Table onte
The Process Model of Problem-Solving Difficulty
Ozge Pala
Etiénne A J.A. Rouwette
Jac A.M. Vennix
Methodology Section
Faculty of Management Sciences
University of Nijmegen
Thomas van A quinostraat 5.0.77
PO Box 9108
6500 HK Nijmegen
The Netherlands
tel +31 24 3611470
fax +31 24 3611599
0.Pala@nsm.kun.nl
<WORK IN PROGRESS>
Abstract
Groups and organizations, or in general multi-actor decision-making groups, frequently come across
complex problems in which neither the problem definition nor the interrelations of parts that make up
the problem are well defined. In these kinds of situations, members of a decision-making group have
disagreements on what the problem is and/or how it should be solved. The study reported in this paper
represents a causal loop diagram, which brings together different causes that lead the group members
into disagreement. In this way features of individual and group decision-making can be integrated in a
coherent framework. By analyzing the problem from a feedback point of view, we hope to clarify the
self-perpetuating quality of these problems. The main feedback loops in this model were identified
with the aim of pointing out key issues to keep in mind for interventions in complex problems. A
small portion of this model was also quantified to show the possible creation of a sustained
disagreement situation.
1, Introduction
Groups and organizations that need to solve a problem together, very frequently, have difficulty in
solving their problems. According to March and Simon (1993), the necessary conditions for inter-
group conflict are a felt need for joint decision-making, difference in goals and/or difference in
perceptions of reality. March and Simon calls these interpersonal inconsistencies in preferences or
identities: “ different people want different things, and not everyone can have everything he or she
desires. Different people have different identities, and their different definitions of appropriate
behavior are mutually inconsistent’. Moreover, groups also face various problems of communication
and coordination which can affect the communication quality or decision-making groups, lead to
stress and motivational problems, and create undesirable or no solutions.
The aim of this paper is to bring together the factors that are at play at multi-actor decision-making
situations. We will relate the processes using a causal loop diagram, identify feedback loops that can
lead to disagreements, and finally, point out key issues to keep in mind while intervening in groups.
This causal loop diagram will form the basis for a formalized system dynamics model.
2. Disagreement
March and Simon (1993) state that differences in goals and perceptions of reality are the necessary
conditions that lead groups or organizations to conflict. Taking this to a more general level, we can
say that disagreement in a problem-solving group arises from differences in the mental models of
individuals. Richardson et. al. (1994) classify types of mental models that are used by individuals to
decide on which action to take into three groups: ends, means, and means/ends models. Ends models
describe the goals people keep in mind when considering interventions in a problematic situation,
while the means model represents the steering points in a problem. Means/ends models capture the
relation between intervention points and goals and reflect assumptions on how a system functions.
The three types of models jointly help people decide on what the problem is and how they can solve
it. So, if people would hold differences in any of these three types, disagreement between the
members of a problem-solving group can arise. We are going to classify disagreements broadly in two
groups: (1) disagreement on problem definition and (2) disagreement on what to do. When the
perceived reality deviates from the desired situation, a problem is said to exist. However, different
people can hold different perceptions of reality as well as different goals/desired situations. Therefore,
we can say that disagreement on a problem definition can arise due to differences in perceptions of
reality and/or differences in what the desired situation (i.e. the ends in Richardson et. al.’s framework)
is. The second type of disagreement is related to how people would like to solve the problem at hand,
i.e. disagreement on what to do. By what to do, we mean the means and strategies that one would like
to use to solve a problem. Strategies and solutions are derived by analyzing the problems we have by
using the assumptions we have on how the world around us functions. Therefore, if people define the
problem in different ways and/or have different cognitive models (i.e. assumptions) on how the
system functions (means/ends model) then this can lead to differences in strategies and solutions they
find.
