In the criminological field, reflexivity in statistics often appears as a methodological problem. Most of the problems emanate from questions of ‘measurement’and are mainly addressed in terms of ‘reliability’, ‘bias’, and ‘error’. There is however another way of discussing quantification in social sciences, but it is rarely used in criminology: the sociology of quantification that invites a particular form of reflexivity. The basic premise of this sociology is that the statistic is not a simple realistic measurement operation, a reflection of reality, but a temporary adaptation to new ‘ways of thinking about society and how to act on it’ (Desrosières 2014). The interpretative framework proposed by this sociology may be summarized in two lines: to analyse the convention underlying the quantification of the social; and, simultaneously, to observe the uses of statistics and networks of actors linked to it. Thus, a sociology of social quantification must lie at the interface of scientific research practices and public policy issues. To prove and to govern (Desrosières, Ibid.) can be seen as two sides of the same operation: quantification.