QUANTITATIVE METHODS
7 Testing
Nicolas Jacquemet
Abstract
Correspondence testing is a quantitative method aimed at measuring discrimination. It involves sending fictitious applications in response to real offers (for example, job offers). By providing and objective measure of discriminatory behaviour, this method is very useful, from a prospective point of view, for designing anti-discrimination policies.
Keywords: Quantitative methods, correspondence, discrimination, anti-discrimination policies, experimental applications
I. How is this method useful for policy evaluation?
Discrimination refers to an unequal treatment on the basis of individual characteristics that should not be relevant to the decision to be made: favouring a male candidate with superior professional skills is not discriminatory; but rejecting a female candidate on the basis of the suspicion that her availability will be less than that of a male candidate with an equivalent profile is indeed discriminatory, because there is no reason to believe that this female candidate matches this stereotype. As such, discrimination is a major source of inequality. Even more than other types of inequality, discrimination is both very costly in economic terms, by depriving the economy of some of its talent; and persistent, because the anticipation of such inequalities of treatment might discourage the discriminated persons and lead them to make choices (of level and branch of education, of career path) which only amplify these initial inequalities.
Despite the importance of the issues at stake, the development of public policies aimed at combating discrimination suffers from a lack of diagnostic elements due to the great difficulty of measuring it. This is the objective of the ” correspondence testing” method.
II. What does this method consist of?
Although this method can be applied to many different sectors of economic activity (housing search, seasonal rentals, master’s students’ applications) and to many different sources of discrimination (religion, sexual preference, socio-economic background, place of residence, disability), this presentation focuses for simplicity on its application to the measurement of hiring discrimination based on gender and/or origin.
This method is designed to provide a measure of the success of different types of applicants according to their socio-demographic characteristics, while at the same neutralising the effect of the intrinsic quality of the applications. Each of these two aims has its own methodological implications.
Constitution of fictitious applications
The success of different types of candidates is observed through the use of artificial applications, sent in response to real job offers circulating on the labour market. The method combines three ingredients: identities, applications and job offers.
The socio-demographic characteristics whose effect is to be measured are conveyed by the identity of the applicant. In order to test both gender discrimi