are applied, the precise mix of methodologies depending on the issue at hand. The use of these
methodologies more or less follows from the nature of theorising. For example, cultural geography is
firmly based in postmodern sociology and anthropology, and from its theoretical premises follows a
methodological approach based on qualitative fieldwork methods or textual analysis. However, in some
cases the core concepts of theoretical contributions turn out to be very hard to operationalise also in
qualitative research designs. For example, the notion of ‘institutional thickness’ (Amin and Thrift, 1992;
Keeble et al. 1999) has been highly influential as a concept in economic geography, but has also been
heavily criticised for being vague concept, which can not be accurately measured, let alone that its impact
on regional development can be determined and tested (Markusen, 1999; Taylor, 2004). More generally,
according to some criticasters, empirical institutionalist approaches show ‘a lack of rigour, lack of
hypothesis testing and ill-defined concepts’ (Martin, 2003, p. 36).2 The contributions of institutional
approaches in economic geography have thus been, primarily and most importantly, theoretical by
suggesting new explanations and mechanisms underlying regional development and the role of
institutions, and in terms of policy, by opening up new discourses, e.g., on the cultural meaning of places
and heritage, and the limited transferability of locally rooted economic production (e.g., Gertler 1997).
It is also important to note that the use of qualitative research methods does not follow from
theoretical premises in all instances. For instance, recent network approaches in institutional economic
geography, including the embeddedness literature and relational economic geography, can, in principle,
very well use quantitative empirical network techniques from quantitative sociology (Burt 1982), and
formal modelling techniques from graph theory (Albert and Barabasi 1999; Barabasi 2002).3 However, in
their programmatic contribution on relational economic geography, Boggs and Rantisi (2003) argued that
‘being relational in practice’ implies, as a rule, a case study approach. Thus, some authors seem to have a
priori objections to the use of quantitative tools, even if theoretical contributions allows for their fruitful
application. It this context, Markusen’s (2003) reply to a number of institutional economic geographers, in
which she pleas to go beyond the oppositional distinction between inductive or deductive research, or
between qualitative or quantitative research, is worth mentioning.
In evolutionary economic geography, quantitative analyses are more common. Taking firms as the
unit of analysis, a natural approach to economic geography is to adopt demographic techniques describing
the entry and exit of firms, and how survival depends on age, location and ancestry (parent company). A
recent example of econometric testing of evolutionary models has been Klepper’s (2002a) study of the
American car industry. In this way, one can assess whether spatial concentration is solely the result of an
endogenous spin-off dynamic, or whether regional differences explain part of the variation in firm’s
success as well. A second methodology, more appropriate to capture spillovers as a source of
agglomeration, is based on graph theory and network analysis. Recent advances in these fields have made
it possible to discriminate between social network distance and pure geographical distance as determinants
of knowledge spillovers (Breschi and Lissoni 2002). By doing so, the black box of knowledge spillovers is
opened, and understanding is gained to what extent spillovers truly emerge from proximity per se, and to
what extent spillovers are bounded by social networks (club goods). Again, one is interested in whether
2 Though cultural studies have become well developed and established in sociology and anthropology, some have
suggested that it still suffers from ‘conceptual imprecision, theoretical ambiguity and empirical open-endedness’ in
economic geography Martin and Sunley (2000, p. 10).
3 An example study in this vein is a recent study by Baum et al. (2003) on networking between Canadian investment
banks using network analysis.
19