Matched-pair field experiments provide the strongest evidence of hiring discrimination
Evidence
The most rigorous way in which to establish the occurrence of discrimination is to conduct a field experiment. In field experiments on discrimination, what the investigators do is to send matched applications from fictitious applicants for actual, real-life job vacancies. The CVs of the two fictitious applications indicate that the applicants have identical work-relevant skills and experience. A large number of these paired applications are sent out to a selection of randomly chosen vacancies and the researchers then compare the respective positive callback rates (for example, invitations to an interview) for applicants with ‘British names’ and those with ‘minority names’. A higher callback rate for fictitious applicants with ‘British names’ than for matched applicants with ‘minority names’ is pretty conclusive evidence of discrimination in recruitment. This is by far the best evidence available for discrimination on grounds of ethnic origins or faith, although it covers only the very first stage of the application process. — Runnymede Trust (ed. Farah Elahi & Omar Khan), Islamophobia, 278
Categorization
- DEFINE: Defines the field-experiment method as the gold standard for establishing discrimination empirically — a methodological anchor for how the chapter’s evidence works.
- XREF: Connects to the reader’s broader economics-of-discrimination literature (the same correspondence-test design used by Bertrand & Mullainathan and Milgrom’s law-merchant field). — connects to: Correspondence-testing field experiments best demonstrate discrimination but mainly cover ethnic rather than religious bias — A. Hussain, N. Meer, Framing Muslim Integration, 283; The sample design could separate race-based from religion-based discrimination experiences
Related
- Correspondence-testing field experiments best demonstrate discrimination but mainly cover ethnic rather than religious bias (similarity 0.92)
- The sample design could separate race-based from religion-based discrimination experiences (similarity 0.82)