71

The usual idea underlying the use of statistical models is that they are rigorous and objective, guaranteeing a higher reliability of results than qualitative analyses. Through the various lures associated with these models, we will show that statistical models are reductive and normative, that they never allow for the validation of a theory, and that they are part of a subjective and discursive approach. Without denying the interest of such models, our proposal will try to convince us that statistical modeling in the human sciences is part of a pragmatist epistemology.