So to summarize, differences in goals and perceptions can lead to different definitions of the
problems. This disagreement, in tum, together with differences in assumptions people hold with
regard to the structure of the system would lead to disagreement on how to solve the problem. Overall
disagreement a group would have is then a function of the disagreements on the problem definition
and what to do in order to solve the problem. Figure 1 shows how differences in perceptions, goals,
and assumptions held by people can lead to disagreements on problems definitions and on what to do,
which in tum jointly create the disagreement.
differences on
desired situation +
(goals)
disagreement on
* problem definition
differences in _ a Se
perceptions of actual disagreement
situation
h.
+
disagreement
on what to do
+
differences in
assumptions on how
the system functions
Figure 1: Main differences in mental models of individuals that lead to disagreement
What is important to examine is the reasons of the differences that lead to disagreements. If we know
what the causes for the differences are then we can better focus our methodologies to alleviate these
causes, There are basically three areas that lead to the divergence of opinions:
* Differences in the usage and interpretation of information leading to differences in
perceptions (e.g. March and Simon 1993, Cartwright and Zander 1953);
* Poor communication quality leading to people talking past each other;
¢ Poor analysis quality (and differences in analysis done) leading to disagreements on what to
do.
In the following subsections we will deal with each of these three areas separately.
2.1 Information
In this subsection, we will explain the effects of information usage on the differences that exist
amongst the group members. As far as the information usage is of concern, the differences in the
perceptions of people can be due to (a) differences in information used; different individuals attend to
and collect different types of information while forming their perceptions of reality. For example, an
production manager and financial manager would tend to be interested different sources of
information and (b) differences in the interpretation of information; even though different individuals
would be presented the same information, how they value this information would differ from one
another based on the characteristics, past etc. of the person who is processing the information (V ennix
1996). Vennix gives the example of the interpretation of the level of an inventory. Some organization
members interpret it as too high, since keeping a stock of inventory is costly, while others see it as too
low since it ensures rapid delivery to customers. Below we will explain both differences in
information used and interpretation in more detail and place them in the causal loop diagram.
Individuals routinely collect and process information. Perceptions of reality are based on this
information. In the environment of organizations, there are many information flows. Neither available
time nor the cognitive capacities of human beings is sufficient to attend to all these information flows.
Complexity of the environment (time pressures, the vast amount of information sources, uncertainties)
coupled with limited information handling capacity of individuals leads to the usage of filters and
simplifying procedures (called heuristics or rules of thumb) in scanning the environment and
processing information (Hogarth, 1987). As a result, only a few of the existing information flows
reach the decision-makers and influence their decision-making. So individuals selectively perceive
their environment and this perception will be a function of the filters used. These filters are largely a
function of individual experiences, mental models, and organizational functions and positions.
Therefore, different individuals will tend to focus on different sources and types of information. Since
information forms the basis of our perceptions of reality, the higher the extent of differences in
information used, the larger will be the differences in perceived reality. Moreover, even if the same
information sources would be used, different people can interpret them in different ways due to
differences in their mental models. Since people hold different assumptions on how the system around
them functions, they would attach different meanings to the same information.
We will now explain the effects of information scanning and processing in more detail (please refer to
figure 2). The first factor that affects the differences in the perceptions is the differences in
information used. As explained above, the differences in information used is largely due to different
filters used by different people. This is called the selective bias. Our filters lead us to select
information in a biased way. The different filters arise due to differences in the functions held by
different individuals, differences in the perceptions', and differences in previous experience’. These
three factors are shown in figure 2 as linked to differences in information used. There are three other
factors that would affect the differences in information used. The first one is the potential number of
information sources available (March and Simon, 1993). If there would be less information sources
then there is a higher probability that the same sources will be picked. The number of available
sources would, in tum, be dependent on the analytical complexity of the problem. If the problem
becomes more in analytical complexity, then it would involve more variables and relationships. To
have information on more factors and relationships, the decision-maker would be exposed to
increased number of information sources. The last but not the least is the extent the information is
channeled within the group. By channeling of information we mean the extent the information is
shared amongst the member of the group. The more the information is shared; more will be amount of
people who are exposed to the same information (March and Simon, 1993). How well the information
is shared in the group is a function of the communication quality of the group. Communication quality
will be discusses in section 3.
