Ventana software support forum

Posted by martin@utalca.cl
Hi,

I have a doubt concerning the definition of ""polarity"" in ""causal loop diagrams"". For instance, the ""+"" is in general said to mean ""when A rises, B will rise, too"" (or ""A decreases, so B, too""). This leads to strange situations like when thinking about population dynamics: ""when the <birthe rate> falls, <POPULATION> will decrease"" and someone protesting that the birth rate cannot make the population decrease.

I think the definition of polarity is ""when A rises, then B will have higher values than it would have had without the rise in A"" (and also the other way around). With this definition, there is no problem in saying ""<birthe rate> ->
POPULATION>"" with a ""-"" polarity: when the birth rate lowers, the
POPULATION>population
will be smaller than it would have been without the decline of the birth rate.

I find this ""dynamic"" interpretatino of polarity practical, but I've not seen it in the books and articles I've found so far (mainly by George Richardson and Kim Warren). So I'd be grateful if someone can point out a fault in my way of understanding polarity or indicate relevant literature.

Thanks,


Martin Schaffernicht
Facultad de Ciencias Empresqariales
Universidad de Talca
Talca - Chile
Posted by martin@utalca.cl
posting date Fri, 07 Apr 2006 08:12:23 -0400

Posted by ""Newton, Paul C"" <Paul.C.Newton2@boeing.com>
Hi,

See the ""Interpretation"" column in Table 5-1 in John Sterman's ""Business Dynamics"" textbook. For positive link polarity he writes ""In the case of accumulations, X adds to Y,"" and for negative link polarity he writes ""In the case of accumulations, X subtracts from Y.""

The classic papers on this topic, and other related ones, are two by George Richardson:

1) ""Problems with causal-loop diagrams"" in System Dynamics Review Vol. 2 No. 2,(Summer 1986):158-170, and

2) ""Problems in causal loop diagrams revisited"" in System Dynamics Review Vol. 13 No. 3, (Fall 1997): 247-252.

George illustrated some of the points in these papers in the development of some sketches toward the end (slides 38-53)of his keynote presentation at the Systems Thinking and Dynamic Modeling in K-12 Education Conference in Skamania, Washington in June 2004, the presentation entitled, ""Thinking about Systems Thinking: How We Improve."" Note particularly his advocacy of using + and - for links influencing stocks, and S and O for the polarities of all other links. You can download this presentation from

http://clexchange.org/conference/cle_2004conference.htm

Paul Newton
Posted by ""Newton, Paul C"" <Paul.C.Newton2@boeing.com>
posting date Sat, 8 Apr 2006 19:59:52 -0700

Posted by John Sterman <jsterman@MIT.EDU>
Martin is exactly right about the polarity of links in a causal map: The correct
definition for a positive link, say, x-->+ y is ""an increase in x causes y to
rise above what it would have been otherwise, and a decrease in x causes y to
fall below what it would have been otherwise"". The definition for a negative
link polarity is similarly ""an increase in x causes y to fall below what it
would have been otherwise, and a decrease in x causes y to rise above what it
would have been otherwise"".

Mathematically, we have

x-->+ y <=> dy/dx > 0 and
x-->- y <=> dy/dx < 0 (where dy/dx indicates the partial derivative)

In the case of flow-stock links where y is a stock and x one of its flows
(such as population and births), then

x-->+ y <=> y = integral(x dt), that is, x adds to the stock y
(x is an inflow to the stock y), and
x-->- y <=> y=integral(-x dt), that is, x subtracts from the
stock y (x is an outflow of y).

The verbal definition above works for all cases at all times.

Note that

1. there is no implication that x actually rises or falls -- it could do either.
The link polarity is a statement about the structure of the system, not a
characterization of its behavior.

2. you cannot tell whether y is actually rising or falling if you know the
change in x. there may be other inputs to y. The link polarity is a conditional
statement of what would happen if there were a change in the input x, holding all
other inputs constant.

