Information Technology Reference
In-Depth Information
-
framing effects (i.e. how the initial exposure to the phenomenon and presen-
tation of the decision-making process can influence the process of modelling);
-
the criteria for choosing subjects, design of data collection and techniques for
data analysis (i.e. how are the instruments of surveys and interviews going to
be actually administered in the field where the decision-makers are located);
-
and testing of the EDTM, which includes the qualitative administration of
the hierarchical model to decision-makers and implementation in an ABSS.
These are important because the model assumes the creation of a tree-like, hier-
archical structure to represent decision-making that is culturally-tuned by sur-
veying a specific groups of individuals [Gladwin, 1989] [Beck, 2000]. That is the
reason for arguing that an EDTM can be systematically tested via an ABSS,
where agents are equipped with the behavioural rules. As the EDTM is based
on qualitative and quantitative evidence, modellers can use this resource to sys-
tematically justify implementation decisions that would otherwise remain unac-
counted. Thus the main contribution by guiding the development of ABSS with
an EDTM is the integration of a methodology that better equip modellers to
deal with qualitative data. Due to the hierarchical nature of an EDTM, it can
be straightforwardly implemented in an ABSS as a series of IF-THENS or using
a system of inference techniques -such as forward chaining in JESS [Doorenbos,
2001] and backward chaining in Prolog [Roy and Haridi, 2004].
4FinlRemrks
This paper is an introduction to the integration of the EDTM qualitative method-
ology, designed to elicit what stakeholders actually do when taking decisions, to
facilitate the bottom-up implementation of behavioural rules in an ABSS. The
result of applying this approach includes a culturally grounded database, con-
taining richly descriptive qualitative and quantitative data, about observations
and reported behaviour of individuals engaged in the same activity and context.
This in turn serves both for building an EDTM and for improving validation of
evidence-driven ABSS. This is because modellers can use EDTM to guide and
justify the implementation of behavioural rules, whilst still being able to use the
model as a baseline for testing. Discussing a complete example of this develop-
ment approach would have been beyond the scope of this paper; yet for an ABSS
that has been built based on the EDTM, refer to [Lucas and Edmonds, 2013].
Acknowledgments.
I would like to express my gratitude for the useful discus-
sions held with Bruce Edmonds and Cathy Urquhart; plus institutional support
from the University College Dublin and the Maastricht School of Management. I
would also like to thanks the comments and suggestions from the two anonymous
reviewers, which are being taken into consideration to develop this paper.