The second factor affecting the differences in perceptions is the differences in interpretation. These
are the differences caused due to the way the information is processed. First of all, the way we
interpret the information is a function of the information we have. Therefore, the more is the
difference in the information people use, the more will be the differences in interpretations.
Differences in interpretation is also a result of the differences people have with respect to the
assumptions on how reality works and what people think the problem is. Individuals seem to
manipulate the collected information to fit it into the way they perceive things (Hogarth, 1987).
If we look at figure 2, we can see that if the differences in information used and the way it is
interpreted are not remedied then it can lead to increased differences in perceptions. If people keep on
using and interpreting information based on the way they perceive the world then each will get more
convinced in their perceptions and the extent of difference in their perceptions will increase.
Therefore, the tools we design should aim at decreasing the differences in the way people use
information.
1 Actually, the way we selectively scan for information is affected by the way we perceive the problem and how
‘we perceive our environment. The model might therefore be extended by including a relation from disagreement
to problem definition.
> Since individuals seem to search for information with already a solution in mind, a link from disagreement on
what to do can be included in the model as well. A solution is based on the previously satisfying alternatives.
This is the heuristic of “rules of thumb” or “representativeness”. And people have the tendency to reject the
information that is disconfirming with the solution they already have in mind (Hogarth, 1987).
(lifferences —————-——_
4 of interest. ~
differences on ~~
desired aicatten
differences in (goals)
fanctons held -
| disagreement on
{+ problem definition.
| "
differences in differences in ¥
previous expanence perceptions of actual eS
2 ‘i go Ce \ disagreement
number f avlibe Sy i AL x 4
Lae \ Z
ee «
4p infomation sources yin & disigreement.
information used ) on what to do
a Ww te
| it
/ f | S.
/ a }
| chanelling of ‘interpretation’ f
\ snips ‘aad = differences in
\ communication _p te group oe assumptions on how
adyven MY Ss se the system functions
‘complexity of the ig
problem
Figure 2: The main causes of differences in perceived reality
2.2. Communication amongst the group members
Communication, or rather miscommunication, is an important reason why things go wrong in
organizations. People holding different viewpoints do not pay attention to what the others in the group
are saying. According to Schein (1987) we may at any time observe the content and the process of
communication in a group. Over time, the recurrent pattems or communication structure may be
identified. As a first start of characterizing communication, we will look at communication process
quality and communication content quality.
Communication process quality represents how the well the process involves factors such as openness
and willingness to listen to each other’s ideas. It is defined by using the following dimensions’: open
communication, clear and understandable communication, equal participation, opportunity to address
issues about which opinions diverged, pragmatic and clear focus, attention to each other’s ideas, and
absence of dominance of the discussion by participants. These dimensions originated from a scale first
used by McCartt and Rohrbaugh (1989) and later by Vennix, Scheper and Willems (1993) and
Rouwette, Fokkema, Van Kuppevelt and Peters (1997). In order to keep the model as simple as
possible, we will reduce these seven dimensions to three variables in the model. Open
communication, clarity and understandability, equality of participation and absence of dominance are
all captured under open communication. Pragmatic and clear focus is included as communication
complexity. Third, there is the motivation to listen to others, which captures the attention to listen to
other's ideas.
Openness of communication is an indicator for communication process quality and therefore has a
Positive influence on communication process quality. However, openness of communication will be
hindered by biased communication, which is a result of lack of motivation to listen to others. If the
communication in the group is not going well, then this would increase the feeling of frustration of
group members leading to stress and less motivation to listen to others which would lead to decreased
communication quality but also to biased communication which would hinder the openness of
communication. Following the theory of procedural justice (Korsgaard, Schweiger and Sapienza,
1995), we can also say that openness of communication also has a positive influence on ‘people feel
° McCartt and Rohrbaugh include “absence of time pressure” which is omitted here because it seems more of an
extemal factor. Even when there is time pressure there can still be a high quality communication. In addition,
the variable ‘stress’ which is included in the model can be taken to represent the effect of time pressure on
individual performance.
acknowledged’, which increases their motivation to listen to others. Moreover, if the number of
parties involved will increase the quality of communication will decrease.