The apparently simpler definitions, e.g., a positive link polarity means ""if x rises,
then y rises"", are incorrect. For example, suppose we have

employee motivation --->+ employee performance +<--------Employee compensation

(let's not debate whether the links are correct -- just assume they are for the sake
of example).

Then a rise in motivation may accompany a decline in performance -- if compensation
falls enough. But it is still always true that a rise in motivation will cause
performance to be greater than it would otherwise have been, consistent with the
correct definitions above.

Similarly, consider the stock-flow structure

births --->+ population -<---deaths

Here, a rise in births tells you nothing about whether population rises or falls.
Population will rise if and only if births exceed deaths, and will fall iff
deaths exceed births. Nevertheless, if births increase, then population will be
higher than it would otherwise have been.

The correct definitions of link polarity and many examples are described in my
textbook (Business Dynamics, Ch. 5).

John Sterman
Posted by John Sterman <jsterman@MIT.EDU>
posting date Sat, 8 Apr 2006 13:12:41 -0400

Posted by ""Bob Cavana"" <Bob.Cavana@vuw.ac.nz>
hello Martin,

You have raised a very important point. The intepretation of the 'polarity'
or 'direction' of the relationship between two variables in a Causal Loop
Diagram (or Influence Diagram) is critical for the correct understanding
of the polarity (or direction) of the complete feedback loop. In Chapter 3
of my book with Kambiz Maani ""Systems Thinking & Modelling: Understanding
Change and Uncertainty"" we discuss these issues.

The situation arises because causal loop diagrams (CLD) use one symbol for
two ideas: an arrow can represent a 'causal influence' and an arrow can
represent an 'addition to or subtraction from' an accumulation. In Figure
3.2 (p27) in our book we provide the following guidelines (where variable
X is at the tail of an arrow (link or influence) and variable Y is at the
head. we use the symbols 's' (same) to be intepreted identically to '+'
(plus) in a CLD; and 'o' (opposite) to be intepreted as a '-' (minus)):

Figure 3.2 Determining types of causal relationship

If X adds to Y, or if a change in X causes a change in Y in the same direction, then use 's' or '+'

If X subtracts from Y, or if a change in X causes a change in Y in the opposite
direction, then use 'o' or '-'

(Based on Richardson, 1997: 249)

(source: Maani KE, Cavana RY (2000). Systems Thinking & Modelling: Understanding
Change and Uncertainty. Pearson Education (New Zealand), Auckland. available from
http://www.pearsoned.co.nz/search/title ... ckSearch=1
or http://www.pegasuscom.com/ )

On p35 of our book we elaborate on the ""Pitfalls in intepreting causal loop
diagrams"". we discuss a small example of a simple CLD with a reinforcing loop (births, population) and balancing loop (population, deaths). This is based
on the work of George Richardson (1997) and Jac Vennix (1996):

Richardson, G. (1997) Problems in CLDs Revisited. System Dynamics Review: 247-252. Vennix, J. (1996) Group Model Building: Facilitating Team Learning using System
Dynamics. John Wiley, Chichester.

I hope this is helpful.

all the best,
Bob

Dr Bob Cavana CMILT
Reader in Desion Sciences
Victoria Management School
Posted by ""Bob Cavana"" <Bob.Cavana@vuw.ac.nz>
posting date Sun, 9 Apr 2006 11:34:03 +1200

Posted by ""Stefan.Grösser"" <stefan.groesser@web.de>
Hi Martin,

You point to an important issue: How to understand the polarity of Causal linkages. The point you mentioned was discussed earlier within the SD-society (especially with respect to the notation of S and O instead of + and -).

For ""the"" current definition of link polarity, I refer to Sterman (2000, Business Dynamics): ""A positive link means that if the cause (in your case: A, the author) increases, the effect (in your case: B, the author) increases above what if would otherwise have been, and if the cause decreases, the effect decreases below what it would otherwise have been"" (p. 139).

I would say this definition of link polarity fits your definition pretty well. Would you agree?

