Senge: “Tests for Building Confidence in System
Dynamics Models”, in Legasto, Augusto A., Jay W. Forrester und James M.
Lyneis (eds.), TIMS Studies in the Management Sciences, Vol. 14, Amsterdam,
1980, pp. 211—226. A discussion of many of these tests (although differently
labeled), as well as additional tests, can be found following a schematically
overview of formal model validation tests in Barlas, Yaman: “Formal aspects of
model validity and validation in system dynamics”, System Dynamics Review,
Vol. 12, No. 3, 1996, p. 189. Also Sterman, John D.: Business Dynamics,
Boston, 2000, pp. 845—901, offers an extensive discussion on validation of
models.
226
1980, pp. 211—216.
extreme policy tests, as well as behavior tests for boundary-adequacy and

sensitivity.227 The last category, tests of policy implications, focus on attempts to
verify that response of a policy change corresponds between the real system and
the model, and to examine the robustness of policy implications. The policy
implication tests include system improvement tests (identifying policies leading
to improvements), changed-behavior-predictions tests, and tests for the
boundary-adequacy and sensitivity of examined policies.228
Model testing is a part of the gradual process of establishing confidence in
the soundness and usefulness of a system dynamics model.229 The entire modeling
process seeks to establish trust in the model, as trust is prerequisite to gaining
learning from the model and establishing confidence in the policy implications
derived from the model. Barlas furthermore discusses validity of a model as
usefulness with respect to purpose, which also involves a discussion of the
validity of the purpose itself.230
When discussing model adequacy and achieving confidence in the model,
Forrester acknowledges that a model is a compromise between adequacy and
time and costs of further improvements.231 A model is always a limited
representation of the real world reflecting a certain worldview, and for social
systems there is no such thing as a perfect and correct model.232 Größler discusses
227
1980, pp. 217—223.
228
Forrester and Senge: “Tests for Building Confidence in System Dynamics
Models”, in Legasto, Forrester, und Lyneis (eds.), TIMS Studies in the
Management Sciences, Vol. 14, Amsterdam, 1980, pp. 224—225.
229
See Barlas, Yaman: “Formal aspects of model validity and validation in system
dynamics”, System Dynamics Review, Vol. 12, No. 3, 1996, p. 188; Forrester
and Senge: “Tests for Building Confidence in System Dynamics Models”, in
Legasto, Forrester, und Lyneis (eds.), TIMS Studies in the Management
Sciences, Vol. 14, Amsterdam, 1980, p. 210.
230
Barlas: “Formal aspects of model validity and validation in system dynamics”,
System Dynamics Review, Vol. 12, No. 3, 1996, p. 184.
231
232
See Sterman, John D.: “All models are wrong: reflections on becoming a systems
scientist”, System Dynamics Review, Vol. 18, No. 4, Winter 2002, p. 522;
bounded rationality in the process view and in the content view.233 In the system
dynamics modeling part of an organizational intervention, this can mainly be
understood as limitations in the modeling process skills and in problem content
skills.
Modeling process skills regards such elements as knowledge acquisition,
system modeling, methods and tools, testing principles, model analysis,
challenging assumptions, fostering discussions and creativity, and theories of
human learning. Also general intervention design skills such as stakeholder
involvement, ensuring agreement in the process, placing the intervention in a
broader organizational development view, and focus on implementation will
influence the rationality in the process view.
Problem content skills regarding elements as the understanding of the
problem at hand, understanding of goal-structures, identification and estimation
of the important parameters, establishment of cause-effect-relationships, etc.
If both modeling skills and problem content skills are high (i.e. relatively
low degree of bounded rationality in both views), it should be expected that the
developing process should create fruitful discussions and that the quality and
utility of the model should be high. If both modeling skills and problem content
skills are low, the risk should be high for both off-track discussions and wrong
models. On the other hand, if modeling skills are high but problem content skills
are low, the modeling process should hopefully open discussions that challenge
some of the wrong perceptions, but the end quality of the model could be
expected to be low. And lastly if problem content skills are high, whereas
modeling skills are low, then it is likely, that the problem would have been solved
better using alternative way of addressing it.
Barlas discusses practical differences for model validity with respect to
different uses of models, although the same general philosophy applies for the
Complex World, Boston, 2000, p. 846; Barlas, Yaman: “Formal aspects of
model validity and validation in system dynamics”, System Dynamics Review,
Vol. 12, No. 3, 1996, p. 187.
233
Größler, Andreas: “A Content and Process View on Bounded Rationality in
System Dynamics”, Systems Research and Behavioral Science, Vol. 21, No. 4,
July/August, 2004, pp. 319—330.
different modes of usage.234 The differences include the limited emphasis on

behavior accuracy testing for models used in theory testing in theoretical
research, and an increased focus on testing against real data in practical
applications focusing on improving real systems. For interactive simulation
games, including management flight simulators, an added step of validity must be
added regarding usage and learning elements, and for participative modeling
processes, testing of the often rather small models is likely to take place partly as
an integral part of the model-building discussions.235 For models addressing
regulators of systems, Ashby has proposed The Law of Requisite Variety, that
only variety can destroy variety, arguing that one of the validity criterion for this
type of model is that the model reflects the complexity and details of the
system.236
c. Policy Formulation and Implementation

Policy formulation includes the creation of entirely new strategies, structures, and
decision rules.237 In Forrester’s six-step modeling process, the three last steps
address policy formulation and implementation, these steps being: ‘Design
Alternative Policies and Structures’, ‘Educate and Debate’, and ‘Implement
Changes in Policies and Structures.’238 The design of alternative policies and
structure is determined based on simulation of alternatives. The alternatives to be
simulated may occur from intuitive insights gained in the previous modeling
phases, as suggestions from the experienced analyst or people from the operating
system, or as a result of a systematic change of parameter-settings. Simulations
234
Barlas, Yaman: “Formal aspects of model validity and validation in system
dynamics”, System Dynamics Review, Vol. 12, No. 3, 1996, pp. 199—200.
235
dynamics”, 1996, pp. 200—201.
236
Ashby, W. Ross: An Introduction to Cybernetics, paperback version, London,
1964, pp. 206—207. A regulator of the air traffic flows around New York is
given as example. See also Conant, Roger C. and W. Ross Ashby: “Every Good
Regulator of a System Must Be a Model of that System”, International Journal of
System Sciences, Vol. 1, No. 2, 1970, pp. 89—97.
237
Complex World, Boston, 2000, p. 104.
238
Dynamics Review, Vol. 10, No. 2, 1994, pp. 245—247.
are stressed throughout system dynamics literature as being an essential part of

the modeling process. Complementary to the many advantages of making mental
models explicit in the system conceptualization and model formulation phases
(discussed earlier in this chapter), the simulation phase offers two essential
rewards: (1) overcoming the limited cognitive ability in handling a larger number
of parameters, feedback-loops and delays, and (2) experimentation is a very
effective way of learning.239 ‘Educate and Debate’ is a step Forrester proposes
aiming at establishing consensus for implementation, addressing both active and
passive change resistance, and is likely to yield questions requiring repeated
work in the previous modeling steps. The last step in Forrester’s six step
modeling process, implementation, is strongly dependent on the success of the
activities in the ‘educate and debate’ step, and besides normal implementation
activities, it is also recommended to ensure later evaluation of the policy changes.
The implementation of the new policies consists of two conceptually
different elements: implementation elements incorporated in the modeling
process, and implementation elements belonging to traditional change
implementation roll-out activities. The implementation elements incorporated in
the modeling process are especially strong in participative modeling approaches,
such as group model building, focusing on involvement and creating improved
and aligned mental models among decision-makers.240 The traditional change
implementation roll-out activities are seldom described in the system dynamics
literature, but are extensively discussed in the literature of organizational
psychology and the normative organization development literature.
239
Environments, Boston, 1993, p. 31; Sterman, John D.: “All models are wrong:
reflections on becoming a systems scientist”, System Dynamics Review, Vol. 18,
No. 4, Winter 2002, p. 522.
240
See Vennix, Jac A. M.: Group Model Building, Chichester, 1996, p. 97.
Although, the importance for designing the modeling process from the very start
with implementation in view is also emphasized in more traditional system
dynamics articles, see Roberts, Edward B.: “Strategies for Effective
Implementation of Complex Corporate Models”, in Edward B. Roberts (ed.):
Managerial Applications of System Dynamics, Cambridge, Massachusetts, 1978,
p. 83
2. Designing System Dynamics Modeling-Based Interventions
a. Experimentation-Based Learning Cycles

A modeling process with a model well focused on the core of the problem at
hand, constructed and simulated according to best practices within the field, and
yielding highly interesting and relevant insights does not necessary result in
organizational change. The implementation challenge has many times proven to
be of such magnitude that otherwise successful modeling studies result in only
very limited change. Based on a discussion of the implementation problems, Weil
offered a list of critical success factors for implementation of system dynamics
projects, which included factors regarding focus and urgency of the problem as
well as a number of factors regarding client involvement.241 Roberts addressed
the problem with the statement that “Organizational changes (or decisions or
policies) do not instantly flow from evidence, deductive logic, and mathematic
optimization.”242 This is, as described earlier in this chapter, in accordance with
most behavioral research emphasizing many other elements beside cognition,
such as emotions, social norm, control beliefs, and group dynamic aspects. A
central characteristic of the implementation challenges in system dynamics
context is that modeling results and insights are difficult to transfer to others.243
Two conceptually different ‘movements’ have crystallized out of the
system dynamics field addressing the difficulties of transferring modeling results:
(1) Participative system dynamics modeling, and (2) Simulation-based
knowledge transfer.
Participative system dynamics modeling is mainly described in the Group
Model Building and Modeling for Learning literature. Participative modeling
seeks to bypass the difficulties of transferring modeling insights from model
builders to decision-makers by involving the decision-makers in the actual
241
Weil, Henry B.: “The Evolution of an Approach for Achieving Implemented
Results from System Dynamics Projects”, in Jørgen Randers (ed.): Elements of
the System Dynamics Method, Cambridge, Connecticut, 1980, p. 290.
242
Roberts, Edward B.: “Strategies for Effective Implementation of Complex
System Dynamics, Cambridge, Massachusetts, 1978, p. 77.
243
Decision Environments, Boston, 1993, p. 31; Vennix, Jac A. M.: Group Model
model-building activities. In research settings, active participation in the

modeling or simulation activities has indicated significantly more enduring
changes in people’s mental models compared to traditional transfer of insights.244
Enduring mental model changes are central to the implementation of sustainable
organizational change.
Simulation-based knowledge transfer includes concepts such as manage-
ment simulators, simulation games, flight simulators, learning laboratories and
managerial practice fields. Here, a user-friendly version of a model is used as an
instrument in a learning context, not as involvement in modeling activities, but as
involvement in simulation activities. The simulation-based knowledge transfer
supports investigation of possible actions without the participant having to face
the consequences of unintended side effects, and accelerating understanding of
the behavior of the underlying system model, which equals getting years of
experiences in a few days training. The effectiveness of the simulations-based
knowledge transfer does not only depend on the quality and relevance of the
underlying model, but also on the process of how it is applied; e.g. structured
formulation of hypotheses (mapping of mental models), as well as creation of
open and creative discussions.
Participative system dynamics modeling typically has a strong exploratory
focus aimed at getting new insight on system behavior of a problem, whereas
simulation-based knowledge transfer is mainly an instrument for transferring
insights from the original model-builders to another group of people.
Nevertheless, both approaches build on the understanding that experimentation is
an effective way of learning, as formulated in The Learning Cycle by Maani and
Cavana (figure B-8).
244
See Vennix, Jac A. M.: Group Model Building, Chichester, 1996, pp. 97—99.
Conceptualization
(intellectual
learning)
Reflection Experimentation
(emotional (action
learning) learning)
Figure B-8: The learning cycle for learning labs 245
Maani and Cavana describe Conceptualization as cognitive learning,

where people develop theories and hypotheses regarding system behavior, as well
as the formalization of learning outcomes from experimentation and reflection.246
Experimentation is described as a learning experience much like real-life
learning, although in an accelerated and risk-free way (“learning by doing” in a
laboratory). Reflection is described as emotional learning allowing participants to
engage with affective elements such as assumptions, attitudes, biases,
resentments etc. Consequently, the learning lab process illustrates the usage of
applying experimentation to move beyond cognitive learning, which according to
the literature as described in subchapter B.II.2. is significant for effective
learning and influencing individual’s attitudes towards behavior.
245
Taken from the figure in Maani, Kambiz E. and Robert Y. Cavana: Systems
Thinking and Modeling – Understanding Change and Complexity, Auckland,
2000, p. 112.
246
In Maani and Cavana: Systems Thinking and Modeling – Understanding Change
and Complexity, 2000, pp. 112—113, the three elements of the learning cycle
are discussed.
b. Knowledge Acquisition in Modeling Projects

Knowledge acquisition is an important aspect of designing system dynamics
interventions. Forrester point out three types of information sources for modeling
projects: mental database, the written database and the numerical database:247
Mental Database
Observation
Experience
Written Database
Numerical Database
Figure B-9: Mental database and decreasing content

of written and numerical databases 248
By ‘mental database’, Forrester means the non-documented information in

people’s minds, based on observations and experience. Forrester stresses the
importance of this type of information, arguing that it is not adequately
appreciated in the management and social sciences.249 The written database and
247
Forrester, Jay W.: “Policies, decisions and information sources for modeling”,
European Journal of Operational Research, Vol. 59, No. 1, 1992, pp. 55—58.
248
Slight modification of figure in Forrester, Jay W.: “Policies, decisions and
information sources for modeling”, European Journal of Operational Research,
Vol. 59, No. 1, 1992, p. 56. Forrester’s figure is also depicted in Luna-Reyes,
L.F. and D. L. Andersen: “Collecting and analysing qualitative data for system
dynamics: methods and models”, System Dynamics Review, Vol. 19, No. 4,
2003, p. 280, along with a discussion on the three types of databases.
249
European Journal of Operational Research, Vol. 59, No. 1, 1992, p. 56.
the numerical database naturally also provide important information to the

modeling process, although caution must be taken as the information has already
been filtered by the writer’s purpose.250 Although for the numerical database in
particular, Warren argues, that companies are often surprised when existing data
are seen in a new perspective due to the modeling efforts, and this actually often
produces new insights in itself.251 In retrieving the mental databases, a number of
different techniques are described in the system dynamics literature. Luna-Reyes
and Andersen provide an extensive overview of methods for collecting and
analyzing qualitative data from the perspective of the different modeling
phases.252 Fey and Trimble as well as Vennix also discuss a larger number of
these techniques, including the different interview forms (fully structured, semi-
structured or unstructured), questionnaires, the Delphi method, and different
versions of the Normative Group Technique.253 Information gathering includes
the opportunity - and the risk - of influencing the persons providing the
information. Especially in participative modeling projects, information gathering
is a balance between structured (and maybe even anonymous) data and
knowledge elicitation from experts and individuals and a team effort among
participators in the modeling process.
In the formulation of a quantitative model, more often than not, the
estimation of some parameters will suffer from lack of explicit or precise data.
Most literature agrees that building on a qualified guess is nearly always better
250
Forrester.: “Policies, decisions and information sources for modeling”, European
Journal of Operational Research, Vol. 59, No. 1, 1992, p. 57.
251
Own notes from Kim Warren’s presentation at the System Dynamics Conference
in Oxford in 2004.
252
Luna-Reyes, L.F. and D. L. Andersen: “Collecting and analysing qualitative data
No. 4, 2003, pp. 271—296. Besides the qualitative methods often mentioned in
the system dynamics literature like interviews, focus and Delphi groups,
observations, etc., the article also include analysis methods like hermeneutics,
discourse analysis, grounded theory, ethnographic decision models and content
analysis.
253
Fey, Willard and John Trimble: “The Evaluation and Development of Knowledge
Acquisition in System Dynamics Studies”, in Proceedings, System Dynamics
Conference, System Dynamics Society, 1992, pp. 173—182; Vennix, Jac A. M.:
Group Model Building, Chichester, 1996, pp. 175 and pp. 187—188.
than leaving out an important parameter in a model.254 Warren argues that in the
everyday business, decisions are made implicit based on judgment and
assumptions of the parameters, where data is missing.255 Asking for qualified
guesses is just a way to make such judgments and assumptions explicit.
c. Designing Participative Modeling Interventions

A system dynamics study will often be part of a larger package addressing an
important business problem.256 Apart from the seldom-used expert models
developed ‘behind the scenes’, most system dynamics projects in organizational
interventions will include some elements of participative modeling. Participative
system dynamics modeling is primarily described in terms of Group Model
Building and Modeling for Learning, and the participative modeling literature in
general position the modeling process as an integral part of management
discussions.257 Vennix, Andersen, and Richardson discuss the goals of Group
Model Building with regards to: (1) creating a climate for team learning, problem
understanding and mental model alignment, (2) fostering consensus, and (3)
creating acceptance and commitment with the decisions.258 In Modeling as
254
Managementsimulatoren”, in: Zeitschrift für Betriebswirtschaft, Ergänzungsheft
3/95, Lernende Unternehmen (Sonderausgabe), 1995, p. 104.
255
256
Weil, Henry B. and Kenneth P. Veit: “Corporate Strategic Thinking: The role of
System Dynamics” in Peter M. Milling and Erich O.K. Zahn (Eds.): Computer-
Based Management of Complex Systems, Proceedings of the 1989 International
Conference on the System Dynamics Society, Stuttgart, 1989, p. 67, discuss how
system dynamics is often part of a larger package addressing a problem, e.g. in a
strategy consulting project.
257
See Vennix, Jac A. M.: Group Model Building, Chichester, 1996, p. 4; Lane,
p. 70; Richmond, Barry: “The Strategic Forum: aligning objectives, strategy and
process”, System Dynamics Review, Vol. 13, No. 2, 1997, p. 131; Morecroft, J.
D. W.: “Executive knowledge, models and learning”, European Journal of
Operational Research, Vol. 59, 1992, p. 13.
258
See Vennix, Jac A. M.: Group Model Building, Chichester, 1996, p. 6;
Andersen, David F., George P. Richardson and Jac A. M. Vennix: “Group model
building: adding more science to the craft”, System Dynamics Review, Vol. 13,
No. 2, 1997, p. 191.
Learning, Lane strongly stresses the learning objectives, such as changing mental
models, improving intuition, allowing risk-free experimentation and revealing
systemic complexity; as vehicles to business improvement.259 Morecroft discusses
the usage of system dynamics in the context of executive dialogue and debate:
changing business environment
recognised strategic issue
Knowledge base &

executive debate and dialogue mental models
Concepts and
theory plus
facilitation
action plans and change
facilitation MAPS
words, diagrams
FRAMEWORK
friendly algebra
MICROWORLD Simulation models

&
gaming simulators
Figure B-10: Maps, frameworks and microworlds 260
259
Lane, David C.: “Modelling as Learning: A consultancy methodology for
enhancing learning in management teams”, European Journal of Operational
Research, Vol. 59, No.1, 1992, p. 72.
260
Taken from the figure in Morecroft, J. D. W.: “Executive knowledge, models
and learning”, European Journal of Operational Research, Vol. 59, 1992, p. 14.
The full article is also printed in the book: Morecroft, John D. W. and John D.
Sterman (eds.): Modeling for Learning Organizations, Portland, Oregon, 1994,
pp. 3—28 (the book is in principle a reprint of the special issue of European
Journal of Operational Research, Vol. 59, No. 1, 1992).
The upper-right rectangles in figure B-10 represent a stack of individual

knowledge bases and mental models, and these will always implicitly be active in
executives recognizing and discussing strategic issues. The lower-right rectangle
represents explicit application of models in the dialogue, with the usage of maps
(examples are written lists, or causal diagrams) and a framework combining maps
with concepts and theory.261 The term Microworlds encompasses in Morecroft’s
understanding all the labels and the connections in figure B-10, and is used for
the whole learning environment.
Figure B-10 does not necessarily imply a dedicated group model building
process, but could insure involvement through individual interviews and model
discussions (to make sure the model reflects each individual’s mental model), and
subsequent usage of the model as a management simulator. In general, though,
most group model building and modeling for learning literature emphasize that
insights are gained during a model-building process, rather than after the model-
building process.262 Andersen, Richardson and Vennix argue for the value of a
group modeling structure, where top management and “doers” are together for an
extended period. Argyris argue, for interventions in general, that the more
difficult an organizational intervention, and the more internal commitment is
necessary for effectiveness, the more the client needs to be directly involved in
the design, execution and monitoring of the changes with an open and
experimental attitude.263 This also contributes to placing the intervention in the
continuous learning and shaping of the organization and its change readiness.264
With the importance of creating shared mental models and establishing
consensus and commitment for decisions, one of the important design decisions
is who to involve in the modeling. If the modeling is used as a tool in strategic
planning, top management is likely to be relevant to involved. If the modeling is
used to address a specific problem, the main stakeholders should be involved.
Also, a project can operate with a structure, that allows for involvement of a large
261
Morecroft, J. D. W.: “Executive knowledge, models and learning”, European
262
See Vennix, Jac A. M.: Group Model Building, Chichester, 1996, pp. 97—99.
263
View, Reading, Massachusetts, 1970, p. 83
264
Schein, Edgar H.: Organisationspsykologi, Danish translation, Herning, 1990,
p. 40; Argyris, Chris: Interventions Theory and Method – A Behavioural Science
View, Reading, Massachusetts, 1970, chapter 1 and 2.
number of people, using a setup with steering committee, a core group, reference
groups etc. 265 The involvement will often be designed with the dual purpose of
acquiring information and ideas, balancing differing perspectives and interests,
and establishing internal commitment.
A way to coordinate the stream of information in a group model building
process is the usage of a Workbook. Vennix describes a Workbook as a
document consisting of text and diagrams, and presenting material as well as
asking questions.266 The Workbook can be seen as an instrument to summarizing
workshops and preparing for the next workshops and is a way to formalize the
information stream in a modeling project. Akkermans discusses the concept of
The Workshop-Workbook Cycle, being a structured, iterative process involving a
number of in-house discussions (among consultants) and client consultations, as
well as Workbook distribution and feedback, between all workshops.267
When designing a modeling intervention, one of the decisions is whether
or not to employ a preliminary model. The advantages include speeding up the
process and reducing participants time investment, but the major risk is to
infringe the degree of the group’s sense of model ownership, and therefore
Vennix stresses the importance of not being defensive about a preliminary model
when discussing and refining the model. 268 Vennix furthermore offers a
schematic overview of factors indicating in which situations it is appropriate to
use a preliminary model, as well as other design decisions such as
communication pattern, usage of soft methods to initiate the process, and
facilitator preparation.269
265
See Akkermans, Henk: Modelling With Managers, Breda, The Netherlands,
1995, p. 87; Vennix, Jac A. M.: Group Model Building, Chichester, 1996,
p. 114.
266
267
Akkermans, Henk: Modelling With Managers, Breda, 1995, p. 89.
268
Vennix, Jac A. M.: Group Model Building, 1996, p. 113.
269
Vennix: Group Model Building, 1996, p. 132.
d. Facilitation of Participative Modeling Interventions

Effectiveness of using modeling in the learning cycle of an organizational
intervention not only depends on intervention design and frameworks, but also on
facilitation skills in knowledge elicitation and the handling of the discussions of
options and consequences.270 One often encounters discussions on how to reduce
dependency of a facilitator through the usage of effective scripts, sub-
components and rigor approaches, but nevertheless most group model building
literature stresses the importance of the quality of the facilitator role in system
dynamics projects. 271
In group model building literature, the facilitation role is frequently called
the role of the modeling team, indicating the usefulness of having more than one
facilitator. Vennix argues that a minimum of two people should guide the
process, a facilitator and a recorder, however the recorder could be from the
client organization.272 Richardson and Andersen suggest that for supporting the
modeling process for large groups addressing weighty problems, five roles that
are best handled by five individuals are: The Facilitator (focus on group
processes and knowledge elicitation), The Modeler/Reflector (focus on model
formulation, and sometimes feeding own sketches back to the group), The
Process Coach (focus on the dynamics of individuals and subgroups), The
Recorder (writing down and sketching group proceedings), and The Gatekeeper
(often a person from the client organization bridging the modeling process and
the client problem).273
270
271
See Andersen, David F and George P. Richardson: “Scripts for group model
building”, System Dynamics Review, Vol. 13, No. 2, 1997, p. 108. See also
No. 2, 1997, p. 195, where furthermore The Gifted Practitioner hypothesis is
stated, indicating that some facilitators are skilled to a degree where choice of
intervention method is less important – and also the other way around - if
facilitation skills are lacking, no methods and tools can assure an effective
intervention.
272
273
Richardson, George P. and David F. Andersen: “Teamwork in group model
Effective learning in teams requires focus on interpersonal relations and

group dynamics.274 The facilitator must strive to create a climate allowing for
open-mindedness, learning and creativity. One aspect of ensuring open-
mindedness is to strive for acceptance of the model by ensuring that the
participants understand all model elements and that the model encompasses all
major elements from each participant’s mental model. Participant acceptance and
understanding of the model is critical for gaining trust in the model, which is a
prerequisite for learning. Vennix discuss the equation, E = Q x A (Effectiveness=
Quality x Acceptance), implying that as long as a model is correct in the larger
perspectives, participant’s acceptance must be strived for, even when it means
compromising on less significant modeling issues.275
Especially in the problem definition phase, the facilitator has to manage
not only idiosyncratic views on the problem held by the individual decision
makers, but also the problem of the decision makers “speaking different
languages.”276 The different use of language, often discussed in terms of codes, is
a critical problem in communication and interpretation processes. Codes are
important in determination of meaning in communication, and creating shared
meaning requires use of similar codes, which is why Scheper recommends giving
considerable attention to codes in the determination of the problem in the early
phase of a group decision process.277 Eco has proposed a widely used model of
communication (figure B-11).
274
View, Reading, Massachusetts, 1970, p. 83.
275
276
Scheper, Willem J.: Group Decision Support Systems, Tilburg/Utrecht, 1991,
p. 23.
277
Scheper.: Group Decision Support Systems, 1991, p. 143.
coded message as interpreted

sender channel addressee message as
meassage expression
content
codes context, codes

circumstance
Figure B-11: A model of communication 278
Spending time on the modeling process to agree on definitions and the

usage of language in understanding problems and issues contributes not only to
align codes between individuals, but should also contribute to the usage of larger
“chunks” of codes. Miller uses an example of a novice in radiotelegraphic code
having to decode every “dit and dah”, whereas more experience will allow him to
deal with each letter as a chunk, and later even operate with words as large
chunks.279 In participative modeling processes, this would mean usage of larger
chunks to allow discussions to build on accumulative insights.
The facilitator role also deals with reducing information filters in
information acquisition. Morecroft discusses the following five different
information filters in decision-making processes: 280
(1) people’s cognitive limitations,
(2) operating goals rewards and incentives,
(3) information, measurement, and communication systems,
(4) organizational and geographical structure, and
(5) tradition, culture, folklore, and leadership.
278
Slight simplification of the model in Eco, Umberto: A Theory of Semiotics,
Bloomington, 1976, p. 141. The model has originally also focus on sub-codes,
which is especially relevant in the discussion of undercoding and overcoding.
The simplification is mainly inspired by the use in Scheper, Willem J.: Group
Decision Support Systems, Dissertation, Tilburg/Utrecht, 1991, p. 143.
279
Vol. 63, No. 2, March 1956, p. 93.
280
Journal of Operational Research, 59, 1992, p. 18.
Although the facilitator is also subject to these filters, conscious considerations of

how to reduce the information filtering is expected to improve the modeling
process.
The last facilitator role to be discussed in this subchapter is the role of
ensuring momentum in the process, which includes handling attitudes to the
project and internal conflicts.281 The Wallow Curve, figure B-12, illustrates an
often-experienced tendency of a peak of irritation and conflict in the project, after
the first few workshops, when the initial period of getting to know the problem
and conceptualize the issues is over.282 The irritation and conflict often reflects
uncertainty regarding the quality and usefulness of the model in the early part of
the design phase, and also that certain dominant project members are trying to
push through their own ideas. It is important to note that the facilitator should not
necessary aim at minimizing the level of cognitive conflicts, as the highest
decision quality often is found in situations where groups have some degree of
conflict, with the reasoning being that disagreement fosters thorough
investigation, information processing and searching for alternatives.283
Furthermore, disagreement and vivid discussions can be a stimulating element in
fostering innovation and creativity.284 Only the extremes should nearly always be
avoided: too much cognitive conflict between group members will result in
ineffective communication, and too little conflict will result in Groupthink.285
281
For further descriptions on facilitation of group model sessions, see Vennix, Jac
A. M.: Group Model Building, Chichester, 1996, pp. 140—171.
282
See Akkermans, Henk: Modelling With Managers, Breda, The Netherlands,
1995, pp. 91—92, where credit for the term The Wallow Curve is given to
McKinsey.
283
Vennix, Jac A. M.: Group Model Building, 1996, p. 154.
284
Numerious examples are seen in the art and music industries.
285
Decision quality as a function of conflict level is often seen depicted as a
reversed U-curve.
problem
understanding
irritatoin &
conflict level
inter- WS1 WS2 WS3 WS4 WS5

views
Figure B-12: The Wallow Curve at work 286
286
Taken from the figure in Akkermans, Henk: Modelling With Managers, Breda,
1995, p. 91.
C. A Case Study Using Participative System Dynamics
Modeling in the Implementation of a Sensitive Change
Project
I. Research Considerations for the Case Study Application