Apart from communication process, communication content can be addressed as well (Schein, 1987).
Janis and Mann (1977) describe the relevant aspects of communication content. In their view,
communication is adequate to the extent it covers all relevant aspects of a decision: all options, goals,
values, risks, and costs. In addition, all positive and negative consequences, conditions and
contingencies should be addressed and all information for weighing options should be analyzed,
resulting in an integration of all relevant information. If these dimensions of a decision are covered,
Janis and Mann speak of vigilant information processing. It is clear that when a problem increases in
analytical complexity (contains more variables, relations or loops) it grows more difficult to cover all
options or goals etc.
Communication process quality and communication content quality are related, as described by Janis
and Mann (1977). In the extreme of groupthink, the very low communication process quality leads to
a low content quality. So it is reasonable to include a relation between quality of process and content.
This is modeled as a direct relation. We also included complexity of communication in the model as a
function of disagreement. Complexity of communication lowers communication quality. Lastly, there
is the indirect effect of information processing quality, which changes under the influence of stress
(included as an element of process quality) and influences communication content quality.
Increase in both communication process and content qualities leads to better results in terms of
decreasing the existing differences between decision-makers. Communication content and process
qualities together determine the overall communication quality. An increased communication quality
enables the real differences of viewpoints to be addressed and would enable sharing of different
aspects of mental models. Therefore, it generates alignment of perceptions and desired situation.
Moreover, with the increase in communication quality the channeling of information in the group can
increase.
Figure 3 includes the variables related to the communication sector in the model. The most important
loop involved in this particular model is the loop that states that the higher the communication quality
the lower will be the differences in perceptions and goals and thus lower will be the difficulty to solve
the problem and as the difficulty to solve the problem decreases the complexity of communication
will decrease leading to even higher quality communication.
tivation to complexityof
«ferences
of intrest.
y lista oothrs — contnunon
|
biased commu |
i | done in
soe” Tae Za, WR | {unions bel
kot nyt i |
Wt Ge | Y |
{ | tao / /\ difrencesin —
{ 4 i \ evs expec
| cgpeatesof communication —_ | ™~ G
\ pneaten process quality “7 .
pole fol ray infomation Mie
cknowledget fiat ‘processing capacity £. differences in
/ 7 numberof avalshe eros
\ vat mon
A ‘afomtion sources {
communication \
content quality / PN
h | lerndesin
commutcat ctonotieg of npr
‘nal \ infomation i differences in
\ analytical assimplions on how
YS complet atthe —/% __the system functions
\, BSs 2 29
\ SS ee
Figure 3: Effects of communication quality
2.3 Problem Analysis
In addition to the use of information and the communication process, the analysis of information also
has an effect on disagreement within a decision making group. Different group members will have
differences in assumptions on how the system functions because most of the time these assumptions
are implicit and it is difficult to communicate assumptions when they stay implicit. Moreover since
people use different analysis procedures to decide on what to do, there occurs a disagreement on what
to do. We will call all these into one variable called the analysis quality. Analysis quality is
determined by several factors such as time pressures, satisfying behavior of individuals. Analytical
complexity is also a variable that would affect the analysis quality. If the problem is high on analytical
complexity then the system can be defined in many different ways. It would be difficult to validate
whose opinion is right. Last but not least, is the effect of communication quality on analysis quality. If
the communication is well in the group, it would allow for easier analysis in terms of identifying the
system structure.
As the analysis quality increases, the assumptions regarding how the system functions, variables and
relationships existing in the real system are made explicit. This would decrease the differences in
assumptions on how the system functions leading to more agreement on what to do (the means).