All the best,

Stefan Groesser

University of Berne, Switzerland
University of St. Gallen, Switzerland
Posted by ""Stefan.Grösser"" <stefan.groesser@web.de>
posting date Sat, 8 Apr 2006 22:51:21 +0200

Posted by Rod MacDonald <Rod@isdps.org>
Martin,

Your point about the problem with polarity in causal loop diagrams is well
taken. I typically start out giving examples of causal connections that do
not involve stocks. I also use the caveat that you identified about ³when
A rises, then B will have higher values that it would have had without the Rise in A.²

George Richardson addressed this issue in an MIT D-Memo file in the mid 70s and it was reprinted in the SD Review in 1997. The citation is below.

Richardson, G. P. (1997) Problems in Causal-Loop Diagrams Revisited. System Dynamics Review 13,3 p. 242 ­ 257.

Sincerely,

Rod MacDonald
Initiative for System Dynamics in the Public Sector
Rockefeller College of Public Affairs and Policy
University at Albany
Posted by Rod MacDonald <Rod@isdps.org>
posting date Sat, 08 Apr 2006 08:26:34 -0400

Posted by matzaball50@aol.com
I believe there is a much deeper problem with causal loop diagrams
and polarity and that is when those diagrams represent social
interactions, they are metaphorical and vey rarely are the 'causes'
of either increases or decreaes due solely to any one interaction
between any two variables.

I actually think naming them 'causal' is a misnomer. 'Reasons for'
might be a more appropriate label

Regards,

Marc
Posted by matzaball50@aol.com
posting date Sun, 09 Apr 2006 10:56:21 -0400

Posted by ""Alan McLucas"" <a.mclucas@adfa.edu.au>
Each of these observations and comments are relevant and (arguably) correct. Unfortunately, much of the confusion about causality and polarity will always remain because of our own SD teaching and practice. The confusion comes about because, as Bob Cavana began to suggest (though he does not explain it this way) ... in causal loop diagrams we do not differentiate between stocks (integrations, accumulations, bathtubs or levels) and what causes them to change, that is, rate-controlling (and related auxiliary) variables. As long as we continue to depict rate variables directly influencing other rate variables without the accumulations being depicted there will be opportunities for confusion.

Jay Forrester's (1994: SDR Vol 10, No. 2-3: 252) comment that ... ""the starting point for model conceptualisation is to first identify the state variables (system stocks, levels or accumulators) then develop the flow rates that cause the state variables to change"", is incisive. Of course, we must be able to communicate ideas about causality to those (senior executives and others) who not want to know about state variables and rate-controlling variables. So, we will continue to use causal loop diagrams even though their shortcomings are well documented. Through our teaching and SD practice we continue to perpetuate this conceptualisation dilemma. However, it is interesting that John's examples of polarity use variables 'employee performance' and 'population', which are unquestionably state variables. Perhaps the way out of this recurring argument is to focus our attention on gaining wide acceptance of an alternate form of causal diagramming, one that clearly differentiates between state variables and the rate (and auxiliary) variables that cause them to change.
Regards,
Alan

Dr Alan McLucas
School of Information Technology and Electrical Engineering, UNSW@ADFA, Australian Defence Force Academy, Northcott Drive, CAMBPELL ACT 2600 AUSTRALIA Posted by ""Alan McLucas"" <a.mclucas@adfa.edu.au> posting date Mon, 10 Apr 2006 12:52:17 +1000

Posted by ""Alan McLucas"" <a.mclucas@adfa.edu.au>
Marc,

What you identify here is fundamental to the nature of relationships between concepts, where concepts are ideas or notions that individuals hold as part of their understanding of the way the world is. This is the realm of cognitive mapping rather than causal mapping, though each form of mapping represent causality in its own way.

