15

This article offers to complement the debate initiated by Dominique Boullier on the influence of big data in social sciences. While it does not question the idea that analysis of numerical phenomena may require the constitution of a new epistemological approach, the paper exposes a complementary point of view: big data also allow to articulate the different social science epistemological traditions, on condition, firstly, of combining new digital technologies and traditional methods, and secondly of mobilizing the canonical concepts of social sciences to analyze numerical phenomena.