Information Technology Reference
In-Depth Information
As a consequence of that, one can find various examples of models developed
that are either mainly theory- or evidence-driven. The earliest approaches to
implement ABSS are undeniably theory-driven, as that is how this research area
begun: by providing a computational alternative for testing -or demonstrating-
theories and hypotheses that could not be tested in real settings. In an attempt
to improve the status quo , the current trend in the ABSS community is rather
skewed towards the preference for implementing evidence-driven models. Due to
the nature of social phenomena, ABSS modellers confront the inherent subjec-
tiveness that accompanies the task of representing -unambiguously, as it has to
be implemented computationally- all relevant social structures and processes.
To code these into an ABSS, one shall adhere to a non-discursive and pre-
cise notation (i.e. a computer language). Experience acquired in using ABSS
development frameworks raise one's awareness that computational experiments
must be clearly designed, specified and correctly implemented so that the tested
hypotheses indeed shed light on the actual phenomena and not only on mod-
els themselves. Thus ABSS modellers are in a position which requires them to
ensure that observed simulation results are due to correct model specifications,
and not artefacts resulting from particular implementations or errors -such as
those discussed in [Galan and Izquierdo, 2005]. However, there are still various
open issues on how should evidence -particularly those qualitative in kind- be
used to inform the development of an ABSS with regards to a social phenomena.
A systematic way of incorporating qualitative data is put forward in this pa-
per, which is by means of integrating an adapted version of the Ethnographic
Decision Tree Modelling (henceforth EDTM) methodology [Gladwin, 1989] in an
evidence-driven approach to develop ABSS ( introduced in [Lucas, 2011]). This
methodological proposal is aimed at strengthening the cross-validation process
[Moss and Edmonds, 2005], from model design to implementation, by grounding
the representation of decision-making processes in a composite model of how indi-
viduals engaged in the same activity and context actually behave. The argument
is that developing an ABSS based on an EDTM promotes a more accountable
validation and detailed justification of how modelling decisions are made. That
is because the result of an EDTM is a hierarchical, tree-like structure that repre-
sents the branching and possible outcomes of the decision-making process based
on data collected directly from the decision-makers. The original EDTM method-
ology grounds the representation of decision-making solely on ethnographic data,
yet the discussed adaptation hereby furthers that by allowing the use of survey
data to account for when subjects are not directly accessible.
2 An Evidence-Driven Approach to Modelling (EDAM)
Both Within and beyond the ABSS, it is generally accepted that the avail-
able development platforms 2 are useful for building models that allow accurate
demonstration and testing of hypothetical scenarios. Notwithstanding the steady
2 Including the following popular ones, each providing their own particular features:
SWARM, NetLogo, Repast, CORMAS, Ascape, SeSAm, SDML, MASON and M4A.
Search WWH ::




Custom Search