4.4 Country-country comparison
The 225 individual Tij-values for each pair of countries following formula (2) are given in
Table 2. Each value in the table is the average over the eight yearly Tij-values in the period
1993-2000. This static representation still gives a fairly good idea of the yearly Tij -values
since no single of the 225 time-series shows a consistent falling or rising trend over time. Put
another way, though a clear integration pattern emerges from the collective of countries, the
country-county dynamics tend to fluctuate over time.
In Table 2 the highest values for Tij are indicated by bold values using an otherwise arbitrary
threshold of -1.00. As expected, using this threshold the strongest collaboration is found for
all intra-national collaborations (i=j) reflecting that all European countries strongly favour
national over European collaboration. The scale effect can now also be observed in greater
detail. Larger countries like France, Germany, Italy, and the UK have a smaller positive bias
values to collaborate nationally (ranging from 1.46 to 1.84) while smaller countries like
Greece, Ireland, Luxembourg, and Portugal, have higher positive bias values to collaborate
nationally (ranging from 4.22 to 6.16).
There are a number of other country-county values that exceed the threshold of -1.00:
Austria-Germany, France-Luxembourg, Belgium-Luxembourg, Finland-Sweden, Belgium-
The Netherlands, Germany-Luxembourg, Belgium-France, Ireland-UK, Belgium-Portugal,
Portugal-Spain, Denmark-Sweden, and Portugal-UK. These results indicate that relative high
propensities to collaborate with another country are very much organised along geographical
lines: high values are typically found for neighbouring states. Also note that in many cases the
countries that collaborate relatively often share a common or similar language. However,
from the data of the Science Citation Index it is not possible to analyse in detail the propensity
of researchers to collaborate with researchers that speak the same language. From the
information contained in publications records, the working language cannot be derived.
TABLE 2 AROUND HERE
5. Discussion
15