Document Type


Date of Original Version



Objectives: Test hypotheses drawn from Smith and Alpert's social conditioning theory that explains biased policing as the result of implicit racial stereotypes. Distinguishing between frisks, external pat-downs, and probable cause searches, we hypothesize that (1) black drivers are more likely than white drivers to be frisked and searched; (2) racial disparity is greater in frisks than searches; (3) racial disparity in frisks, but not searches, is conditional upon the racial composition of the community; and (4) that drivers' race is not related to the productivity of searches. Methods: Data are all traffic stops made by the Rhode Island State Police in 2006, exclusive of those in which a search was mandatory. Multinomial and binary logistic regression is employed to estimate models of frisks, searches, search productivity and to test the conditional effect of community context. Results: Each of the four hypotheses is supported. Conclusion: Biased policing is largely the product of implicit stereotypes that are activated in contexts in which black drivers appear out of place and in police actions that require quick decisions providing little time to monitor cognitions. This insight has important implications for police training. Because of limitations in this study, additional research that distinguishes frisks and searches is needed.


M. Lilliana Gonzalez is from the Department of Computer Science and Statistics.

Leo Carroll is from the Department of Psychology.