Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Researchers commonly measure discrimination by comparing responses to multiple fictional applicants sent to the same vacancy. I find evidence that these applications interact. Using data from several existing experiments, I find that applicants randomly assigned to compete against higher quality applicant pools receive more callbacks. In the presence of such spillovers, many experiments confound discrimination against an individual's characteristics with employers’ responses to the composition of the applicant pool. Under one reasonable set of assumptions, adjusting for applicant pool composition increases measured discrimination by 30% on average. Avoiding experimental designs that stratify treatment assignment by vacancy can eliminate such confounding.