Wednesday, July 23, 2014

Sex Offender

Rape is a crime.

Unfortunately, it's a crime unlike any other. Before I mansplain why, a brief review of hypothesis-testing errors.

First, a reminder of how to construct a null hypothesis. In the most basic form, start with a control group that receives no interventions, they're all business-as-usual. Then get yourself a treatment group, those subjects who are otherwise identical to the control group but for one change, the treatment of interest. Your null hypothesis should be that the treatment produces no effect. The aim of inquiry is to reject the null hypothesis.

This is where error comes in and why the results of hypothesis tests sound like they were written by a mental patient. We don't say "X causes Y", but rather something closer to "with 95% confidence, we reject the null hypothesis that X has no effect on Y". Similarly, we don't say "X doesn't cause Y", but rather "with 95% confidence, we cannot reject the null hypothesis that X has no effect on Y." The wordiness and the careful construction of the result statements are explicit nods to the limits of statistical inference. Hypothesis testing is still, for all its sophistication, a bit like rolling one of those fancy Dungeons and Dragons dice: every now and again, you roll a natural 20. In other words, sometimes we erroneously reject a true null hypothesis (Type I) or erroneously fail to reject a false null hypothesis (Type II). A type I error suggests that X causes Y even when it actually doesn't, and a type II error suggests that X doesn't cause Y even when it actually does.

What does this have to do with rape? Well, the null hypothesis is that X did not rape Y, and the job of the criminal justice system is to attempt to reject this null hypothesis. In the case of rape, this task is rife with epistemological confusion, at least for marginal cases. Consent, or at least a plebeian notion of consent, is murky, especially when one or more parties is intoxicated. Add to that the barbaric habits of victim-blaming and perpetrator-excusing (particularly when it comes to cases involving strong social class disparities) and it shouldn't be too surprising when conscientious folks want to work to rid the system of type II errors, ultimately with the aim of effectively deterring incident rates.

Hence the rise of sex offender registries. Hence the rise of minimum mandatory sentences. Hence the conflation of forcible rape with other sex offenses (such as consensual sex between people slightly on opposite sides of statutory bright lines or public nudity or snapchatting nudes). The purpose of all these measures is to raise the relative price of sexual deviancy in an effort to discourage the deviant behavior. It's plain vanilla ordinary law and economics, straight out of Becker or Epstein.

The costs, of course, are an increase in the incidence of type I errors, where kids end up permanently on sex offender registries. Thanks to typical precautionary heuristics, landing on one of those registries severely curtails someone's options. Offenders can't buy property in certain areas, there are lots of places that won't hire registrees, and forget registered fathers volunteering at school, or even picking their kids up from soccer practice.

Still, it's a matter of trade-offs. Maybe swapping one type of error for another is acceptable to the median voter. Maybe the concentrated costs and diffuse benefits of registries are good on net for communities. Maybe. Still, it's worth considering what institutional failures prevent rape from being handled the same basic way as, say, armed robbery. It is my impression that police aren't terribly inclined to say to a B&E victim "well, maybe you shouldn't have been flaunting your living room like that" or to an arson victim, "did you leave the house with a gasoline can in your garage this morning?"

Euvoluntary sexual encounters can be mutually felicitous. Coercive sexual encounters not so. In between are a wide galaxy of indeterminacy, and it seems reckless to assume that even a well-meaning legislature, far from the specifics of the thing can adequately anticipate the nature and details of such encounters, much less to accurately prescribe punishments that both fit the crime and are suitable to basic common-sense justice.

Much of the world (not just the US) has a nasty problem with how the crime of rape is handled. It seems an affront to justice to stymie progress by handicapping the evolution of the common law with restrictive ex ante legislative tinkering. Our euvoluntary future deserves better.

No comments:

Post a Comment

Do you have suggestions on where we could find more examples of this phenomenon?