An Interpretive Approach to Drawing Causal L oop Diagrams
Mostafa J afari* Roozbeh Hesam Amiri ”” Atieh Bourouni °
jafari@ iust.ac.ir amiri_r@ ind.iust.ac.ir bourouni@ ind.iust.ac.ir
2. ¢ Department of Industrial Engineering, Iran University of
Science and Technology (IUST), Narmak, Tehran, Iran
Abstract
Causal loop diagram largely influences the effectiveness of system dynamics. The
complex interpretive nature of management problems makes it difficult to recognize all the
existing causal loop relations. In order to build system dynamics models for ill-defined
problems, "Group Model-Building" is developed. As discussed by Vennix, one source of these
messy situations is different perceptions of individuals. In this paper, we develop an
interpretive approach to drawing causal loop diagrams assuming that there are different
perceptions about same concepts and the analyst is closely engaged with finding most agreed
causal relationships.
Keywords: causal loop diagram; system dynamics; interpretive approach; group model building;
soft system dynamics; Decision Making Trial and Evaluation Laboratory (DEMATEL)
I. Introduction
In many complicated problem situations, determining causal relationships is a very
complex process because in every organization managing means interpreting and reacting to
interacting event and ideas of the real world (Checkland, 2001). From this point of view,
there is no unique definition of problem, but each individual has his own perspective in
defining and interpreting a problem situation (Lane and Oliva, 1998). The difference is
between hard and soft systems thinking. The hard point of view of systems has an
“objectivist” that consider problems as independent of individual’s point of views and
interpretation. The soft systems thinking has a “subjective” that take into consider the
importance of participant’s perception (Rosenhead and Mingers, 2001).
According to Vennix (1999) one of the individual sources of messy problems is
“perception and reality construction". When working with teams, one's perception is affected
by his professional background or position in organization.
Group model building is discussed by Vennix et al. (1990, 1992, 1993, 1997), Vennix
(1995, 1996), Luna-Reyes et al. (2006), Andersen et al. (1997a , 1997b), Rouwette et al.
(2002) and Visser (2007).
The organization of the paper is as follows: in the second section, interpretive systems
thinking and practice is presented. In the third section, different types of problems in
management science are discussed in order to show the importance of interpretive approaches
and group model building methods. In the forth to sixth section, three steps of an interpretive
approach to drawing causal loop diagrams are depicted.
* Corresponding author. Tel.: +98 912 5356487.
II. Interactive systems thinking and practice
The interpretive systems approach is frequently referred to as “soft systems thinking”
because it gives pride of place to people rather than to technology, structure or organization.
In contrast to the functionalist approach, its primary area of concern is perceptions, values,
beliefs and interests. It accepts that multiple perceptions of reality exist, and sometimes come
into conflict, and wants to help managers and consultants to work successfully in a
pluralistic” environment of this kind.
Interpretive approaches do not assume that organizations are just ‘‘human machines’’ in
which people are organized according to their functions, all of which are, geared to some
unitary objective. Instead, they assume that people may, rightly or wrongly, fight their own
comer rather than be subsumes into some overarching objective. Thus these approaches make
different assumptions about the nature of organizations. Soft approaches stress the
importance of organizational and individual learning. They stress that, when people face
problematic situations this is a chance for them to learn how to cope with such circumstances
in such a way that their performance is improved.
In hard approaches, it is typically assumed that a model is a would-be representation of
part of the real world. By contrast, in soft approaches the idea is that models are developed so
as to allow people to think through their own positions and to engage in debate with others
about possible action.
Some interpretive systems approaches are Interactive Management (Warfield and
Cardenas, 1994), Social Systems Design (Churchman, 1979), Strategic Assumption Surfacing
and Testing (Mason and Mitroff, 1981; Mitroff and Emshoff, 1979; Mitroff, Emshoff, and
Kilmann, 1979), Social Systems Science (Ackoff, 1977), Soft Systems Methodology
(Checkland, 1976, 1981; Checkland and Scholes, 1990; Checkland and Holwell, 1998), Soft
Systems Thinking (Senge, 1990), Soft Operation Research, Soft Cybernetics, Soft System
Dynamics (Lane, 2000).
"Soft systems thinking" is heavily influenced by the “root metaphor” of contextualism. So
itis very important to have a common sense about relevant aspects of the nature of problem.