This would bring us to the model shown in figure 4.
pe
pa tees —————_ ~
comity f i ~
a San ra “ti a RS
x | desired situation
fue ' | we at oN
L ast eG a / incon ld me :
, Wr Gaal wy f+ \
| [A tga / teen Ay
\ Nos 4 F é previous experience < oe =
\ communicabon braces —arg oe Se ‘situation
‘communication \
content quality /
;
2 EX
\ process quality A AL Ae / \
wore, / 7\ wales . “si wel | a
. | RSPCA umber of avaiable. ierencs i hal
A Gitte i pY e
-
ewnctsin,
{ntrpetaioa*—— 7
— ( on aie smn
‘analytical eed ~ assumptions on how
complexity of the ‘id the system functions
. a Za
= ~~ if
— +e
‘extent al ferences
invents
\
a anal
tet of exces ofthe a
——p assumptions variables and_—
icing reionsips
ieaplas \ ai
~prematire
Figure 4: Overall model linking the effects of information usage, communication quality, and analysis
quality to disagreement
The model in figure 4 integrates information, communication and analysis factors discussed in the
previous sections. In the following section we will address the feedback loops included in the
structure.
3. The main feedback loops
In Figure 5, you can see a simplified version of figure 4. It can be seen that there are three positive
feedback loops that can create an ever-increasing disagreement. These loops go through differences in
goals, perceptions, and assumptions. If no intervention is taken at an already existing disagreement
situation then the increased complexity of communication can lead to low communication and
analysis qualities catalyzing the differences in goals, perceptions, and assumptions. Moreover, if the
disagreement on the problem definition is not addressed, this can lead individuals to focus on different
information sources getting more and more convinced of their own point of view.
communication complexity of
toe process quality communication _ 4
communication
content iw Gs
ji
+ sa
communication
ity —__ lifferences in disagreement
disagreement on
goals Fs prob def + T+
a
a +
differences in . ;
information used differences in
ST erntions
+ analysis +
quality,
differences in
assumptions
Figure 5: Positive feedback loops.
Other important loops can be seen at the communication process quality part of the model. Unless
precautions are taken, quality of communication amongst individuals that have differences of opinions
can degrade since people will be stressed and not be motivated to listen to each other. These processes
can be seen in figure 6.
disagreement
+
smobvation to complexity of
senor communication
biased commi
ae communication
——_- process quality
people a F information
cknowiedget processing capacity
‘communication a
content: Dg
exit in rs
i
: completo the
wy mb
Figure 6: Positive feedback loops around communication quality
This means that when we intervene with groups to resolve disagreements, one has to pay attention to
using
* Effective of communication procedures,
* Effective methods to clarify problem and share information, and
+ Effective methods to elicit implicit assumptions and analyze the problem.
4. Initial quantification attempt
An initial attempt was made to start with the quantification. The portion modeled is related to the
disagreement in problem definition. As can be seen in figure 7, the only portion that is included is the
differences in information used, differences in interpretation, differences in goals, and differences in
perceptions.
realizaton_time differences_in_goals
additonal_effecl_of ai) in_pere
ceffect_of dif in_functags
citt_due fo info_ditt Hem Rep.
<if_in_interpretation_info
normal_probabilty_
iff_inafo_usee
‘avg_channeling,
Figure 7: Initial formalized model
If we run the model, we see that without precautions, it is possible to end up with sustained
disagreement on problem definitions. In figure 8, four runs are presented. The initial values of
differences in goals and perceptions are assumed to be 0. This represents the assumption that initially
within the group everybody share the same goals and perceptions and hence, the potential to define a
possible problem in exactly the same way. This would mean that there would be no disagreement on
the problem definition. Such a situation would represent 100% consistence within the
group/organization.
The first two runs show the simulation of a hypothetical situation where everybody in the group
assumed to have the same function’. In run 1, the average channeling of information in the group is
assumed to be 100%, meaning that everybody gets to see all the information that is has been collected.
