This paper reports on an experiment concerning the social construction of statistical definitions, where the first census of Higher Education Institutions in Europe has been developed. It conceptualizes the construction of indicators as a social process of definitions and boundaries' negotiation, involving value judgments, social and political opinions, as well as practical interests and power strategies of actors. The paper exemplifies this process on three issues, namely the social demand for establishing a census, the controversy concerning the definition of a perimeter as well as the selection of indicators, and the nature of comparability judgments. We first conclude that the socio-political dimension has to be explicitly taken into account when designing statistical systems; second, that social scientists involved in this process need to openly recognize the conflicts around the definition of indicators; third, that the objectified and taken for granted status of indicators makes them a powerful instrument to influence policy decisions and, that indicator designers need to make their own value judgments and interests fully transparent.