In a cognitive map a 'causal' link between concepts has the sense of 'leads to'. Consider an example where two concepts are linked in a causal way in a cognitive map: 'actively partaking in the practice of smoking cigarettes' (leads to) 'the onset of heart disease later in life'. This might be the belief of an individual. This is less emphatic in its meaning than stating as we might in a causal loop diagram that 'Smoking' (a rate variable) causes 'Heart Disease' (a state variable).

In the cognitive map the 'causal' link acknowledges what we think about the nature of effect: not everybody who smokes cigarettes will suffer from heart disease, though everybody who smokes will be affected to some extent and if we smoke we can expect to develop heart disease.

Cognitive mapping is another tool we should have in our systems thinking multi-methodology toolbox.

Often we need to capture ideas about a problem through cognitive mapping, to inform thinking at higher level of aggregation where we wish to formulate
'dynamic hypotheses' (Sterman, 2000) in terms of variables. Alternatively,
when working at the higher level of aggregation, we need to be able to revert to a more detailed level of thinking, using cognitive mapping (for example). You might look at McLucas, A.C., 2001, 'Decision Making: Risk Management, Systems Thinking and Situation Awareness', Argos Press, Canberra. http://www.argospress.com/books/risk-ma ... /index.htm
This book uses case studies to explain how cognitive mapping, causal loop diagrams and influence diagrams can aid systems thinking. It explains the differences between the techniques.

Regards,
Alan

Dr Alan McLucas
School of Information Technology and Electrical Engineering, UNSW@ADFA, Australian Defence Force Academy, Northcott Drive, CAMBPELL ACT 2600 AUSTRALIA Posted by ""Alan McLucas"" <a.mclucas@adfa.edu.au> posting date Tue, 11 Apr 2006 09:14:58 +1000

Posted by Martin Schaffernicht <martin@utalca.cl>
Hi,

the contributions of Marc and Alan rise two questions in me:

1. how to compare ""cognitive maps"" and ""casual loop diagrams""?
2. why is the use of ""influence diagrams"" not more widespread?


On 1: a typical ""cognitive map"" consists of nodes and links with a polarity sign
and (sometimes) an indicator of strength. The nodes are statements that Eden
and Ackermann (www.phrontis.com) describe this way:

""an account of a problem is broken down into its constituent elements - usually
distinct phrases of 10-12 words [...]. These are treated as distinct concepts which
are then reconnected to represent the account in a graphical format. This reveals
the pattern of reasoning about a problem in a way that linear text cannot.""

As I understood it up to now, the difference is that in causal maps the nodes
are not neccessarily ""variables"" but may also contain ""events"". Another
difference is that in causal maps, the concept of ""loop"" is not as important as
in ""causal LOOP diagrams"" (CLD). I always thought that a CLD is the
articulation of beliefs (""mental model""). Also I believed that in CLD, the
casual links do not neccessarily mean ""determines in all the cases""; if the link
means so strong an attribution, then the nodes should probably be formulated
very carfully. In the smoking example, one would not want the model to be
refuted only because sometimes smoking does not lead to heart disease. How
would it be then to state ""smoking causes the likelihood of heart desease to rise""?

Are there more differences?

On 2: one of the first ""SD"" texts I happened to read was ""Systems Enquiry - a
System Dynamics Approach"" y Eric Wolstenholme (Wiley, 1990), where extensive use
is made of ""influence diagrams"". This type of diagrams is very much like CLDs
but it distinguishes between rate and accumulation variables and defines the
rule that a rate variable may only influence an accumuation variable (and vice
versa). (There are ""organizational boundaries, too, but this is not so relevant
here.) This has also been the last time I saw ""influence diagrams"" used. Why
aren't they used?

""Saludos"",

Martin Schaffernicht
Posted by Martin Schaffernicht <martin@utalca.cl>
posting date Tue, 11 Apr 2006 16:04:55 +0200

Posted by ""Bob Cavana"" <Bob.Cavana@vuw.ac.nz>
hello Martin,

you raise a couple of extra interesting questions on this discussion topic: "" 1. how to compare ""cognitive maps"" and ""casual loop diagrams""?
2. why is the use of ""influence diagrams"" not more widespread?""

