Statistical Proofs and Demonstrations in Social Sciences: 20-21 Program

At a time when the experimental sciences question the limits of the practices and standards of classical statistics and of the frequentist paradigm, this debate is still not very present in the social sciences, in particular in France. However, statistical proof in these disciplines is characterized by certain peculiarities (non-experimental data, personal data, high collection costs, etc.) which make difficult the solutions usually recommended (pre-registration, publication of null results, replication, meta-analyzes , etc.).

The seminar Statistical Proofs and Demonstrations in Social Sciences aims to take stock of the state of thinking and their consequences for the practices of quantitative research in the social sciences, by examining in 2020-2021 alternative methods of analysis (statistics Bayesian, machine learning, so-called “narrative” methods) and their links to more “classic” analysis methods (networks and geometric analyzes).

The seminar is open to advanced students, mastering the basics of quantitative reasoning in the social sciences, as well as to doctoral students and teacher-researchers. s wishing to deepen their thinking and training in quantification.

Full program: test-statistique-en-sciences-sociales