The application of system dynamics modeling has proven useful in exploratory
organizational interventions because it supports learning, alignment of mental
models, group decision-making and the creation of commitment. Consequently,
system dynamics has contributed to the formulation of a number of strategic
initiatives in corporate settings. The initial question being investigated in this
dissertation is: is it purposeful to apply system dynamics modeling in the
implementation of strategic initiatives?287 The research is mainly based on
discussions of literature, but is also complemented by a case study in the action
research tradition with dual focus on the implementation of planned change as
well as knowledge development.288
Voss, Tsikriktsis, and Frohlich categorize case research based on purpose:
exploration, theory building, theory testing, or theory extension/refinement.289 In
these terms, the purpose of the present case study is theory building, since no
particular existing theory regarding the use of participative modeling in a planned
change organizational intervention viewpoint supports the research. Furthermore,
287
An alternative scientific approach could have been to investigate the challenges
of bridging the two problem-solving cycles in organizational change projects,
and to evaluate and compare the usage of a number of different methods and
tools in such a setting. The arguments for the applied scientific approach include
the decision to make one in-depth case study in action research tradition, rather
than to spread focus and efforts over a number of field studies.
288
For definitions of action research, see Cummings, Thomas G. and Christopher G.
Worley: Organizational Development and Change, Ohio, 2001, p. 23.
289
Voss, Chris, Nikos Tsikriktsis, and Mark Frohlich: “Case research in operations
management”, International Journal of Operations & Production Management,
Vol. 22, No. 2, 2002, p. 198.
84 C. Case Study
the research approach involving a single site case study does not have the design
characteristics necessary to qualify for testing a theory. 290 Discussing whether to
use one or more case studies in a research project, Stake argues for the benefit of
maximizing knowledge through focusing on a single, in-depth case study, rather
than to disperse focus over a number of case studies.291 Other researchers,
however, champion the viewpoint that theory building should preferably be based
on insights from a larger number of case studies.292 The main argument for the
decision to use only one case study in the present research is the iterative seek-
and-learn relationship that has taken place between the case study and the
theoretical discussion on how best to apply modeling in change management.293
Furthermore, a survey approach was not considered a possible alternative, due to
lack of known cases utilizing system dynamics modeling in change management
context. As an alternative to a multiple case study approach or a survey approach,
the single-site case study approach puts the spotlight on one instance to be
investigated in more detail and thereby concentrates efforts rather than trying to
290
This fits well with the quote from Andersen, David F.; George P. Richardson
and Jac A. M. Vennix: “Group model building: adding more science to the
craft”, System Dynamics Review, Vol. 13, No. 2, 1997, p. 196: “case studies are
only suitable to generate hypotheses, not to test them rigorously.”
291
Stake, Robert E.: The Art of Case Study Research, Thousand Oaks, 1995,
pp. 4—5, argues that a good instrumental case study not necessarily needs to
examine a typical case, as an unusual case helps illustrate matters often
overlooked in typical cases. Furthermore, he argues that even when designing a
collection of cases, representation is often difficult to defend.
292
See Eisenhardt, Kathleen M.: “Building Theories from Case Study Research”,
Academy of Management Review, Vol. 14, No. 4, 1989, pp. 532—550; and
Leonard-Barton, Dorothy: “A Dual Methodology for Case Studies: Synergistic
use of a Longitudinal Single Site with Replicated Multiple Sites”,
Organizational Science, Vol. 1, No. 3, 1990, pp. 248—266. It should be noted,
that even though the present dissertation includes only one case study, it
indirectly benefits from a larger number of case studies from the group model
building literature, as many of these have elements of a change management
focus.
293
See also discussion on exploratory case studies, Yin, Robert K.: Case study
Research, 3 rd edition, Thousand Oaks, 2003, p. 23.
C. Case Study 85
cover a large number of instances. The aim is to “illuminate the general by

looking at the particular”.294
The design of the case study is based on literature studies as well as the
external facilitator’s experiences of organizational interventions due to 10 years
of management consulting experience in business process reengineering and
change management.295 The literature studies formed the first basis for the theory
development on how to benefit from system dynamics modeling in change
management context.296 During the case study, the practical insights gained
served in parallel as inspiration on where and how to focus further literature
studies. The literature studies include primarily literature from the field of system
dynamics, the field of organizational psychology, and descriptive management
literature regarding implementation of strategic initiatives, often called change
management literature.
The literature of system dynamics contributes in general to the case study
with a dynamic perspective on problem understanding and individual learning,
with the emphasis on model-building based on stock and flow structures, and
with computer simulations of aggregate time-series. Group Model Building
literature contributes in particular with focus on modeling as a framework for
discussions facilitating the sharing and aligning of mental models. The core of
the model used in the case study consists of two aging chains, and the issues of
delays constitute a central part of the analysis. Respecting the scientific
foundation and the large number of proven successful system dynamics
interventions, the case study does not intend to change or challenge the system
dynamics modeling process as described in the leading system dynamics
literature, but rather the case study seeks to place the modeling process in a
different context: applying system dynamics modeling in planned change
organizational intervention.
294
Denscombe, Martyn: The Good Research Guide for small-scale social research
projects, Philadelphia, 2 nd edition, 2003, p. 30.
295
The external facilitator (the author of this dissertation) has most of her
management consulting experience from IBM Management Consulting and
Deloitte & Touche Consulting Group (now split into Deloitte Consulting and
Braxton).
296
Yin, Robert K.: Case study Research, 3 rd edition, Thousand Oaks, 2003, p. 28,
emphasizes the importance of theory development as a part of case study design,
regardless of whether the purpose of the case study is to develop or test theories.
86 C. Case Study
The literature of organizational psychology contributes in particular to

theories of action research, individual learning processes and group psychology.
Action research takes the theoretical point of departure that dynamic systems
such as organizations can best be examined using carefully planned, concept-
based interventions. Thus, the evaluation cannot involve traditional evaluation
schemes, such as the use of control groups, as the complex social environment is
not controllable to a degree that allows isolation of the true variables.
Furthermore, all evaluation processes of human systems influence the system,
and must be designed in such a way that academic evaluation is not counter-
productive with regards to the desired results of the intervention. This is in
accordance with a premise of action research, namely that action research
projects should always be adding value to a given organizational intervention,
and therefore not differ from the ethics underlying consulting and counseling
projects.297 The theories regarding individual learning and group processes are
mainly included as a focus on creation of changes in attitudes and intentions,
which also play a major part in the applied evaluation framework.
Change management literature offers recommendations on creating the
foundation for the rollout of the strategic initiative, including thoughts on the
development of change leaders.298 The case study also benefits from the more
traditional change management virtues such as intervention planning, stakeholder
management, and implementation planning and review:299
• Intervention planning includes the definition of business objectives and
targets, the framing of the intervention, the identification of consultation
relationships, roles and responsibilities in the project organization, and
297
Schein, Edgar H.: Organisationspsykologi, Danish translation, Herning, 1990,
p. 253.
298
Vol. 33, No. 3, 2001, p. 45. See also Roberto, Michael and Lynne Levesque:
“The Art of Making Change Stick”, MIT Sloan Management Review, Summer
2005, Vol. 46, No. 4, Summer 2005, p. 56, where modeling is relevant with
regards to what they call chartering and learning.
299
Intervention planning, stakeholder management and implementation planning
and review are three conceptually different, but also strongly interdependent
tasks.
C. Case Study 87
time and budget planning.300 Thus, this activity - as the other two
activities discussed next - contains both system dynamics and non-
system dynamics parts, for instance the coordination with related
projects and activities.
• Stakeholder management involves a thorough analysis of all the major
interest groups and individuals who have significant influence—directly
or indirectly—on the success of the intervention. Focus is on the
stakeholders’ interests and power, their importance for solution design
and implementation, and on relevant means of involvement and
communication.301 Stakeholder analysis is vital to intervention planning,
both to secure relevant parameters to be included in the process, and to
secure appropriate involvement of and communication with
stakeholders, including employees.302
• Implementation planning and review deals with the planning of the
implementation, including an action plan, communication plan and a
clear delegation of responsibilities.303 The communication plan develops
300
Roberto, Michael and Lynne Levesque: “The Art of Making Change Stick”, MIT
Sloan Management Review, Summer 2005, Vol. 46, No. 4, Summer 2005,
pp. 55—56, describe intervention planning (although calling it chartering). In
Koningswieset, Roswita and Alexander Exner: Systemische Intervention,
Stuttgart, 1998, it is throughout the book discussed how intervention planning
can be seen as creating an overall architecture and design of the intervention,
including selection of relevant techniques.
301
Flood, Robert L.: Solving Problem Solving, 1995, p. 20, places ‘organizational
politics’ as one of the four key dimensions to take into consideration in a whole
organizational system view, the three others being organizational processes,
design and culture. Borum, Finn: Strategier for organisationsændringer,
Copenhagen, 1995, pp. 79—92, offers a discussion on politics in change
processes.
302
For discussions on “Employee Involvement”, see both Cummings, Thomas G.
and Christopher G. Worley: Organizational Development and Change, Ohio,
2001, p. 317; and Thun, Jörn-Henrik, Peter M. Milling, and Uwe Schwellbach:
“The Impact of Total Employee Involvement on Time-based Manufacturing“, in
“What Really Matters in Operations Management“, Proceedings of the European
Operations Management Association, 8th International Annual Conference,
2001, pp. 133—135.
303
For implementation actions, focus is typical on the practical adjustments of
business processes and organizational design with single-loop orientation (result
88 C. Case Study
over the course of the intervention and includes elements such as

motivating change and the communication of visions, results,
implementation plan and successes. This activity furthermore includes
the establishment of procedures for reviews and corrective actions.
The case study addresses problems and issues in bridging between the top-
management launching a strategic initiative and the sustainable rollout of the
organizational change. The main arguments for selecting this specific case
include: (1) that the change initiative represented important and classical change
management challenges such as significant change resistance, strong emotional
attitudes, and that the decision to launch the initiative was made without
involving the key managers responsible for the implementation, and (2) the
change project had characteristics making it likely to gain from the insights from
a system dynamics model, due to the importance of understanding the influence
of delays in an aging chain. Furthermore, the writer of this dissertation had
extensive knowledge of the company and easy access to the decision-makers, due
to previous engagements with the company as well as due to personal
relationships. The existence of personal relationships allowed for increased top
executive involvement in the design of the modeling process, and also allowed
for extensive theoretical discussions with the project owners on the practical
implications of the usage of system dynamics modeling in corporate settings.304
However, personal relationships also create biases in terms of design
opportunities as well as participant perceptions of the researcher’s objectivity,
which has to be taken into account especially in the evaluation of the case study.
As a conclusion on the research considerations: the main research idea has
been to investigate the usefulness of participative system dynamics modeling in
orientation) rather than double-loop orientation (process orientation), see

Borum, Finn: Strategier for organisationsændringer, Copenhagen, 1995, p. 21
and p. 37. Although, having the implementation plan including procedures for
review and corrective actions, the scene will be set for later double-loop
learning. The implementation plan must take into consideration all four
organizational dimensions (processes, design, culture and politics) as described
Flood, Robert L.: Solving Problem Solving, 1995, p. 20.
304
Stake, Robert E.: The Art of Case Study Research, Thousand Oaks, 1995, p. 4,
recommends to “pick cases which are easy to get to and hospitable to our
inquiry.”
C. Case Study 89
the bridging of the two problem-solving cycles: the cycle of diagnostics and
decision-making, and the cycle of change management.305 In other words, the
dissertation investigates the utilization of proven, successful system dynamics
methods and tools in a new, specific context. An alternative scientific approach
could have been to investigate the challenges of bridging the two problem-
solving cycles in organizational change projects, and to evaluate and compare the
usage of a number of different methods and tools in such a setting. The
arguments for the applied scientific approach include the decision to make one
in-depth case study in action research tradition, rather than to spread focus and
efforts over a number of field studies. A survey approach was not considered a
possible alternative, due to lack of known cases utilizing system dynamics
modeling in change management context. Regarding the use of the case study, it
is stressed that the purpose primarily is to generate inspiration to the focus of the
research due to the seek-and-learn nature of the case study, seeking for theory
building in the action research tradition. Additionally, the case study serves as a
practical illustration in chapter D, in the discussion of the usage of system
dynamics modeling in the implementation of strategic initiatives. The next
subchapter describes the case itself: the problem context, the intervention
planning, the model, and its results. Following the case description comes a
subchapter outlining the evaluation framework for the case study as well as the
actual evaluation.
II. Case Study: Refining and Implementing a Location Strategy
1. The Problem and its Context

The Case Study Company is a major, international company, which is a market
leader in its main product area. Research and development (R&D) is a large and
critical part of the company’s sustainable competitive advantage, reflected by the
fact that approximately every third employee works in R&D. The company has a
strong tradition of employee empowerment and is a relatively un-hierarchical
305
For a discussion on the two cycles of the problem-solving process,
see chapter A.I.
90 C. Case Study
organization. The case study was carried out in one of the major R&D divisions,
consisting of a number of rather different R&D business units.
Strategically, the company works towards creation of a truly global
company in terms of the entire value chain distribution, and as part of these
efforts the board of directors launched a strategic initiative regarding a balanced
location strategy.306 For R&D, specific targets were decided pertaining to the
distribution of the number of R&D employees placed in high-cost countries (e.g.
the USA and Western Europe) vs. the number of R&D employees placed in low-
cost countries (e.g. India, China, Eastern Europe). At the time of the launch of
the initiative, fall 2004, the company had significantly more R&D employees in
high-cost countries with the consequence of relatively high development costs.
The cost of an R&D employee in a high-cost country was at this time
approximately four times the cost of an R&D employee in a low-cost country. It
was a sensitive issue due to the fear among employees in high-cost countries that
the future could bring reductions of staff in high-cost locations as seen in many
other companies in the USA and Western Europe. The situation at the case study
company, however, included strong growth expectations and the company did not
intend to weaken existing high-cost locations in terms of the absolute numbers of
employees. The expansion of low-cost R&D locations should reflect a worldwide
growth of the R&D to gain speed in time-to-market and an extended product
portfolio.307 The motivation was not only to increase capacity and cost-efficiency,
but also to strengthen the local presence in growing markets such as China and
India with a view to increased future sales. New employees in low-cost locations
were primarily supposed to take over existing tasks from high-cost locations.
306
Sloan Management Review, Summer 2005, Vol. 46, No. 4, Summer 2005, p. 53,
define strategic initiatives as “corporate programs aimed at creating new
business processes or transforming existing ones to accomplish major goals,
such as enhancing productivity or improving customer service.” The new
location strategy has major impact on both R&D processes, delivery processes,
and sales support processes as well as the company’s risk profile, currency
spread, and organizational culture elements, which is why it was perceived as a
major strategic initiative in the company.
307
Perlitz, Manfred: Internationales Management, 5 th edition, Stuttgart, 2004, p. 72,
discusses factors for trade between high-cost and low-cost countries. These
factors are also relevant for decisions on distribution of employment.
C. Case Study 91
This way the company wanted to free capacity of experienced senior R&D
employees in high-cost countries to be used in new, challenging R&D projects.
The objective of the organizational intervention consisted of both ‘harder’
and ‘softer’ elements. The softer elements included the creation of acceptance
and commitment to the launched strategic initiative among key implementers, and
the harder elements included understanding the most influential parameters in
building up capacity in low-cost locations, with special focus on productivity and
costs. The process should seek to define the ‘ideal’ implementation approach,
balancing board expectations regarding a reduced cost/capacity ratio with an
effective and realistic implementation plan. The strategic initiative was intended
to stimulate a reinforcing growth loop as depicted in Figure C-1. As such, the
project context was not to explore and identify potential new strategies or
policies, but to implement a given decision in the best possible way and to gain
commitment and understanding from the key implementers. Thus, the
intervention was more of a change management project than a policy formulation
project.
strengthening low-cost
locations without reducing
high-cost locations
+ + +
+ maintained
company improved reduced
time-to- unit-price motivation in
growth high-cost
market of R&D
+ locations
+
improved
competitiveness
+
+
Figure C-1: The reinforcing growth loop underlying the intervention 308
308
Causal-loop diagrams are normally constructed in such a way that the variable
names do not indicate the direction of change. In Figure C-1, the variable names
include the direction, e.g. company growth, reduced unit-price etc. This was
done for communication purposes, emphasizing how the reinforcing growth loop
was intended to work.
92 C. Case Study
The board of directors launched the strategic initiative as an integral part

of the yearly business planning and budgeting process. Although, before
incorporating the R&D location strategy in the budget, board approval was
needed of the detailed implementation plan. The plan should be detailed with
regards to the R&D demand (tasks and project to be carried out in each location),
the planned staffing in each location, and planning of the transformation (hiring,
transfer of tasks, training, process changes and communication strategy).
Consequently, the people responsible for making the changes happen needed to
be thoroughly involved in the planning. The board also gave specific targets for
the planning: at year-end 2007, 25% of the division’s R&D personal should be
based in low-cost locations (as opposed to 10% at year-end 2004), and the new
strategy should not negatively influence the existing preliminary 3 years cost
budget. Also, negative productivity effects in 2005 should be avoided.
The board recognized the significant change management challenges
following the launch of the initiative. Transfer of jobs to low-cost locations was
an issue often discussed at that time in the high-cost society in general, and also
at the company’s high-cost locations. For this reason, a decision was made and
clearly stated and repeated in corporate communication: “there will be no lay-
off’s due to the location strategy.” Naturally, this was only one piece in the
change management puzzle, as many concerns remained amongst the employees:
What will happen in the long run? What will the consequences be for my job
situation in the short run? Can “they” deliver the same high quality “we” deliver?
Is it effective to “spread” tasks? These questions, and many more, would have to
be addressed in the change process. Furthermore, for the middle management
responsible for carrying out the changes a number of political challenges existed
with regards to individual power bases and interests.309
309
Snabe, Birgitte, Andreas Größler, and Peter M. Milling: “Policies and Politics of
Establishing R&D Capacity in Low-Cost Locations”, Tagungsband, GWS-
Tagung, Greifswald, in print, 2006, focus especially on the political challenges
in the case study.
C. Case Study 93
2. Intervention Process
The location strategy project was first initiated with a project team in each of the
five business units in the division, but due to lack of consistency and efficiency,
and lack of structure in the coordination and communication between the teams,
it was decided to develop a shared, formal model on an abstract and highly
aggregated level, aiming towards:
• creating a structured and ‘objective’ frame for the rather emotional and
diverse discussions;
• establishing a forum for exchange of experiences and best practices;
• refining and making operational the strategy outlined by the board;
• improving the change process effectiveness.
The change imperative was stated as: “right now is the right time to hire
people in the low-cost locations, because due to company growth, it can be done
right now without staff reduction at high-cost locations, and the expected results
are improved competitiveness and further company growth, also securing jobs at
high-cost locations in the future.” This was a difficult message to communicate
because of the fear of jobs moving from high-cost to low-cost locations.
An external facilitator from Mannheim University (the author of this
dissertation) was brought in as process coach and modeling facilitator, based on a
participative modeling approach. For system dynamics modeling, the Vensim
modeling environment was used.
Targeting and planning the intervention initially involved a discussion
with the project owners about the problem, the business objectives, the
intervention objectives, the dynamic hypotheses, and the suitability of applying
system dynamics to the problem. A preliminary model was built to frame the
modeling project, and also for the project owners to feel confident that major
results from the model were consistent with the directions of the strategy
expressed in the objectives and targets. This preliminary model was the basis for
the decision to move forward with modeling.
94 C. Case Study
DIVISION LEVEL
3.
3. Consolidate,
Consolidate, Balance
Balance &
&
1.
1. Vision
Vision & Modeling
Meta Modeling Coordinate
Agree on objectives & guiding principles Reduce limitations/barriers
Agreement on shared,
- formal model Set division targets
Set and agree critical parameters Balance between areas
Define ideal strategy Identify critical success factors
Agree on communication strategy
BUS. AREA LEVEL
4.
4. Detailed
Detailed Planning
Planning
2.
2. Area
Area Strategy
Strategy & Execution
Execution
Evaluate ideal strategy against reality Detailed planning
Define Area strategy based on ideal strategy Align with budget process 2005
Identify limitations/barriers Build 3 year plan (quarterly breakdown)
Business cases based on agreed parameters Detailed communication plan
Identify short term wins
Figure C-2: Intervention process as communicated in the project
The modeling intervention was coordinated with other projects; most

importantly the business planning and budgeting process, in which detailed
Excel-models existed in each business unit. Figure C-2 illustrates the change
process with visioning and modeling on division level combined with project and
implementation ownership on business area level.
The major stakeholders were identified early in the process, and the senior
vice presidents responsible for the five business units were consulted about the
plan for the system intervention in order to get their input as well as their
commitment. The process was agreed upon, and the project organization was
established. The major stakeholders are the management team of the division
(Senior Vice Presidents each responsible for a R&D business unit), the project
owners (division chief controller and the COO of the division), the core project
team chosen to find the proper strategy (called location champions), corporate
controlling, corporate management (the board of directors), the world-wide
corporate location strategy responsible, and all the day-to-day managers with
high influence on the implementation of the strategy. Furthermore, all employees
of the company are stakeholders in a communication point of view. Project
planning included the establishment of roles and responsibilities in the project;
see table C-1 for an overview:
C. Case Study 95
Project Participants Roles and Responsibilities

Division COO & Project owners
Division Controller - responsible for total system intervention
- establishment of business objectives and targets
- establishment of preliminary model
- facilitators of all modeling sessions and meetings
Leaders of the Decision-makers on division and business unit level
Business Units - acceptance of the process, objectives and principles
(Senior Vice - approval of the parameter-settings
Presidents) - approval of the business cases and implementation
plans
Business Unit Project core team
project participants - improvement of preliminary model and parameter-
(Location champions) setting
- identification of most important parameters
- establishment of detailed business cases
- detailed planning of implementation
Managers with high Change leaders in the implementation phase
influence on - involvement in implementation planning
implementation - execution of implementation
Location Strategy Corporate location strategy coordinator
Corporate Coordinator - coordination of company-wide parameter-settings
- discussion on company-wide learning
Corporate Controlling Decision-makers on corporate level
& - corporate goals and strategies
Management Board - approval of business cases
Table C-1: Roles and responsibilities as defined in the project
The solution design activities consisted of a larger number of meetings and

workshops with a variety of agendas around the problem. Only approximately
half of the activities directly involved modeling or simulation, but all meetings
had some kind of impact on the model, its parameters and/or the process of
implementing modeling results.
The intervention facilitation role was split between the modeling facilitator
and a discussion facilitator, who was one of the project owners, with in-dept
knowledge of the company, the problem and of the intervention goals.
96 C. Case Study
The modeling and simulation process with the core project team served as
a cognitive framework for objective discussions of the problem – challenging
preconceived perceptions and aiming towards reducing the tendency often known
from budgeting and business planning processes, that each stakeholder to some
degree primarily looks after his or her business unit’s interests rather than
corporate objectives. Using a shared model on a highly aggregated level moved
the focus and discussion towards a holistic view. In this process some important
new aspects – including one additional stock and a number of additional
parameters – were added to the preliminary model. Also some parameters and
relationships, which through simulations showed only little impact, were removed
in order to simplify the model. For communication purposes, the modeling
process was focused on developing a relatively simple model that could provide a
picture of the overall behavior of the problem-system without including too many
details, as overview was considered more important than detailed correctness.
The focus on keeping the model simple was also due to the role of the system
dynamics model role as a ‘meta-model’, creating cross-unit overview, and being
complemented by a detailed excel-model in each business unit with the format
needed in the budgeting and business planning process.
The parameter setting was a cornerstone in the change-process, as these
agreed parameters formed the basis for each business unit’s business case in
phase 2 of the intervention process. The discussion of the parameters often
resulted in a discussion on how the strategy could and should be executed, as the
parameter setting reflected implementation decisions; e.g. the logistics and cost
model of traveling, how to structure knowledge transfer, etc.
Model testing was done partly behind the scenes by the modeling
facilitator using some of the most respected sources as guides and checklists”,310
partly during the modeling and simulation efforts, as “validation is the process of
establishing confidence in the soundness and usefulness of a model“.311 Also
310
dynamics”, System Dynamics Review, Vol. 12, No. 3, 1996, pp. 183—210; and
1980, pp. 209—228.
311
Dynamics Models, 1980, p. 210.
C. Case Study 97
bearing in mind Forrester’s view on model validity: “Model validity is a relative

matter. The usefulness of a mathematical simulation model should be judged in
comparison with the mental image or other abstract model which could be used
instead.”312
Through the simulation of different scenarios the discussion focused on
the most influential parameters and causal relations of the problem, and possible
improvement options. Based on the modeling and simulation, a presentation with
the most important learning was developed to document the major insight
reached by the core project team. This presentation, together with the Vensim
model and the business unit excel models, was used in communicating with the
other stakeholders: to both communicate modeling results and to receive the
stakeholders input on the model, the parameters and the insights gained.
In the last workshop of the case study project, the core team established a
set of critical success factors for the implementation. The concept of critical
success factors was originally introduced by Rockart as a method for aligning
business and IT strategy, but has developed to a broader usage on both strategic
levels and in organizational interventions.313 The critical success factors served as
feedback to the Board. One of the critical success factors emphasized the need
for cost-orientation rather than head-count orientation in the company
governance model. Another critical success factor concerned the cross-
organizational coordination of the detailed implementation plans per business
(also involving managers from outside of the division). The detailed
implementation planning included determination of R&D tasks to be moved from
each business unit in high-cost countries and furthermore was explicit and
specific about the future R&D tasks of the affected employees. The detailed
implementation plan, and a detailed communication strategy and plan, were
considered to be the cornerstones in securing motivation and morale among
employees. The implementation plan furthermore included clear delegation of
responsibility for improvement of the parameters identified to be the most
312
Forrester, Jay W.: Principles of Systems, Cambridge, 1968, chapter 3, p. 4.
313
Rockart, John F.: “Chief executives define their own data needs”, Harvard
Business Review, March-April 1979, pp. 81—93. For a discussion on the usage
of critical success factors, see also Snabe, Birgitte and Jakob F. Ølgaard:
Informationsstrategi, Master Thesis at The Institute of Mathematical Statistics
and Operations Research, Technical University of Denmark, Lyngby, 1992,
pp. 25—30.
98 C. Case Study
influential ones. Reviews and follow-ups were integrated in the existing

performance measurement system.
3. The Model and Selected Simulations

The modeling purpose was to find an effective and realistic plan for reducing the
cost/capacity ratio under board guidelines of cost growth only due to inflation in
high-cost and low-cost countries, and an increase from 10% to 25% of low-cost
headcount compared to the total headcount of the division.
The basic structure of the model is based on two separate aging chains
(see figure C-3), each being an extended version of Sterman’s “two-level
promotion chain”.314 The aging chain on the right represents the high-cost
locations (abbreviated HC); the aging chain on the left represents the low-cost
locations (abbreviated LC). In each aging chain this model had originally three
basic stages: “New hired FTE”, who are newly hired employees (FTE = Full
Time Equivalents) spending their time in class-room training learning the
development tools of the company; “Rookie FTE”, who are employees working
on development projects with reduced productivity, and then finally “Productive
FTE”, who are fully productive employees.315 In the low-cost aging chain, an
additional stock was added: a stock containing Rookies spending time on taking
314
See Sterman, John D.: Business Dynamics, Boston, 2000, p. 491. Sterman
operates with only two levels in his promotion chain, with employees leaving
both levels. The location strategy model operates with 3 and 4 levels, with
employees only leaving the latest stage as this reflects the historical data well.
Other examples of system dynamics models of aging chains include Martinez,
Ignacio J. and Luis F Luna: “The Dynamics of Best Practices: A Structural
Approach”, at CD-ROM of Proceedings, System Dynamics Conference, System
Dynamics Society, 2001, which operates with three levels of practitioners:
junior, intermediate, and advanced.
315
In the model as well as the case description, the abbreviation FTE (Full Time
Equivalents) is used to indicate number of full-time employees. The abbreviation
is an established term in the case study company, used to adjust budget numbers
for employees working part-time.
handover capacity
reduction
<HC hire> <HC quit>
C. Case Study
HC CAPACITY USE
ON HAND-OVER
LC PRODUCTIVITY
LC replacing HC quit REDUCTION
LC ROOKIE HC ROOKIE
PRODUCTIVITY production per month PRODUCTIVITY
ADDITIONAL
GROWTH REPLACEMENT
IN HC VS. LC
LC QUIT HC QUIT
FRACTION FRACTION HC ROOKIE HC TRAINING <HC
TIME TIME quit>
LC new hire LC LC HC
LC H-O-R LC 1c quit quit
New hired 1a 1b Rookie Productive Productive Rookie New hired
FTE LC
FTE LC FTE LC FTE LC FTE HC HC FTE HC HC FTE HC HC
LC repl. hire LC 2a LC job-trained trained hire
2b
<LC quit> <New hired FTE LC> total FTE HC

total FTE LC LC ONGOING
TRAVEL COST
LC PERSON COST HC PERSON COST
INCREASE RATE
INCREASE RATE
LC TRAINING TIME
LC TRAINING LC person cost HC person cost
LC TOTAL ROOKIE TIME COST HC TRAINING
COST
HAND-OVER FRAKTION cost per month
OF LC ROOKIE TIME
LC flow input
LC HAND-OVER TRAVEL COST
<H-O-R FTE LC>
handover
travel costs
Figure C-3: The location strategy model