Q.1 -Regarding question 1, i would recommend that you take a look at the
work Ed Mares and i did on demonstrating how to construct a policy
argument from the perspective of critical thinking and how to convert
this into a causal loop diagram using the tools of systems thinking
and system dynamics. in this paper we examine the issue of New Zealand
Customs Service taking effective measures against the entry of anthrax
into New Zealand. We isolate the key concepts and use them to construct a
conceptual diagram (similar to a cognitive map), and then operationalise
the concepts (ie convert the 'concepts' to 'variables') and form a causal
loop diagram. this paper was published in System Dynamics Review:

Cavana, R.Y. and Mares, E.D. 2004. Integrating critical thinking and
systems thinking: From premises to causal loops. System Dynamics Review,
20(3): 223-235.

Q.2 - as regards your 2nd question, i suggest you refer to Geoff Coyle's
excellent book on 'System Dynamics Modelling' for a very comprehensive
treatment of Influence Diagrams (in particular: Chapter 2 'Influence
Diagrams', pp18-47; and Chapter 3 'Influence Diagrams Case Studies',
pp48-83). the full reference is:

Coyle RG. 1996. System Dynamics Modelling: A Practical Approach. Chapman
& Hall, London.

A lot of SD people use the terms 'influence diagram' and 'causal loop
diagram' interchangeably. However there are some differences, particularly
in emphasis. In drawing influence diagrams, Geoff suggests a number of
conventions, including using a solid line (link) to represent physical
flows and broken lines (links) to represent information, actions and
behavioural effects ( eg Coyle, 1996, pp 21 & 51). This might be a very
useful convention to adopt when drawing causal loop diagrams, particularly
in conjunction with the definitions of the symbols +/s and -/o discussed
in my previous posting on this topic [REPLY Polarity in Causal Loop
Diagrams (SD5892)].

however, an 'influence diagram' has a slightly different meaning in the
field of 'Decision Analysis', where an influence diagram is a ""network
of nodes connected by arrows"" (Daellenbach, 2002, p146). R. Howard uses
'influence diagrams' in 'place of decision trees or in parallel to
decision trees'. The appropriate references are:

Daellenbach HG, Flood RL. 2002. The informed student guide to management
science. Thomson, London (pp146-146 by H.Daellenbach on 'Influence Diagrams'.

Howard RA. 1990. From influence to relevance to knowledge. in RM Oliver and
JQ Smith (eds). Influence Diagrams, Belief Nets and Decision Analysis.
Wiley, Chichester.

i hope these comments are useful.

all the best,
Bob

Dr Bob Cavana CMILT
Reader in Decision Sciences
Victoria Management School
Posted by ""Bob Cavana"" <Bob.Cavana@vuw.ac.nz>
posting date Thu, 13 Apr 2006 10:34:16 +1200

Posted by ""Iason Anagnostopoulos"" <jason.anagnostopoulos@strath.ac.uk>

about how to compare cognitive maps and causal loops diagrams ,

I think I could add some more comments,

Initially, i think that cognitive maps or causal maps can be very useful
in order to inform causal loops diagrams in terms of a more explicit
social and political theory (Eden and Ackermann 1998) and a more clear
understanding of the imporatnce of the interaction of mental models in
the understanding of the system structure (Morecroft 1994)

Secondly , I would like to point that the term cognitive mapping refers
to an individual map which can have on average about 80 -100 concepts ,
while the term causal map (Eden and Ackermann 1998, Bryson et al. 2004)
refers to map constructed by a group using either the weaving of the
individuals maps or the construction of the maps together in a group
workshop Oval Mapping Technique or a networked computer workshop

A causal map can produce about 800-1000 concepts or even more, which can
be categorised as goals, strategies issues ( concerns for the future) ,
options and distinctive competencies

The distinctive competencies may inform concepts which are sustaining
themselves by having feedback loops , (Eden and Ackermann 2000)

