Police officers often claim to be searching motorists so as to deter crime rather than to catch as many criminals as possible. According to a new study by economists Jeff Dominitz and John Knowles , published in the November 2006 issue of the Economic Journal , if this is the case, unbiased police will focus their searches on groups that would be most deterred by the threat of higher search rates. What''s more, unbiased policing no longer implies that guilt rates of motorists will be equalised across racial groups.
Public outrage and legal retribution over racial profiling by police officers usually revolves around the apparent injustice implied by motor vehicle search rate statistics.
When the rate at which a motorist is singled out by police turns out to be far higher for black motorists than for whites, both the press and the courts see evidence of racist police practices.
But according to a recent line of economic research, the public may be barking up the wrong tree, statistically speaking: these studies argue that racially imbalanced search rates do not imply racist policing. A far better indicator of racism may be the fraction of searched motorists who police find carrying contraband, usually drugs and guns.
The study by Professors Dominitz and Knowles shows how this approach can be applied in cases where the police claim to be searching motorists in order to minimise crime.
The original work in this line – by Knowles, Nicola Persico and Petra Todd – argued that if the job of the police is to catch as many guilty motorists as possible, then racism against black people would be signalled by lower find rates when the police searched black motorists. Unbiased police would see that they could catch more guilty drivers by spending less time searching black motorists and searching more whites instead.
When these researchers examined the court-mandated collection of search statistics by Maryland state troopers, they found that although black people were three times as likely as white people to be searched, the find rates were the same for blacks and whites. They argued that this finding indicates the police were not racially biased, and that black motorists had a higher tendency to carry contraband. (This is because the threat of search deters motorists from carrying contraband; black motorists, if searched at a higher rate, would respond by carrying drugs or guns less often.)
Equality of find rates appears, however, to be the exception rather than the rule. More generally, find rates are substantially lower for minority motorists. Recent examples include Missouri and Florida, where state-wide hit rates on Hispanic motorists were about half that of white motorists; Minneapolis, where these searches were just over one-third as successful; and Pennsylvania, where these searches were less than twothirds as successful.
Less extreme but still substantial disparities are found when comparing searches of black motorists to those of white motorists in these jurisdictions. If police officers were indeed supposed to be maximising find rates, then these statistics would indicate widespread discrimination against black motorists.
In the new study, Dominitz and Knowles argue that in many cases the police claim to be searching motorists in order to deter crime, rather than to catch as many criminals as possible. In such cases, unbiased police focus their searches on the group that would be most deterred by the threat of higher search rates. Unbiased policing no longer implies that guilt rates of motorists will be equalised across groups, so the question becomes a matter of finding conditions under which a lower find rate for black motorists still indicates racism.
The analysis by Dominitz and Knowles reveals an interesting implication: the ''crimeminimisation''
defence hinges on how similar members of a given group are to each other in their tendency to be deterred from crime. If the police search black motorists at a higher rate, and these searches result in lower find rates, then racism can be inferred if the white motorists are more similar to each other than are black motorists.
The reason is that the more similar are members of a group, the more effective will be drug interdiction efforts targeted towards that group – so unbiased, crime-minimising police would never choose to search more intensely a group that is more disparate in terms of crime tendency unless that group was carrying contraband at a higher rate than the rest of the population.
Similarity can be measured using characteristics that predict crime but are not immediately visible to the police, such as income, marital status and employment.
While the new study does not take a stand on whether crime-minimisation is more prevalent than arrest-maximisation, the results imply that the new find-rate approach may be more widely applicable than the previous research suggests.
''Crime Minimisation and Racial Bias: What Can We Learn from Police Search Data?'' by Jeff Dominitz and John Knowles is published in the November 2006 issue of the Economic Journal.