Adaptive Correspondence Experiments
Correspondence experiments probe for discrimination by manipulating employer perceptions of applicant characteristics. We consider the gains from dynamically adapting the number and characteristics of fictitious applications to the sequence of employer responses received so far. Calibrating the employer callback process to data from a recent correspondence experiment by Nunley et al. (2015), we find it is possible to cut the number of applications required to detect a fixed number of discriminatory jobs roughly in half relative to the static benchmark design that sends the same number and mix of applications to all jobs. These gains are achieved primarily from abandoning jobs with very low callback probabilities and those that demonstrate a willingness to call black applicants.
We thank seminar participants at the 2021 American Economic Association annual meetings. A version of this paper is scheduled to appear in the May 2021 AEA Papers and Proceedings. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.