However , for me the concern would be the following

If somebody would like to use causal maps in order to conceptualise
qualiative system dynamics (causal loop diagrams) , he/she should find
a way to appreciate the differences between the constuct of a cognitve
map or a causal map and that of a causal loop diagram

I think that the first issue would be to find the difference between facts
and the opinions of the system , and later find a way to explore the dynamics
hidden there

Possibly, somebody could start from those constructs which represent a more
physical structure ( staff ) or the level of the service , and then he/she
try to explore the perception of the members of the group for the system
( how the level of the expecation about the service of the system interacts
with the staff we have etc)

However, the process from one method to another may require a process of
mental challenge for the individual or the group. When the mental challenge
is big, then this may affect the ownership

THis is because the individual or the group should understand and appeciate
the difference between the two methods , the difference of assertion of reality
(construct) and the description in flows level description as a fact (system
dynamics variables) ,

For example, one construct in congitve or causal map may say ( ie. crack the
requirement problem- have sufficient and qualified staff) , however this should
be translated as two variables in system dynaics such as 1) (sufficient) staff
2)qualified staff ,

however one question to get asked is what is the role of recruitment to have
qualified staff , does the proces of the recruitment is going to bring some
new members of the staff which is going to increase the qualifications or it is
just a description which should further negotiated to incorporate a more
informative process for system dynamics when the system dynamics developement
is going to take place

Further to that, some constructs may inform a situation ( sustain the good level
of service quality) which may inform a negative feedback loop, or some event (i.e.
fight the growing demands on X) , but such process require a more explicit
representation of a structure which represents a system dynamics purpose

Last but not least, taking into account that the causal or strategy maps have too
many constructs the question to get answered is , how could somebody explore feedback
loops

In system dynamics there are a number of ways to do that (Randers 1980, Wolstenholme
1990, Sterman 2000, Vennix 1996)

of course there are a number of ways to analyse the maps by finding the most
busy concepts or cental or potent ones ,

i have the perception that the process is not very easy per se especially
when the problem does not focus on a well known feedback loop structure or
a known reference mode most familiar with a traditional system dynamics
kind of problem with an explicit purpose ,

I would be very glad if somebody could describe how could elicit feedback
loops or which ways could use in order to do so

Thanks a lot

Jason
Posted by ""Iason Anagnostopoulos"" <jason.anagnostopoulos@strath.ac.uk>
posting date Wed, 12 Apr 2006 17:56:35 +0100

Posted by <richarp@bellsouth.net>
Hi all,

Martin's first question about comparing cognitive maps is one
of the main challenges in my dissertation research. The maps
I am concerned with are not the concept maps described by Eden
and Acherman. They are causal maps similar to (but more
simplified) the causal maps used for modelling.

I am assessing the effect systems-based instruction has on
students' ability to interpret information about a complex
system. To do this, I am comparing (among other things)
students' causal maps describing a hypothetical fishing
controversy.

Participants pick from a stack of 36 cards, each expressing a
potentially important variable (e.g. fish population, fishing
effort, water quality, etc.) They are also allowed to add any
variable they feel they need to describe the system. Then, I
guide them through a series of exercises during which they
create their own causal maps. I the links are labelled only as
positive or negative. I avoid strength indicators for the sake
of simplicity, as the participants are by design not knowledgable
about the system outside of a short article I provide.

I am now in the process of evaluating data from a both a systems
group and a control group comparing values such as:

1. Link to node ratio--providing a proxy for the ""weblike"" shape
of a participant's map, as opposed to a ""linear"" shape.

2. Percent including at least one feedback loop.

3. Similarity index--for any two maps, take the ratio of the number
of mutual links to the number of unique links. I use this value to
compare the participants' maps to expert maps.

My original hope was to use ecological indices (e.g. Finn cycling
index, Ulanowicz's ascendancy) to compare the maps. However, those
indices require a conserved flow from through nodes (e.g. mass or
energy). Influence is not conserved.

