In this study, it has been proposed to understand integration as the degree in which interaction
patterns among countries are biased. Integration has been measured by the difference between
the observed frequency matrix of interactions and the matrix that would have resulted from
random interactions. The main novelty of the approach holds that it takes into account both
the interactions between countries and the interactions within each country. In this way, the
measurement of integration among countries is adjusted for differences in the size of the
national systems.
The indicator has been applied to data on multiple address publications in the EU to analyse
the process of integration of the European science system. Using data on institutional
addresses in records of the Science Citation Index for the period 1993-2000, results show that
the process of European integration has indeed occurred. It is also found that the integration
process has not been the result of a falling bias of countries to collaborate nationally, but
solely the result of a falling bias in the choice of partner in European collaboration.
Furthermore, the size of a country correlates with the degree of integration of a country,
which indicates that larger countries have contributed most to the process of integration. The
latter result has been related to scale advantages arising from diversification in large countries
and network externalities stemming from language. These explanations are largely suggestive
and merit further theoretical and empirical elaboration.
From a policy evaluation perspective, the results suggest that European science policy has led
to a more evenly distribution of European partnerships, but has not led to a “substitution” of
national for European partnerships. This is not to say that European science policy has not
succeeded. On the contrary, for what concerns European collaboration the bias in partner
selection has steadily decreased. European funding, including the equability conditions
attached to it, can be expected to have a substantial effect on a more evenly collaboration. The
fact that the bias to collaborate nationally has not decreased may well reflect the effect of
national science policies aiming to increase national collaboration15 rather than the
ineffectiveness of European science policy as such. This is in line with the observation that
from a budgetary point of view, European science policy has not been successful: until now,
member states still account for 95 percent of expenditures on public civil R&D in the
European Union (Banchoff, 2002).
15 For example, the Dutch government has promoted the creation of national research schools in all
disciplines during this period.
16