Economic Geography

Frequently theory or models underlay the research
approach, and motivate the type of data sought
In contrast to the above point, where off-the shelf data were used to undertake
a particular type of analysis, it has also been common in my work to have a model
form the underpinning for a particular piece of analysis, and to then go gather
data to be used with this model. I’ve undertaken many economic impact studies
that are structured in this way, most of them using the Washington State
input–output model, or a reduced form of it for a sub-state region. A good exam-
ple of this type of work are the economic impact studies undertaken for ArtsFund,
an organization in Seattle that collects from corporate donors funds that are
passed to non-profit arts organizations in our region. ArtsFund has sponsored
three economic impact studies for King County arts organizations, based on the
years 1992, 1997, and 2003. Each of these studies has involved extensive survey
Approaching research methods in economic geography
189

---

research, including a major survey of patrons and a survey of arts organizations
(Beyers and GMA Research Corporation 2004). Patrons were asked about
spending in relation to their arts experience, but were also asked a number of
other questions, including open-ended qualitative questions regarding the role of
the arts in the community and to them personally. My role has been to help design
these studies, and to do the numerical analysis of the results of the surveys.
In the case of these economic impact studies there is a clear model that is being
‘fit’ through the gathering and use of particular data. However, in other cases one
has a sense of the ‘model’ that you are seeking to fit data to, but cannot be sure
about exactly how to represent the ‘model’. For example, one of the goals of the
producer service project referred to above was to evaluate the flexibility thesis in
the context of the producer services (Christopherson 1989; Gertler 1988). This
‘model’ of flexible production was not only debated, it was a ‘soft’ concept
compared to the input–output model that is a set of linear equations. We explored
various facets of the flexibility issue in our project, including changes in the mix of
full-time, part-time, and contractual workers (we found a modest increase in contin-
gent work), the way in which new jobs were approached, the use of outside special-
ists, collaboration, and the evolution of what services were offered (Beyers and
Lindahl 1999). In this work we had in the back of our minds the flexibility model
that was in the popular literature, but were providing a test of it guided by the partic-
ularities of our own research agenda. The point here is this: there are many differ-
ent types of models, which range from very precisely defined mathematical systems
to general frameworks that have some orderly properties, but are not codified
with rigid structures. In our research we need to be embracing these different
frameworks with data appropriate to the type of model we are developing.