One can also compare the use of specific nodes in each students map,
looking at which variables tend to effect the system the most and
which are most affected by other variables in the system.

If anyone has any other methods for comparing such maps, I would be
interested in hearing them.

Thanks,
Richard
Posted by <richarp@bellsouth.net>
posting date Wed, 12 Apr 2006 10:21:05 -0400

Posted by Martin Schaffernicht <martin@utalca.cl>
Hi Richard,

I'd be interested in knowing more about your research.

I do not know about methods to assess similarity, but there are some papers
about the assessment of /differences/ (in some way this may be the other side of
the coin):

* Langfield-Smith, K. and Wirth, A. 1992. Measuring differences between
cognitive maps, Journal of Operational Research *43*(12): 1135-1150
* Langan-Fox, J., Code, Sh. and Langfield-Smith, K. 2000. Team mental
models: techniques, methods and analytic approaches, Human Factors
*42*(2): 242-271
* Langan-Fox, J.,Wirth, A., Code, Sh., Langfield-Smith, K. and Wirth, An.
2001. Analyzing shared and team mental models, International Journal of
Industrial Ergonomics *28*: 99-112
* Markóvski, L and Goldberg, J. 1995. A method for eliciting and comparing
causal maps, Journal of management *21*(2), p. 305-333

The ""distance ratio"" method is used with ""cognitive maps"" and takes into account
that which ""exists"" in cognitive maps. I believe the concept ""feedback loop""
does not exist explicitly, at least not in the way it does inside the SD
worldview. So if you think that attributions like ""loop"" (also ""delay"" and
""non-linearity"") are important for your study (your point 2), I'd like to know
how useful you find this approach.

Good luck,
Martin
Posted by Martin Schaffernicht <martin@utalca.cl>
posting date Thu, 13 Apr 2006 21:40:47 +0200

Posted by Jonah Fogel <jfogel@utk.edu>
In reference to Richard's post:

I'm also trying to compare cognitive maps. In my case, I'm assessing
the similarity between individuals' cognitive structures before and
after a group model building workshop series. As Vennix (1996) points out, we don't have much empirical evidence to show that individuals are
forming consensus during group model building or how they are able to do
it. Cognitive mapping is able to capture both the content and
organization of information. Therefore, comparing cognitive maps among
stakeholders in a group model building workshop should help us to ""see""
where consensus is, where it is not, and possible knowledge gaps that
the group isn't addressing. This research has implications for conflict
management, disciplinary integration of multidisciplinary research, and
social change efforts.

My method asks participants to individually answer a 'focus question'
such as ""what are the forces that are moving the state variable of
interest away from the desired future condition?"" Participants then
pool their answers to form a 'master list' of concepts describing
the dynamics of the system. Then participants each individually
choose 10 factors (or some other set number across individuals) from
the master list that most clearly are related to the change in
question. Using these chosen factors individuals then use a card
sorting technique to construct their personal cognitive map of the
system. Following the group model building workshop sessions
individuals are asked to repeat the personal cognitive mapping to
test for changes.

The comparison of cognitive maps is done using a social network
analysis program known as UCINet. Similarities can then be computed
to find sub-groups within the larger group context and predictive
variables (such as income, education, level of participation, trust,
etc.) can be tested for significance.

Richard, were you using a computer program for your analysis or were
you computing by hand? I'd be happy to talk more about this work if
anyone is interested.

Lastly, I used Vensim to build a stock-flow model. The model is being
developed from the group built model and may be used for policy testing
purposes. The model and my data analysis for the cognitive mapping is
not finished yet...but soon...

be well,
Jonah Fogel

PhD Candidate

University of Tennessee
Forestry, Wildlife, and Fisheries
274 Ellington Hall
Knoxville, TN, 37996
Posted by Jonah Fogel <jfogel@utk.edu>
posting date Thu, 13 Apr 2006 23:40:19 -0400