Measurement errors: A principal investigator-agent approach

This paper interprets the production of survey data as a labor market under asymmetric information. Measurement errors correspond to erroneous supply of observations by sample members. This paper discusses how to lower measurement errors using incentives that reward sample members for errorless supply and assesses the impact of such incentives on the sampling distributions of common estimators. When measurement errors are elastic with respect to such incentives, the latter can reduce bias even when information from the validation is not used and increase statistical efficiency when it is. We test our results on a survey of US physicians.