Wednesday, August 20, 2014
I've recently been working on an article on preventive law enforcement methods. By preventive law enforcement methods, I mean any kind of pre-crime intervention that has the goal of preventing crime before it actually occurs. A good example is law enforcers interrupting a larceny while the attempt is ongoing. Another common example is a simple Terry stop.
While thinking about these methods, I thought that one of the most important costs that come with the use of such methods is increased wrongful conviction or wrongful intervention costs. I collectively refer to these and similar errors as type-I errors. It seemed quite intuitive to me that the use of preventive methods would increase the frequency of type-I errors. This is because, convictions for interrupted attempts, or interventions due to suspicions, require less evidence than convictions for completed crimes. Therefore, preventive methods, compared to non-preventive methods, are likely to increase type-I errors.
A friend, after I ran my ideas by him, told me that this need not be true: preventive enforcement also reduces the amount of evidence available, therefore, such methods can reduce type-I errors. After thinking about his comment, I agreed that as a theoretical matter the 'evidence reduction effect' can be greater than the 'reduction in the amount of evidence required'.
However, as an empirical matter, I still believe that the more frequent use of preventive methods is very likely to increase type-I errors for two primary reasons. First, preventive methods increase the number of people who are candidates for being subject to type-I errors: everyone can be subjected to a Terry stop, the same is not true for post-crime investigations. Moreover, the unfortunate presence and frequency of law enforcer misconduct suggests that allowing preventive methods to be used liberally is likely to increase the instances in which police officer can misuse their powers.
I am unable to locate conclusive data for my assumption. However, the numbers quoted in Floyd v. City of New York appear quite convincing:
-4.4 million Terry stops were conducted between 2004-2012 by the NYPD.
-By 2009, for 36% of the stops the officer could not state what specific crime he was suspecting
-94% of all stops did not result in an arrest.
-52% of them were followed by a protective frisk for weapons. 1.5% of the frisks resulted in the finding of a weapon.
-About 350,000 stops led to a search into the person's clothing, based on the officer's belief that the person was concealing a weapon, or an object immediately perceived to be contraband. 91% of the time, the object was not a weapon. 86% of the time the object was not contraband.
I am not aware of any study that compares error rates associated with preventive vs. non-preventive enforcement methods. But, are we safe to assume that preventive law enforcement methods increase type-I errors?
This is of course not to suggest that we should only focus on costs, and not focus at all on benefits. Such methods surely result in benefits by preventing harm before it happens. In fact, are these benefits not, at least theoretically, exactly what we should be balancing against type-I error costs in figuring out how often preventive enforcement methods ought to be used?