Einstein`s Effect on University Rankings
Transcription
Einstein`s Effect on University Rankings
Einstein’s Effect on University Rankings Lettre envoyée à la revue Science le 14 septembre 2007 Par Yves Gingras Directeur scientifique de l’Observatoire des sciences et des technologies Professeur au département d’histoire de l’Université du Québec à Montréal Titulaire de la Chaire de recherche du Canada en histoire et sociologie des sciences Membre du Centre interuniversitaire de recherche sur la science et la technologie The story on university rankings (Science, 24 August, p. 1026), was interesting in many ways, but the most important lesson to draw from the sordid debate between the Free University of Berlin and Humboldt University to appropriate the name of Einstein for themselves was missing: the very fact that the presence in the “indicator of quality” of a 1921 Nobel Prize could change the ranking of a university (85 years later!) by more than 100 places, is in itself the comic (or tragic) proof that this “indicator” in the “Shanghai Ranking” is bogus. The basic methodology of evaluation demands that the chosen indicator be 1) adequate to the object, which is not the case here since Einstein’s Nobel tells us nothing about the quality of these universities today; 2) homogeneous in its measure and 3) that its value does not change erratically because of a minute variation of one component of the indicator for organizations known to be very inertial and that cannot change rapidly in one or two years. It remains an open question for social psychologists to explain why university administrators care so much about the Shanghai Ranking, known to be difficult to reproduce (1), instead of using the more simple and homogenous (and often converging) indicators of citations, publications and research investments which provide sufficient information to get a fair idea of the place of an institution in the field of research. Adding apples and oranges to get a “synthetic fruit” cannot do justice to any of the ranked organizations, which are better measured along many separate dimensions. (1) R.V. Florian, Scientometrics, 72, 25, 2007. END