5. Even though they represent a large share of the sample(s), Japanese centres do not drive the results significantly. To check robustness, the analysis was run without the Japanese CoEs and the results differed only slightly from the whole sample of CoEs. Whenever specific results obtained without Japanese CoEs were found to differ substantially (i.e. more than 15 percentage points) from the whole sample, the difference is noted in a footnote.
6. This problem is not unique to this project. In the context of the revision of the Frascati Manual(www.oecd.org/sti/inno/frascati-manual-revision.htm). efforts are under way at the OECD to improve the reliability and comparability of official R&D data on higher education expenditure on R&D (HERD). Measurement challenges are linked to the difficulty of separately identifying R&D from other activities in this sector.
7. The size of the research teams ranges from a minimum of 3 to a maximum of 460 researchers. Since data on employed staff were in some cases of uncertain quality. outliers (funding per researcher over USD 500 000) were eliminated from the computation of the average size of research teams. The optional field for comments revealed that providing exact data on staff was difficult in some cases. Some respondents noted that staff in the CoE are employed by the host institution(s) and/or partner(s). while in other cases the post-doctoral and/or PhD candidates were not solely attached to the CoE. Therefore. these data should be interpreted with care.
8. 2005 is used as a threshold for the age of the whole CoEs sample since it splits the sample into those whose age is above/below the average age in the data examined. CoEsSB are defined as young (old) if their age is below (above) the sample average of 9 years. The same applies for CoEsLB for which the average age is 6 years.
9. Within the category “technical sciences” the following areas are grouped: natural sciences. engineering and technology. medical sciences and health and agricultural sciences.
10. The questionnaire obtained information about the three most frequent activities for each CoE. Figure 3.5 shows the sum of all preferences for each option proposed in the survey as a percentage of the total.
11. These groups were not equally represented in terms of the number of responses and are based on qualitative responses.
12. This number reaches around 18 FTE researchers when Japanese CoEs are removed from the sample of CoEs.
13. Data are available for 161 CoEsLB and 44 CoEsSB samples.
14. These percentages are the sum of CoEs replying positively to both the first and second item in Figure 3.14.
15. This percentage reaches 90% when Japanese CoEs are removed from the main sample.
16. This percentage reaches 66% when Japanese CoEs are removed from the whole sample.
17. One respondent mentioned that interdisciplinary research is not yet adequately represented in the funding agency's evaluation/review boards and that it is easier for researchers to submit proposals for disciplinary rather than interdisciplinary research projects.
18. In some countries, the global financial crisis led to severe funding cuts, deteriorating working conditions for researchers (i.e. salary cuts) and few career prospects. One respondent mentioned that it seems almost certain that the current situation will lead to a significant fall in both research activity and quality in the short to medium term.
European Commission (2013), She Figures 2012: Gender in Research and Innovation, Luxembourg: Publications Office of the European Union.
OECD (2012), Transferable Skills Training for Researchers: Supporting Career Development and Research, OECD Publishing. doi: 10.1787/9789264179721-en
OECD (2010), Performance-based Funding for Public Research in Tertiary Education Institutions, OECD Publishing doi : 10.1787/9789264094611-en.