Page 16 -

1 Introduction 9

1.3.2 Methodology Part

The next part on methodology consists of 11 chapters that aim to guide the
reader through the process of simulating complex social phenomena. It starts with
two approaches to designing and building simulation models: formal, i.e. using
approaches from computer science (Chap. 6), and informal (Chap. 5). The former
is more appropriate where the goals and specification of the proposed simulation
are known and fixed, while the latter is more appropriate in the case where possible
models are being explored, in other words when the simulation model one wants
cannot be specified in advance.
However carefully a modeller designs and constructs such models they are
complex entities, which are difficult to understand completely. The next (Chap. 7)
guides the reader through the ways in which a simulation model can be checked
to ensure that it conforms to the programmer’s intentions for it. Chapter 8 looks at
the importance of ontological structure for agent-based simulations, contrasting this
with approaches that have almost no apriori structure. It also takes one through
some of the ways of formalising and checking this structure.
Once one has a simulation model one is happy with, then one needs to decide
what runs of the model are needed to make one’s point. Chapter 11 tackles this
subject giving firm guidelines to ensure one has the right “power” that enables
the required distinctions to be made, but avoiding showing misleading levels of
significance.
Three chapters in this part are concerned with the results of simulations.
Chapter 9 concentrates on the validation of simulation models: the many ways
in which a model and the possible outputs from simulation runs can be related
to data as a check that it is correct for its purpose. Chapter 10 explores ways of
analysing and visualising simulation results, which is vital if the programmer or a
wider audience are to understand what is happening within complex simulations.
Chapter 14 looks at the broader question of the meaning and import of simulations,
in other words the philosophy of social simulation including what sort of theorising
they imply.
Two other chapters consider separate aspects but ones that will grow in impor-
tance over time. Chapter 12 looks at participatory approaches to simulation, that
is, ways of involving stakeholders more directly in the model specification and/or
development process. This is very different to an approach where the simulation
model is built by expert researchers who judge success by the correspondence with
data sets and can almost become an intervention within a social process rather
than a representation of it. Chapter 13 investigates how analytic approaches can
be combined with simulation approaches, both using analytics to approximate and
understand a simulation model and using simulation to test the assumptions within
an analytic model.