III. Why an interpretive approach?
It is not always possible for an analyst to recognize and draw different causal relations in
a complex system using an “objectivist” approach. It is recognized that there should be no
single analyst, but a process of debate should take place among different actors (Guimares
Pereira et al, 2005).
According to Ackoff (1974, 1979) problems in management science are three points on a
spectrum: puzzles, problems and messes (Figure 1). A “puzzle” is a set of circumstances in
which there is no ambiguity whatsoever once some thought has been given to what is
happening or needs to be done. Although the nature of puzzles is simple, they are not always
simple to solve. A problem is more complicated than a puzzle, but less complicated than a
mess. This complication stems from the fact that a problem has no single answer that is
definitely known to be correct. A mess is a set of circumstances in which there is extreme
ambiguity and in which there may well be disagreement. In a puzzle, there is complete
agreement about the nature of the puzzle (a single correct definition) and also a single correct
solution. In a mess there is a whole range of possible definitions and descriptions of what is
going on, and there may be no way of knowing whether a solution, as such, exists at all.
In management science, problems, puzzles and messes are regarded as social constructs.
This does not mean that every aspect of the issues to be faced is within the mind of the
analyst or the participants but the interpretation of those "facts" is less certain, and different
people may interpret the same ‘‘facts’’ in different ways. Some of these interpretations may
tum out to be wrong, in the sense that they cannot be defended.
Puzzles Problems Messes
1
Formulation ' Agreed Agreed Arguable !
i
1
Solution Agreed Arguable Arguable t
Figure 1: Puzzles, problems and messes (adopted from Ackoff)
Lane (2000) has argued that system dynamics is very different from hard systems
thinking. Even on the basis of the classic texts of Forrester it is less austerely “objective” than
is often represented. If one considers recent work by Wolstenholme, Senge and Lane, and the
various craft skills that have grown up around the modeling, then it simply cannot be
considered as “hard”, or “optimizing”, or “deterministic.” At the same time, Lane makes no
pretence, and would not wish to, that system dynamics is a “soft” method in the style of SSM.
As discussed by Lane (1999), Forrester's ideas operate at the level of method not social
theory so system dynamics, though not wedded to a particular social theoretic paradigm, can
be re-crafted for use within different paradigms.
Vennix’s (1996) work on “group model building” centers on building system dynamics
models with teams in order to improve their performance when tackling strategic, messy
problems. As problems become more complex it is clear that any individual can have only a
limited view of their nature and causes. Group model building seeks to build on the natural
tendency people have to think in terms of causal processes in order to systematically elicit
and integrate the limited individual mental models into a more holistic view of the problem.
As the result is a shared system dynamics model, this can then be used to explore the
dynamics of the holistic view.
The client is involved throughout the model building process. The first step is to
construct a preliminary system dynamics model on the basis of individual interviews of
participants or the study of research reports and policy documents. This model is then further
refined, in consultation with the individuals involved, before being presented at a group
session. During the group session the team seeks to elaborate the model to bring it to a point
where the dynamic complexity of their view of the problem situation can be explored. This
process depends crucially upon the facilitator. This facilitator needs a thorough knowledge of
system dynamics and must also exhibit the right attitudes, skills and tasks. If all goes well,
the model building process will lead to the team learning their way to a shared social reality.
“Group model building” as discussed above is about identifying networks of related
variables rather than simple causal chains. It is rare for people to see more than one cause of a
problem.
An interpretive systems approach to solve different interpretation of stakeholders is
presented. This approach uses Decision Making Trial and Evaluation Laboratory
(DEMATEL) method (Gabus & Fontela, 1972, 1973) in order to find a common sense of
most important concepts from causal loop diagrams.
IV. Step 1: Drawing causal loop diagram for each individual
In this step, a skilled interviewer individually draws causal loop diagrams for each
stakeholder after a deep interview; Sterman (2000) suggests 15 important guidelines for
causal loop diagrams:
1. Each link must represent causal relationship between the variables. Y ou must not
include correlations between variables.
2. Be sure to label the polarity of every link in your diagrams. Label the polarity of
important feedback loops in your diagrams using the definitions in Table 1 to
help you determine whether the links are positive or negative.