In this case, the effect of having the same functions and seeing all the information available is no
differences in terms of either the goals or the perceptions. Thus, the disagreement on problem
definition will not exist. In run 2, the channeling of information is 0%, meaning that information that
is collected by one member of the group is not available to others. In this case, there is a slight
difference in perceptions leading to a slight increase in disagreement on problem definition.
The last two runs show a case where the differences in functions are at its maximum level. This would
Tepresent a situation in which all the members of the group have different functions. Runs 3 and 4
show situations where the channeling of information in the group is 100% and 0%, respectively. As
* Differences in previous experiences which is in the CLD is assumed to be part of this variable.
can be seen the differences in functions lead to higher differences in goals. The channeling of
information, on the other hand, moderates the effect of differences in functions on differences in
goals. If the information is shared in the group to its fullest extent then the effects of having different
functions disappears because everybody gets to see all the information collected. On the other hand, if
the information is not shared then differences in perceptions increase to higher values. In either of the
Tuns, we can see that disagreement on problem definition persist.
a2
g
3 8S 104
Zl 4
c
£ £05)
o
g 3 i
2 g a >
5 5 00
5 é
s 0.5-$-$$_+-_++——,
~_det
2
g
6
c
6
2
G
2
a
8
xy
i}
Figure 8: Model runs with the initial basic model
The quantification effort needs to include other sections of the model and be further validated before
more definite conclusions be drawn.
4. Conclusions
In this paper, an attempt was made to formulate an integrated framework of decision-making in
complex problems. Groups that need to come to joint-decision are often faced with disagreements.
These disagreements can arise due to differences in perceptions of reality, differences in goals
(desired conditions), and/or differences in assumptions held with respect to how the system functions.
The main cause of differences in perceptions is related to how information is used and processed by
the individuals of the group. The differences in the types or sources of information used by
individuals and the differences in how people interpret information are the basic causes for differences
in perceptions. Together with the differences in perceptions and goals, the communication problems
are also crucial for the (non-existence of disagreements.
We have put together all these factors in a causal loop diagram. The diagram sheds light onto crucial
processes that are going on in multi-actor decision-making. First of all, if the differences in
information use are not addressed, then the perceptions of reality can end up diverging from each
other. The reason for this is that individuals will tend to look for information that will prove their
point of view to be correct and they will (un)consciously avoid challenging information.
Communication plays a very important role in these kind of groups. If it is not paid attention to then
increase stress and motivational problems can lead to further divergence and thus, the problem
situation will not be resolved. During these processes, one should pay attention to making the implicit
assumptions individuals hold explicit, so that the other team members can share them as well.
The next stage in this project will be building a quantified model of these processes.
References
Hickson , DJ. RJ. Butler, D. Cray, G.R. Mallory, D.C. Wilson. 1986. Top decisions: strategic
decision making in organizations. Oxford, UK: Basil Blackwell.
Hogarth R. 1987. Judgment and Choice. John Wiley and Sons.
Janis, 1.1. and L. Mann 1977. Decision making. A psychological analysis of conflict, choice and
commitment. New Y ork: The Free Press.
March J. 1994. A Premier in Decision Making: How Decisions Happen. New Y ork: The Free Press.
March J and H. Simon. 1993. Organizations. Second edition. Oxford, UK: Blackwell Publisher.
McCart, A. and J. Rohrbaugh. 1989. Evaluating group decision support effectiveness: a performance
study of decision conferencing. Decision support systems, vol 5, 243 253.
Rouwette, E.A.J.A., Fokkema, E., Kuppevelt, H.H.J.J. van, Peters, V.A.M. 1998 . Measuring Marco
Polis Management Game's influence on market orientations. Simulation & Gaming, 29(4), 420431.
Schein, E.H. 1987. Process consultation Vol. 2. Addison Wesley.
Vennix, J.A.M., Scheper, W., and Willems, R. 1993. Group Model Building: What Does the Client
Think of It? Proceedings of the 1993 International System Dynamics Conference. Cancun, Mexico,
534 543.
Back to the Top
