Racial Bias in Police Investigations

Law enforcement discrimination studies are usually qualified by selection concerns regarding which encounters transpire between individuals and police. This paper overcomes these problems by examining automobile crash investigations by a State Police Department. Because officers are dispatched to investigate crashes on the basis of factors unrelated to driver race, these interactions are effectively exogenous. I show that, conditional on the Census Block Group area of a crash, the race of the investigating officer is uncorrelated with that of the driver. For these investigations, I find that police officers exhibit significant bias in issuing citations to other-race drivers for both moving and nonmoving violations, though no bias is evident for felony violations such as hit-and-run. Because racial bias is present even for easily observable offenses such as expired vehicle registration, these findings are consistent with preference-based rather than statistical discrimination. JEL: J7, K42