We exploit a simple but realistic model of trooper behavior to design empirical tests that address the following two questions. Are police monolithic in their search behavior? Is racial proﬁling in motor vehicle searches motivated by troopers’ desire for eﬀective policing (statistical discrimination) or by their racial prejudice (racism)? Our tests require data sets with race information about both the motorists and troopers. When applied to vehicle stop and search data from Florida, our tests can soundly reject the null hypothesis that troopers of diﬀerent races are monolithic in their search behavior, but fail to reject the null hypothesis that none of the racial groups of troopers are racially prejudiced.
Anwar, Shamena and Fang, Hanming, "An Alternative Test of Racial Prejudice in Motor Vehicle Searches: Theory and Evidence" (2004). Cowles Foundation Discussion Papers. 1741.