Improving Compliance in Experimental Studies of Discrimination

\noindent In experimental studies of discrimination, researchers often manipulate cues of social identities such as race to isolate their discriminatory effects, holding all else constant. However, these studies cannot distinguish between \textit{noncompliance}, in which respondents do not observe, acknowledge, or update their beliefs about the social identity being signaled, and \textit{nondiscrimination}, in which respondents update their beliefs based on the signal but do not discriminate. Focusing on experiments addressing racial discrimination, we find evidence from a pre-registered, nationally representative survey experiment that racialized name cues suffer from compliance as low as 33%, especially for Black treatments. Adding pre-tested racialized pictures and resume items improves compliance to 95% on average and reduces the variance in compliance across racial treatment groups. Our simulation studies show that this noncompliance tends to attenuate the estimated effects of race by as much as 85\%, implying that racial discrimination may be much deeper than decades of experiments have suggested.