99
100 C. Case Study
over tasks from high-cost countries, which will be the case for all new employees
in low-cost countries who are not merely replacing people who have left a
position at a low-cost location. This stock is called H-O-R FTE (Hand-Over-
Rookies). These employees have zero productivity and, as they are spending time
physically with those HC employees, whose tasks they are taking over, they
furthermore tax time from fully productive employees in high-cost countries in
the process of knowledge transfer.
The two main decision points influencing the changes in stock levels in the
model are:
1. The rate “HC hire”, where only a share of the employees leaving high-
cost locations will be replaced at a high-cost location, depending on the
factor called REPLACEMENT OF HC VS. LC in the model. Those
headcounts not being replaced in HC will be replaced in LC.316
2. The rate “LC new hire”, consisting partly of replacement of employees
leaving the stock of productive LC employees, partly of headcounts not
being replaced in HC, and partly of the additional new employees
joining the division (called ADDITIONAL GROWTH in the model). All
additional new employees are allocated to low-cost countries, based on
a growth factor relative to the stock of experienced employees in low-
cost. 317
It is important to note that tasks will be moved from high-cost to low-cost
locations in bulk. Employees in high-cost locations, who have transferred their
tasks to low-cost locations, will either take over tasks from a person leaving the
division or take part in new R&D projects within the division. Tasks moved to
low-cost locations enable new and innovative tasks in high-cost locations; i.e.
does not lower productivity in high-cost locations.
316
Due to the non-replacements, the number of employees can be reduced in some
locations. However, the total number of employees in high-cost locations will
stay roughly stable due to necessary hiring in other functions in the division.
317
In Sterman, John D.: Business Dynamics, Boston, 2000, p. 491, the “two-level
promotion chain” has a growth factor related to the total number of employees,
but in the location strategy model it makes more sense to base the growth on
fully productive FTE’s, due to ramp-up limitations (ratio between experienced
staff and new staff).
C. Case Study 101
Based on parameter settings, the model will calculate the development in

stock levels as well as the two central model outputs: cost per month and
production per month. For a detailed description of model parameters, see
appendix A. For model equations, see appendix B.
The stocks in the model are initialized in equilibrium (hire rate = quit rate
for both low-cost and high-cost locations) through a distribution of the total
number of employees for both low-cost and high-cost countries to their
respective stocks of newly hired, rookies and experienced FTE’s. The number of
newly hired employees, plus rookie employees, plus experienced employees,
equals the total number of employees. The distribution into these three categories
is a calculation based on quit rate, training time and rookie time. For further
description on stock initializations, see appendix C.
In the modeling process, the facilitation and communication function was
prioritized over model correctness; especially in terms of using parameters
directly useful in the budgeting process, and also in keeping the model as small
as possible to avoid a “black box” effect. The preliminary model (see appendix
D) had “set the stage” for the main trends of the model, and the model refinement
process had the dual objective of getting acceptance in the model as well as
adjusting and improving the model. Model-wise, the process included adding 1
new stock (the hand-over-rookies), a few new causal relations (centered on
productivity reductions due to handover tasks as well as low experience level in
low-cost locations), and removing a number of parameters less significant for
investigating the decision parameters. The removing of the parameters was done
to simplify the model.318
A number of design decisions had to be made in the modeling process.
The growth rate ADDITIONAL GROWTH could have been modeled reflecting a
goal-seeking structure based on the discrepancy between the actual value and the
goal-value for the fraction of LC employees out of total number employee.319
318
Examples of removed parameters: the number of productive days is higher in LC
is than in HC, but these parameters were removed from the model (and added in
a comment instead), as it was decided to let the higher productivity in HC equals
out with additional coordination overhead. Also the cost of hire was removed
from the model, as they proved to be insignificant in simulations.
319
In this case, it should be used in combination with a MIN-function in the LC
new hire-rate to secure that the number of newly hired employees did not exceed
102 C. Case Study
This would have made the policy of how many additional employees to hire
endogenous.320 However, in order to keep the model simple and with focus on the
few, most important parameters, the growth rate is simply implemented as a
constant (fixed for the first period, then gradually decreasing to zero after the 36th
month). For the same reasons, as well as to allow some degree of cost overrun, it
was decided not to handle the cost restrictions endogenously, but to incorporate
cost as an explicit auxiliary.
In Sterman’s Two-Level Promotion Chain, the rate of employees moving
from a ‘rookie’ state to an ‘experienced’ state was modeled as a fraction of the
number of Rookies321. In the location strategy model the rates between stages are
calculated as delay-functions of the inflow-rates. For the training period, a high-
order delay was used to imitate a pipeline delay, as this is a fixed period of time
for each employee. For the period of being a rookie, a lower order delay was
used, to reflect the variability in the learning curve for individuals. The
variability in the difficulties of the task areas is not explicit in the model, but is
considered in stipulating the average time for employees being rookies. Having
outflow-rates as a function of inflow-rates (rate-on-rate modeling) places the case
study model in a slight conflict with the tendency among system dynamic
practitioners to avoid rate-on-rate modeling, and rather calculate outflow-rates as
a fraction of stock-level. The reason for the tendency to avoid rate-on-rate
modeling can perhaps be seen partly as motivated by prevailing attention to
macro-trends of systems with continuous parameter development (rather than
systems with steps in inputs), and partly motivated by a historical tradition based
on Forrester’s “Principles of System”, which can be dated back to a time where
delay-functions constituted a seriously challenging workload for computers.322
For the location strategy model, the new strategy with a step in the hiring rate at
the very beginning is being investigated, and the use of the delay-function better
reflects the true pattern of employee-flows in the start-up period. Due to the use
of rate-on-rate modeling, a fixed period of time will pass before the first
a realistic level compared to number of experienced employees (the ramp-up

limitations).
320
See chapter B.III.1 for a discussion on endogenous and exogenous explanations
for system behavior.
321
Sterman, John D.: Business Dynamics: Systems Thinking and Modeling for a
Complex World, Boston, 2000, p. 491.
322
See Forrester, Jay W.: Principles of Systems, Cambridge, 1968, chapter 5, p. 10.
C. Case Study 103
additional new employees flow into the rookie state (e.g. 3 months) and an even
longer time will pass (e.g. 9 months) before the number of experienced
employees is beginning to increase. This is intuitively more acceptable for group
modeling participants, as they are not trained in system dynamics modeling, and
therefore could have trouble abstracting from the number of experienced
employees beginning to increase as early as the very first month. Interestingly
enough, though, the major trends and learning – even regarding year 1 - are the
same with both ways of modeling the delays, see appendix E.
The figures C-4, C-5 and C-6 show the behavior of selected variables for four
simulation runs. Naturally, a large number of simulations took place in the
workshops to investigate the importance of the influence of the different
parameter settings and the consequences of different scenarios, but the selected
simulation runs give an impression of some of the key learning from the model.
The selected simulation runs are labeled: “INI”, “Base run”, “40% HC
replacement”, and “Faster training and hand-over”. Below is a short description
of the parameter setting in each of the simulations:
• INI: In this run, no changes ot the number of employees in low-cost or

high-cost take place (hire rate = quit rate in both aging chains). The
simulation run therefore reflects the development in costs and
production as could be expected in a zero-growth situation, based on
stipulation of the cost and production related input parameters. The
production naturally stays constant, whereas the costs increase due to
the expected increases in employee costs.
• Base run: This simulation run reflects a parameter setting with

temporary, strongly reduced replacement hiring in high-cost, in order to
finance the build-up of resources in low-cost locations. Compared to
INI, two principal changes are made: The REPLACEMENT IN HC VS. LC is
set to 0.2 in the first 36 months; where after it is set to 1, indicating full
replacement of employees leaving. Furthermore, the ADDITIONAL
GROWTH in low-cost locations is set to = 3% per month for the first year
(equals more than 40% ramp-up in year 1), then linear decreasing to 0
after 36 months. The efforts not being used to train new people in high-
cost allow the total new hiring in low-cost to be higher than the
additional growth (due to high-cost employees job-training new low-cost
hires rather than new high-cost hires).
104 C. Case Study
• 40% HC replacement: The parameter setting in this run is only slightly

changed compared to the INI simulation run. The only change is that the
REPLACEMENT IN HC VS. LC is set to 0.4 in the first 36 months; as
opposed to 0.2. This change is made to investigate the consequences of
allowing more external people to be hired in high-cost locations
compared to the base run.
• Faster training and hand-over: The parameter setting in this run is also
very like the parameter setting in the Base run. The only difference is
that the classroom training is reduced from three months to two months
for low-cost new hires (requiring a more intensive training program), as
well as reducing the fraction of the rookie time to be spent 1-on-1 to
handover tasks with high-cost employees. This fraction is reduced from
50% (equaling 3 months) to 33% (equaling 2 months), and this will
require that high-cost employees also have to work more intensively in
the hand-over period and it also calls for an improved process for the
hand-over of tasks.
The following three figures show major results of the four selected runs:
LC FTE fraction of total FTE

0.4
2 4 2 4 2 4 2 4 2
2 3 4 3 3 3 3
0.2 23 4 2 3 4
4
4 2 3
4 2 3
1 2 3 1 1 1 1 1 1 1 1 1 1
0
0 6 12 18 24 30 36 42 48 54 60
Months
INI 1 1 1 1 1 1 1 Dmnl
Base run 2 2 2 2 2 2 2 Dmnl
40% HC replacement 3 3 3 3 3 3 Dmnl
Faster training and hand-over 4 4 4 4 4 Dmnl
Figure C-4: Fraction of employees in low-cost countries compared to

total number of employees in the division
C. Case Study 105
Index Cost per Production

1.2
1.15 1
1
1
1
1
1.1 1
1
1 23
1 2 34
1 34
1.05 1 3 23 2 3 2 3 2 3 2 34 234 2
3 1 3 1 2 3 23 2
3 2 3 2 4 4 4 4 4 4
4 1 4 4 4
2
1
0 6 12 18 24 30 36 42 48 54 60
Months
INI 1 1 1 1 1 1 1 1 1 1 1 Dmnl
Base run 2 2 2 2 2 2 2 2 2 2 2 Dmnl
40% HC replacement 3 3 3 3 3 3 3 3 3 Dmnl
Faster training and hand-over 4 4 4 4 4 4 4 4 Dmnl
Figure C-5: Development in cost-index for an average productive unit

(e.g. cost for one employee for a fixed period)
Index Production per Month

1.1 4 4 4
3 2 3
4 2
4
3
4 2
4
3
4
1.05 2
4
3
4 2
4
3
4 2
4 3
1 1 2 2 1 1 1 1 1 1 1 1 1 1 1 1
4
0 6 12 18 24 30 36 42 48 54 60
Months
INI 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Dmnl
Base run 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Dmnl
40% HC replacement 3 3 3 3 3 3 3 3 3 3 3 3 Dmnl
Faster training and hand-over 4 4 4 4 4 4 4 4 4 4 Dmnl
Figure C-6: Development of productivity index for an average unit

(e.g. output/month/unit)
106 C. Case Study
To some extent, interpreting model results is always a subjective exercise.

This is especially true of the scenario simulation in this case, as the stipulation of
how much each input-parameter can realistically vary is a matter of
argumentation and consensus within the group of participants. The following is a
description of the core group’s understanding of key model results:
• Hand-over efforts have a strong “worse-before-better” effect on
productivity, and action must be taken to optimize this process—even if
this results in higher travel costs. Although this insight did not have
much ‘newness’ value to the business units, it was very valuable to have
a model that distinctly and clearly ‘proved’ the matter.323 A workshop
was arranged with corporate controlling, to make the point clear that
even though the division receives a relatively large number of additional
head-counts in year 1, the division will have no additional productivity
in year 1, but rather a slightly reduced productivity due to hand-over
efforts.
• Not replacing all employees in high-cost locations for a three-year
period is the only practical way to build up resources at low-cost
locations in a (rather) cost-neutral scenario. This was a politically
sensitive discussion. The model helped to make the discussion more
objective compared to the very emotional and unstructured discussions
taking place regarding this topic before the modeling part of the project
was introduced.
• Reducing training time has an accumulative productivity effect. With
the large number of new employees, investments should be made to
optimize their training—even if this results in higher training and travel
costs.
The list above concentrates on the insights directly supported by model

simulation. Additionally, learning and changing of attitudes and intentions also
323
The “worse-before-better“ effect is widely recognized in change management
literature. The effect is also illustrated by the means of system dynamics models,
e.g. Repenning, Nelson P. and John D. Sterman: Nobody Ever Gets Credit for
Improvement”, California Management Review, Vol.43, No.4, Summer 2001,
p. 74.
C. Case Study 107
took place in a more indirect way due to discussions during model refinement
and parameter setting. Examples are:324
• The alignment of actual numbers to be used in the business cases (also
called the Excel-models). The business cases were often large ‘black-
box’ models. By having a system dynamics model on a highly
aggregated level, the most important parameters were made transparent.
An example of this is the employee turnover rate in low cost countries
that in the different business cases was set to be in the range of 10% to
25% based on personal assumptions, although historical figures showed
to be approximately 7%.
• The strategy could be accelerated by a shift in the entire company from
headcount orientation towards cost orientation.
• Identification of detailed transfer planning as a critical success factor.
The after-transfer solution for involved high-cost location employees
was perceived to be especially important in order to secure motivation
and morale. This included planning for replacements in high-cost
locations being made by colleagues who have handed over tasks to low-
cost locations.
• A number of best practice experiences were exchanged between the
business units because some units already had more experience in
building up capacity in low-cost locations. The modeling sessions also
served as a forum for the discussion of topics such as bridgeheads,
hiring strategies, and practical aspects of making new employees
productive as quickly as possible.
324
Further examples are discussed in chapter D.
108 C. Case Study
III. Evaluation of the Case Study
1. A Framework for Evaluating the Effectiveness and Efficiency of the

Case Study
Being a real-time single site case, with no test group for comparison, means that
even if the intervention yields good results, it is impossible to know whether
other intervention mechanisms would have yielded even better results. The
evaluation framework takes the theoretical point of departure in the evaluation
frameworks developed by Huz and Rouvette; and is adapted to the research focus
and the data collection conditions of the case study.325
The evaluation is structured in two sections: evaluation of outcomes, as
well as evaluation of method and comparative conditions that may explain the
intervention’s effectiveness.
Evaluation of outcome:
• Evaluation of outcome on an individual level
• Evaluation of outcome on a group level
• Evaluation of outcome on an organizational level
Evaluation of method and comparative conditions:
•Method evaluation of the use of system dynamics modeling compared to
other approaches
• Method evaluation of the specifics of modeling for strategy
implementation compared to other system dynamics modeling purposes
• Comparative conditions that may explain the intervention’s effectiveness
(both context comparative conditions and mechanism comparative
conditions)
The evaluation of both the outcome and method is relevant to identifying
the benefits of the intervention. It gives some guidance for the effectiveness of
the applied method, but - because of the nature of a seek-and-learn case study
325
Huz, Steven, David F. Andersen, George P. Richardsen and Roger Boothroyd:
“A framework for evaluating systems thinking interventions: an experimental
approach to mental health system change”, System Dynamics Review, Vol. 13,
No. 2, 1997, pp. 149—169; Rouwette, Etiënne: Group model building as mutual
persuasion, The Netherlands, 2003, pp. 68—95. The framework of Rouwette
partly is based upon the work of Huz.
C. Case Study 109
with no test group - it will not give data to support or reject hypotheses about
relative effectiveness compared to alternative mechanisms. The value of the
evaluation is therefore strongly connected with the search for explanations and a
focus on the question of why certain outcomes might happen.
Data on the evaluation variables was collected from three sources: (1)
observations by the facilitator, (2) semi-structured interviews, and (3)
questionnaires among both core team members and non-core project participants.
The usefulness of the sources differs between the measurement variables, but
each source adds some information to all variables regarding outcome and
method. In the analysis, the appropriateness of the different sources for each
measure variable is taken into account. For observations and interviews, focus
was especially on potential manipulation and biases.326 Significant bias must be
expected to occur due to the mental model of the observer and interviewer (the
author of this dissertation).327 The usage of a structured evaluation framework
lower the bias, but the mental model will inevitable be a “filter” in the selection
and interpretation of observations and interview results. However, as the purpose
of the case study is theory building rather than theory testing, the bias is less
crucial for the research value.
The questionnaires are closely linked to the evaluation framework, and are
designed to be answered anonymously. They are furthermore short (1 page), in
order to increase response rate, and they aim to provide unambiguous and
objective questions about participants opinions regarding both intervention
output and mechanism on a 1 to 7 scale (from strongly disagree to strongly agree,
4 being neutral). The questions refer directly to the measure variables in the
evaluation framework, to reduce bias due to the researcher’s own pre-coded view
of the research. In addition, the use of checklists for how to use and design
questionnaires was applied in an attempt to reduced bias.328 The questionnaires
326
Yin, Robert K.: Case study Research, 3 rd edition, Thousand Oaks, 2003,
pp. 93—96, discusses the problems of participant observations. Kvale, Steinar:
InterViews: An Introduction to Qualitative Research Interviewing, Thousand
Oaks, 1996, pp. 235—252, discusses the reliability and validity of interviews.
327
For discussions on self-reinforcing of mental models, see Bakken, Bent E.:
Learning and Transfer of Understanding in Dynamics Decision Environments,
Boston, 1993. pp. 29—30; Argyris, Chris: Reasoning, Learning, and Action –
Individual and Organizational, San Francisco, 1982, p. 39.
328
Denscombe, Martyn: The Good Research Guide for small-scale social research
projects, Philadelphia, 2 nd edition, 2003, chapter 9.
110 C. Case Study
were pre-tested with both a project owner and a research colleague. The
questionnaire was given to the five core-team members, to the three most
involved steering committee members, and to three other participants, who were
not as involved in the project, but had only been exposed to the model in one or
two meetings. The latter group received a reduced version of the questionnaire,
as they did not have an overview of the complete project. The interviews are
structured around the same measure variables as the questionnaires, but with
open questions, and the interviews were conducted with the two project owners.
The facilitator’s own observations are also decoded in the same structure as the
questionnaires and the interviews.
The following tables (C-2a to C-2e) illustrate the applied framework’s sources
for data collection.
C. Case Study 111
Measure Questionnaire Interviews Observations

variables self assessments with project
owners
Personal reactions “I believe it was Project owners’ Attitudes in
to the modeling useful to include the perception of modeling sessions
process model in the participants atti- and other meetings
(see also under project” tude towards the
method) process
Learning gained, “I gained interesting Project owners’ Changes in
and changes in learning from the perception of positions in the
goal structures model” insights gained discussions
and mental by individuals (and pre/post tests
models of change
ambassadors)
Commitment to “I agree with the Project owners’ Whether
the outcome of recommendations perception of the participants in
the modeling derived from the commitment subsequent
sessions model – and will act among meetings actively
accordingly” participants argued for the
results
Changes in “The modeling Project owners’ Observations
behavior affected some of my perception of the regarding behavior
decisions” business case’s
alignment with
the modeling
results
Table C-2a: Main sources for evaluation of outcomes on individual level

112 C. Case Study

owners
Group “The meta-model Project owners’ Whether the
communication was a useful perception of the modeling process
framework communication created open
facilitating during modeling discussions and
discussions” sessions exchange of views
Consensus “The modeling Project owners’ Whether the
process helped perception of group seemed to
building a shared group consensus get closer in
view of the established opinions
location strategy” through the regarding the
modeling process strategy
Common language “The modeling Project owners’ Agreement on
efforts helped perception of using the same
creating a common creation of a terms – also
language for the common language outside the
location strategy” through the modeling sessions
modeling sessions
Transfer of “The meta-model Project owners’ Effectiveness in
insights was a useful tool perception of the transfer of
in the presentation usefulness of the insights to non-
of the ideal model in transfer core project
location strategy” of insights participants
Table C-2b: Main sources for evaluation of outcomes on group level

C. Case Study 113

owners
System changes “I believe the Project owners’ If decision is

recommendations perception of the included in
from the modeling boards reaction to budgets and
process will be the overall business
implemented” recommendations plans
Results “I believe the Project owners’ Business results

recommendations expectations
will have positive regarding business
business impact” benefits
Table C-2c: Main sources for evaluation of outcomes on organization level

114 C. Case Study

owners
Efficiency “The use of Project owners’ n.a.
(compared to modeling increased perception of the
normal project the efficiency of the efficiency of the
execution in the project process” process in general
case company)
Efficiency “Using modeling in Project owners’ Project progress
(compared to this case was more perception of the compared to other
other approaches efficient compared efficiency – types of consulting
or methods) to other compared to if approaches
approaches” other approaches (highly subjective)
had been used
Quality in results “Using modeling in Project owners’ The importance of
this case created perception of the insights gained
higher quality quality of the (highly subjective)
results compared to results compared
other approaches” to if other
approaches had
been used
Further use of SD “I intend to use Project owners’ Later actual system
modeling in other perception of the dynamics projects
relevant change general trust in
projects” the model
Table C-2d: Main sources for evaluation of system dynamics compared to other approaches
C. Case Study 115

owners
Intervention driven (not included) Project owners’ How the initially
by business perception of the established
objectives and importance of objectives and
targets initial business targets influenced
objectives and the process
targets
Project framing “It was useful to Project owners’ Possible conflicts
(effectiveness) start with a 1st draft perception of the concerning model
of the model to usage of a boundaries
kick off the preliminary
process” model
Project framing “I believe the Project owners’ Attitudes in
(consequences for model reflects the perception of the modeling sessions
model trust and core of the project and other meetings
ownership) problem” participants’ and
own trust &
ownership
Structured (not included) Project owners’ Implementers
involvement of perception of the influence on the
implementers importance intervention
Table C-2e: Main sources for evaluation of the usage of system dynamics
in a change management context
Evaluation of comparative conditions that may explain intervention effectiveness

(both context comparative conditions and mechanism comparative conditions)
will only depend on observation, as neither questionnaire nor interviews included
this.
116 C. Case Study
2. Conclusions on Case Study Effectiveness and Efficiency

A schematic overview of observations made during the intervention as
well as results from the interviews are to be found in appendix F. The case study
focused on the role of a system dynamics modeling process used in bridging
between the top-management launching a new location strategy, and the planning
and implementation of the new strategy. A highly aggregated system dynamics
model proved useful for this purpose, and participants accepted it as an abstract
representation of the subject matter. This model was preliminary established by
the project owners (representatives for the decision-makers), and later elaborated
and improved during the participative modeling process. It has to be clarified that
in the case study, the actual strategy decision was made without using system
dynamics modeling and the model was built post hoc in order to secure that
model outcomes are broadly consistent with these decisions.
The modeling process allowed participants to exchange their experiences
and ideas in a relatively objective way, i.e. without relying too much on gut
feelings, and supplemented the discussions on the change imperative, reducing
most of the participants’ initial fears concerning the location strategy.
Simultaneously, a common ground for discussions was laid by recurring on a
formal model that was open for inspection, critique and change. Within the broad
strategic objectives set by the board, participants of the modeling process could
refine and alter the actual policies that resulted from the strategy. Participants’
reactions in the evaluation of the case intervention emphasized that they
acknowledged this possibility. Another remark was that participants appreciated
the rather ‘politics-free’ atmosphere that was achieved by working with a system
dynamics model. This can be interpreted as a hint that members of organizations
are willing to accept strategies even when their results seem to contradict their
initial personal wishes as long as the process is transparent and some sort of
involvement is secured. Especially due to the sensitive nature of the topic, it was
interesting to observe how the modeling and simulation efforts helped to direct
the discussions and facilitate individual learning. However, especially in the
beginning of the project, there was a tendency among core team members to
think of modeling (including quantification) as an additional task, increasing the
workload in an already stressed period.
Project participants clearly experienced single-loop learning, and
indications of double loop learning included changes in attitudes towards the
strategy and underlying assumptions. From the case company perspective, it
118 C. Case Study
Question n mean sd
I believe it was useful to include the model in the
8 6.00 0.76
project
I gained interesting learning from the
11 5.82 0.87
model
I agree with the recommendations derived from the
8 5.00 1.41
model and will act accordingly
The modeling affected some of my
8 4.63 1.85
decisions
The model was a useful framework facilitating
11 6.18 0.87
discussions
The modeling process helped building a shared
11 5.73 0.90
view of the location strategy
The modeling efforts helped creating a common
11 5.27 1.01
language for the location strategy
The model was a useful tool in the presentation of
11 5.09 1.14
the ideal location strategy
I believe the recommendations from the modeling
11 4.27 1.19
process will be implemented
I believe the recommendations will have positive
11 4.91 1.22
business impact
The use of modeling increased the efficiency of the
8 5.13 1.13
project process
Using modeling in this case was more efficient
8 5.13 1.13
compared to other approaches
Using modeling in this case created higher quality
8 5.38 0.92
results compared to other approaches
I intend to use modeling in other
8 5.00 0.93
change projects
I believe the model reflects the core
8 5.38 1.19
of the problem
It was useful to start the modeling with a
8 6.25 0.71
preliminary model to kick-start the process
Table C-3: Questionnaire results

D. The Usage and Utility of Participative Modeling in
Change Management
While system dynamics modeling is traditionally used in exploratory
organizational interventions, the core of this dissertation is to investigate the
usefulness of system dynamics modeling, when the purpose is to implement an
already outlined strategic initiative. Regardless of whether a strategic initiative is
derived from a specific strategic problem-solving process or from the ongoing
strategy-forming process of the company, the implementation faces the
challenges of change management.329 In this chapter, the discussion of
participative modeling efforts used in supporting change management is
structured into three sections: discussion of intervention context, discussion of
the intervention process, and discussion of intervention outcomes.330 The
intervention context is understood primarily as problem and organizational
characteristics. The intervention process is described in terms of business
objectives and targets, structured development of change leaders, the design of
the change process, and facilitation of modeling and simulation sessions. Finally,
the intervention outcomes are discussed in terms of transfer of insights and
ownership from decision-makers to implementers, refining and aligning strategies
through scenario testing, and organizational learning.
Examples from the case study will be used as examples in the discussions.
These examples are not included as a means of verifications of any point made,
but merely serve as illustrations.
329
In Warren, Kim: “Improving strategic management with the fundamental
principles of system dynamics”, System Dynamics Review, Vol. 21, No. 4,
Winter 2005, p. 329, differentiation of strategic management is made between
one-off challenges and the continuous direction of enterprise strategy.
330
The ‘context – process – outcome’ structure is inspired by the ‘context –
mechanism – outcome’ configurator as described Rouwette, Etiënne: Group
model building as mutual persuasion, Nijmegen, 2003, pp. 87—92; and also by
the ‘intervention – organization – effect’ model (a variant of a classical S-O-R
model) as described in Borum, Finn: Strategier for organisationsændringer,
Copenhagen, 1995, p. 56.
120 D. The Usage and Utility of Participative Modeling in Change Management
I. Context Factors Relevant for Deciding on Usage of Modeling in

Change Management
Implementing strategic initiatives calls for decisions on change strategies and

resulting change management approaches.331 Palmer and Dunford discuss change
management approaches in terms of controlling vs. shaping change.332 It seems
reasonable to assume that modeling is ill suited for change approaches primarily
focused on controlling, as a modeling process is a learning process allowing for
involvement and influence. For change approaches aiming at shaping intended
change outcomes, the usage of system dynamics modeling offers a way to involve
key implementers with the dual purpose of transferring insights from decision
makers to implementers, and having the implementers refine the implementation
process. Anderson and Anderson discuss the need for the creation of change
leaders when starting organizational changes, addressing the issue of change
process to be something that managers have to lead rather than manage.333 Leavitt
discusses the dual role of middle management as managers (respecting
hierarchical decisions) and leaders (leading people).334 The creation of change
leaders requires (on top of a number of personal requirements such as
communication skills, interpersonal skills, etc.) establishment of internal
commitment amongst the people responsible for implementing the change.
Creation of such commitment is often centered on the creation of awareness,
consensus, and confidence regarding the goals and the change process.335
Palmer and Dunford refer to approaches aiming at shaping intended
change outcomes for ‘change management as coaching’, which is the change
331
For a discussion on change strategy taxonomies, see chapter A.II.
332
Palmer, Ian and Richard Dunford: “Who says change can be managed?
Positions, perspectives and problematics”, Strategic Change, Volume 11, August
2002, p. 244.
333
Vol. 33, No. 3, 2001, p. 45.
334
Leavitt, Harold J.: Top Down – Why Hierarchies Are Here to Stay and How to
Manage Them More Effectively, Boston, 2005, pp. 164—168.
335
According to Akkermans, Henk: Modelling With Managers, Breda, p. 20, the
efforts of establishing internal commitment in organizational interventions are
often centered on the creation of awareness, consensus, and confidence
regarding the goals and the change process
D. The Usage and Utility of Participative Modeling in Change Management 121
approach that overlaps most with organizational development and action

research, and the strategy typically recommended in the prescriptive, normative
management literature on change management.336 They also introduce a change
approach called ‘change management as interpreting’ dealing with shaping
partially intended outcome, placing managers in the role of sense-makers in
organizational environments where different meanings and related change
intentions are competing to be implemented. Winch and Derrick discuss the
interpreting type of intervention dealing with knotty problems, where not only the
problem or system to be studied is characterized by structural and dynamic
complexity, but the complexity also characterizes the intervention itself.337 For
this change approach, the usage of system dynamics modeling has a stronger
traditional exploratory focus, compared to usage in coaching approaches aimed at
transforming the managers responsible for the implementation into change
leaders pursuing already outlined targets.
Besides the change approach, the characteristics of the problem at hand
also determine the appropriateness of using system dynamics modeling. The
model in the case study is a simple system dynamics model primarily focused on
the problematics of delays in aging chains, displaying no significant feedback
loops. The insights gained in the project were only to a small extent directly
related to the behavior of the model. The main part of the learning and consensus
building took place as a result of the process around the model building, e.g. the
quantification and the discussions of the importance of the different parameters
leading to discussions of best practices. It is difficult to know, however, if
problems requiring much more complex models can also benefit from using
336
Although action research has it origin from micro-organizational issues with
much of the earlier literature primarily discussing changes of group behavior, the
field of organizational development has included a whole range of techniques to
adjust the theories of action research to be suitable for large-scale changes in
organizations. Practically all of the literature from Lewin, Argyris and Schein
used in this dissertation discuss mostly interventions on group level. For the
evolution of action research in the organizational development literature, see
Palmer, Ian and Richard Dunford: “Who says change can be managed?
Positions, perspectives and problematics”, Strategic Change, Volume 11, August
2002, p. 247.
337
Winch, Graham and Sonja Derrick: “Flexible Study Processes in ‘Knotty’ System
Dynamics Projects”, Journal of Systems Research and Behavioral Science,
Vol. 23, No. 4, in print, 2006, pp. 1—2.
system dynamics modeling in a change management perspective, although it

seems fair to assume that such complexity will result in a number of challenges
regarding barriers for participants to understand the model as well as the greater
time investment needed. Through their practical consulting experiences in group
model building Eskinasi and Fokkema have observed what they call ‘a striking
correlation’ between a high level of details in the model and client
disappointment and lack of success.338 Eskinasi and Fokkema’s experiences deal
with models in exploratory settings rather than change management settings,
which might even be a more acute argument for using simple models in change
management, as participants would be expected to accept more involvement in
projects dedicated to a problem to be explored. Although indications exist for
preferring relatively simple models in change management purposes, one should
not forget the basic argument for using system dynamics in the first place: the
usage of modeling to compensate for the human brain’s difficulties in coping
with feed-back structures and delays.339 For more complex problems, it could be
interesting to investigate whether a combination of an exploratory modeling cycle
involving top executives, and a change management modeling cycle involving the
key implementers, would allow for improved sustainability in problem-solving
for more complex problems.
It is also important to assess the organizational situation to determine the
appropriateness of using modeling. Flood and Jackson discuss the industrial
relations literature on the political relationships in business organizations
between individuals and groups using the three metaphors: unitary, pluralist and
coercive relationships.340 These three metaphors are useful when assessing the
338
Eskinasi, Martijn and Eppie Fokkema: “Bursting the myth of making easy
modeling money”, unpublished working paper (permission granted from the
authors), Proceedings, Second European System Dynamics Workshop,
Nijmegen, 2005, p. 63.
339
Example: implementation of a new business process might benefit from other
modeling mechanisms such as the ARIS toolkit as described in Scheer, August-
Wilhem: Business Process Reengineering – Reference Models for Industrial
Enterprises, 2nd edition, Berlin, 1994.
340
Flood, Robert L. and Michael C. Jackson: Creative Problem Solving – Total
Systems Intervention, Chichester, 1991, p. 12. For a discussion on power,
conflict and organizational differences, see also Argyris, Chris: Interventions
Theory and Method – A Behavioural Science View, Reading, Massachusetts,
1970, p. 81.
issues of interests, conflict and power (see table D-1). From a change
management viewpoint, modeling is primarily relevant in a pre-dominantly
pluralist situation. In a dominant unitary situation, change can be managed
without investing time on modeling. In a dominant coercive situation, it is
illusionary to expect the gain the necessary consensus and trust in the modeling
sessions.
Unitary Pluralist Coercive

