Révéler les composantes de la discrimination à partir des taux de rappel

In order to measure hiring discrimination, researchers carry out field experiments : several applications from fictitious individuals are sent in response to the same job offers, and the comparison of their callback rates by employers gives a measure of discrimination in access to an interview. This method reaches certain limits in terms of possible interpretation, when several explanatory mechanisms of discrimination are at work simultaneously. We propose a methodology that solves these identification problems. We apply this method to gender discrimination in three construction professions : Masonry, plumbing and electricity. In our sample, we find the existence of significant discrimination that could be explained differently in each profession.