D’abord les données, ensuite la méthode? Big data et déterminisme en sciences sociales

While large quantities of data have long been used by social science researchers, for example survey questionnaires, the use of massive and heterogeneous digital data, or “big data” is more and more frequent. As theory is abandoned in the search for correlations, is this multitude of data promoting a new form of determinism? On the contrary, the history of the social sciences demonstrates that the increase in data available has led to a gradual phasing out of a determinist hypothesis inherited from the natural sciences, to the benefit of a methodological empowerment based on statistical modelling. In this context, this article demonstrates that the emphasis on the size of the big data is not necessarily indicative of a return to determinism; instead it tends to reveal the present mismatch between the characteristics of these massive quantities of data and the methods and infrastructures in the social sciences.