(modeling might not (modeling ill suited)
be worth the effort)
Interest Common objectives Diverging group Oppositional and
- a well integrated interests with the contradictory interests
team organization as a – rival forces
mutual focal point –
loose coalition
Conflict Rare and transient Inherent, but may Inevitable and likely
well to lead to radical
have positive aspects change of whole
structure
Power Replaced by Medium through Unequally distributed
conceptions such as which conflict of thus allowing
leadership and interest may domination,
control be resolved subjugation and so on
Table D-1: Political characteristics of situations in terms of the issues of
interest, conflict, and power341
In the case study, the Board of Directors exercised some degree of

coercive power in the outlining of the overall objective. It was not a question of
whether a new location strategy should take place, but only of how to implement
the decision. This was clearly against the initial personal interest of a number of
341
The table is based on the table in Flood, Robert L. and Michael C. Jackson:
Creative Problem Solving – Total Systems Intervention, Chichester, 1991, p. 13.
The indication of modeling relevance is a modification of the table.
the people responsible for implementing the new structure and processes. In this
situation, the overall business strategy and objectives of the companies was used
to seek interest alignment, and the modeling process was used as the main
mechanism ensuring involvement and focusing the discussions on both the
usefulness and the practical implications of the decided change.342 The level of
conflict was relatively high, resulting in strong engagement in the discussions,
although not of a magnitude preventing a constructive atmosphere after the
initiation of a structured and framed process. Within the frame given by the
outlined targets, the project participants had a high degree of empowerment.
Even radical new ideas on how to conduct business were discussed, recognizing
the need for creativity to come from all levels in the organization, and not as a
top-down process.343
A modeling process will inevitably be part of the social constructivism in
the setting.344 In the social constructivism typology by Wenneberg, ‘the critical
perspective’ and ‘the ethical reflection’ provide particularly interesting
perspectives on modeling.345 In the case study, at a point in time the core team
discussed whether to include the parameter ‘motivation’: having capacity build-
up in low-cost locations resulting in reduced motivation among high-cost
employees and consequently reducing productivity in high-cost locations. From a
critical perspective, this might be an actually existing underlying social construct,
but the project owners did not want to legitimize such behavior, which can be
342
See Snabe, Birgitte and Andreas Größler: “System Dynamics Modelling for
Strategy Implementation: Case Study and Issues”, Journal of Systems Research
and Behavioral Science, Vol. 23, No. 4, in print, 2006, pp. 16—17, for a
discussion on ‘manipulation’ when refining and implementing a given decision.
343
See Hamel, Gary: Leading the Revolution, Boston, 2000, p. 280, opposing the
idea that new strategies, innovation and change should always start from the top.
Throughout the book it is discussed how innovation and radical ideas on how to
change the way a company does business should also come from ‘activists’ from
inside the company; with activists being people dedicated to rule-bursting and
daring unconventional business.
344
Kieser, Alfred: “Kontruktivistische Ansätze”, in Alfred Kieser (ed.):
Organisationstheorien, 3rd edition, Stuttgart, 1999, p. 288, discusses social
constructivism as how human communication and interaction produce a social
reality that appears as the objective reality.
345
Wenneberg, Søren B.: “Socialkonstruktivisme som videnskabsteori – Sisyfos’
videnskab”, Online Paper, Institut for Ledelse, Politik og Filosofi, Copenhagen,
2002, pp. 8—9.
seen as an ethical reflection, taking responsibility for influencing social

constructs. A pragmatic solution from the project owners was to agree on the
importance of motivation and to establish the guideline that scenarios should only
be simulated with parameter settings within a range where motivation was not
reduced; and in this way to avoid including ‘motivation’ explicit in the model.346
When discussing social constructivism in a modeling perspective, Van der
Smagt criticizes the field of system dynamics and group model building for
having a one-sided causal model construction view with focus on creating
consensus about the representation (truth seeking), and typically ignoring a
constitutive model construction view dealing with definition legitimacy
(negotiation).347 For system dynamics modeling applied in change management
purposes, the constitutive view is particularly important, as executives often have
strong commitment to shaping change. The use of a preliminary model is a
central instrument in framing and targeting the intervention, allowing project
owners to ‘set the scene’ aiming at aligning expected insights and results from the
modeling process with the purpose of the intervention. Furthermore, the use of
a preliminary model not only frames the intervention, but also gives the project
owners an opportunity to advocate a certain worldview. Example: in the case
study company, there was a tendency among high-cost location employees to
question the efficiency and the quality of colleagues in the low-cost locations.
The project owners clearly upfront explained, that they disagreed in this causal
relation, and during the modeling process it was discussed and agreed, that lower
productivity should only be modeled as a consequence of less years of
experience. This can be seen as a commitment to constitutional change, where the
project owners seek to influence the mental models of modeling participants. A
modeling project might be a well-suited mechanism when executives want to
346
Negative consequences on organizational trust among employees has been seen
to be avoided even in cases of both redeployment and relocation of employees,
which was interpreted as being due to factors such as acceptance of the
arguments for change, perceived organizational support, and fairness; see Ferres,
Natalie, Julia Connell, and Anthony Travaglione: “The effect of future
redeployment on organizational trust”, Strategic Change, Vol. 14, March-April
2005, pp. 87—88.
347
Van der Smagt, Ton: “Causation and Constitution in System Dynamics;
Modelling a Socially Constituted World”, Journal of Systems Research and
Behavioral Science, Vol. 23, No. 4, in print, 2006, pp. 13—14. Van de Smagt,
p. 1, furthermore argues: “causal models blind us for constitutional change.”
impose changes in participants’ worldview (or Weltanchauung).348 Constitutional

issues can be a conscious and integrated part of establishing preliminary models,
both in the argued causal relations and with regards to underlying values or
beliefs. Furthermore, the project owners’ involvement as a steering committee
throughout the participative modeling process allows for constitutional
negotiations in the model construction. 349 On the other hand, if an organization
doesn’t also strongly encourage the truth-seeking perspective, it is doubtful that
modeling efforts can create learning and consensus.
In a change management setting, models are used to transfer and refine
insight rather than develop new insight. The target group is not made up of top
executives with extensive years of experience, but employees with less
experience in the overall system’s causal relations and behavior. Employees from
different business units often lack the overall picture, and in the view of Lyneis,
companies far too often experience underperformance due to policies not being
aligned between functional areas of the company and policies not being aligned
with corporate goal-settings.350 System dynamics modeling can help to make
some of these misalignments explicit, which is especially relevant for people with
only partial responsibilities in the company.351 Often top executives themselves
find the insights yielded by system dynamics models to be intuitively correct, but
348
See Checkland, Peter: “Systems Thinking, Systems Practice”, Chichester, 1993,
p. 219, for a discussion on the importance of the concept of Weltanschauung in
both ‘hard’ and ‘soft’ system methodologies.
349
Van der Smagt, Ton: “Causation and Constitution in System Dynamics:
Modelling a Socially Constituted World”, 2006, p. 14, argues that causal models
do not even ‘invite’ to think about constitutional issues. This might very well be
true for the modeling workshops, but in the preparation of the modeling
processes (especially in drafting the preliminary model), model owners have a
clear opportunity to include structured commitment to constitution.
350
See Lyneis, James M.: Corporate Planning and Policy Design: A System
Dynamics Approach, Boston, 1980, pp. 6—9. Lyneis furthermore emphasize
corporate policies not being sufficiently robust to handle changing conditions.
351
In Thygesen, Henriette H.: System Dynamics in Action, Copenhagen, 2004,
p. 189, it is argued that the system dynamics modeling process assist in creating
“an atmosphere of shared reality” between project participants.
with little newness, due to the executives own extensive, intuitive understanding
of the overall system behavior.352
A final comment regarding context factors relevant to the appropriateness
of using system dynamics modeling concerns skills, attitudes and traditions
among project participants. It is relevant to consider whether the employees have
skills and attitudes suitable for the usage of a highly conceptual and mathematical
approach. Also, the process must correspond with how the organization normally
handles change, solves problems, and involves employees. In the case study,
most participants were mathematical skilled people; many were engineers. The
question is whether modeling is only relevant for this type of company. An
argument against this viewpoint is that in most companies, major change projects
make use of extensive Excel-models to plan, track and optimize. Thus, formal
models are part of strategic change initiatives anyway. Implementation of change
projects as a part of business planning must be able to give input to the budgets
for the coming quarters and years. Therefore, managers responsible for leading
change are generally used to models, used to mathematical approaches, and often
appreciate the advantages that system dynamics offers compared to black-box
models. Furthermore, people who want to hide weak arguments in huge
datasheets will have a problem. Through the system dynamics modeling process,
the numbers from the business cases in each business unit came out in the open,
and “the approach made it difficult for people to play politics”. The structure of
the Excel-based business cases often made them into ‘black box’ models, where
it was difficult for people other than the model owners to see the critical
parameter setting through the ‘jungle’ of detailed information and hidden
formulas. The system dynamics modeling process served as a ‘meta-model’,
aiming at having a model on a high aggregation level only including the most
significant parameters—creating transparency even beyond the people doing the
actual modeling.
352
A former Forrester student, who has worked five years with system dynamics
modeling in McKinsey, has stated this viewpoint. The CEO of the case study
company has also stated that viewpoint.
II. Process Considerations
The design of the modeling process is a sub-set and closely interrelated to the
overall intervention design, with its stakeholder analysis, communication strategy
and plan, adjustment of business processes and measurement systems etc. In the
case study, although recognizing the interdependencies as well as the iterative
nature both internally between the steps and with regards to the overall
intervention design, eight conceptual modeling steps crystallize:
Step 1
Frame and Design Modeling Process
Step 2
Present Preliminary Model as Kick-off
Step 3
Adjust and Refine Model
Step 4
Discuss and Stipulate Parameters
Step 5
Analyze Sensitivity and Identify Critical Parameters
Step 6
Simulate Scenarios
Step 7
Plan for Improvements of Critical Parameters
Step 8
Provide Feedback to Initial Decision-Makers
Figure D-1: Conceptual modeling steps in the case study

The first step, Frame and Design Modeling Process, explicitly formulated
the given decisions (‘what is not for discussion’), respectively the degrees of
decision freedom in the implementation and modeling process (‘what is open for
discussion’). The preliminary model had a central role in step 1, both in the
investigation of the appropriateness of using a model and in the process of having
project owners feel comfortable with expected key learning from the model. The
business objectives and the targets in combination with the preliminary model
clearly framed the modeling process, and because of transparency of process
intentions, the project owners expected no perceptions of manipulation.
In the second step, the preliminary model was used to initiate the modeling
workshops with core project participants, who represented some of the key
managers responsible for implementation. Also, a number of other stakeholders
were introduced to the model in early versions, in order to allow for a sense of
involvement and for getting input, and to strive for both effectiveness and
efficiency in the process. The third step, Adjust and Refine Model, was intended
to serve the purpose of participants gaining trust in the model, as well as
improving the model. Allowing for model adjustments, together with later steps,
was seen as important for giving modeling participants true influence, which is an
important factor in getting commitment in the change process. In step four,
Discuss and Stipulate Parameters, the parameters were discussed in the group.
Some of the parameters were highly subjective, e.g. how long does it take for a
rookie employee to be fully productive, and it was perceived as important for
participant trust, that these estimations were made without interference from top
executives. The quantification process served as a means for aligning parameters
between business units, to be used in the different business units in the detailed
business cases. Also, the parameter estimations initiated discussion on best
practices across the business units.
The fifth step, Analyze Sensitivity and Identify Critical Parameters,
together with the sixth step, Simulate Scenarios, enabled the change leaders to
identify the crucial parameters, where improvement initiatives would have
significant impact.353 The main purpose was to focus the discussion on the critical
parameters in a way leading to focused implementation decisions. The facilitator
played an important role in turning barriers and issues into improvement
353
Example: the group realized the importance of reducing the elapsed time of
handing over tasks. If this had been a top-down request, it could very well have
caused significant change resistance.
opportunities requiring innovative solutions. It should be noted, however, that

other modeling processes could be building on models, where the implementation
decisions and improvement opportunities are not directly linked to parameters,
but could be linked to possible variations in model structure or formulas. If this
would be the case, both critical parameters and underlying decisions from
structure and formulas could be relevant to prioritize.
The seventh step, Plan for Improvements of Critical Parameters, was
closely interlinked with the development of detailed implementation plans in
each business unit. Focus was on creating a constructive and action-oriented
discussion on high-impact improvement opportunity. Although the entire
modeling process aimed at aligning mental models, this step seemed especially
important in the overcoming of critical barriers to change and the development of
a shared cross-organizational implementation plan. Step 8, Provide Feedback to
Initial Decision-Makers, partly took the form as a series of check-points with the
project owners, and partly took the form of a formal feedback formulated as a set
of critical success factors indicating policy and governing adjustments important
for sustainable and successful implementation. The feedback cycle has the twin
aim of providing the top executives with important information on policy changes
as well as of encouraging motivation among modeling participants through
influence and involvement.
Figure D-1, with the eight conceptual modeling steps, relates to the case
study and is not stated as a generic modeling process for modeling in change
management context. The figure merely gives some indication of potential
relevant steps for change-management-oriented modeling, and interventions
dealing with more complex problems, different magnitude of the number of key
stakeholders, different selected change strategy, or other differences, would be
expected to call for adjustments in the modeling process. The next four sub-
chapters discuss in more general terms significant process considerations
regarding the usage of system dynamics modeling in change programs. The
discussions will mainly be founded on the theory discussions from the previous
chapters and on additional normative change management literature.
Furthermore, the discussions will include fairly detailed case study insights, when
these can serve the aim of illustration. The four sub-chapters are selected trying
to pinpoint some of the main differentiators between modeling in exploratory
settings vs. modeling in a change management context.354 The four sub-chapters

are: (1) Business objectives and targets directing and framing the intervention,
(2) Structured development of change leaders, (3) Designing of the change
process, and (4) Facilitation of modeling and simulation sessions.
1. Business Objectives and Targets Directing and Framing the

Intervention
Compared to the traditionally exploratory system dynamics modeling processes,
system dynamics modeling for change management limits the problem-solving
process: the question of ‘what to do’ has already been answered, ‘how to do it’ is
the problem that is addressed by the modeling process. In corporate settings, the
outline for change management projects will often be business objectives and
targets stated by the top executives. Seeing the modeling process as a journey of
learning, the business objectives and targets give the direction and the frame for
the journey and the modeling and simulation activities assist in understanding
implementation challenges and the creation of commitment and alignment
between the individual participants.
For process improvements rather than reengineering, expectations of
improvements also depend on the maturity of the process. For a process that has
already undertaken many improvements programs, fine-tuning is expected rather
than improvement leaps.355 The usage of a system dynamics model, however, will
often be applied to deal with the deeper understanding of system behavior, and
will consequently typically have broader impact than merely continuous process
improvement.
354
The selection of four significant areas is made to focus the discussion and to
avoid the impossible mission of writing a ‘manual’ on how to apply system
dynamics modeling in change management context.
355
In Schneiderman, Arthur M.: “Setting Quality Goals: Use observed rates of
continuous improvement to position targets”, Quality Progress, April 1988,
p. 56, rational goal setting is discussed in terms of the individual company’s
situation. The term “half-life“ is introduced, indicating that a constant time
factor exist for reaching half of the theoretical possible process improvement at
any given time (comparable with the behavior of radioactive decay).
With a modeling process being a journey of learning, a central challenge

in a change management context is to ensure that the learning is very likely to
correspond with the intended change. In this respect, the use of a preliminary
model can draft the process. Akkermans discusses the usage of a preliminary
model as a means to create a certain focus in a modeling process.356 In the case
study, the preliminary model was constructed to illustrate the expected behavior
of the new location strategy. During the development of the preliminary model in
the case study, the very first results already took form as the project owners
gained some interesting insights. Something first considered as a potential
mistake in the model turned out to be an important insight: it became clear that
one decision, which had recently been made, had a stronger negative impact in
year 1 than anticipated. A decision was therefore made to modify the course of
action, and make the transition over a longer time-horizon. In addition, the
preliminary model appropriately illustrated the high-level idea behind the new
strategy, which was a prerequisite for the project owners to initiate the modeling
activities. The project owners had no intentions to start a group model building
process from clean sheets of paper with the risk of losing control, but wanted to
see a preliminary model that in broad terms supported their relatively firm
viewpoint on the future direction of the company. It is likely be a general trend in
corporate environments that executives have a clear view of the direction they
want to drive a given change. Executives could therefore be expected only to
initiate a modeling process, if they are shown a model whose behavior
corresponds with the insights that they want to transfer to the implementers, and
if the model at the same time seems to be an effective and appropriate “tool” to
demonstrate and investigate the problem. Furthermore, lack of experience and
comfort with participative modeling processes could result in the fear of a project
resulting in a model with hidden errors or having the problem being addressed or
conceptualized erroneously. In the case study, top executive initial trust in the
modeling process was gained through the preliminary model.
From the viewpoint of modeling participants, the modeling process in a
change management context is restricted to some degree, compared to the
exploratory freedom in traditional modeling studies. Akkermans discusses the
possible reduced participant ownership of the model, when using a preliminary
model, compared to an ‘empty whiteboard’ approach.357 Regarding the location
356
Akkermans, Henk: Modelling With Managers, Breda, 1995, pp. 116—117.
357
Akkermans: Modelling With Managers, 1995, p. 116.
strategy case study, one extreme viewpoint could be that no free exploration of
all possible scenarios took place, no free discussions over sound and just business
objectives were held, and only limited hierarchy-free conversations resulted.
Sterman discusses situations, where clients use models to support already reached
conclusions, recommending modelers to “speak truth to power” and get “a better
client,” and Borum discusses the importance of free, informed choices in
establishing commitment in change processes. 358 A different viewpoint on the
same matter could be that everyone in the process was informed about the
expected goals of the project, no-one was given the impression that he or she had
more influence on the direction of the strategy than was actually the case, and no-
one did generally criticize the task of implementing a strategy that was decided
by someone else. On the contrary, everyone agreed that by modeling, structure
was given to a usually chaotic and emotionally charged process and with the
system dynamics model a basis for discussion existed. Furthermore, free,
informed choices took place both with regards to the model refinement,
parameter settings, and the decisions on how to implement the strategy. This was
a deliberate part of enhancing participant ownership of the model.
It can be argued that the context in which the case study took place reflects
a rather typical and frequently occurring set-up in corporate environments with
which organizational members do not necessarily disagree. In the evaluation
interviews, there were some remarks concerning the fact that participants in the
process were quite informed about their options to change things. Most agreed
that their task was to refine and implement a given strategy, not to re-formulate
the strategy. However, for others the modeling process served as a disciplinary
tool, as the following quote from a participant shows: “A few participants did not
agree with the business objectives and for that reason they also did not agree with
the process, but nevertheless the process forced them in the decided direction,
and through the modeling they gained some of the insights motivating the
intervention in the first place”. However, this can be true for any intervention
approach and for many situations in organizations: people in business firms are
well aware of the fact that they sometimes have to implement things they would
358
Complex World, Boston, 2000, p. 85; Borum, Finn: Strategier for
organisationsændringer, Copenhagen, 1995, p. 58. It is important to note, that
Sterman’s recommendation is made in a modeling context different from change
management.
rather not. In interviews with the project owners, the issue of possible
‘manipulation’ was addressed explicitly. The project owners did not agree with
this viewpoint, as it is normal business that the top executives outline a strategy
and the level below is responsible for implementing it. In their view, top
executives as a part of their role and responsibility exercise legitimate use of
power, which has noting to do with manipulation because it is openly stated
upfront. Thus, there are frequently decisions in larger organizations that are not
open for discussion. It is difficult to say whether the problematic of a framed
intervention had a negative impact on the participants’ ownership and trust in the
model. The questionnaires do not explicitly include questions regarding this
possible impact of a preliminary model, due to the problem of a measure that
influences the system (in this case, creating negative attitudes by explicitly
raising the issue).
As a final remark on the role of business objectives and targets, it is
relevant to consider the common trends in the literature discussing change
management. In general, clear objectives and targets are seen as important and
necessary elements in creating planned change. Therefore, when modeling efforts
are used within the context of change management it could be seen as quite
natural that the process is framed by business objectives and targets.359
Furthermore, the literature also emphasizes the iterative nature of problem
solving and strategic processes, implying the need for implementation issues and
new insights gained in an implementation process feeding back to the strategy
forming or decision-making phases.360 In the case study, feedback of modeling
insights to the original decision-makers took place as formally stated critical
success factors, implying needed adjustments in the overall company governance
model.361
359
A parallel can be drawn to the importance of clear objectives in learning and
experimentation situations supported by simulators, see Größler, Andreas:
“Don’t Let History Repeat Itself: Methodological Issues Concerning the Use of
Simulators in Teaching and Experimentation”, System Dynamics Review,
Volume 20, Number 3, Fall 2004, pp. 268—269.
360
See discussion in chapter A on the diagnostics and decision-making cycle, and
the change management cycle, in problem-solving processes.
361
One example is the importance of head-count focus vs. cost focus in the
governance procedures, which is discussed elsewhere in this dissertation.
2. Structured Development of Change Leaders

The more difficult a change program is and the more internal commitment is
necessary for effectiveness, the more the key employees need to be involved in
the design, execution and monitoring of the changes in order to ensure
sustainable change.362 In many consulting projects, a practical solution on how to
establish such involvement is to strive for the development of what could be
called ‘change leaders.’363 In this context, change leaders are the critical managers
from across the organizational structure and hierarchy with the highest direct
influence on implementation. The term ‘reference group’ is sometimes used in
the literature for a similar type of employee involvement in projects, but often
reference groups have a broader scope than the establishment of change leaders,
also encompassing other groups of stakeholders.364 Anderson and Anderson
discuss the creation of change leaders in what seems to be a somewhat different
context, focused on the leadership skills among the top executives formally
responsibility for the change process.365
362
Argyris, Chris: Interventions Theory and Method – A behavioural Science View,
Reading, Massachusetts, 1970, p. 83.
363
This observation is based on a number of consulting projects, which the author
has carried out as consultant or engagement manager in IBM Management
Consulting and Deloitte Consulting Group between 1993 and 2002. Andersen
Consulting uses a similar approach (although it uses other terms), see
Lochmann, Hans-Dieter und Michaela Rüsch-Kornasoff: “Organization Change
Strategy – Ein wesentlicher Baustein des Reengineerings”, in Manfred Perlitz,
Andreas Offinger, Michael Reinhardt and Klaus Schug (eds.): Reengineering
zwischen Anspruch und Wirklichkeit, Wiesbaden, 1996, pp. 329—340.
364
Involvement of reference groups is often discussed in change management
literature with regards to the capture of important information and input, and in
order to ease a later implementation by avoiding negative attitudes to changes
only due to the fact that some people feel offended that they were not at all
involved in the project. In the approach “The Reference Group” by Jørgen
Randers, referees are involved primarily through interviews and workshops, see
Rouwette, Etiënne: Group model building as mutual persuasion, Nijmegen,
2003, p. 44. In group model building approaches emphasis is rather on direct
participation in modeling and simulations efforts.
365
Vol. 33, No. 3, 2001, p. 45.
The purpose of development of change leaders is to transfer project

ownership from the top executives who have decided and launched a change
project to the managers being responsible of implementing it. This could be done
with the dual objective of:
(1) using the change leaders’ input to establish effective action and
communication plans, and maybe to refine or alter the change program,366
and
(2) get the change leaders committed to the change, taking into account the
full range of cognitive, affective, and conative elements, social norms
elements, and perceived control elements.367
Discussing change leaders, it is interesting to draw a parallel with the

usage core team members, as seen in many Business Reengineering projects,
where individuals are taken out of their previous environment, to use their skills
and experiences in the creation of reengineered business processes.368 In
reengineering projects, the role of the core team members has a strong diagnostic
focus of the problem solving process, whereas the use of change leaders has a
strong change management focus building upon the individual change leaders’
366
The structured involvement of change leaders is a way to exercise what Klein
calls ‘sympathetic understanding’ regarding the change resistance, seeking
valuable input about the nature of the system that is going to change, potentially
motivating a modification of the change itself or the change implementation
process, see Klein, Donald: “Some Notes on the Dynamics of Resistance to
Change: The Defender Role”, in Warren G. Bennis, Kenneth D. Benne and
Robert Chin: The Planning of Change, 4 th edition (first published in 1966), New
York, 1985, p. 103.
367
The five elements are taken from Ajzen’s framework for theories of planned
behavior; see discussion in chapter B.II.1. See also Schein, Edgar H.: Process
Consultation, Boston, 2000, part I, p. 68, emphasizing the importance that the
handover from the decision-makers to the implementers should be carefully
planned to avoid communication breakdown.
368
As seen in Business Reengineering projects, where individuals are taken out of
their previous environment, to use their skills and experiences in the creation of
reengineered business processes Bungard, Walter: “Zur Implementierungs-
problematik bei Business-Reengineering Projekten”, in Manfred Perlitz, Andreas
Offinger, Michael Reinhardt and Klaus Schug (eds.): Reengineering zwischen
Anspruch und Wirklichkeit, Wiesbaden, 1996, pp. 264—265.
existing organizational platform.369 Compared to projects where project members

are selected primarily based on knowledge, experience, and formal responsi-
bilities, change leaders are selected using additional criteria. Change leaders are
also selected based on peer network and informal power-bases, which means that
not only managers formally responsible for the implementation, but also
managers with strong informal influence are relevant for consideration as change
leaders.370 The change leaders will (if it works out successfully) function as early
adaptors of the change, and this fits well with the thoughts of Schein, who
recommends that the personal power-base of individuals, their connections with
peer employees, as well as their change readiness should influence with whom to
start the change process.371 These individuals have the potential to positively
influence the change process in informal ways, and alternatively (if not being
involved) they might contribute to the stiffness in the organizational change
resistance. For the later reason, it might be relevant to involve powerful
employees even with low change readiness, primarily to avoid or reduce their
counter-productiveness in the implementation phase. The learning and
establishment of commitment strived for through a modeling process might be
especially relevant for such individuals.
Involving change leaders in change planning activities often encompasses
key employees from different hierarchical levels. In this regard, the planning of
activities and forums should take into concern the challenges regarding
hierarchical-free discourse, which is an area that still troubles the field of social
science despite the fact that significant research has already been made in this
369
Although it should be noted, that even though the change leaders are going to
implement an already outlined strategic initiative, the implementation in itself
also involves creative problem-solving. Projects involving system dynamics will
be likely to regard discontinuous change, requiring the change leaders to
navigate in a new landscape, where new and innovative solutions are required.
370
It should be noted that having change leaders also being informal change leaders
implies a broader definition of the term change leader compared to the use in
Vol. 33, No. 3, 2001, pp. 40—48, where focus is mostly on top executives.
371
Schein, Edgar H.: Organisationspsykology, Danish translation, Herning, 1990,
p. 257.
area.372 A number of arguments exist for ‘educating’ a group or a number of

groups of change leaders, as opposed to educate the change leaders individually.
Decisions made by individuals in group settings, even regarding the individual’s
own goals, seems to have a significantly more endurable behavioral effect
compared to those made in a 1-on-1 lecturing setting.373 Also, there are strong
indications that it is easier to change the ideology and social practice of a small
group handled together than of single individuals.374 Furthermore, most
individuals do not want to divagate too far from the standards of the group they
belong to or wish to belong to.375 Therefore, the feeling of belonging to a
‘prestigious’ group of change leaders might be important. To avoid the change
leaders “falling back” into old norms or behavior, the new group ties to the other
change leaders could be expected to be of importance. Also, the selection process
of change leaders might benefit from considerations regarding the individual’s
personality characteristics, such as whether he or she easily feels uncomfortable
representing new norms or behavior. In the efforts to strive for creating
sustainable change, the change leaders could also be expected to influence the
social norms in broader organizational contexts, and the discussions within the
group of change leaders could be expected to positively influence the individual
change leaders perceived behavioral control.376
Gladwell presents an interesting analogy for the structured usage of
change leaders in change management, proposing that the spread of messages and
behavior in social systems can be compared to epidemics spreading a virus.377 An
372
Jöns, Ingela: Managementstrategien und Organisationswandel, Mannheim
University, 1995, p. 157.
373
Lewin, Kurt: “Group Decision and Social Change (first published in Newcomb
and Hartley’s Readings in social psychology, 1948, pp. 330—341), in Martin
Gold: The Complete Social Scientist – A Kurt Lewin Reader, Washington,
American Psychological Association, 1999, pp. 276—279.
374
Lewin, Kurt: “Group Decision and Social Change”, in Martin Gold: The
Complete Social Scientist – A Kurt Lewin Reader, Washington, 1999, p. 273.
375
Complete Social Scientist – A Kurt Lewin Reader, 1999, p. 281.
376
For a discussion on the importance of social norms and perceived behavioral
control in the changing of behavior, see Ajzen, Icek: Attitudes, Personality and
Behavior, Chicago, 1988, pp. 121—133.
377
Gladwell, Malcolm: The Tipping Point, paperback edition, New York, 2002,
p. 9.
epidemic will spread widely as a result of the behavior of the people transmitting
the virus, the characteristics of the virus, and environmental factors.378 Using this
analogy in a change management context gives rise to conscious reflections
regarding: who to involve, how to make the message and behavior ‘stick’ and
how to create settings and conditions that support the change process. In terms of
the use and selection of change leaders, Gladwell in particular has some
interesting thoughts regarding the first aspect: who to involve. He has formulated
the ‘Law of the Few,’ arguing that the effective spread of messages and behavior
often depends on relatively few individuals in a social system, namely individuals
with special skills regarding relations with others, regarding information
accumulation or regarding persuasion.379 The people with special skills regarding
relations with other people are called ‘Connectors.’ Compared to other
individuals, they know and interact with significantly more people. They are the
social glue of an organization, and they spread messages. The people who
accumulate information, the ‘Mavens’ are the one’s whose opinion most people
will take very seriously. The last group, the ‘Salesmen,’ consist of those who are
able to persuade unconvinced colleagues. Gladwell argues, that all three types of
personalities are critical to involve if a change process is to take advantages of
the ‘word-of-mouth’ phenomena.
Gladwell’s thoughts on creating change as initiating epidemics has some
similarities with Hamel’s 8-step process on how to “revolt” the business by
creating a movement within the organization, although Hamel’s process is a
bottom-up approach to be used by innovative employees rather than a top-down
change approach. The 8-step process includes “infecting” others with the idea,
creating coalitions, and working closely together with representatives from across
the organization.380
The structured development of change leaders should also be viewed in
terms of organizational learning. An organizational intervention cannot be seen
independently of the ongoing organizational development. The single
intervention is highly dependent on the flexibility and change readiness of the
individuals involved and the existing system structures. The intervention, on the
378
When an epidemic starts to spread widely, it has passed, what Gladwell calls
‘the tipping point, see Gladwell, Malcolm: The Tipping Point, paperback
edition, New York, 2002, p. 18.
379
Gladwell, Malcolm: The Tipping Point, New York, 2002, p. 19 and pp. 30—88.
380
Hamel, Gary: Leading the Revolution, Boston, 2000, pp. 187—206.
other hand, is also a part of the continuous learning and shaping of the
organization, its change readiness, and its change capabilities, why it seems
relevant to have the intervention planning including organizational development
considerations as well as considerations of each change leader’s individual
personal transition.381 The usage of modeling can help in the establishment of
understanding and commitment among change leaders regarding a change
process. But other elements are important for the effectiveness of change leaders,
mainly regarding general leadership skills such as interpersonal relations and
communication skills, e.g. presenting ideas in a way that activates feelings and
makes the message memorable.382
3. Designing the Change Process