Determine the loop polarity
Name your loops
Indicate important delays in causal links
Indicate variable names
Use curved lines for information feedbacks
Make important loops follow circular or oval paths
Organize your diagrams to minimize crossed lines
0. Don't put circles, hexagons, or other symbols around the variables in causal
diagrams.
11. Iterate. Y ou will have to redraw your diagrams, often many times, to find the best
layout.
12. Choose the right level of aggregation
13. Don't put all the loops into one large diagram
14. Make the goals of negative loops explicit
15. Distinguish between actual and perceived conditions
ore SiS? Greco
Table 1: Link polarity (Sterman, 2000)
Symbol Interpretation Mathematics Example
ATT else equal, if X increases av/ax >0
(decreases), then Y increases in the case of
+ (decreases) above (below) lati na
X—TPY __whatit would have been. aia Effort > Results
In the case of accumulations, X =
Tn the ca Y [4-305 + ¥,
All else equal, if X increases avlax <0
(decreases), then Y decreases In the case of -
- (increases) below (above) what. accumulations,
X——PY it would have been
In the case of accumulations, X =f
subtracts from Y. " fc APs Is PNG
Frustration Results
V. Step 2: Working with diagrams (Identifying the root metaphor)
After drawing causal loop diagrams for each individual, they should be analyzed. In
order to start analyzing, the first step is to identify the key concepts from diagrams. These key
concepts will be used in workshops.
Drawing individual causal loops -— Here, it is supposed that we have eight people (S; to Ss)
who have knowledge about a problem. An expert interviewer can draw these causal loop
diagrams and related matrixes (Table 2). Each number in the matrix shows the relationship
between two concepts. For example, a, =1 shows that there is a positive relationship
between iandj.
Tl OOononee
ow ood ooococesa coowtcd oooced
ocooolo on oolo cocooolo on oolo
ocoolonso ocoolono cocooloun ocooluano
cooloun ooloono on looo ooloono
a luooo a looond a louwxno a looonso
MSR sae OSB eS MSR ien SSS Ps
aS ar at ar
Ww Wy Ww Wy
< NL, < \ < <
oe
ot
OOo
Ss
oO
o
i)
4
Het BsaScies
oO
oo ooc.o oO oo oO
aon colo oot colo
ooolono oooloono
coloocno ooluonr
a lounwosd a loous
me RSaAcs GSEs
a? ay
Ww
© Ww
fo a a ~
fil, f %
( yf a
tL \ /
i uw
s <
a |
Oo
4
i)
o
oo l
colo
loo
oon
Soe
So
So
o
i)
Determining the key concepts — According to interviewees’ matrixes, we have
3 +P, +P; +P; +P, +P,
2
1
P, +P, +P, +P, +P; +P, +
T=
OOOOoOooIooooo
Soe NOAH So
PehesirSoe
=
The T matrix is normalized as the DEMATEL method suggested, so we get matrix M as
mp osg0 0 O10 g
2220 o111 0 0.444 ota
(222 0.111 0 0 0.333 0.2220]
“fp.222 0.222 0.1110 oO 0
0.333 0.111 0.2220 0.1110
pp.111 0 0.222 01110 0
According to DEMATEL method, Q is calculated as
Q =M x(I -M)*
709 2.083 0.522 0.341 1.289 0.491
734 1.195 0.528 0.329 1.233 0.498
[D725 1.24 0.423 0.31 1.106 0.5771
“1.623 1.088 0.391 0.183 0.684 0.284}
[51 1181 0.468 0.434 0.738 0.4281]
p42 0.628 0.418 0.238 0.465 0.214
Table 2 shows both "direct influence" and "indirect influence" which are calculated from
the Q matrix.
Table 2: DEMATEL direct and indirect influences
concept direct influence index indirectinfluence index total influence normalized total influence
A 5.435 3.721 9.156 0.193
B 4.517 7.415 11.932 0.251
Cc 4.381 2.75 7.131 0.150
D 3.253 1.835 5.088 0.107
E 3.759 5.515 9.274 0.195
F 2.383 2.492 4.875 0.103
According to the normalized total influence, we can sort concepts from most important
to least important one.
