Evaluation et politiques : Y a-t-il de bons indicateurs pour la recherche ?

Evaluation through benchmarking processes is expanding quickly in firms as well as in national administrations, international organisations (or the construction of Europe). They cover a wide spectrum of domains, research as well economic, social or employment issues. Their basic tool is the use of indicators of performance (such as the global rate of employment to evaluate the performance of national employment policies). They draw their political legitimacy and acceptance from the fact that these are figures (hence a priori objective and non debatable). Issues are, in reality, much more complicated. Using indicators as a tool for governance is not only substituting technique for politics in the sake of political neutrality; it is at the same, even if often inadvertently, making politics through technical choice. By using some key European exemples, I put in contrast two conceptions for the building and use of indicators, an instrumental one (derived from New Public Management) and an ethical one (derived from the works of Amartya Sen). Then I try to apply these conceptions towards research evaluation and to sketch a pluralist approach to evaluation in terms of objectives, actors and methods.