Looking for models and frameworks for the design of change processes, most
approaches still build upon Lewin’s change model and the action research
theories, suggesting a general framework for planned change in organizations
with four basic activities:383
1. Entering and contracting (initial data gathering and committing resources)
2. Diagnosing
3. Planning and implementing change
4. Evaluating and institutionalizing change
381
See Schein, Edgar H.: Organisations Psychology, Herning, Forlaget systime,
1990, p. 40; Argyris, Chris: Interventions Theory and Method – A Behavioural
Science View, Reading, Massachusetts, 1970, chapter 1 and 2.
382
McKee, Robert: “Storytelling That Moves People”, Harvard Business Review,
June 2003, p. 52.
383
and Change, Ohio, 2001, p. 28. This framework fits also well with generic
consulting methodologies; e.g. a phase model of reengineering projects
described by Perlitz: (1) initializing, incl. project initialization and project
understanding, (2) problem analysis and redesign, (3) implementation, and (4)
anchoring and continuous development, see Perlitz, Manfred, Jürgen Bufka,
Andreas Offinger, Michael Reinhardt, und Klaus Schug: “Reengineering-
Projekte erfolgreich umsetzen – Ergebnisse einer Erfolgsfaktorenstudie”, in
Perlitz, Manfred, Andreas Offinger, Michael Reinhardt and Klaus Schug (eds.):
Reengineering zwischen Anspruch und Wirklichkeit, Wiesbaden, 1996, p. 186.
The dedicated change management literature focuses especially on the

third step in Lewin’s model: planning and implementing change. The theory of
planned change, with the “unfreezing-movement-freezing” process, takes as point
of departure the need for weakening existing behavior or attitudes, establishment
of feelings of dislike for the present situation, and also establishment of
psychological feeling of safeness in the change process.384 The establishment of
dislike for the present situation is often discussed using terms such as ‘change
imperative’, ‘burning platform’, and ‘establishing sense of urgency’, and De
Geus and Dörner discuss the usage of both positive and negative goals for
motivating change.385 The change imperative functions as a mechanism to open
minds to change and to lower change resistance. Change resistance, or rather
overcoming change resistance, is a central element in the discussions of planned
change and implementation of organizational changes.386. Doppler and
Lauterburg give three causes of change resistance: (1) the goals, background or
motives for change are not understood, (2) the people concerned understand what
is said but do not believe it, and (3) the message is understood and believed, but
the people concerned will not or cannot comply with the change.387 Furthermore,
they summarize symptoms for change resistance in two dimensions: verbal vs.
nonverbal and active vs. passive, see table D-2.
In the case study regarding the implementation of a location strategy, a
large number of the symptoms in table D-2 could be observed directly or
indirectly. The descriptions of goals, background and motives for the change
(described in chapter C.II.1 and C.II.2) are intuitively easy to dismiss as “wishful
384
For further discussion on the unfreezing-movement-freezing process, see chapter
A.II as well as Schein, Edgar H.: Organisationspsykologi, Herning, 1990,
pp. 254—255.
385
See Kotter, John P: Leading Change, Boston, 1996, pp. 35—49; de Geus, Arie
P.: “Planning as Learning”, Harvard Business Review, Vol. 66, No. 2, March-
April 1988, pp. 70—74; and Dörner, Dietrich: The Logic of Failure, New York,
1996, pp. 49—54.
386
Chin, Robert and Kenneth D. Benne: “General Strategies for Effecting Changes
in Human Systems”, in Bennis, Warren G., Kenneth D. Benne and Robert Chin:
The Planning of Change, 4 th edition, New York, 1985, p. 22.
387
Doppler, Klaus and Christoph Lauterburg: Change Management – Den
Unternehmenswandel gestalten, 10 th edition, New York, 2002, p. 324.
thinking from the Board.”388 For the people expected to lead the change as well as
for the employees as such, the project owners found that all the three main causes
for change resistance listed by Doppler and Lauterburg were relevant when
designing the change process: addressing whether the reasons for change were
understood, addressing whether the reasons were believed, and addressing
whether people were willing and able to conform with the change.
Verbal Nonverbal
(Talk) (Behavior)
Active OPPOSITION REBELLION

(aggressive) Augmenting against Disturbances
Accusations Controversy
Threats Intrigue
Polemic Rumors
Stubborn formalism Building of cliques
Passive EVASIVENESS DEMOTIVATION

(escape) Silence Lack of attention
Treat it as unimportant Lack of energy
Make it look stupid Absence
Make fun of it Withdrawal
Discuss less important issues Illness
Table D-2: Generic symptoms of change resistance389
In the early years of the 21st century, normative management literature has
increased attention devoted to the affective side of planning and implementing
388
- The change imperative was stated as: “right now is the right time to hire people
in the low-cost locations, because due to company growth, it can be done right
now without staff reduction at high-cost locations, and the expected results are
improved competitiveness and further company growth, also securing jobs at
high-cost locations in the future.”
389
Own translation and modification from Doppler, Klaus and Christoph
Lauterburg: Change Management – Den Unternehmenswandel gestalten, 10th
edition, New York, 2002, p. 326.
change. Kotter and Cohen have introduced the term “see-feel-change” arguing
that, “people less change what they do because they are given analysis that shifts
their thinking than because they are shown a truth that influences their
feelings.”390 Roberto and Levesque use similar expressions, arguing for the “show
me” rather than the “tell me” approach.391 To make people “see” includes
communicating the visions and the need for change using emotionally engaging
approaches, such as storytelling, symbolic actions, a video of an angry customer
or an off-site event.392 The usage of participative system dynamics modeling fits
well with the affective oriented change approaches. The theories of system
dynamics are based upon not only a cognitive, but also an affective and ‘seeing is
believing’ learning approach, supporting the learning process by stimulating
experimentation and simulating experience. The importance of the latter is
expressed by Brown as follows: “It’s never enough to just tell people about some
new insight. Rather, you have to get them experience it in a way that evokes its
power and possibility”.393
A modeling process fosters involvement and participation, which is widely
recognized as some of the most effective strategies for overcoming change
resistance.394 Although dealing with change in large organizations, there is the
problem of scalability. For large-scale organizational change it seems rather
unrealistic to have a significant number of the implicated employees participate
390
See Kotter, John P. and Dan S. Cohen: The Heart of Change, Boston, 2002, p. 1.
As the book is practical rather than research oriented, there is no discussion of
the “see-feel-change” in terms of cognitive, affective and conative elements.
391
Roberto, Michael and Lynne Levesque: “The Art of Making Change Stick”,
MIT Sloan Management Review, Summer 2005, Vol. 46, No. 4, Summer 2005,
p. 56.
392
See Kotter and Cohen: The Heart of Change, 2002, p. 181; Roberto, Michael
and Lynne Levesque: “The Art of Making Change Stick”, MIT Sloan
Management Review, Summer 2005, Vol. 46, No. 4, Summer 2005, pp. 56—57.
Feelings facilitating change include faith, trust, optimism, urgency, reality based
pride, passion, excitement, hope, and enthusiasm, whereas feelings like anger,
false pride, pessimism, arrogance, cynicism, panic, exhaustion, insecurity, and
anxiety undermine change, see Kotter and Cohen: The Heart of Change, 2002,
p. 180.
393
Brown, John Seely: Research That Reinvents the Corporation, Harvard Business
Review, August 2002, p. 108.
394
and Change, Ohio, 2001, p. 158.
in modeling efforts.395 In the case study, participative modeling did not support
the general learning process of the organization, but targeted the key managers
responsible for the implementation, by aiming to developing these individuals
into change leaders. The modeling efforts addressed the initial parts of the
change management process – or what could be called bridging the launch of the
strategic initiative and the implementation efforts. Roberto and Levesque have
proposed a change framework arguing that certain parts of the change processes
must be planned well before actually starting the main intervention; hence the
name Four Antecedent Processes (figure D-2).
CHARTERING
The process by which the organization defines the initiative’s purpose, its scope and the way
people will work with one another on the program.
• Boundary Setting – Definition of scope of initiative
• Team Design – Definition of roles, responsibilities, norms and ground rules for teamwork
How managers develop, test and refine ideas through experimentation before full-scale
LEARNING
rollout..
• Discovery – Data and information gathering to define goals of initiative and means of
achieving objectives
• Experimentation – Testing and refinement of initiative prior to full-scale rollout.
The use of symbolism, metaphors and compelling stories to engage hearts as well as
MOBILIZING
minds in order to build commitment to the project.

• Storytelling – Use of stories and metaphors to create compelling accounts about need
for initiative and explain specific changes
• Symbolic actions – Use of symbols to reinforce creditability and legitimacy of core team
and its message
A series of activities aimed at reshaping the organizational context, including a

REALIGNING
redefinition of roles and reporting relationships as well as new approaches to monitoring,

measurement and compensation.
• Job Redesign – Alteration of underlying structures and processes that support jobs.
• Performance Management – Invention of new metrics to measure effectiveness of initiative
and incorporation of the metrics into employee appraisal process
Figure D-2: Four antecedent processes in organizational interventions 396
395
Although some of the same underlying mechanisms can be applied in larger
scale in the form of gaming environments, simulators, etc.
396
The figure is taken from Michael and Lynne Levesque: “The Art of Making
Change Stick”, MIT Sloan Management Review, Summer 2005, Vol. 46, No. 4,
Participative modeling is especially relevant in the second of the four

antecedent processes, the learning process, as both individual and group learning
is the essence of modeling. Planning of the modeling process is also closely
interlinked with the chartering process, with focus on traditional change
management activities such as total intervention design and planning, traditional
stakeholder analysis, strategies for communication and involvement of key
stakeholders.397 Regarding the mobilizing process, a modeling process has the
potential of providing a deeper understanding of the problem as well as
simulating experience, which will contribute to change leaders’ ability to perform
effective leadership. However, the individual leaders’ ability to perform effective
communication and their interpersonal skills must be expected to be the main
factors for the effectiveness of the mobilization phase.398 When it comes to the
realigning process, the modeling process has offered the possibility of create a
deeper understanding of the problem, and cross-organizational discussion in the
search for best practice solutions and an aligned implementation plan.
Furthermore, actual model insights will be likely to influence the realigning
process through input to both job redesign and performance measurements. All in
all, modeling seems to have the potential to contribute to the establishment of the
foundation for change roll-out, and due to its contribution to the development of
a cross-organizational shared mental model, it also seems to have the potential to
support the refreezing or, in more modern terms, the establishment of sustainable
change.
With the modeling process being an integral part of a planned change
process, it is necessary to ensure that the process to ‘stays on track’, with an
ongoing alignment between modeling learning and the objectives of the change
project. The Workbook methodology as described in the Group Model Building
literature offers a way to ensure communication along the process internally
among modeling participants as well as with stakeholders who do not take part in
the modeling sessions. When modeling for change management purposes, it also
Summer 2005, p. 56 and p. 60. The four processes are stated to be critical to
successful change; although the need for clear objectives, sound project
management, accountability and control systems are also stressed (p. 55).
397
An example of this kind of planning is described in the case study description of
the intervention design, see description chapter C.II.2.
398
The importance of leaders presenting ideas in a way that activate feelings and
make the message memorable is also discussed in McKee, Robert: “Storytelling
That Moves People”, Harvard Business Review, June 2003, p. 52.
seems fair to assume that at least one of the facilitators in the modeling projects
must be able to represent the viewpoints from the top executives. Furthermore,
traditional project steering mechanisms such as frequent check-points with a
steering committee would seem to be a likely approach in securing alignment
between modeling learning and change objectives. Although such aligning is a
balancing act: on the one hand to get unbiased input from and discussions within
the group, on the other hand to control the scope and orientation of the modeling
process.
For change projects dealing with more complex or cross-organizational
problems involving highly specialized managers, the cognitive capabilities of the
individuals must be expected to yield special challenges. Kieser and Koch
discuss the problem of shared knowledge when the individuals have difficulties
overlooking the possible combinations of knowledge, and they present case study
findings suggesting the relevance of thinking in terms of re-combining knowledge
rather than sharing knowledge.399 For organizational learning in organizations
with high specialization, Kieser and Koch also call for a knowledge integration
mechanism that does not rely on cross-learning, and discuss the benefits of
simulated prototyping, simulated experience, joint thought experiments et cetera,
and in their conclusions they emphasize the importance of teams and
communities in creative knowledge creation.400 It could be an area of further
research to investigate the possible role of system dynamics modeling in such a
context; maybe in a combination of a top executive “meta-modeling” process in
iterations with change leaders participating in modeling processes involving one
or more sub-models. For complex problems, the involvement of change leaders is
important in the search for sustainable change, but it must also be expected to be
crucial to involve the employees representing the critical specialized
knowledge.401
399
Kieser, Alfred and Ulrich Koch: Organizational Learning through Rule
Adaptation: From the Behavioral Theory to Transactive Organizational
Learning, Mannheim, 2000, p. 12 and pp. 15—16.
400
Kieser and Koch: Organizational Learning through Rule Adaptation, 2000,
pp. 18—27.
401
See Forrester, Jay W.: “Policies, decisions and information sources for
modeling”, European Journal of Operational Research, Vol. 59, No. 1, 1992,
pp. 42—63. See also Kieser and Koch: Organizational Learning through Rule
Adaptation, Mannheim, 2000, p. 19, discussing the transactive memory concept
As reducing change resistance is a core challenge in the actual roll-out of

change initiatives, a counterproductive element is the ‘worse-before-better’
effect. Implementing change typically means changing routines, which requires
the employees to perform their job in new ways, and consequently can be used as
an argument against the usefulness of the change in the early phases of the
changes.402 A number of system dynamics models have been published exploring
and describing the effect of an initial lowering in performance due to
improvement programs, including a model by Repenning and Sterman simulating
a ‘working smarter strategy’ in creating and sustaining process improvement.
Jacobsen and Samuel have modeled planned organizational change, considering
the change process in terms of gap between current performance and target
(assuming a certain gap will make a planned change process necessary), the
simplicity of the change program, the cost of the change program, the pacing of
the change program, the organizational change resistance, and lastly also two
factors making employees accept a change program, namely employee
involvement in the change and inducements (e.g. bonuses and other
compensatory mechanisms).403 A major insight indicated by the use of this system
dynamics model is that any planned organizational change will temporarily
impair the organization’s performance. In the case study, the model clearly
showed this ‘worse-before-better’ effect and thereby had a major role in
expectation settings for the roll-out phase.
in terms of meta-knowledge or directory knowledge: knowledge about where to

find knowledge.
402
Repenning, Nelson P. and John D. Sterman.: Nobody Ever Gets Credit for
Improvement”, California Management Review, Vol.43, No.4, Summer 2001,
p. 74.
403
Jacobsen, Chanoch and Yitzhak Samuel: “Planned Organizational Change:
Theory, Model, Data and Simulation”, in Milling, Peter M. and Erich O.K. Zahn
(eds.): Computer-Based Management of Complex Systems, Proceedings of the
1989 International Conference of the System Dynamics Society, Heidelberg,
1989, pp. 104—118.
4. Facilitation of modeling and simulation sessions

Kotter’s 8-stage process of change and Roberto and Levesque’s Four Antecedent
Processes both address the questions that, according to Doppler and Lauterburg,
all employees ask themselves about a planned change initiative:404
1. Why and for what reasons do we need to change? (What are the goals?
Are all the motivations made explicit or are there some hidden ones? Is the
change really important)
2. Am I able? (Is it too difficult? Is it possible for me to do? What are my
chances for success?)
3. Am I willing? (What’s in it for me? Is it interesting? Will I loose benefits
or other factors that I value like colleagues, career opportunities etc.?)
In the modeling discussions and the change process around them, these
three questions are important when explicitly or implicitly addressing the change
resistance for each participant. The more open and honest the workshop
atmosphere is, the more directly the questions can be addressed. In general, the
atmosphere of the modeling sessions is important. Gladwell describes a number
of scientific research projects conveying that a positive atmosphere (e.g. a smile
from a charismatic person or physical head-nodding among the receivers)
influences the receiver’s likelihood to agreeing with a message.405 The
organizational situation and interpersonal factors influence the magnitude of the
challenge for the facilitator to establish a constructive atmosphere. If the change
is due to present and serious problems, for example, managers are likely to
exhibit behavior including hiding information, secrecy and denial, blame,
avoidance and/or passivity and helplessness.406 In such an atmosphere, the
404
Own translation from Doppler, Klaus and Christoph Lauterburg: Change
Management – Den Unternehmenswandelgestalten, 10th edition, New York,
2002, pp. 326—327.
405
Gladwell, Malcolm: The Tipping Point, paperback edition, New York, Back Bay
Books, 2002, pp. 74—87.
406
Kanter, Rosabeth Moss: “Leadership and the Psychology of Turnarounds”,
Harvard Business Review, June 2003, p. 61.
facilitator’s job of establishing trust in the process is challenging, but also crucial
for team members to be willing to contribute and cooperate.407
Although a good atmosphere improves the learning potential, an ‘extre-
mely’ good atmosphere also constitutes facilitation challenges. In groups
characterized by group cohesiveness and consensus, groupthink can occur with
lack of critical thinking, resulting in limiting discussions to few scenarios without
due consideration of alternative scenarios or possible alternative gains.408
Furthermore, groupthink often leads to not attaining sufficient information (even
from experts within the organization), ignoring facts not supporting the favorite
scenarios, dismissing feedback or new information that should lead to changing
earlier group decisions, as well as underestimating implementation challenges.409
In a change management context, the preliminary model and intervention
objectives and targets frame the modeling project, but still, within the framework,
such a project also involves innovative problem-solving and decision-making
regarding the implementation strategy and plan. O’Connor and McDermott stress
the importance of widening perspectives by stimulating creativity, having a
variety of different viewpoints represented, and getting different sorts of
feedback.410 The facilitator furthermore has the option to use different curiosity
stimulating mechanisms, as curiosity makes people question mental models and
generate new perspectives411 With regards to creativity and curiosity, creative
problem-solving literature often emphasizes the importance of “asking stupid
407
Henttonen, Kaisa and Kirsimarja Blomqvist: “Managing distance in a global
virtual team: the evolution of trust through technology-mediated relational
communication”, Strategic Change, Vol. 14, March-April 2005, p. 108. Here,
trust is discussed in respect to ‘normal’ teams in ‘normal’ projects, but
Henttonen and Kirsimarja later also discus evolution of trust in virtual teams
(which was actually one of the issues discussed among project participant the
case study outlining how to implement the location strategy).
408
Poor decision-making as result of groupthink is described in Janis, Irving L.:
“Groupthink: The Problems of Conformity” (original printed in Psychology
Today, Nov. 1971, pp. 271—279), in Morgan, Gareth: Creative Organization
Theory, Newbury Park, California, 1989, pp. 225—227.
409
Janis: “Groupthink: The Problems of Conformity”, in Morgan, Gareth: Creative
Organization Theory, Newbury Park, California, 1989, p. 227.
410
O’Connor, Joseph and Ian McDermott: The Art of System Thinking – Essential
skills for creativity and problem solving, London, 1997, p. 141.
411
O’Connor and McDermott: The Art of System Thinking – Essential skills for
creativity and problem solving, London, 1997, p. 141.
questions” in an open an honest atmosphere. Leonard discusses the value of

making constructive use of tensions between organizational units in the search for
innovations, and calls for mechanisms to “translate across different languages and
finding ways of encouraging the depersonalization of conflicting perspectives.”412
This is also related to the importance of aligning language and codes in change
projects.413 Conflicting perspectives can follow from conflicting policies and
objectives, structural misalignments, and from individual conflicting cognitive
and affective perceptions. Warren describes the example of sales-driven policies
vs. earnings-driven policies, where a sales-driven policy (with corresponding
investment in marketing) can be better in the long term, whereas the earnings-
driven policy (focusing on cost cutting) would seem better in the short term.414
Structural misalignments include conflicting measurement systems between both
departments and individuals, resulting in focus on own performance measures
rather that ‘the big picture.’ Differing cognitive and affective perceptions
influence the alignment of goals and objectives on the individual level. In the
case study an interesting workshop observation is, that even the core-group
person showing the strongest personal disagreement with the location strategy
objectives became involved in committed and vital discussions after a very short
time, showing a positive attitude in the sessions. One of the explanations could be
that he could not ‘resist the fun of modeling’ when he took part in sessions, as he
and the other core members were very mathematically skilled and interested
individuals. Another explanation could be that modeling efforts offer a cognitive
framework for reducing personal barriers for involvement and honesty in
discussions.415 For this reason, modeling might have particular benefits in diverse
412
Leonard, Dorothy: Wellsprings of Knowledge – Building and Sustaining the
Sources of Innovation, Boston, 1998, pp. 74—75.
413
See discussion in chapter B.III.2.d.
414
415
This has some similarities with how astrologers use horoscopes as a virtual
reality, where humans re-arrange their perception of their own life, see Munk,
Kirstine: “In the Airy Spaces of Our Minds…: Cosmology and ritual design in
modern, Western astrology”, in York, Michael (ed.): Nature, Religion, and
Culture, London, in print, 2006, p. 27. Munk refers to the Danish astrologer and
psychotherapist Pia Balk-Møller, who suggests that the use of a chart
(a horoscope) makes a discussion ‘safer’, as the secrets are already on the table,
and Munk furthermore stresses the importance of imaginative involvement in
discussions of charts, where insight “not only has to be understood
intellectually, but also has to be imagined and felt” (pp. 29—30).
teams. Distefano and Maznevski present research, showing that diverse teams
often perform either significant worse (more often) or significant better (less
often) compared to homogeneous teams.416 Managed well, diverse teams have
potential to outperform homogeneous teams, and modeling efforts might be a
relevant approach in such a setting due to possibilities of lowering personal
barriers as well as due to providing structure to otherwise diverse discussions.
The creation of an open and honest modeling atmosphere, which at the
same time has some degree of tension and investigative interest, has much to do
with creating a platform for learning and directing discussions, not with regards
to efficient model completion, but with regards to allowing exchange of
viewpoints and aligning mental models. In scenario simulations, focus should be
on understanding why the model produces the given behavior, rather than using
the model as “an evaluator” of different scenarios. The insights from discussion
of the model behavior can lead to new insights. For real life evaluations, Farson
and Keyes discuss the importance of moving beyond both success and failure,
analyzing the underlying reasons with less focus on evaluation and more focus on
interpretation.417 Being rewarded or complimented for success can actually be as
de-motivating as criticism, whereas most humans are motivated by getting a
deeper insight into their problems. In model understanding, not taking model
results at face value, but using them to generate more discussion on model
behavior encourages the long process of testing and gaining trust in the model,
and avoiding the risk of false trust in the model. It is important that there are
enough group activities to enable the necessary discussions and that each
individual has been sufficient involved in the model building to gain both a sense
of “ownership” and to gain trust in the model. Participant’s trust in the model is a
prerequisite for establishing learning, commitment, and alignment of mental
models. This is important when striving for sustainable change built upon a new
shared mental model.418
Effective facilitation depends not only on fostering discussions, but also
on making sure that the process relies on available facts. In the case study, the
416
Distefano, Joseph J. and Martha L. Maznevski: “Creating Value with Diverse
Teams in Global Management”, Organizational Dynamics, Vol. 29, No. 1, p. 46.
417
Quoted in Farson, Richard and Ralph Keyes: “The Failure-Tolerant Leader”,
Harvard Business Review, August 2002, pp. 66—67.
418
For a discussion of the importance of shared mental models in organizational
learning, see chapter B.II.3.
initial setting of each of the parameters involved issues that could have been
discussed for hours in both workshops and other related meetings. Opinions on
parameters were often very different within the core project team. As an example,
it was a widely accepted ‘fact’ among many of the project participants that
employees in low-cost countries very often stayed only 1–2 years, because as
soon as they attained experience in R&D, they could get a better-paid job in a
high-cost country. But with the ‘forced quantification process’ with parameter
stipulations, facts came on the table, as the data were actually available,
documenting a very low employee turnover in the low-cost countries.419 The
parameter stipulations furthermore served to set benchmarks between business
units, and this approach enabled cross-business-unit knowledge and experience
exchange. For example, one business unit already had high-scale experience with
building up resources in low-cost countries, reflecting very low hiring costs due
to procedures that the other business units decided to adopt. The modeling
approach this way served as a forum for transfer of best practices, which was one
of the objectives of the modeling project in the first place.
In the case study, the quantification of the model and the simulation of
scenarios helped to provide an understanding of which parameters most strongly
influenced the effectiveness of the strategy. Based on this understanding, the
facilitator directed the discussion on how the strategy could and should be
executed, as the parameter setting reflected implementation decisions. For
example, the discussion was focused on reducing the time spent on hand-over of
tasks, resulting in discussions on structuring knowledge transfer both during and
after the initial transfer period. The discussions also focused on how to reduce
both time and costs of classroom training. Without modeling, the intuitive choice
could have been to focus primarily on the costs, including the travel costs, but
due to the model-guided new understanding of the influence of the different
parameters, increased focus was placed on reducing the time spent on classroom
training. For system dynamics practitioners, the model seems very simple, but it
is interesting to notice that the project team first tried to handle the problem with
the use of a normal Excel-spreadsheet, which became a complicated “black box”,
419
Christensen, Søren and Jan Molin: Organisationskulturer, Copenhagen, 1987,
p. 27, discuss myths as inward, empty explanations used to legitimize certain
behavior. See also Ackoff’s morale: “There is nothing so deceptive as an
apparent truth“: Ackoff, Russel L.: The Art of Problem Solving – Accompanied
by Ackoff’s Fables, New York, 1978, p. 84.
where it was difficult to see and understand significance of the different

parameters in their influence of model behavior. Not just identifying possible
parameter values, but using the modeling workshop to discuss how to ‘optimize’
the most influential parameters was seen as a cornerstone in the development of
the implementation plan of the new location strategy.
Briggs, de Vreede, and Nunamaker Jr. discuss in general terms the need
for facilitation as a barrier to the diffusion of the usage of group support systems;
arguing that despite the documented advantages of such systems, economic and
political factors result in facilitators not remaining long term at facilities
supporting non-routine, ad hoc projects.420 To attempt to overcome this barrier,
they propose the concept of ‘thinkLet’, packaging facilitation skills in a pre-
configured process aimed at supporting company-specific decision processes.421
The packaging of facilitation skills is supposed to allow practitioners without
broad facilitation competencies to guide a “repeatable, predictable pattern of
collaboration among people working towards a goal.”422 Whether this would also
be suitable for system dynamics modeling in change management settings is an
area for further research. In the case study, the questionnaire results scored
relatively low on expectations to use modeling in other projects, despite the
higher score on the usefulness of using modeling.423 One could expect this to be
connected with the problem of facilitator accessibility.
As a special comment on the facilitation challenges, it is worth noting the
rare, but interesting, problematic of ‘flirting-with-disaster.’ Salge and Milling
discuss the “flirting-with disaster” behavior in the Chernobyl on-line operations,
where bypassing of rules resulted only in positive experiences, and therefore
420
Briggs, Robert, Gert-Jan de Vreede, and Jay F. Nunamaker Jr.: “Collaboration
Engineering with ThinkLets to Pursue Sustained Success with Group Systems”,
Journal of Management Information Systems, Vol. 19, No. 4, Spring 2003,
p. 32.
421
Briggs, de Vreede, and Nunamaker Jr.: “Collaboration Engineering with
ThinkLets to Pursue Sustained Success with Group Systems”, Journal of
Management Information Systems, Vol. 19, No. 4, Spring 2003, p. 45.
422
Briggs, de Vreede, and Nunamaker Jr.: “Collaboration Engineering with
ThinkLets to Pursue Sustained Success with Group Systems”, Journal of
Management Information Systems, Vol. 19, No. 4, Spring 2003, p. 46.
423
See table C3 in chapter C.III.2.
slowly became de facto routine.424 Although the effect is discussed in terms of

learning in real systems, one could anticipate the same type of unwanted learning
taking part in simulation environments. In simulation environments, this ‘flirting-
with-disaster’ might even be seen relatively more frequently, as the consequences
would be only of a virtual nature. If the probability of the disaster occurring is
small, requiring very specific parameter combinations, then even extensive
simulations could take place without resulting in the experience of negative
consequences. Although not qualifying for such a strong word as ‘disaster’, the
case study yielded a glimpse of such behavior. As discussed before, the core
project team should only simulate scenarios within a range of parameter settings,
where no significant negative motivational responses would be anticipated.
Nevertheless, it is very likely that some of the simulations had parameter settings
that flirted with the risk of a company-wide decrease in motivation. As the final
implementation decisions were based on qualitative, subjective evaluations of
organizational acceptance, the ‘flirting-with-disaster’ scenarios had no
consequences for the present case study, but in other modeling projects the
context could be imagined to be of such character, that the facilitator should pay
attention to possible unwanted learning effects.
In general, the literature stresses the importance of facilitation for the
success of participative modeling processes.425 In the case study, it was perceived
that one of the facilitator’s most crucial tasks was, while balancing the allowing
for learning and the pursuing of efficiency, to keep the discussion focused on
solving the critical implementation issues. Having the group of change leaders
taking responsibility for identifying as well as constructively solving the critical
implementation issues, commitment and ownership for the implementation plan
seemed to be building up.
424
Salge, Markus and Peter Milling: “Who is to blame, the operator or the
designer? Two stages of human failure in the Chernobyl accident”, System
Dynamics Review, Vol. 22, in print, 2006, the figures 8 and 9.
425
See discussion in chapter B.III.2.d.
III. Outcomes of Participative Modeling Efforts in the

