3 RESEARCH APPROACH
3.2 Research Framework
3.2.2 FCM-based Scenario Development
The expert panels were used to develop the FCM-based scenarios for wind
energy. Members of the FCM EP were asked to provide their causal maps and
highlight the factors that may affect deployment of wind energy on a large scale
in Pakistan. Miles and Keenan recommend that the researcher should provide
background material to the participants of a scenario workshop so that they have
similar background information [211]. Therefore, introductory information was
provided to the experts so that they could understand the context of the research
along with instructions to develop causal maps. The agenda and the handout
documents of the FCM scenario workshop are attached as Appendix A to this
dissertation. The handout highlights the purpose of the workshop, focus
questions, detailed instructions for the construction of causal maps, and
examples of FCMs for the experts. The agenda and the handout documents
were provided to the participants four weeks prior to the workshop.
Individual FCMs were obtained from the FCM EP members prior to the
scenario workshop. The experts were asked to look into the social, technological,
economic, environmental, and political aspects while identifying the concepts
individual maps were combined into an integrated FCM by the researcher. In the
workshop, the purpose of the research, the basic principles of scenario planning,
and FCM-based scenarios were also explained to the members of the expert
panel. During the workshop, the participants reviewed and critiqued the
integrated FCM and highlighted the input vectors for creating FCM-based
scenarios.
For the development of the FCM, the following steps recommended in the
literature were followed [152, 201, 334]:
1. Identify and define the important factors:
a. Write down issues on Post-its; and
b. Cluster these issues as a map and discuss their importance.
2. Define the causal link between these factors:
a. Identify factors which are linked together;
b. Determine that the relationships is positive or negative; and
c. Define relative strength of the relationships by assigning causal
weights using a 5-point Likert-type scale, with values that range
from 1, representing a very weak causal link, to 5, representing a
very strong causal link.
3. Review and discuss the combined / integrated FCM:
a. The moderator individually obtains the FCMs from every expert
prior to the workshop and combines those into an integrated FCM.
a. Paste red dots on the most uncertain factors (5 red dots are
provided to every expert).
5. Identify plausible input vectors consisting of the most important factors
from the integrated FCM:
a. The moderator will provide a tabular worksheet highlighting the
critical scenario drivers at the top of the each column and indicating
the number of conceivable development variations of each scenario
driver; and
b. Combine the development variations into plausible strands (input
vectors) using markers of different colors.
Combining of Multiple FCMs
Multiple FCMs can be combined together to produce a joint effect and capture
the opinions of multiple experts together in one map for further analysis [163].
The combined FCM is considered more useful than an individual FCM because
the information is obtained from a multiplicity of sources [324]. After combining
the FCMs, the experts are asked again to review the integrated FCM and
highlight the most uncertain factors/concepts.
An example of combing multiple FCMs is shown in Figure 10, developed for a
pilot study project [11]. The central objective of this integrated FCM is to
investigate the factors that will cause the large scale deployment of wind energy
FCMs from seven experts and taking the average of the causal weights. The
concepts highlighted by a continuous boundary line are identified by all of the
experts (from concept 1 to concept 14 and concept 16), whereas the remaining
concepts highlighted by a dotted line are identified by two experts.
Taber and Siegel proposed a method for combining multiple FCMs. This
method computes the expert credibility weights based on the Hamming distance
between the inferences vectors obtained from the FCMs of various experts [324,
325]. The integrated FCM shown in Figure 10 is composed of 20 concepts,
where 15 concepts are identified by all of the experts and two experts identified
five additional concepts highlighted by a dotted line. It was found that the
credibility weight of the two experts who identified additional concepts is reduced
because they differed from the majority. The credibility weight of the five experts,
who proposed the same concepts, is 0.90. Whereas, the credibility weight of the
two experts who identified additional important concepts in their FCMs is 0.75.
Thus, this method estimates a lesser expert credibility weight for those experts
who differ and disagree from the majority. For scenario planning, it is critical to
collect diverse input from multiple experts and identify the weak signals that have
the potential to play a vital role in the future. Therefore, this approach is not a
suitable approach for combining FCMs for scenario planning.
In the other method, multiple FCMs are combined by taking the average of
the causal weights. This is another commonly employed technique and it has
152]. Both approaches presented in the literature for combining multiple FCMs
are explained in Appendix B.
Figure 10: Integrated causal map/FCM for deployment of wind energy
After forming the integrated FCM, the experts were asked to identify the most
uncertain and important scenario drivers (concepts), because it is important to
ranking the scenario drivers, the morphological analysis was used to develop the
input vectors.
FCM Simulation
The input vectors are used for conducting the simulation and generating the
FCM-based scenarios. Although, a few important scenario drivers are used to
form the input vectors; but in the FCM simulation all of the scenario drivers
(concepts) in the FCM model are considered for generating scenarios. It happens
because when a concept changes its state, it affects all concepts that are
causally dependent on it, and this process depends on the direction and strength
of the causal link [152]. The newly activated concepts may further influence other
concepts which they causally affect and this activation spreads in a non-linear
fashion in the FCM model until the system attains a stable state [152]. Due to the
meta-rules, it is also possible that in some cases several input vectors may lead
to the same final system state [156]. The FCM simulations can be used to
experiment with different input vectors and compare their outcome [156].
Therefore, it helps to deal with a complex situation and holistically evaluate all
concepts of interest. Moreover, despite using the input vectors consisting of the
critical scenario drivers, all drivers/concepts and their causal links in the FCM
model are considered during the development of FCM-based scenarios.
The FCM simulation is performed until the output vector is stabilized. It is
squashing function is applied after every multiplication as a threshold function to
the output vector. A simple binary squashing function is used which squeezes the
result of multiplication in the interval of (0, 1). For n number of concepts, the input
vector is 1 by n, the FCM adjacency matrix is n by n, and the output vector is 1
by n [324]. The new output vector is again multiplied with the FCM adjacency matrix and this process is repeated until the multiplication results in equilibrium
[152, 296, 324]. As a result, the system is settled down and stabilized, and then
new matrix multiplications will result in the same output state vector. Implications
of the FCM model are analyzed by clamping different concepts and the vector
and adjacency matrix multiplication procedure, to assess the effects of these
perturbations on the state of a model [2]. Thus, the FCM simulation process
provides a holistic overview and investigates the internal dynamics of the model.
FCM Validation
The literature recommends that every step should be validated through the
experts as an ongoing activity [156, 317]. Participation of the experts help to
address the validity and acceptability aspects of the model [133]. It is important to
accurately translate experts’ feedback in the FCM model [156] and weak
facilitation may lead to poor quality of the model [109]. Therefore, it is critical to
strictly follow the FCM modeling guidelines through a high quality process [155,
156]. The steps taken to ensure validity of the FCM model developed for this