Should the police use race as one of the characteristics upon which they make decisions about stopping and searching motorists or pedestrians? (The question assumes, of course, that the police are not operating from a description of a specific individual believed to be involved in a crime.) Among those who have answered “no” to a question of this sort is our nation’s Attorney General. Others think that the practice is OK, as long as it is consistent with efficient policing: after all, you wouldn’t want to focus lots of law enforcement on groups that are rare offenders, such as elderly women. But is it right that a black driver on I-95 in Maryland in the late 1990s was five times more likely to be subject to a search than was a white driver?
Those who take the “efficient policing” position often say that the disproportionate number of stops is OK, as long as the probability that a searched motorist is carrying contraband (in the case of anti-drug enforcement, the aim of most of the highway searches) is about the same for blacks as for whites. (This probability is sometimes called the “hit rate.”) By this reckoning, if only 5 percent of the blacks who are searched are found to be carrying drugs, while 20 percent of the whites searched are carrying, then the racial disparities in searches are not consistent with efficient policing and should be curtailed, eliminated, or reversed. On the other hand, if the hit rate for searches is about 20 percent for both groups, then the use of race as an indicator might be acceptable.
But I and my co-author, Michael Alexeev of Indiana University, think that this standard “efficient policing” story is mistaken, for reasons that I will mention after the “Continue reading…” link.
Our paper (available from Mike’s website) argues, among other things, that comparative hit rates offer essentially no information as to whether racial profiling is “efficient” or not. Efficiency in policing is a tricky concept, and for most notions of what efficient policing would look like, hit rates just don’t matter. To quote from the paper itself — apologies in advance for the econ-ese — we “identify three mechanisms through which claims of efficient profiling can falter. First, efficiency in serving the goal of maximizing arrests does not imply efficiency in terms of minimizing illegal behavior. Second, efficient profiling by individual police officers can lead to socially excessive profiling in the aggregate. And third, to the extent that profiling reduces cooperation between the targeted group and the police, efficient profiling from the point of view of maximizing arrests will be inefficient in terms of maximizing convictions, or in minimizing illegal activity.” Hit rates that are similar among different races suggest that individual police officers are behaving efficiently with respect to the goal of maximizing arrests. But that is a far cry from representing socially efficient policing.
And finally, I’d like to publicly apologize to my long-suffering friend and co-author, Professor Alexeev, for sitting on the revisions to our Racial Profiling paper for, oh, approximately one year. And that is just the delay for the most recent round of revisions. But you know, Mike, I had to blog.
Comments are closed.