Implementation of Change Programs
Discussing how big, hierarchical organizations manage change, Leavitt

states, that many “do a good job of bringing new ideas into their
organizational tents, but then they do a poor job of weaving those ideas into
their cultures. Instead, they dump that grab bag of changes onto the labs of
their middle managers, expecting them to integrate and implement the whole
potpourri.” 426 A modeling project provides the opportunity to avoid such
‘dumping’ of change initiatives, and the following sub-chapters discuss
possible outcomes of participative modeling projects used from the outset of
larger change projects in hierarchical corporations.
1. Modeling and simulation as a tool for transfering insights and

ownership from decision-makers to implementers
Taking theoretical point of departure in the two cycles of the problem-solving
process as described in chapter A, modeling efforts in a change management
context focus on bridging the diagnostics and decision-making cycle with the
change management cycle.427 The diagnostics and decision-making cycle is in
itself a learning process among decision-makers. The decision-makers identify
and discuss problems in a groping process, where, especially for high-
performance management teams, they develop a shared mental model of the
problems at hand.428 Following diagnostics and decision-making, formulating and
launching a change project inevitably means involving more people, who have
not taken part in the learning journey of the diagnostics and decision-making
426
Manage Them More Effectively, Boston, 2005, p. 98.
427
The discussion in chapter A.I on the two cycles of problem-solving, Diagnostics
and Decision-Making (cycle I) and Change Management (cycle II), is based on
the problem solving process described in Schein, Edgar H.: Process
Consultation, Boston, 2000, part I, p. 61.
428
Eisenhardt, Kathleen M.: “Strategy as Strategic Decision Making”, Sloan
Management Review, Spring 1999, pp. 66—67. Eisenhardt calls the mental
model alignment “building collective intuition.”
cycle. A modeling process can utilize a system dynamics model in bridging the
two cycles of the problem-solving process, transferring the main insights
motivating the change and illustrating the intended outcome of the change
project. Furthermore, by allowing for model refinement, parameter stipulation,
and scenario testing with a high degree of empowerment among the modeling
participants, the modeling process serves as an instrument ensuring the
implementers true involvement and influence in the change process. Involvement
and influence are often discussed as essential mechanisms in lowering change
resistance and ensuring the commitment among key employees necessary for
creating sustainable change.
A modeling process will impact the transfer of insights and ownership at
individual, group, and organizational level. On the individual level, learning of
causal relationships will influence cognitive structures, and if designed well, the
process in a group setting could be expected to also influence affective and
conative elements as well as social norms and perceived behavioral control for
the project participants.429 Consequently, based on the framework of Ajzen, one
could expect behavioral changes.430 In the case study, the participants’
questionnaire answers scored relatively high on the usefulness of including the
model in the project, its usefulness in facilitating discussions, the usefulness of
starting the project with a preliminary model, individual learning in general, and
the building of a shared view.431 From the project owners’ perspective one of the
main objectives of the modeling process was establishing consensus about the
change imperative, as significant change obstacles existed of both cognitive and
affective characteristics.432 A year and a half after the modeling process, it was
429
This conclusion is based on the discussions throughout the dissertation,
supported by both the organizational psychology literature and the normative,
prescriptive change management literature.
430
Ajzen, Icek: Attitudes, Personality and Behavior, Chicago, 1988, p. 133. See
also discussion in chapter B.II.1.
431
At the same time, the lowest score was given to expectations of actual
implementation of the recommendations. However, in follow-up discussions this
gave the impression of being due to pessimistic expectations to the required shift
from headcount-orientation to cost-orientation, see further description in
chapter C.
432
The short version of the change imperative being that in times of company
growth it is possible to build up capacity in low-cost locations without reducing
high-cost locations - and this is a necessary strategy for the continued
observed, that the recommendations derived from the modeling process were
implemented at an even higher speed than outlined in the implementation plan.
Although many influential factors exist, the project owners give significant credit
to the modeling project’s influence on the change leaders commitment to the
change. At group level, the modeling process allowed for discussion and
exchange of experiences, and the high-level aggregated model secured alignment
of mental models and codes among key implementers across the organization. At
organizational level, the change affected significantly more people than were
involved in the modeling project. However, indirect influence takes place
through the development of true change leaders.
2. Refining and Aligning Implementation Plans Through Scenario

Simulation
Apart from the transfer of insights and ownership, a more tangible outcome of the
modeling usage is the development and cross-organizational alignment of
implementation plans. In the case study, the conceptual modeling step ‘Discuss
and Stipulate Parameter’ contributed to the alignment of parameters across the
different business cases in the different business units.433 The step ‘Analyze
Sensitivity and Identify Critical Parameters’ focused the discussion on the critical
parameters, which was seen as contributing to the effectiveness and efficiency of
the project. For instance, the time horizon was deliberately set to five years to
focus the discussion on the present implementation challenges. It was well known
to the executives that employees feared the long-term consequences for the
number of jobs in high-cost locations. Deliberately limiting the modeling time
horizon to five years focused the discussion on the implementation efforts, rather
than on potential long-term policies and effects. Although one could argue, that
this was a way of ‘hiding’ a relevant discussion from the official meetings (and
leaving this issue to the unofficial meetings and coffee-breaks), it was also a way
to set up a barrier against a very emotional discussion that popped up in most
meetings and mentally blocked participants from addressing the implementation
competitiveness of the company without determining the distant future of the

company.
433
For an overview of the conceptual modeling steps in the case study, see chapter
D.II.
challenge at hand. Furthermore, the long-term effects were indirectly addressed,

as the modeling efforts and the focused discussions actually established a sense
of urgency: The need to build up capacity in low-cost locations in times of
growth, seeking competitive advantages and securing jobs in high-cost locations
in the long run. The step ‘Simulate Scenarios’ helped to focus attention on the
critical improvement opportunities of the implementation strategy and helped in
the discussion of the appropriateness of the different alternative scenarios.434 The
step ‘Plan for Improvements of Critical Parameters’ stimulated discussions on
practical solutions to improvement opportunities and contributed to a cross-
organizational alignment of implementation plans.435
In the case study, the modeling process served as a framework for
discussion and exchange of experiences important for decisions on the actual
detailed implementation plans. However, even though the model was perceived
as addressing the core of the problem, a number of additional issues and aspects
were discussed without any direct link to the model or the modeling process, but
they were nevertheless important for reaching the objectives and targets as well
as for developing true change leaders. It seams fair to assume, that it is a general
trend that a model can only address a certain part of a given problem at hand, and
that an intervention should always include sessions and discussions not directly
linked to the modeling and simulation efforts.
3. Organizational Learning in Change Management Oriented Modeling
In the system dynamics tradition, modeling projects should not only aim at
creating insights of relevance in a given intervention, but also strive for the
general improvement of system thinking skills among the modeling project
434
Snabe Birgitte: Brugtvognskoncept for Skandinavisk Motor Co., Thesis, Det
erhvervsøkonomiske diplomstudium, Copenhagen Business School,
Copenhagen, 1994, p. 9 and p. 89, discusses scenario evaluation and selection
among alternative scenarios in terms of strategic fit, profitability, risk,
competitor reactions and reversibility.
435
A discussion of the importance and the challenges of cross-organizational
alignment of robust policies can be found in Lyneis, James M.: Corporate
Planning and Policy Design: A System Dynamics Approach, Massachusetts,
1980, pp. 6—9.
participants. With the development of change leaders being an essential part of

modeling efforts in a change management context, the mere experience of a
system-oriented approach is likely to contribute to the development of system
thinking skills. However, change-management-oriented modeling is more focused
on developing practical plans compared to traditional exploratory modeling, for
which reason it seems fair to expect less significant improvement of system
thinking skills. Taking a broader view of organizational learning, a modeling
process in a change management context contributes to the participants’ general
understanding of cross-organizational systems due to the process of refining and
discussing a shared model with colleagues from other parts of the organization.
This could often be expected to be a learning experience contributing to a more
holistic understanding of the company. With a modeling process involving a
larger number of participants developing a shared mental model, this constitutes
clear elements of organizational learning.
E. Targeted Participative Modeling in Change
Management
System dynamics modeling in organizations has been applied to identify strategic

objectives, to explore potential strategic behavior modes and, ultimately, to
formulate improved organizational policies to achieve strategic advantage.
Although implementation of improved policies has always been on the agenda in
the field of system dynamics, the literature rarely discusses how system dynamics
modeling can also be used in the change process of implementing already
outlined strategic initiatives.436 The purpose of this dissertation has been to
investigate the usefulness of system dynamics modeling used in a change
management context, as opposed to the often seen usage of system dynamics
modeling in an exploratory context. The change management context is defined
here as the implementation process of a planned change initiative with already
identified objectives and targets directing and framing the modeling intervention.
Chapter A discussed the need for and the challenges in organizational
interventions, and illustrates a problem-solving process of iterative stages with
two conceptual cycles: the diagnostic and decision-making cycle, and the change
management cycle.437 The practical implications of the two conceptual cycles
include issues of transfer of insights and ownerships, as it will often be the case,
that different people are responsible for the diagnostic and decision-making
process (often top executives) and the change management process (primarily the
436
For processes, where modeling has been used designing alternative policies and
structures, Forrester discusses the need for educating and debating with people
who will be involved the implementation. The challenge often includes changing
deeply embedded policies and emotional beliefs, where “it is not that people
disagree with the goals, but rather how to achieve them.“ See Forrester, Jay W.:
“System dynamics, system thinking, and soft OR”, System Dynamics Review,
Vol. 10, No. 2, 1994, p. 247.
437
The two cycles of the problem-solving process are inspired from the problem-
solving process in Schein, Edgar H.: Process Consultation, Boston, 2000, part I,
p. 61.
162 E. Targeted Participative Modeling in Change Management
operational managers). Chapter A also discussed taxonomies for change

strategies, and furthermore places the usage of system dynamics modeling in an
organizational intervention perspective. Chapter B mainly discussed the theories
underlying the usage of system dynamics, as well as methods and procedures for
the use of system dynamics modeling in organizational interventions. The system
dynamics theories, methods, and procedures described in chapter B draw from
the natural science literature as well as cognitive and social psychology,
economics, and other social sciences literature, and constitute the main
foundation for the research presented in this dissertation.
Chapter C and D concentrated on the usage of system dynamics modeling
in a change management context, from both a case study and a literature based
perspective. The case study is a single-site case study in the action research
tradition, where modeling efforts were used for the purpose of implementing a
strategic initiative. Objectives and targets for the intervention were established in
the decision-making process leading to the launching of the new location
strategy, and a preliminary model was established prior to the initiation of the
modeling project to focus and frame the modeling efforts. The model served as a
cognitive framework for discussing the change imperative as well as
implementation issues. The planned change intervention had a strong focus on
developing the managers responsible for the implementation into change leaders
through true involvement and influence in the change process, including an
iteration with the top executives based on the insight gained in the modeling
process. Also, the practical development and alignment of cross-organizational
implementation plans was central in the process, where scenario simulations
supported discussions and exchange of experiences important for decisions on
the actual detailed implementation plans. Based on the evaluation feedback
received, the case study yielded indications of valuable outcomes at individual
level (e.g., understanding the importance and the motivation for the change), at
group level (e.g., facilitation of communication, exchange of ideas and
experiences, and alignment of mental models), and at organizational level (e.g.,
insights incorporated in actual budgets and business plans across the
organization). From the case company perspective, it seems fair to conclude that
the modeling process was an effective and efficient way of refining a strategy that
fulfils board objectives as well as preparing the grounds for sustainable
implementation. Especially due to the sensitive nature of the topic, it was
interesting to observe how the modeling and simulation efforts helped to direct
and focus the discussions and facilitate individual learning. However, it is not
E. Targeted Participative Modeling in Change Management 163
possible to know whether a different approach would have been more effective or
efficient, as the case study was a single-site study without any test group.
The usage of system dynamics modeling for change management purposes
could be called a ‘targeted participative modeling’ process, addressing how to
implement a strategic initiative with already established objectives and targets.
The managers responsible for an implementation process often played only a
small part, if any, in the decision-making process from which the strategic
initiative originated. Targeted participative modeling looks especially promising
with regards to:
(1) The usage of modeling and simulation as a tool to transfer insights and
ownership from decision-makers to implementers.438 In the pursuit of
sustainable change, iterations with top executives might be relevant,
with the dual objective of adjusting the strategic initiative according to
implementation issues and giving the managers responsible for
implementation true influence on the entire change process. 439
(2) The refining and aligning of cross-organizational implementation plans
through scenario simulation. Although simulations should not serve as
‘an evaluator’ of scenarios, they can aid in the investigation of expected
system behavior for alternative scenarios and, as such, support decision-
making at the operational level critical for successful implementation.
The positive indications of both the effectiveness and efficiency yielding
from the case study could give rise to expectations to potential significant usage
of targeted participative modeling, as larger organization often launch strategic
initiatives involving the need to bridge the decision-making process with the
implementation process. A question arising from the case study is whether a
modeling approach as used in the case study would also work if the problem at
hand had been significantly more complex. For some complex problems, a
business simulator might prove more suitable compared to a modeling process in
438
With regards to the transfer of insights, a targeted participative modeling process
has certain similarities with the usage of management simulators.
439
Communication skills, interpersonal skills and other personal leadership skills
influence effectiveness in the change process, together with company factors
including change readiness, change capabilities and culture, see Cummings,
Thomas G. and Christopher G. Worley: Organizational Development and
Change, Ohio, 2001, p. 144.
164 E. Targeted Participative Modeling in Change Management
a change management context. Alternatively, a modeling process supporting the

total problem-solving process might yield additional benefits; i.e. modeling used
in an iterative set-up supporting both the decision-making and the change
management cycle. Further research in the design and evaluation of targeted
participative modeling processes could address this question, as well as barriers
for practical application (e.g., the challenge of acceptance and availability of
modeling facilitators). The evaluation of targeted participative modeling
processes constitutes a significant challenge due to the complexity of social
systems and the difficulties in measuring learning processes.440 However, the
indications of effectiveness and efficiency of the usage of targeted participative
modeling discussed in this dissertation improve the prospects for further research
into balanced constitutional and causal modeling processes, as opposed to the
one-sided commitment to causal modeling typically seen in the field of system
dynamics.441
440
Some of the same challenges are seen in the design and evaluation of business
simulators, see Größler, Andreas: Entwicklungsprozess und Evaluation von
Unternehmenssimulation für lernende Unternehmen, Frankfurt am Main, 2000,
p. 178.
441
See Van der Smagt, Ton: “Causation and Constitution in System Dynamics:
Modelling a Socially Constituted World”, Journal of Systems Research and
Behavioral Science, Vol. 23, No. 4, in print, 2006, pp. 13—14, as well as the
discussion in chapter D.I.
Appendices
Appendix A: Model Parameter Descriptions........……………………. 167
Appendix B: Model Equations…………………………………………… 171
Appendix C: Equations for Stock Initializations ………………………… 183
Appendix D Preliminary Model ………………………………………… 185
Appendix E: Model without ‘Rate-on-Rate’ Modeling………………….. 187
Appendix F: Facilitator Observations

and Key Quotes From Interviews………………………….. 189
Appendix A: Model Parameter Descriptions
The following three tables give a short description of the model input
parameters. The first table describes parameters particularly relevant to
high-cost locations. The next table describes parameters particularly relevant
to low-cost locations, and the third table describes parameters mainly
relevant for transfer of tasks and build-up of employees in low-cost
locations.
Parameter (and Units) Description
INI TOTAL FTE HC Number of R&D employees in High Cost locations

(Person) (new + rookies + experienced by Year end 2004)
HC QUIT FRACTION The fraction of people quitting HC locations

(1/Month) (internal rotation in the company not included)
HC TRAINING TIME Training time (basic new-hire training)

(Month)
HC TRAINING COST Basic new-hire training costs per person per month
(EUR/(Person*Month))
HC ROOKIE TIME On-the-job training

(Month)
HC ROOKIE Productivity factor for Rookies in HC

PRODUCTIVITY (Rookies are inexperienced and therefore have
(Dimensionless) reduced productivity)
INI HC AVERAGE Total cost (including travel, rent, licenses etc.)

PERSON COSTS allocated per HC FTE by Year-end 2004.
(EUR/(Person*Month)) Salaries are only XX% of this
HC PERSON COST Increase rate in HC person costs

INCREASE RATE
(1/Month)
Appendix A, Table 1: Parameters mainly relevant to high-cost locations
168 Appendix A
INI TOTAL FTE LC Number of R&D employees in Low Cost locations

(Person) (new + rookies + experienced by Year-end 2004)
LC QUIT FRACTION Fraction of people leaving LC locations

(1/Month)
LC TRAINING TIME Training time for the basic training of newly hired
(Month) employees (classroom training)
LC TRAINING COST Basic training cost for newly hired employees

(EUR/(Person*Month)) (per person per month)
LC TOTAL ROOKIE Time spent on on-the-job training after finished

TIME classroom training. This includes both the time as
(Month) normal rookie and the time as hand-over rookie
LC ROOKIE Productivity factor for Rookies (although only in the

PRODUCTIVITY part of the time NOT used to hand over tasks)
(Dimensionless)
INI LC AVERAGE Total cost (excluding travel) allocated to LC FTE by

PERSON COSTS Year-end 2004.
(EUR/(Person*Month)) Salaries are only XX% of this.
The cost will grow with the rate below
LC PERSON COST Increase rate in LC person cost

INCREASE RATE
(1/Month)
LC ONGOING TRAVEL Average Travel + Hotel + Rented cars etc.

COST for all LC employees
LC PRODUCTIVITY Productivity reduction due to low average experience

REDUCTION (1-2 years in LC vs. 5-10 years in HC)
(Dimensionless)
Appendix A, Table 2: Parameters mainly relevant to low-cost locations
Appendix A 169
REPLACEMENT Factor for how many of HC quit who will be

IN HC VS. LC replaced in HC. When a person leaves a high-cost
(Dimensionless) location, the job is either given to a new-hired
employee, or (when replacement rate lower than 1)
the job is given to an employee who has transferred
his or her job to a low-cost location.
The replacement rate is only below 1 for a limited

period of time (to finance the initial build up in low-
cost locations).
ADDITIONAL Factor for the ramp-up of productive LC employees

GROWTH (fixed for the first period, then gradually decreasing
(Dimensionless) to zero after the 36 th month)
HAND-OVER FRACTION The fraction of the time a newly hired LC Rookie

OF LC ROOKIE TIME spends on hand-over tasks (as opposed to normal
(Dimensionless) rookie on-the-job training). Example: if the
parameter is 1/3, and the total LC Rookie time is 6
months, an LC New Hire will after finished
classroom training spend 2 months as Hand-Over-
Rookie (with zero productivity), and 4 months as
“normal rookie” (with the normal LC ROOKIE
PRODUCTIVITY).
LC HANDOVER TRAVEL Travel costs related to hand-over, i.e. only effect

COST hand-over fraction of Rookies.
HC CAPACITY USE ON 1-on-1 hand-over will result in XX% reduction of

LC TRAINING the HC employee’s productive time
(Dimensionless) (LC employee will have 0 productivity)
Appendix A, Table 3: Parameters mainly relevant for transfer of tasks and
build-up of employees in low-cost locations
Appendix B: Model Equations
The two following tables give an overview of the significant equations in the
model. The first table describes the main equations for the rates influencing
stock levels, and the next table describes the main equations for calculation
of the monthly production, and well as monthly costs. After the two tables, a
complete list of model equations is to be found.
Main equations for the rates influencing stock levels
HC hire = HC quits * REPLACEMENT IN HC VS. LC
LC replacement hire = LC quit

Comment: This hiring only compensates for the employee turn-over
LC new hire =
+ Productive FTE LC*ADDITIONAL GROWTH
+ LC replacing HC quit
Comment:
LC replacing HC quit = HC quit – HC hire
The rates between stages are calculated as delay-functions of the inflow-rates.

For the training period, a high-order delay was used to imitate a pipeline delay, as
this is a fixed period of time for each employee. For the period as a Rookie, a
lower order delay was used, to reflect the variability in the learning curve for
individuals
Appendix B, Table 1: Main equations influencing stock levels

172 Appendix B
Main equations for production per month and cost per month
Handover capacity reduction =

H-O-R FTE LC * HC capacity use on hand-over
Comment:
Each H-O-R (Hand-Over-Rookie) in a low-cost location will need physically
to sit together with the person from whom he or she is taking over tasks. For
this reason, the experienced person in the high-cost location will have
reduced capacity (an average stipulated in the parameter: HC capacity use on
hand-over)
Productivity per month =
+ Productive FTE HC
+ HC ROOKIE PRODUCTIVITY*Rookie FTE HC
- handover capacity reduction
+ Productive FTE LC*(1-LC PRODUCTIVITY REDUCTION)
+ LC ROOKIE PRODUCTIVITY*Rookie FTE LC
Comments:
1) Unit of productivity is person/month (the case company often uses man-
year, man-month or man-days as unit for projects or production)
2) Number of productive days in LC higher than in HC. This is not included
in the model, but “equals out” with coordination overhead
Cost per month =
+HC person cost*total FTE HC
+New hired FTE HC*HC TRAINING COST
+(LC person cost+ LC ONGOING TRAVEL COST)*total FTE LC
+New hired FTE LC*LC TRAINING COST
+handover travel costs
Comment:
HC person costs and LC person costs develop over time with the person cost
increase rate. In Vensim this is can be managed by treating person costs as
stocks:
input rate = PERSON COST INCREASE RATE*LC person cost
(an initialized with the initial values of the person costs)
Appendix B, Table 2: Main equations influencing production per month
Appendix B 173
Starting below is a complete list of the model equations from the case study
model. According to the agreement with the case study company, the most
confidential numbers (such as average employee costs etc.) have been made
unrecognizable, through having ‘NN’ replacing the first digits. A number of the
parameters do not relate directly to the model, but serve the purpose of producing
nice output graphs.
List of equations (generated in Vensim):
Formulas Dimension
ADDITIONAL GROWTH=
(0.03*PULSE(0, 61)
+RAMP(-0.03/36, 12, 48)) Dmnl
LC new hire=
+(Productive FTE LC*ADDITIONAL GROWTH)
+LC replacing HC quit
+"HC micro-site optim."
+"LC micro- site optim." Person/Month
INI LC AVERAGE PERSON COST=

NN000/12
EUR/(Person*Month)
LC PERSON COST INCREASE RATE=

+(LN(1.15)/12)*PULSE(0, 24)
+(LN(1.1)/12)*PULSE(24, 36) 1/Month
LC person cost= INTEG (

+LC PERSON COST INCREASE RATE*LC person cost,
INI LC AVERAGE PERSON COST) EUR/(Month*Person)
174 Appendix B
"HC job-trained"=
DELAY N(HC trained, HC ROOKIE TIME,
Rookie FTE HC/HC ROOKIE TIME, 3) Person/Month
Productive FTE HC= INTEG (

+"HC job-trained"-HC quit-"HC micro-site optim.",
INI TOTAL FTE HC/
(1+(HC ROOKIE TIME+HC TRAINING TIME)*HC QUIT FRACTION))
Person
"LC repl. hire"=

LC quit Person/Month
LC 2a=
DELAY N(
("LC repl. hire"),
LC TRAINING TIME,
(New hired FTE LC/LC TRAINING TIME)
, 12) Person/Month
LC 1a=
DELAY N(
LC new hire,
LC TRAINING TIME,
0, 12) Person/Month
New hired FTE LC= INTEG (

+"LC repl. hire"+LC new hire
-LC 2a-LC 1a,
LC TRAINING TIME/
(LC TRAINING TIME+LC TOTAL ROOKIE TIME)
*INI TOTAL FTE LC/
(1+(1/((LC TRAINING TIME+LC TOTAL ROOKIE TIME)
*LC QUIT FRACTION)))) Person
cost per month=

+HC person cost*total FTE HC
+New hired FTE HC*HC TRAINING COST
Appendix B 175
+(LC person cost+LC ONGOING TRAVEL COST)*total FTE LC

+New hired FTE LC*LC TRAINING COST
+handover travel costs EUR/Month
Rookie FTE LC= INTEG (

+LC 2a
+LC 1b
-LC 2b-LC 1c,
LC TOTAL ROOKIE TIME/
(LC TRAINING TIME+LC TOTAL ROOKIE TIME)
*INI TOTAL FTE LC/
(1+(1/((LC TRAINING TIME+LC TOTAL ROOKIE TIME)
*LC QUIT FRACTION)))) Person
LC 2b=
DELAY N(LC 2a, LC TOTAL ROOKIE TIME,
Rookie FTE LC/LC TOTAL ROOKIE TIME, 3) Person/Month
Productive FTE LC= INTEG (

+LC 2b+LC 1c
-"LC micro- site optim."-LC quit,
INI TOTAL FTE LC/
(1+(LC TOTAL ROOKIE TIME+LC TRAINING TIME)*LC QUIT
FRACTION)) Person
"H-O-R FTE LC"= INTEG (

LC 1a-LC 1b, 0) Person
LC 1b=
DELAY N(LC 1a, LC TOTAL ROOKIE TIME*
"HAND-OVER FRAKTION OF LC ROOKIE TIME", 0, 12) Person/Month
LC 1c=
DELAY N(LC 1b, LC TOTAL ROOKIE TIME*
(1-"HAND-OVER FRAKTION OF LC ROOKIE TIME"), 0, 3) Person/Month
LC QUIT FRACTION=
0.07/12 1/Month
176 Appendix B
handover capacity reduction=

"H-O-R FTE LC"*"HC CAPACITY USE ON HAND-OVER" Person
handover travel costs=

"H-O-R FTE LC"*"LC HAND-OVER TRAVEL COST" EUR/Month
total FTE LC=

New hired FTE LC+Productive FTE LC
+Rookie FTE LC+"H-O-R FTE LC" Person
LC TOTAL ROOKIE TIME=

