bias towards national collaboration persisted. 12 From this, we can conclude that only the
second hypothesis regarding the convergence of inter-national biases is confirmed, while no
evidence is found that the bias to collaborate nationally, has declined.
FIGURE 3 AROUND HERE
4.3 Country comparison
The integration values for each of the fifteen member states are plotted in Fig. 4 following
formula (6). Differences between countries are quite pronounced. In particular, that the degree
of integration is closely related to the size of a country. The three largest countries (UK,
Germany, France) have the lowest Ti-values indicating the highest degrees of integration
while smaller countries (Greece, Finland, Portugal, Ireland) have the highest Ti-values
indicating the lowest degrees of integration. 13
FIGURE 4 AROUND HERE
The strong correlation between the size of a country and its level of integration can be further
analysed by plotting the yearly qi -values with the corresponding yearly Ti -values of each
country (Fig. 5). Note again here that ‘size’ does not refer to more conventional measures
such as the number of researchers in a country, but to a country’s share in collaborations (qi )
as it is directly derived from the marginal totals of the collaboration matrix. The shape of the
function that best explains the scatter plot in Fig. 5 could is asymptotic rather than linear
suggesting that scale effects are marginally decreasing. The propensity to collaborate
internationally thus tends to rise with country size, but decreasingly so. Importantly, the
12 Given the largely unchanged positive Ti=j-values and the trend of the negative Ti j-values towards
zero, the falling trend in T-values Fig. 1 must be understood as resulting from a rising share of
international collaboration as a percentage of all collaborations. The share of international
collaborations ∑∑qij (i≠j) has indeed risen fom 0.137 to 0.161, while the share of intranational
collaboration ∑∑qij (i=j) has fallen from 0.863 to 0.839.
13 The result on scale effects is empirically not conflicting with Katz’s (2000) result that smaller
countries tend to engage more often in international collaboration, because Katz (2000) made use of an
indicator that did not relate the amount of international collaborations to the amount of national
collaborations.
13