Table 3: Total influence
normalized total
influence
0.251
0.195
0.193
0.150
0.107
0.103
concept
mMoO>mw
It should be mentioned that these important concepts will be used in workshop sessions
in order to have a more effective discussion about the concepts that are more important than
others. In this example concept B, E, A and C are selected to be discussed.
In another approach we can mention all of the concepts in workshop sessions form B to
F,
Analyzing clusters - Since practical causal loop diagrams may contain a lot of concepts,
even when produced from one person, some way is needed to support their use. One
important feature of such analysis is the idea of a cluster of concepts. These are sets of
concepts that are similar in some way and could, in some sense, be more or less separated
from the rest of the diagram. Clearly, if a diagram contains concepts that are all strongly
interlinked, it may not be fruitful to attempt this sort of analysis. This might be the case if the
ratio of links to nodes is high.
A cluster indicates that there is an issue of some importance that may have an effect
rather greater than just on a single input and output link. Underlying the cluster identification
is the notion that ‘‘language is the common currency of organizational life.’’ That is, people’s
words have meanings, and a good starting point is to assume that, though the meanings will
change over time, the same words may have more or less the same meaning. Clusters can be
formed around the names and words that are used - this explains the importance of capturing
the words used by the interviewee.
VI. Step 3: Sessions and workshops
The idea of a workshop is to gain commitment to agreed and negotiated action. The
consultant is, as would be expected in this negotiative approach, not just a neutral facilitator,
but also has interests and may have other expertise.
When planning for a workshop the facilitator must establish a clear set of workshop
objectives and should anticipate the potential workshop stages that might be useful. This
process design should be negotiated with the client, during which the facilitator and client
both learn their way into the problem situation and the issues that need to be tackled.
Most teams of managers have a shared life or culture of some kind if the team is at all
cohesive, and may not be over-welcoming of attempts to question their ideas. It is thus vital
that the consultant starts these workshops in a carefully planned and facilitative mode. Given
that many senior managers are extremely powerful personalities then this may be easier said
than done! Perhaps the best approach is for the consultant to be explicit with the group about
her role as chair.
In these sessions most important concepts (from step 2) should be discussed. The goal is
to negotiate with different actors and finally draw some most agreed causal loop diagram of
the system.
VII. Conclusion
System dynamics modelers look forward to an approach for drawing causal loop
diagrams which can consider different perceptions of people. Hence, we have developed an
interpretive approach. With the proposed approach, the complex causal relationships between
concepts are discovered, key concepts are identified using DEMATEL, groups and team are
created and after a negotiative discussion, the most agreed causal loop diagram is drawn.
VIII. References
Ackoff R.L. (1974), Redesigning the Future: A Systems Approach to Societal Planning.
John Wiley & Sons, New Y ork.
Ackoff, R.L. (1977), Optimization + objectivity = opt out, European Joumal of
Operational Research, 1: 1.
Ackoff R.L. (1979), The future of operational research is past. Journal of the Operational
Research Society, 30(2), 93-104.
Andersen DF, Richardson GP (1997a), Scripts for group model building. Syst. Dyn.
Rev. 13, 107-129.
Andersen DF, Richardson GP, Vennix JAM (1997b), Group model building: adding
more science to the craft. Syst. Dyn. Rev. 13, 187-201.
Checkland, P.B. (1976), Towards a systems-based methodology for real-world problem-
solving, in: Systems Behaviour, J. Beishon and G. Peters, eds., Harper and Row, London, pp.
51-77.
Checkland, P.B. (1981), Systems Thinking, Systems Practice, Wiley, Chichester.
Checkland, P.B., and Scholes, P. (1990), Soft Systems Methodology in Action. Wiley,
Chichester.
Checkland, P.B. and Holwell, S. (1998), Information, Systems and Information Systems.
Wiley, Chichester.
Checkland, P. (2001), Soft System Methodology. In Rosenhead, J., Mingers J. (eds).
Rational Analysis for a Problematic World. John Wiley and Sons, Chichester, UK.
Churchman, C.W. (1979), Paradise regained: a hope for the future of systems design
education, in: Education in Systems Science, B.A. Bayraktar, H. Muller-Merbach, J.E.
Roberts, and M.G. Simpson, eds., Taylor and Francis, London, pp. 17-22.