6 Month
production per month=

(+Productive FTE HC
+HC ROOKIE PRODUCTIVITY*Rookie FTE HC
- handover capacity reduction
+((1-LC PRODUCTIVITY REDUCTION)*Productive FTE LC)
+LC ROOKIE PRODUCTIVITY*Rookie FTE LC)
*make unit per month Person/Month
"HAND-OVER FRAKTION OF LC ROOKIE TIME"=

0.333 Dmnl
output old year cost=

+IF THEN ELSE(Time=4, Nice Graph Year Cost/TIME STEP, 0)
EUR/Month
Appendix B 177
input year cost=

+IF THEN ELSE(Time=0, cost per month*12*STY month/TIME STEP, 0)
+IF THEN ELSE(Time=12, Year cost/TIME STEP, 0)
+IF THEN ELSE(Time=48, Year cost/TIME STEP, 0) EUR/Month
Nice Graph Year Cost= INTEG (

input year cost-output old year cost, 0) EUR
adding=
production per month*1/12 Person/Month
summing=
cost per month EUR/Month
LC FTE fraction of total FTE=

total FTE LC/total FTE Dmnl
LC replacing HC quit=
HC quit-HC hire Person/Month
LC quit=
Productive FTE LC*LC QUIT FRACTION Person/Month
"HC CAPACITY USE ON HAND-OVER"=

0.5 Dmnl
LC TRAINING COST=
NN00 EUR/(Person*Month)
"LC micro- site optim."=

(1-HC FRACTION OF MICROSITES)*"MICRO-SITE OPTIMIZATION"
Person/Month
LC PRODUCTIVITY REDUCTION=
0.2 +0*0.2*PULSE(0, 48) Dmnl
178 Appendix B
LC ONGOING TRAVEL COST=

"LC HAND-OVER TRAVEL COST"=

cost per production=

cost per month/production per month EUR/Person
total FTE=
total FTE HC+total FTE LC Person
HC quit=
Productive FTE HC*HC QUIT FRACTION Person/Month
HC hire=
HC quit*"REPLACEMENT IN HC VS. LC" Person/Month
"MICRO-SITE OPTIMIZATION"=
0*(NN*PULSE(1, 1)+NN*PULSE(12, 1)
+NN*PULSE(24, 1)) Person/Month
HC FRACTION OF MICROSITES=
0.5 Dmnl
"HC micro-site optim."=

"MICRO-SITE OPTIMIZATION"
*HC FRACTION OF MICROSITES Person/Month
STY month=
1 Month
HC person cost= INTEG (

+HC PERSON COST INCREASE RATE*HC person cost,
INI HC AVERAGE PERSON COSTS) EUR/(Month*Person)
INDEX production per month=

(production per month*STY month)/startindex production per month Dmnl
Appendix B 179
INI HC AVERAGE PERSON COSTS=

NN0000/12 EUR/(Month*Person)
HC PERSON COST INCREASE RATE=

LN(1.025)/12 1/Month
INDEX cost per month=

(cost per month*STY month)/startindex cost per month Dmnl
startindex production per month= INTEG (

production per month-production per month,
production per month*STY month) Person
INDEX cost per production=

INDEX cost per month/INDEX production per month Dmnl
startindex cost per month= INTEG (

cost per month-cost per month,
cost per month*STY month) EUR
INI TOTAL FTE HC=

NN20 Person
Year production= INTEG (

adding-reset production every year, 0) Person
make unit per month=

1 1/Month
reset production every year=

+IF THEN ELSE(Time=12, Year production/TIME STEP, 0)
+IF THEN ELSE(Time=48, Year production/TIME STEP, 0) Person/Month
Year cost= INTEG (

+summing-reset cost every year, 0) EUR
180 Appendix B
reset cost every year=

+IF THEN ELSE(Time=48, Year cost/TIME STEP, 0) EUR/Month
HC trained=
DELAY N(HC hire, HC TRAINING TIME,
New hired FTE HC/HC TRAINING TIME, 12) Person/Month
LC ROOKIE PRODUCTIVITY=
0.5 Dmnl
HC TRAINING COS
LC TRAINING TIME=
2 Month
New hired FTE HC= INTEG (

HC hire-HC trained,
HC TRAINING TIME/(HC TRAINING TIME+HC ROOKIE TIME)
*INI TOTAL FTE HC
/(1+(1/((HC TRAINING TIME+HC ROOKIE TIME)
*HC QUIT FRACTION)))) Person
HC ROOKIE PRODUCTIVITY=
0.5 Dmnl
HC TRAINING TIME=
3 Month
"REPLACEMENT IN HC VS. LC"=

0.4*(PULSE(0, 12)
+PULSE(12, 12)
+PULSE(24, 12))
+1*(PULSE(36, 12)
+PULSE(48, 12)) Dmnl
Appendix B 181
total FTE HC=

Rookie FTE HC+Productive FTE HC+New hired FTE HC Person
HC ROOKIE TIME=
6 Month
Rookie FTE HC= INTEG (

HC trained-"HC job-trained",
HC ROOKIE TIME/(HC TRAINING TIME+HC ROOKIE TIME)
*INI TOTAL FTE HC
/(1+(1/((HC TRAINING TIME+HC ROOKIE TIME)
*HC QUIT FRACTION)))) Person
HC QUIT FRACTION=
0.0NN/12 1/Month
INI TOTAL FTE LC=

NN0 Person
FINAL TIME = 60 Month

~ The final time for the simulation.
INITIAL TIME = 0 Month

~ The initial time for the simulation.
SAVEPER = TIME STEP Month

~ The frequency with which output is stored.
TIME STEP = 0.25 Month

~ The time step for the simulation.
Appendix C: Equations for Stock Initializations
The stocks in the model are initialized based on the condition for equilibrium,
where the total number of hirings in both low cost locations and high cost
locations equals the number of employees leaving.
For both aging chains, the three stocks to be initialized are: Productive FTE, New
hire FTE, and Rookie FTE.
The model input parameters used to calculate the initial values are:
INI total FTE, QUIT FRACTION, TRAINING TIME, and ROOKIE TIME.
Three equilibrium equations are:

(1): INI total FTE = Productive FTE + New hire FTE + Rookie FTE
(2): New hire FTE+ Rookie FTE =

Productive FTE * QUIT FRACTION * (TRAINING TIME+ROOKIE TIME)
(3): New Hire FTE =

(TRAINING TIME/(TRAINING TIME+ROOKIE TIME) )
*(New hire FTE+ Rookie FTE)
resulting in the following stock initializations (see workings on next page):

• Productive FTE (ini) = INI TOTAL FTE
/(1+(ROOKIE TIME+TRAINING TIME) *QUIT FRACTION)
• New hire FTE (ini) =
(TRAINING TIME / (TRAINING TIME+TOTAL ROOKIE TIME))
* INI TOTAL FTE
/(1+(1/((TRAINING TIME+TOTAL ROOKIE TIME)*QUIT FRACTION)))
• Rookie FTE (ini) =

(ROOKIE TIME/(TRAINING TIME+ROOKIE TIME))
*INI TOTAL FTE
/(1+(1/((TRAINING TIME+ROOKIE TIME)*QUIT FRACTION)))
184
Initializing of the stocks in equilibrium (both LC and HC) :
The three equilibrium equations:

(1): INI total FTE = Productive FTE + New hire FTE + Rookie FTE
(2): Productive FTE * QUIT FRACTION * (TRAINING TIME+ROOKIE TIME) = New hire FTE+ Rookie FTE
(3): New Hire FTE = (TRAINING TIME/(TRAINING TIME+ROOKIE TIME) )*(New hire FTE+ Rookie FTE)
gives when New hire FTE + Rookie FTE from equation (2) is entered into equation (1):
<=> INI total FTE = Productive FTE + Productive FTE * QUIT FRACTION *(TRAINING TIME+ROOKIE TIME)
<=> Productive FTE = INI total FTE / (1+(TRAINING TIME+ROOKIE TIME)* QUIT FRACTION
and also, when the result of Productive FTE from equation (2) is entered to equation (1):
INI total FTE = ((New hire FTE+ Rookie FTE) / QUIT FRACTION * (TRAINING TIME+ROOKIE TIME))+New hire FTE + Rookie
FTE <=> INI total FTE = (New hire FTE + Rookie FTE) * (1+(1/ QUIT FRACTION * (TRAINING TIME+ROOKIE TIME) )
<=> (New hire FTE + Rookie FTE) = INI total FTE / (1+(1/ QUIT FRACTION * (TRAINING TIME+ROOKIE TIME) )
And the initial values of each of New hire FTE and Rookie FTE are then found by the equation (3):
New hire FTE = LC TRAINING TIME/(LC TRAINING TIME+LC ROOKIE TIME)**INI TOTAL FTE LC/(1+(1/((LC TRAINING
TIME+LC ROOKIE TIME)*LC QUIT FRACTION)))
(and furthermore is the stock of LC H-O-R Rookies initialized with 0

Appendix C
INI TOTAL FTE LC
INI TOTAL FTE HC

INI HC AVERAGE LC ROOKIE
PERSON COSTS PRODUCTION
FRACTION HC ROOKIE
production per month PRODUCTION
ADDITIONAL FRACTION
<HC micro-site
GROWTH
optim.> REPLACEMENT
FACTOR HC
LC TRAINING HC QUIT HC TRAINING
TIME LC ROOKIE LC QUIT <HC quit>
<LC quit FRACTION HC ROOKIE TIME
TIME FRACTION TIME
&
siteopt.> HC quit
New hired Rookie Productive Productive Rookie New hired
Appendix D: Preliminary Model
LC FTE LC LC FTE LC LC FTE LC LC quit FTE HC HC FTE HC HC FTE HC HC

hire trained job-trained & siteopt. HC micro- job-trained trained hire
MICRO -SITEsite optim.
OPTIMIZATION
total FTE LC LC FRACTION OF
MICROSITES
total FTE HC
HC PERSON COST
LC PERSON DEVELOPMENT
<HC hire> TRANSFER COST RATE
<HC quit> FACTOR HC person cost
LC TRAINING HC TRAINING
COST
LC HIRE COST cost per month
COST HC HIRE
COST
Appendix D, Figure 1: The preliminary model in the case study

Appendix E: Model without ‘Rate-on-Rate’ Modeling
The next page (Appendix E, Figure 2) is an adjusted version of the model,
where the out-flow rates from the stocks are modeled as simple fractions of
the level-values. This way, rate-on-rate modeling is avoided, which is often
recommended in the literature. Some of the stocks had to be split up in two
parts, in order to use this approach. Figure 1 shows a simulation run with the
same parameter setting as used in the simulation runs in Chapter C. It should
be noted that the main trends – and thereby the main model insights – are the
same as in the original model, also for year 1, even though the adjusted
model results in the new hiring policy influencing the stock of experienced
employees from nearly the very beginning.
Index Cost per Production

1.2
1.15
1
1
1.1 1
1
3 2
1 3 2 4
1.05 3 3 3 4
1 2 2 2 2
32 4 4 4
4 4
41
1
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60
Months
INI 1 1 1 1 1 Dmnl
Base run 2 2 2 2 2 Dmnl
40% HC replacement 3 3 3 3 Dmnl
Faster training and hand-over 4 4 4 4 Dmnl
Appendix E, Figure 1: Simulation run of adjusted model (avoiding rate-on-rate)

188
handover capacity INI TOTAL FTE LC

reduction INI TOTAL FTE HC
<HC hire> <HC quit> HC CAPACITY USE INI HC AVERAGE
ON HAND-OVER PERSON COSTS
LC PRODUCTIVITY
INI LC AVERAGE
REDUCTION
PERSON COST
LC ROOKIE
non-replaced HC quit
PRODUCTIVITY HC ROOKIE
production per month
PRODUCTIVITY REPLACEMENT
ADDITIONAL IN HC VS. LC
GROWTH
LC QUIT HC QUIT HC ROOKIE HC TRAINING
FRACTION <HC
FRACTION TIME TIME quit>
LC add. hire LC A1 LC A2
New hired H-O-R Rookie LC A3 LC HC
FTE LC quit quit
FTE LC A A FTE LC A Productive Productive Rookie New hired
New hired Rookie FTE LC FTE HC HC FTE HC HC FTE HC HC
FTE LC B LC B1 FTE LC BLC B2 job-trained trained hire
LC repl. hire
<LC quit> LC ONGOING total FTE HC

total FTE LC TRAVEL COST
(arrows hidden) HC PERSON COST
LC PERSON COST
LC TRAINING TIME INCREASE RATE
INCREASE RATE
LC TRAINING LC person cost HC person cost HC TRAINING

LC TOTAL ROOKIE TIME COST COST
HAND-OVER FRAKTION cost per month
LC flow input
OF LC ROOKIE TIME
LC HAND-OVER TRAVEL COST
handover
<H-O-R FTE LC A> travel costs
Appendix E, Figure 2: Model without rate-on-rate modeling

Appendix E
Appendix F: Facilitator Observations
and Key Quotes From Interviews
The observations and interview quotes are structured according to the
evaluation framework described in chapter C.III. When nothing is indicated,
observations and interview quotes were made during the project, i.e. fourth
quarter 2004.
Personal reactions to the modeling process:

Very early in the modeling process, an engaged and vital Observation
discussion started, showing a positive attitude in the
sessions. A few persons were resistant to the process, due
to disagreement with the intervention objectives. It is an
interesting point, however, that not even the core-group
person disagreeing with the objectives of the process
could “resist the fun of modeling” when he took part in
modeling sessions, as he and the other core members
were very mathematically skilled and interested
individuals.
“A few participants did not agree with the business Interview

objectives, and did therefore never really buy in”
Gain of learning, and changes in goal structures and

mental models:
In the development of the preliminary model, the very
first results already took form as the project owners Observation
gained some interesting insights. Something first
considered as a potential mistake in the model turned out
to be an important insight, and it became clear that one
decision, that had just been made, had a stronger negative
impact in year 1 than anticipated, and it was therefore
decided to modify the decision, and make the transition
over a longer time-span.
Through the discussions and model simulations in the Observation

modeling workshops, the core project team gained
190 Appendix F
insights and exchanged experience relating to the

location strategy. Also, the model was a framework for
the setting of parameters to be used in the business case
in each of the business areas.
Through investigations of effects of the changes in the Observation

different parameters of the model, the core team
identified effective optimization opportunities, as well as
sensitivity risks. Some of the most important insights
gained were the understanding of the ‘reinforcing growth
loop’ motivating the intervention, of how relatively few
non-replacements in high-cost countries could
compensate for the costs of building up the required
volume of R&D employees in low-cost countries, as well
as distinct benefits of reducing training and hand-over-
time compared to reducing costs of training and
traveling.
Often, simulations served as an “eye-opener”. In a few Observation

cases a parameter was perceived as “not possible to
reduce”, but through simulations with increased value, a
strong impact was seen, and the individuals then opened
up for discussion on what it would take to optimize a
certain parameter, e.g. reduce hand-over time.
The initial setting of each of the most important Observation

parameters could be discussed for hours in both
workshops and other related meetings. For example, it
was a widely accepted “fact” among many of the project
participants, that employees in low-cost countries often
stayed for only 1-2 years, because as soon as they
attained experience in R&D, they could get a higher
paying job in a high-cost country. Through the parameter
stipulation, facts came on the table, documenting a very
low employee turnover in the low-cost countries. (This
could be viewed in the context of Ackoff’s morale:
“There is nothing so deceptive as an apparent truth”).
Appendix F 191
In the modeling sessions, especially in the parameter Observation

stipulation, knowledge and experience exchange took
place across the business units. Especially one business
unit had already high-scale experience with build-up of
resources in low-cost countries. Using this modeling
approach served as a forum for the transfer of best
practices.
“The preliminary model was important to get Interview

confirmation on the feasibility of the objectives, and the
preliminary model also gave a better understanding of the
dynamics of the problem”
“Parameter discussions were effective in challenging Interview

assumptions”
“Simulations were strong in showing the importance of Interview

the different parameters”
“The exchange of Best Practices was one of the Interview

objectives for starting a cross-business unit process in the
first place”
Commitment to the outcome of the modeling sessions:

Modeling participants often argued supporting the Observation
insights gained in the modeling, when presenting the
results in other meetings – but there were also a few
examples, when this was not the case, primarily in
situations with divergence between insights and
personal interests.
Changes in behavior:
“In general the team members developed business cases Interview
in compliance with the modeling insights and results,
with only one exception”
192 Appendix F
Group communication:
The discussion seemed to be both very structured and Observation
very open and frank. The result-oriented process,
however, did not leave time to go into depth in all of the
relevant discussions, but due to the structure, most of the
time invested by the participants in discussions was used
very effectively.
A couple of times, the modeling helped to take focus Observation

away from discussions, when simulations proved the low
importance of a parameter. Therefore, there was little
relevance of the continuous discussion about the exact
stipulation of the value.
“The discussion improved radically compared to the Interview

rather unstructured communication we had in the project,
before we decided to use system dynamics. The model
directed the discussions back to the core of the problem”
Communication and consensus:

Opinions on parameters were often very different within Observation
the core project team, and the model proved to function
as a structure for fact-finding and alignment of
perceptions.
“The approach makes it difficult for people to play Interview

politics”
Consensus:
Opinions on the importance of different causal- Observation
relationships differed initially, but through the model-
building process a more shared understanding of the
problem and its dynamics was created.
The discussion of the parameters often initiated longer Observation

discussions on how the strategy could and should be
Appendix F 193
executed, as the parameter setting reflected

implementation decisions; e.g. the logistic and cost
model of traveling, how to structure knowledge transfer,
etc.
Common language:
Within the core project team, there was a tendency to Observation
increased alignment, but this was difficult to transfer to
non-core members in the relatively short meetings with
these people. Parts of the language did spread to some
extent, such as “one employee is one employee”
regardless of the type of location. The factor for reduced
productivity in low-cost countries only reflects a lower
average experience-level. A stronger outcome on this
dimension would have required a less result-oriented
process, with more time to in-depth discussion.
“Even more of an effect – especially outside the core Interview

team – would have been better”
Transfer of insights:
The model clearly confirmed some viewpoints that the Observation
project team wanted to communicate to the board and the
corporate controlling. Whereas these insights did not
have much “newness” value, it was very valuable to have
a model that distinctly and clearly “proved” the matter.
These type of insight included the worse-before-better
effect, implying that the division even receiving a
relatively large number of additional head-counts in year
1, would have no additional productivity, but rather a
slightly reduced productivity. Also, the model showed
very clearly that even the relatively large growth in the
fraction of low-cost employees compared to the total
number of employees, does not result in a decreased cost
per produced development hour, as the inflation has
stronger influence than the benefits to be realized through
a location strategy of the discussed scope.
194 Appendix F
Running a few simulations appeared to be very Observation

convincing in the discussions with non-core stakeholders.
“the model made the strategy very transparent, with clear Interview
definitions – and was better than words for
communication”
System changes:
The board accepted the business cases developed in the Observation
project, and the hirings for 2005 went approx. as
planned. Also, the business case was implemented in the
3 years business plan, and the execution should follow
the plans.
The business unit that disagreed most with the project Observation
objectives has actually been “overperforming” year-end
2005 with regards to hirings in low-cost locations.
“The business cases are approved by the board, and Interview

incorporated in the budgets and business plans”
Results:
1½ years after the approval of the detailed Observation
implementation plans, the FTE ratio between low-cost (Q2 2006)
and high-cost locations has increased beyond the plan.
This must also be seen in relation to the fact, that due to
the success of the company, more hiring were needed and
approved than originally planned.
“The modeling clearly helped in getting the managers Interview

committed to the strategy to use the current success to (Q2 2006)
build up capacity in low-cost locations, seeking to avoid
having to lay off employees in less fortunate periods.”
Appendix F 195
Efficiency:
The project kept established deadlines. Some disturbance Observation
and discussion took place due to the fact, that the
intervention also encompassed many elements not
included in the modeling.
Especially in the beginning of the project, there was a Observation

tendency among core team members to think of the
modeling as an additional task, increasing the workload
in an already stressed period of time. But on the other
hand, the model helped both to structure and to facilitate
the discussions, which are likely to have reduced the
overall time spent by the core team. To obtain this
efficiency, however, a lot of efforts were made in
workshop and meeting preparation by the facilitator and
project owners were made.
Overall, it seemed very efficient to use the chosen Observation

software to make a shared model on a high abstraction
level, with easy simulation options. Some improvement
in the software, however, would be welcomed, as quite a
lot of “behind the scenes” work was needed in order to
create nice and effective output-graphs in separate views,
avoiding waisting time and thereby creating irritation
towards the modeling efforts.
Also, even relatively small changes in structures could be
very time-consuming to implement in the chosen
software (Vensim).
“It was a very structured and effective process” Interview
“The project progressed even better than planned due to Interview

discipline and focus in the process”
“Maybe even too efficient: more difficult to act Interview

politically in the budget-negotiation” (said with a smile)
196 Appendix F
Quality in results:
For SD practitioners, the model seems very simple, but it Observation
should be recognized that the project team first tried to
handle the problem using a normal Excel-model, which
became a complicated “black box”, where it was difficult
to see and understand how the different parameters
influenced the model.
Further use of SD:

One of the project owners has been partly involved in a Observation
later project using system dynamics. This later modeling (Q2 2006)
project was initiated in another part of the organization,
and due to the positive experiences in the location
strategy project, the project owner positively supported
the idea in his new role as project sponsor.
Intervention driven by business objectives and targets:

The intervention was initiated with clear objectives and Observation
targets (directions from the board). Only a modeling
process supporting this type of intervention was
considered by the project owners. No participants
questioned this circumstance.
“Most of our strategic projects are initiated with very Interview

clear business objectives and targets”
Project framing:
The project owners had no intentions of starting a group Observation
model building process from clean sheets of paper with
the risk of losing control. This might be a general trend in
the corporate environment; that executives have a clear
view of the direction they want to drive a given change,
and that they will not take the risk that a model could
show contradicting results, which in their view could be
Appendix F 197
due to hidden errors in the model or the problem being

addressed or conceptualized erroneously. Trust in the
modeling process was gained through the preliminary
model.
Compared to exploratory modeling, the targeted Observation

participative modeling approach restricts the problem-
solving process (with regards to “what to do,” not in
“how to do”). It is difficult to say whether this had
negative impacts on the participants’ ownership and trust
in the model. The questionnaires do not explicitly include
questions regarding this possible impact of a preliminary
model, due to the problem of measurements influencing
the system (in this case creating negative attitudes).
The preliminary model confirmed some intuitive

expectations of the project owners, and showed to be an Observation
effective mean of communicating these cause-effect
relationships, which was a cornerstone in continuing the
modeling efforts.
“A few participants did not agree with the business

objectives and for that reason also not with the process, Interview
but nevertheless the process forced them in the decided
direction, and through the modeling they gained some of
the insights motivating the intervention in the first place”
“We were open about the premises for the process, and Interview
participants should therefore not feel in the slightest way
manipulated” (This was the answer to a question, if the
use of a preliminary model and fixed business objectives
could have caused the participants to feel somewhat
manipulated)
“Initially I was a bit skeptical, but along the process I Interview

started to trust the model”
198 Appendix F
“We got were we wanted to” Interview
“The modeling helped changing the focus from ‘seeing Interview

only problems’ to discussing sustainable and fair
execution”
Context comparative conditions:
The problem was more politically sensitive than truly Observation

messy. There were clearly defined objectives and targets.
The case company has a strong tradition for employee Observation

empowerment and is a relatively un-hierarchical
organization.
Attitude to intervention: it was a top-down decision to Observation

initiate the intervention, initially against the “true wish”
of many of the participants, although most of them could
agree with the rationale behind the intervention.
There was a technical environment with young and Observation

highly educated people with a tradition of mathematical
and “rational” problem solving. All participants were
perceived as high-performers and have been with the
company for years.
Mechanism comparative conditions:
A preliminary quantitative model was used to investigate Observation

whether a model was appropriate to illustrate the change
objectives. The preliminary model showed what main
learning to anticipate.
The modeling process was focused on developing a Observation

relatively simple model that could illustrate an idea of the
overall behavior of the problem-system without including
Appendix F 199
too many details, partly because overview was

considered more important than detailed correctness
(avoiding black-box effect), partly due to the fact that the
system dynamics Vensim model was complemented by a
more detailed excel-model with the format needed in the
budgeting and business planning. The result was a model
that was relatively easy to explain in even 1-2 hour
meetings.
The observer (and facilitator) primarily had a theoretical 2nd order

foundation for SD modeling, with only little SD observation
modeling experience, but has more than 10 years of
planned organizational intervention experience, including
other types of modeling.
The observer (and facilitator) had personal relationships 2nd order

with company executives, which could be expected to observation
influence both ‘positively’ with regards to access to
information and dialogues with the decision-makers, and
‘negatively’ with regards to creating biases. Although
being a highly subjective observation, the observer had
the impression that most of the participants were rather
indifferent to the existence of personal relationships,
which might be due to the fact that most of the
participants were high-placed managers themselves with
a high degree of self-confidence.
Bibliography
Ackoff, Russel L. : The Art of Problem Solving – Accompanied by Ackoff’s Fables,

New York, (John Wiley & Sons) 1978.
Ajzen, Icek: Attitudes, Personality and Behavior, Chicago, (The Dorsey Press) 1988.
Ajzen, Icek and Martin Fishbein: “The Prediction of Behavior from Attitudinal and
Normative Variables”, Journal of Experimental Social Psychology, No. 6, 1970,
pp. 466—487.
Akkermans, Henk: Modelling With Managers, Breda, The Netherlands, 1995.
No. 2, 1997, pp. 187—201.
Andersen, David F and George P. Richardson: “Scripts for group model building”,
System Dynamics Review, Vol. 13, No. 2, 1997, pp. 107—129.
Anderson, Linda A. and Dean Anderson : “Awake at the Wheel: Moving beyond
Vol. 33, No. 3, 2001, pp. 40—48.
Argyris, Chris: Interventions Theory and Method – A Behavioural Science View,

Reading, Massachusetts, (Addison-Wesley Publishing Company) 1970.
Argyris, Chris: Reasoning, Learning, and Action – Individual and Organizational,

San Francisco, (Jossey-Bass Publishers) 1982.
Argyris, Chris: On Organisational Learning, 2 nd edition, Oxford, (Blackwell

Publishers) 1999.
202 Bibliography
Argyris, Chris and Donald Schön : “ Organizational Learning: A theory of Action

Perspective” (first printed as part of book with the same title in 1978) , reprint in
Gareth Morgan: Creative Organization Theory, Newbury Park, California,
(SAGE Publications) 1989, pp. 139—142.
Ashby, W. Ross: An Introduction to Cybernetics, paperback version, London,

(Methuen & Co. Ltd.) 1964.

Environments, Boston, (MIT dissertation) 1993.
Barlas, Yaman : “Formal aspects of model validity and validation in system

dynamics”, System Dynamics Review, Vol. 12, No. 3, 1996, pp. 183—210.
Baron, Jonathan : Thinking and Deciding, 3 rd edition, Cambridge, (Cambridge

University Press) 2000.
Beer, Stafford : Beyond Dispute – The Invention of Team Syntegrity, Chichester,

(John Wiley & Sons) 1994.
Bennis, Warren G., Kenneth D. Benne and Robert Chin : The Planning of Change,
4 th edition, New York, (CBS College Publishing) 1985.
Berger, Ulrike and Isolde Bernhard-Mehlich : “Die Verhaltenswissenschaftliche

Entscheidungstheorie”, in Alfred Kieser (ed.): Organisationstheorien, 3 rd edition,
Stuttgart, (Kohlhammer) 1999, pp. 133—168.
Bonabeau, Eric: “Don’t Trust Your Guts”, Harvard Business Review, May 2003,
pp. 116—123.
Borum, Finn : Strategier for organisationsændringer, Copenhagen, (Handelshøj-

skolens Forlag) 1995.
Brown, John Seely: Research That Reinvents the Corporation, Harvard Business
Review, August 2002, pp. 105—114.
Briggs, Robert, Gert-Jan de Vreede, and Jay F. Nunamaker Jr .: “Collaboration

Engineering with ThinkLets to Pursue Sustained Success with Group Systems”,
Journal of Management Information Systems, Vol. 19, No. 4, Spring 2003,
pp. 31—63.
Bibliography 203
Bungard, Walter and Catrin Niethammer: “Psychologische Aspekte des Change

Management im interorganisationalen Kontext”, in Walter Bungard, Jürgen
Fleischer, Holger Nohr, Dieter Spath and Erich Zahn (eds.), Customer
Knowledge Management, Stuttgart, 2003, pp. 105—117.
Bungard, Walter: “Zur Implementierungsproblematik bei Business-Reengineering

Projekten”, in Manfred Perlitz, Andreas Offinger, Michael Reinhardt and Klaus
Schug (eds.): Reengineering zwischen Anspruch und Wirklichkeit, Wiesbaden,
(Gabler Verlag) 1996, pp. 253—273.
Checkland, Peter: “Soft System Methodology”, in Jonathan Rosenhead (ed.):

Rational Analysis for a Problematic World, Chichester, (John Wiley & Sons)
1989, pp. 21—42.
Checkland, Peter: “Systems Thinking, Systems Practice”, Chichester, (John Wiley

& Sons) 1993.
Chin, Robert and Kenneth D. Benne: “General Strategies for Effecting Changes in
Human Systems”, in Warren G. Bennis, Kenneth D. Benne and Robert Chin: The
Planning of Change, 4 th edition, New York, (CBS College Publishing) 1985,
pp. 22—45.
Christensen, Søren and Jan Molin : Organisationskulturer, Copenhagen, (Akademisk

Forlag) 1987.

and Change, Ohio, (South-Western College Publishing) 2001.
Conant, Roger C. and W. Ross Ashby: “Every Good Regulator of a System Must Be
a Model of that System”, International Journal of System Sciences, Vol. 1,
No. 2, 1970, pp. 89—97.
Davenport, Thomas H.: Process Innovation, Boston, (Harvard Business School

Press) 1993.
Denscombe, Martyn : The Good Research Guide for small-scale social research
projects, 2 nd edition, Philadelphia, (Open University Press) 2003.
De Bono, Edward : Lateral Thinking for Management, England, (American

Management Association) 1971.
204 Bibliography
de Geus, Arie P.: “Planning as Learning”, Harvard Business Review, Vol. 66,
No. 2, March-April 1988, pp. 70—74.
de Geus, Arie P.: The Living Company, Boston, (Longview Publishing Limited)
1997.
Distefano, Joseph J. and Martha L. Maznevski : “Creating Value with Diverse Teams
in Global Management”, Organizational Dynamics, Vol. 29, No. 1,
pp. 45—63.
Doppler, Klaus and Christoph Lauterburg : Change Management – Den

Unternehmenswandel gestalten, 10 th edition, Frankfurt/Main, (Campus Verlag)
2002.
Doyle, James K. and David N. Ford : “Mental models concepts for system dynamics
research”, System Dynamics Review, Vol. 14, No. 1, Spring 1998,
pp. 3—29.
Doyle, James K. and David N. Ford : “Mental models concepts revisited: some
Winter 1999, pp. 411—415.
Dörner, Dietrich : The Logic of Failure, New York, (Metropolitan Books) 1996.
Eco, Umberto : A Theory of Semiotics, Bloomington, (Indiana University Press)

1976.
Eden, Colin : “Using cognitive mapping for strategic options development and
analysis”, in Jonathan Rosenhead (ed.): Rational Analysis for a Problematic
World, Chichester, (John Wiley & Sons) 1989, pp. 21—42.
Eden, Colin : “Cognitive mapping and problem structuring for system dynamics
1994, pp. 257—276.
Eisenhardt, Kathleen M.: “Building Theories from Case Study Research”, Academy
of Management Review, Vol. 14, No. 4, 1989, pp. 532—550.
Eisenhardt, Kathleen M.: “Strategy as Strategic Decision Making”, Sloan

Management Review, Spring 1999, pp. 65—72.
Bibliography 205
Eskinasi, Martijn and Eppie Fokkema: “Bursting the myth of making easy modeling
money”, unpublished working paper (permission to quote granted from the
authors), Proceedings, Second European System Dynamics Workshop,
Nijmegen, 2005, pp. 62—66.
Farson, Richard and Ralph Keyes: “The Failure-Tolerant Leader”, Harvard Business
Review, August 2002, pp. 64—71.
Ferres, Natalie, Julia Connell, and Anthony Travaglione: “The effect of future
redeployment on organizational trust”, Strategic Change, Vol. 14, March-April
2005, pp. 77—91.
Fey, Willard and John Trimble: “The Evaluation and Development of Knowledge
Acquisition in System Dynamics Studies”, in Proceedings, System Dynamics
Conference, System Dynamics Society, 1992, pp. 173—182.
Fine, Charles H. : “Clockspeed-based strategies for Supply Chain Design”,

Production and Operation Management, Vol. 9, No. 3, 2000, pp. 213—221.
Flood, Robert L. and Michael C. Jackson : Creative Problem Solving: Total Systems
Intervention, Chichester, (John Wiley & Sons) 1991.
Flood, Robert L.: Solving Problem Solving, Chichester, (John Wiley & Sons) 1995.
Dynamics Models”, in Augusto A. Legasto, Jay W. Forrester und James M.
1980, pp. 209—228.
Forrester, Jay W.: Principles of Systems, Cambridge, (MIT Press) 1968.
Forrester, Jay W.: Industrial Dynamics, Cambridge, (MIT Press) 1961.

Papers of Jay W. Forrester – Foreword by Gordon S. Brown, Cambridge,
(Wright-Allen Press, inc.) 1975, pp. 211—244.
European Journal of Operational Research, Vol. 59, No. 1, 1992, pp. 42—63.
206 Bibliography
Gladwell, Malcolm: The Tipping Point, paperback edition, New York, (Back Bay
Books) 2002.
Gold, Martin : The Complete Social Scientist: A Kurt Lewin Reader, Washington,
(American Psychological Association) 1999.
Größler, Andreas: Entwicklungsprozess und Evaluation von Unternehmens-

simulation für lernende Unternehmen, Frankfurt am Main, (Peter Lang) 2000.
Größler, Andreas, Peter Milling and Graham Winch : “Perspectives on rationality in

system dynamics: a workshop report and open research questions”, System
Größler, Andreas: “Don’t Let History Repeat Itself: Methodological Issues

Concerning the Use of Simulators in Teaching and Experimentation”, System
Dynamics Review, Volume 20, Number 3, Fall 2004, pp. 263—274.
Größler, Andreas: “A Content and Process View on Bounded Rationality in System

Dynamics”, Systems Research and Behavioral Science, Vol. 21, No. 4,
July/August, 2004, pp. 319—330.
Hamel, Gary: Leading the Revolution, Boston, (Harvard Business School Press)
2000
Henttonen, Kaisa and Kirsimarja Blomqvist: “Managing distance in a global

virtual team: the evolution of trust through technology-mediated
relational communication”, Strategic Change, Vol. 14, March-April 2005,
pp. 107—119.
Hodgson, A. M.: “Hexagons for system thinking”, European Journal of Operational

Research, Vol. 59, No. 1, 1992, pp. 123—136.
Homer, Jack B. : “Why we iterate: scientific modeling in theory and practice”,

System Dynamics Review, Vol. 12, No. 1, Spring 1996, pp. 1—19.
Hogarth, Robin : Judgment and Choice – The Psychology of Decision, 2 nd edition,

Chicago, (John Wiley & Sons) 1987.
Bibliography 207
Huff, Anne S. and Rhonda Kay Reger: “A Review of Strategic Process Research”,
Journal of Management, Vol. 13, No. 2, 1987, pp. 211—236.
Huz, Steven, David F. Andersen, George P. Richardson and Roger Boothroyd : “A

framework for evaluating systems thinking interventions: an experimental
approach to mental health system change”, System Dynamics Review, Vol. 13,
No. 2, 1997, pp. 149—169.
Imai, Masaaki : Kaizen: Der Schlüssel zum Erfolg der Japaner im Wettbewerb,
german translation, München, (Wirtschaftsverlag Langen Müller Herbig) 1992.
Jacobsen, Chanoch and Yitzhak Samuel: “ Planned Organizational Change: Theory,

Model, Data and Simulation”, in Peter M. Milling and Erich O.K. Zahn (eds.):
Computer-Based Management of Complex Systems, Proceedings of the 1989
International Conference of the System Dynamics Society, Heidelberg,
(Springer-Verlag) 1989, pp. 104—118.
Janis, Irving L. : “Groupthink: The Problems of Conformity” (original printed in

Psychology Today, Nov. 1971, pp. 271—279), reprint in Gareth Morgan:
Creative Organization Theory, Newbury Park, California, (SAGE Publications)
1989, pp. 224—228.
Johnson-Laird, P. N.: Mental Models, Cambridge, (Cambridge University Press)

1983.
Jöns, Ingela : Managementstrategien und Organisationswandel, Mannheim, (Mann-

heim University Dissertation) 1995.
Kahnemann, Daniel and Amos Tversky: “On the Psychology of Prediction”,

Psychological Review, Vol. 80, No. 4, July 1973, pp. 237—251.
Kampmann, Christian P. E. : Feedback complexity and market adjustment – An

experimental approach, Boston, (MIT Dissertation) 1992.
Kanter, Rosabeth Moss: “Leadership and the Psychology of Turnarounds”, Harvard

Business Review, June 2003, pp. 58—67.
Keough, Mark and Andrew Doman : “The CEO as organization designer: An

interview with Professor Jay W. Forrester, the founder of system dynamics”, The
McKinsey Quarterly, No. 2, 1992, pp. 3—30.
208 Bibliography
Kieser, Alfred : “Human Relations-Bewegung und Organisationspsycholgie”, in

Alfred Kieser (ed.): Organisationstheorien, 3 rd edition, Stuttgart, (Kohlhammer)
1999, pp. 101—131.
Kieser, Alfred : “Kontruktivistische Ansätze”, in Alfred Kieser (ed.):

Organisationstheorien, 3 rd edition, Stuttgart, (Kohlhammer) 1999,
pp. 287—318.
Kieser, Alfred and Ulrich Koch : Organizational Learning through Rule Adaptation –
From the Behavioral Theory to Transactive Organizational Learning, Mannheim,
2000.
System Dynamics Review, Vol. 10, Nos. 2-4, Summer-Fall 1997, pp. 277—290.
Klein, Donald : “Some Notes on the Dynamics of Resistance to Change: The

Defender Role”, in Warren G. Bennis, Kenneth D. Benne and Robert Chin
(eds.): The Planning of Change, 4 th edition, New York, (CBS College
Publishing) 1985, pp. 98—105.
Koningswieset, Roswita and Alexander Exner: Systemische Intervention, Stuttgart,

(Klett-Cotta) 1998.
Kotter, John P : Leading Change, Boston, (Harvard Business School Press) 1996.
Kotter, John P. and Dan S. Cohen : The Heart of Change, Boston, (Harvard Business
School Press) 2002.
Kotler, Philip : Marketing Management, 7 th edition, New Jersey, (Prentice-Hall, Inc.)

1991.
Kvale, Steinar: InterViews – An Introduction to Qualitative Research Interviewing,

Thousand Oaks, (SAGE Publications) 1996.
Lane, David C. : “Modelling as Learning: A consultancy methodology for enhancing

learning in management teams”, European Journal of Operational Research,
Vol. 59, No. 1, 1992, pp. 64—84.
Lane, David C. : “Friendly amendment: A commentary on Doyle and Ford’s

proposed re-definition of ‘mental model’”, Systems Dynamics Review, Vol. 15,
No. 2, Summer 1999, pp. 185—194.
Bibliography 209
Lane, David C. : “Should System Dynamics be Described as a ‘Hard’ or

‘Deterministic’ System Approach?”, Systems Research and Behavioral Science,
Vol. 17, 2000, pp. 3—22.
Manage Them More Effectively, Boston, (Harvard Business School Press) 2005.
Leavitt, Harold J., William R. Dill, and Henry B Eyring: The Organizational World
– A systematic view of managers and management, New York, (Harcourt Brace
Jovanovich, Inc.) 1973.
Leonard-Barton, Dorothy: “A Dual Methodology for Case Studies: Synergistic use

of a Longitudinal Single Site with Replicated Multiple Sites ”, Organizational
Science, Vol. 1, No. 3, 1990, pp. 248—266.
Leonard, Dorothy: Wellsprings of Knowledge – Building and Sustaining the

Sources of Innovation, Boston, (Harvard Business School Press) 1998
Lewin, Kurt: “Group Decision and Social Change” (first published in Newcomb and
Hartley’s Readings in social psychology, 1948, pp. 330—341), in Martin Gold:
The Complete Social Scientist – A Kurt Lewin Reader, Washington, (American
Psychological Association) 1999, pp. 265—291.
Lochmann, Hans-Dieter und Michaela Rüsch-Kornasoff : “Organization Change

Strategy: Ein wesentlicher Baustein des Reengineerings”, in Manfred Perlitz,
Andreas Offinger, Michael Reinhardt and Klaus Schug (eds.): Reengineering
zwischen Anspruch und Wirklichkeit, Wiesbaden, (Gabler Verlag) 1996,
pp. 329—340.
Luna-Reyes, Luis Felipe and Deborah Lines Andersen : “Collecting and analysing
qualitative data for system dynamics: methods and models”, System Dynamics
Review, Vol. 19, No. 4, 2003, pp. 271—296.
Lübcke, Poul (ed.): Politikens filosofi leksikon, Copenhagen, (Politikens forlag)

2001.
Lyneis, James M.: Corporate Planning and Policy Design – A System Dynamics
Approach, Boston, (Pugh-Roberts Associates, Inc.) 1980.
210 Bibliography
Maani, Kambiz E. and Robert Y. Cavana : Systems Thinking and Modeling –

Understanding Change and Complexity; Auckland, (Person Education New
Zealand, Lt.) 2000.
Maier, Frank und Andreas Größler : ”What are we talking about? A taxonomy of
Computer Simulations to Support Learning”, System Dynamics Review, Vol. 16,
No. 2, 2000, pp. 135—148.
Maier, Frank: Die Integration wissens- und modellbasierter Konzepte zur

Entscheidungsunterstützung im Innovationsmanagement, Berlin, (Duncker &
Humblot), 1995.
Martinez, Ignacio J and George P. Richardson: “An Expert View on the System
Dynamics Modeling Process: Concurrences and Divergences Searching for Best
Practices in System Dynamics Modeling”, at CD-ROM of Proceedings, System
Dynamics Conference, System Dynamics Society, 2002.
Martinez, Ignacio J. and Luis F Luna : “The Dynamics of Best Practices: A Struc-
tural Approach”, at CD-ROM of Proceedings, System Dynamics Conference,
System Dynamics Society, 2001.
McKee, Robert: “Storytelling That Moves People”, Harvard Business Review, June
2003, pp. 51—55.
McPherson III, L. Fillmore: “Organizational Change: An Industrial Dynamics

Approach”, in Edward B. Roberts (ed.): Managerial Applications of System
Dynamics, Cambridge, Massachusetts, (Productivity Press) 1978,
pp. 447—459.
Miller, George A.: “The Magical Number Seven, Plus Minus Two: Some Limits on
Our Capacity for Processing Information”, The Psychological Review, Vol. 63,
No. 2, March 1956, pp. 81—97.
Milling, Peter: Systemtheoretische Grundlagen zur Plannung der Unternehmens-

politik, Berlin, (Verlag Duncker & Humblot) 1981.
Milling, Peter: “Leitmotive des System-Dynamics-Ansatzes”, Wirtshaftswissen-

schaftliches Studium, Vol. 10, 1984, pp. 507—513.
Bibliography 211
Milling, Peter M. and Erich O.K. Zahn (edt.): Computer-Based Management of

Complex Systems, Proceedings of the 1989 International Conference of the
System Dynamics Society, Heidelberg, (Springer-Verlag) 1989.

Managementsimulatoren”, in: Zeitschrift für Betriebswirtschaft, Ergänzungs-
heft 3/95, Lernende Unternehmen (Sonderausgabe), 1995, pp. 93—112.
Milling, Peter: “Modeling Innovation Processes for Decision Support and

Management Simulation”, System Dynamics Review, Vol. 12, No. 3, 1996,
pp. 211—234.

Systemen”, in Peter Milling (ed.): Entscheiden in komplexen Systemen,
Wirtschaftskybernetik und Systemanalyse, Band 20, Berlin, (Duncker &
Humblot) 2002.
Mintzberg, Henry: The Rise and Fall of Strategic Planning, New York, (The Free
Press) 1994.
Morecroft, J. D. W. : “Executive knowledge, models and learning”, European Journal

of Operational Research, Vol. 59, Vol. 1, 1992, pp. 9—27.
Morecroft, John D. W. and John D Sterman (ed.): Modeling for Learning

Organizations, Portland, Oregon, (Productivity Press) 1994.
Morgan, Gareth : Creative Organization Theory, Newbury Park, California, (SAGE

Publications) 1989.
Munk, Kirstine: “In the Airy Spaces of Our Minds…: Cosmology and ritual design
in modern, Western astrology”, in Michael York (ed.): Nature, Religion, and
Culture, in print, London, (Equinox) 2006.
Newell, Allen and Herbert A. Simon : Human Problem Solving, Englewood Cliffs,
New Jersey, (Prentice-Hall, Inc.) 1972.
O’Connor, Joseph and Ian McDermott : The Art of System Thinking – Essential
skills for creativity and problem solving, London, (Thorsons) 1997.
212 Bibliography
Palmer, Ian and Richard Dunford : “Who says change can be managed? Positions,
perspectives and problematics”, Strategic Change, Volume 11, August 2002,
pp. 243—251.
Pearson, Alan : “You Drive for the Show but you Putt for the Dough”, appendix in
Stafford Beer: Beyond Dispute: The Invention of Team Syntegrity, Chichester,
(John Wiley & Sons) 1994, pp. 311—322.
Perlitz, Manfred, Jürgen Bufka, Andreas Offinger, Michael Reinhardt, und Klaus
Schug : “Reengineering-Projekte erfolgreich umsetzen: Ergebnisse einer
Erfolgsfaktorenstudie”, in Manfred Perlitz, Andreas Offinger, Michael Reinhardt
and Klaus Schug (eds.), Reengineering zwischen Anspruch und Wirklichkeit,
Wiesbaden, (Gabler Verlag) 1996, pp. 181—208.
Perlitz, Manfred : Internationales Management, 5 th edition, Stuttgart, (Lucius &

Lucius) 2004.
Randers, Jørgen : “Guidelines for Model Conceptualization”, in Jørgen Randers

(ed.): Elements of the System Dynamics Method, Cambridge, Connecticut,
(Productivity Press) 1980, pp. 117—139.
Rao, Abhijit: “Recognition of Conative and Affective Behavior in Web Learning

using Digital Gestures”, North America Web-Based Learning Conference,
Online Proceedings, New Brunswick, 2001.
Repenning, Nelson P. and John D. Sterman : Nobody Ever Gets Credit for Fixing
Problems that Never Happened: Creating and Sustaining Process Improvement”,
California Management Review, Vol.43, No.4, Summer 2001, pp. 64—88.
Richardson, George P. and Alexander L. Pugh : Introduction to System Dynamics

Modeling with DYNAMO, Cambridge, (MIT Press) 1981.
Richardson, George P. and David F. Andersen: “Teamwork in group model

Richmond, Barry: “The Strategic Forum: aligning objectives, strategy and process”,
System Dynamics Review, Vol. 13, No. 2, 1997, pp. 131—148.
Bibliography 213
Sloan Management Review, Summer 2005, Vol. 46, No. 4, Summer 2005,
pp. 53—60.
Roberts, Edward B .: “Strategies for Effective Implementation of Complex

System Dynamics, Cambridge, Massachusetts, (Productivity Press) 1978,
pp. 77—85.
Rockart, John F .: “Chief executives define their own data needs”, Harvard Business
Review, March-April 1979, pp. 81—93.
Rosenhead, Jonathan (ed.): Rational Analysis for a Problematic World, Chichester,

(John Wiley & Sons) 1989.
Rouwette, Etiënne: Group model building as mutual persuasion, Nijmegen, (Wolf

Legal Publishers) 2003.
Salge, Markus and Peter Milling : “Who is to blame, the operator or the designer?
Two stages of human failure in the Chernobyl accident”, System Dynamics
Review, Vol. 22, in print, 2006
Schein, Edgar H. : Organisationspsykologi, Danish translation, Herning, (Forlaget

systime) 1990.
Schein, Edgar H.: Process Consultation – MAOM Capstone Course for the
University of Phoenix, Boston, (Person Custom Publishing) 2000.
Schneiderman, Arthur M.: “Setting Quality Goals: Use observed rates of continuous
improvement to position targets”, Quality Progress, April 1988, pp. 51—57.
Scheer, August-Wilhem: Business Process Reengineering – Reference Models for

Industrial Enterprises, 2 nd edition, Berlin, (Springer-Verlag) 1994.
Scheper, Willem J.: Group Decision Support Systems, Tilburg/Utrecht,

(Dissertation) 1991.
Scherer, Andreas G.: “Kritik der Organisation oder Organisation der Kritik?
Wissenschaftstheoretische Bemerkungen zum Umgang mit Organisations-
theorien”, in Alfred Kieser (ed.): Organisationstheorien, 3 rd edition, Stuttgart,
(Kohlhammer) 1999, pp. 1—37.
214 Bibliography
Senge, Peter M. and John D. Sterman : “Systems thinking and organizational

learning: Acting locally and thinking globally in the organization of the
future”, European Journal of Operational Research, Vol. 59, No. 1, 1992,
pp. 137—150.
Senge, Peter M.: The Fifth Discipline, paperback edition, New York, (Currency
Doubleday) 1994.
Senge, Peter M. : The Dance of Change, New York, (Currency Doubleday) 1999.
Simon, Herbert A.: “Rational Decision Making in Business Organizations”, The

American Economic Review, Vol. 69, No. 4, 1979, pp. 493—513.
Simon, Herbert A.: The Science of the Artificial, 3 rd edition, Cambridge, (The MIT
Press) 1996.
Snabe, Birgitte and Jakob F. Ølgaard : Informationsstrategi, Master Thesis, Institute

of Mathematical Statistics and Operations Research, The Technical University of
Denmark, Lyngby, 1992.
Snabe, Birgitte: Brugtvognskoncept for Skandinavisk Motor Co., Thesis, Det

erhvervsøkonomiske diplomstudium, The Copenhagen Business School,
Copenhagen, 1994.
Snabe, Birgitte and Andreas Größler : “System Dynamics Modelling for Strategy
Implementation: Case Study and Issues”, Journal of Systems Research and
Behavioral Science, Vol. 23, No. 4, 2006, pp. 467—481.
Snabe, Birgitte, Andreas Größler, and Peter M. Milling : “Policies and Politics of
Establishing R&D Capacity in Low-Cost Locations”, Tagungsband, GWS-
Tagung 2005, Greifswald, in print, 2006.
Stake, Robert E.: The Art of Case Study Research, Thousand Oaks, (SAGE
Publications) 1995.
Sterman, John D.: “ Misperceptions of Feedback in Dynamic Decision Making”, in

Peter M. Milling and Erich O.K. Zahn (eds.): Computer-Based Management of
Complex Systems, Proceedings of the 1989 International Conference of the
System Dynamics Society, Heidelberg, (Springer-Verlag) 1989, pp. 21—31.
Bibliography 215
Complex World, Boston, (McGraw Hill Higher Education) 2000.
Sterman, John D.: “All models are wrong: reflections on becoming a

systems scientist”, System Dynamics Review, Vol. 18, No. 4, Winter 2002,
pp. 501—531.
Taylor, Frederick W.: The Principles of Scientific Management, New York, first
time published in 1911, (Norton & Company) 1967.
Takahara, Yasuhiko : “A Formal Model of Organization”, in Singo Takahashi,

Kyoichi Kijima and Ryo Sato (eds.): Applied General Systems Research on
Organizations, Tokyo, Springer, 2004, pp. 3—35.
Thun, Jörn-Henrik: Die zeitbasierte Fertigungsstrategie, Wiesbaden, (Mannheim

University Dissertation) 2002.
Thun, Jörn-Henrik, Peter M. Milling, and Uwe Schwellbach : ”The Impact of Total
Employee Involvement on Time-based Manufacturing“, in Kate Blackmon, Steve
Brown, Paul Cousins, Andrew Graves, Christine Harland, Richard Lamming,
and Harvey Maylor (eds.): ”What Really Matters in Operations Management“,
Bath, 2001, pp. 131—143.
Tilebein, Meike: “Eine strukturwissenschaftliche Betrachtung von Diversity

Management”, Tagungsband, GWS-Tagung 2005, Greifswald, in print, 2006.
Tversky, Amos: “Additivity, utility and subjective probability”, in Ward Edwards

and Amos Tversky (eds.): Decision Making, Middlesex, (Penguin Modern
Psychology) 1967, pp. 208—238.
Thygesen, Henriette H.: System Dynamics in Action, Copenhagen, (Samfunds-

litteratur) 2004.
Van der Smagt, Ton : “Causation and Constitution in System Dynamics; Modelling a
Socially Constituted World”, Journal of Systems Research and Behavioral
Science, Vol. 23, No. 4, in print, 2006.
216 Bibliography
Vennix, Jac A. M, David F. Andersen, George P. Richardson and John Rohrbaugh :

“Model-building for group decision support: Issues and alternatives in
knowledge elicitation”, European Journal of Operational Research, Vol. 59,
No. 1, 1992, pp. 28—41.
Vennix, Jac A. M. : Group Model Building, Chichester, (John Wiley & Sons) 1996.
Voss, Chris, Nikos Tsikriktsis, and Mark Frohlich : “Case research in operations
management”, International Journal of Operations & Production Management,
Vol. 22, No. 2, 2002, pp. 195—219.
Warren, Kim: Competitive Strategy Dynamics, Chichester, (John Wiley & Sons)
2002.
Warren, Kim: “Improving strategic management with the fundamental principles of

system dynamics”, System Dynamics Review, Vol. 21, No. 4, Winter 2005,
pp. 329—350.
Webster’s Encyclopedic Unabridged Dictionary of the English Language, New

York, (Portland House) 1989.
Weil, Henry B.: “The Evolution of an Approach for Achieving Implemented Results
from System Dynamics Projects”, in Jørgen Randers (ed.): Elements of the
System Dynamics Method, Cambridge, Connecticut, (Productivity Press) 1980,
pp. 269—291.
Weil, Henry B. and Kenneth P. Veit: “Corporate Strategic Thinking: The role of
System Dynamics” in Peter M. Milling and Erich O.K. Zahn (Eds.): Computer-
Based Management of Complex Systems, Proceedings of the 1989 International
Conference on the System Dynamics Society, Stuttgart, (Springer-Verlag) 1989,
pp. 66—75.
Wenneberg, Søren B .: “Socialkonstruktivisme som videnskabsteori: Sifyfos’

videnskab”, Online Paper, Institut for Ledelse, Politik og Filosofi, Copenhagen,
2002.
Winch, Graham and Sonja Derrick: “Flexible Study Processes in ‘Knotty’ System
Dynamics Projects”, Journal of Systems Research and Behavioral Science,
Vol. 23, No. 4, in print, 2006.
Bibliography 217
Wolstenholme, Eric F .: “The definition and application of a stepwise approach to

model conceptualisation and analysis”, European Journal of Operational
Research, Vol. 59, 1992, pp. 123—136.
Wolstenholme, Eric: “Using generic system archetypes to support thinking and

modelling”, System Dynamics Review, Vol. 20, No. 4, Winter 2004,
pp. 341—356.
Yin, Robert K.: Case study Research, 3 rd edition, Thousand Oaks, (SAGE
Publications) 2003.
Zock, Alexander: “A critical review of the use of System Dynamics for

organizational consulting projects”, at CD-ROM of Proceedings, System
Dynamics Conference, System Dynamics Society, Oxford, 2004.

The Usage of System Dynamics in Organizational Interventions

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

The Usage of System Dynamics in Organizational Interventions

Hochgeladen von

Copyright:

Verfügbare Formate

Birgitte Snabe

The Usage of System Dynamics

With a foreword by Prof. Dr. Peter Milling

Dissertation Universität Mannheim, 2006

flage Dezember 1997

Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt.

Die Wiedergabe von Gebrauchsnamen, Handelsnamen, Warenbezeichnungen usw. in diesem

initially individual objectives and strategies into a generally accepted

Foreword .............................................................................................. VII

List of Figures ...................................................................................... XV

List of Tables..................................................................................... XVII

A. Organizational Intervention Skills as Corporate Competence........ 1

II. Foundation and Strategies for Planned Change Interventions....... 10

III. The Usage of System Dynamics Modeling

B. Conceptual Foundation for the Usage of System Dynamics ......... 27

III. The Development Process of System Dynamics Models

C. A Case Study Using Participative System Dynamics Modeling

I. Research Considerations for the Case Study Application ............. 83

II. Case Study: Refining and Implementing a Location Strategy ....... 89

III. Evaluation of the Case Study .....................................................108

D. The Usage and Utility of Participative Modeling

I. Context Factors Relevant for Deciding on

II. Process Considerations ..............................................................128

III. Outcomes of Participative Modeling Efforts in the

E. Targeted Participative Modeling in Change Management...........161

Figure A-1: The Basic Model of Corporations ...................................... 2

Figure A-2: Model of stages of problem solving ................................... 5

Figure A-3a: Diagnostics and decision-making (cycle I) ......................... 6

Figure A-3b: Change management (cycle II) ........................................... 6

Figure A-4: Goal-seeking system ........................................................ 16

Figure B-1: Accumulative levels of models in the usage of system

Figure B-2: Limited linear perception of system ................................. 36

Figure B-3: Theory of planned behavior ............................................. 43

Figure B-4: Different limited linear perceptions of a system ............... 47

Figure B-5: Mental models as instruments between actual systems

Figure B-6: The basic structure of organizational learning .................. 54

Figure B-7: Formal models supporting organizational learning ........... 56

Figure B-8: The learning cycle for learning labs ................................. 70

Figure B-9: Mental database and decreasing content

Figure B-10: Maps, frameworks and microworlds................................. 74

Figure B-11: A model of communication .............................................. 79

Figure B-12: The Wallow Curve at work .............................................. 81

Figure C-1: The reinforcing growth loop underlying the

Figure C-2: Intervention process as communicated in the project ........ 94

Figure C-3: The location strategy model ............................................. 99

Figure C-4: Fraction of employees in low-cost countries compared

Figure C-5: Development in cost-index for an average productive

Figure C-6: Development of productivity index for an average unit

Figure D-2: Four antecedent processes in organizational

Appendix D, Figure 1: The preliminary model in the case study....……. 185

Appendix E, Figure 1: Simulation run of adjusted model

Appendix E, Figure 2: Model without rate-on-rate modeling……...…… 188

Table C-1: Roles and responsibilities as defined in the project........…. 95

Table C-2a: Main sources for evaluation of outcomes on

Table C-2c: Main sources for evaluation of outcomes

Table C-2d: Main sources for evaluation of system dynamics

Table C-3: Questionnaire results……..………………………….……. 118

Table D-1: Political characteristics of situations in terms of the

Table D-2: Generic symptoms of change resistance..………………… 142

Appendix A, Table 1: Parameters relevant to high-cost locations……… 167

Appendix A, Table 2: Parameters relevant to low-cost locations………. 168

Appendix A, Table 3: Parameters mainly relevant for transfer of tasks

Appendix B, Table 2: Main equations influencing production/month….. 172

I. The Need for and the Challenges of Organizational