Gabus, A., & Fontela, E. (1972), World problems, an invitation to further thought within
the framework of DEMATEL. Switzerland, Geneva: Battelle Geneva Research Centre.
Gabus, A., & Fontela, E. (1973), Perceptions of the world problematique:
Communication procedure, communicating with those bearing collective responsibility
(DEMATEL report no. 1). Switzerland Geneva: Battelle Geneva Research Centre.
Guimares Pereira, A., Corral Quintana, S., Funtowicz, S. (2005), GOUVERNe: new
trends in decision support system for groundwater governance issues. Environmental
Modelling and Software: 20, 111-118.
Lane, D. and Oliva, R. (1998), The greater whole: Towards a synthesis of system
dynamics and soft system methodology. European Journal of Operational Research: 107,
214- 235.
Lane, D. (1999), Social theory and system dynamics practice. European Journal of
Operational Research 113, 501-527.
Lane, D. (2000), Should systems dynamics be described as a ‘hard’ or ‘deterministic’
systems approach? Syst. Res. 17, 3-22.
Luna-Reyes LF, Martinez-Moyano IJ, Pardo TA, Cresswell AM, Andersen DF,
Richardson GP (2006), Anatomy of a group model-building intervention: building dynamic
theory from case study research. Syst. Dyn. Rev. 22, 291-320.
Mason, R.O., and Mitroff, 1.1. (1981), Challenging Strategic Planning Assumptions. John
Wiley and Sons, Chichester.
Mitroff, I.I., and Emshoff, J.R., (1979), On strategic assumption-making: a dialectical
approach to policy and planning, Academy of Management Review, 4:1.
Mitroff, L1., Emshoff, J.R. and Kilmann, R.H. (1979), Assumption analysis: a
methodology for strategic problemsolving, Man. Sci., 25-583.
Rosenhead, J., and Mingers, J. (2001), A New Paradigm for Analysis. In Rosenhead, J.,
Mingers J. (eds) (2001), Rational Analysis for a Problematic World. John Wiley and Sons,
Chichester, UK.
Rouwette EAJA, Vennix JAM, van Mullekom T (2002), Group model building
effectiveness: review of assessment studies. Syst. Dyn. Rev. 18, 5-45.
Senge, P.M. (1990), The Fifth Discipline: the Art and Practice of the Leaming
Organization, Random House, London.
Sterman, John. (2000) Busyness Dynamics — systems thinking and modeling for a
complex world, John Wiley.
Vennix JAM, Gubbels JW, Post D, Poppen HJ. (1990), A structured approach to
knowledge elicitation in conceptual model-building. System Dynamics Review 6: 194-208.
Vennix JAM, Andersen DF, Richardson GP, Rohrbaugh J. (1992), Model-building for
group decision support: issues and alternatives in knowledge elicitation. In Modelling for
Learning, special issue of the European Journal of operational Research (Morecroft JDW,
Steman JD (eds)) 59(1): 28-41.
Vennix JAM, ScheperW Willems R. (1993), Group model-building: what does the client
think of it? In The Role of Strategic Modelling in International Competitiveness, Proceedings
of the 1993 International System Dynamics Conference, Sepeda E, Machuca J (eds). Cancun:
Mexico; 534-543.
Vennix JAM, Gubbels JW. (1994), Knowledge elicitation in conceptual model building:
a case study in modeling a regional Dutch health care system. In Modeling for Learning
Organisations, Morecroft JDW, Sterman JD (eds). Productivity Press: Portland, 121-146.
Vennix JAM. (1995), Building consensus in strategic decision making: insights from the
process of group model building. Group Decision and Negotiation 4(4): 335-355.
Vennix JAM. (1996), Group Model Building: Facilitating Team Learning Using System
Dynamics. Wiley: Chichester.
Vennix JAM, Andersen DF, Richardson GP (1997), Foreword: Group model building,
art, and science. System Dynamics Review Vol. 13, No. 2: 103-106.
Vennix JAM (1999), Group model-building: tackling messy problems. Syst. Dyn. Rev.
15, 379-401.
Visser M (2007), System dynamics and group facilitation: contributions from
communication theory. Syst. Dyn. Rev. 23, 453-463
Warfield, J.N. and Cardenas, A.R. (1994), A Handbook of Interactive Management,
Iowa State University Press, Iowa.
