Sexual Assault, Standards of Evidence, and Type 1 vs Type 2 Error

Rape is a difficult crime to prove according to the standard of evidence used in U.S. courts of criminal law, which is that a perpetrator's guilt must be proven beyond a reasonable doubt. Chances are that in most cases of sexual assault, there will only be two witnesses, the rapist and the victim. Physical evidence of the crime will probably be scant. Whatever evidence does exist will fade with time, meaning that victims have a short window in which to report the crime and have even the ghost of a chance of something happening. Even if a victim reports the crime immediately and seeks medical attention, chances are that evidence will only be able to prove that the victim had sex recently. Maybe there will be signs that violence occurred (bruising, or lacerations), or perhaps evidence of a date-rape drug in their system, but probably not.

In these cases where evidence of the crime is scant, there remains only one piece of evidence that a crime occurred: the accusation of the victim. People often express outrage that rape victims are so aggressively interrogated about the credibility of their story, including being asked about things like what they were wearing and their sexual history. It's certainly right to point out that none of that has a bearing on whether or not a crime actually occurred. But the task at hand for law enforcement is to investigate whether a crime probably occurred, and they have to have rely on evidence. Because the victim's word is the only evidence in most cases, they have no choice but to try to evaluate the victim's credibility.

All of this makes it pretty unlikely for most rapes to lead to an arrest, much less a conviction, when the prosecution faces such a high evidentiary burden (beyond a reasonable doubt). Few rapes are reported to police, probably because victims understand the basic quandary they face. They can choose to have their reputation and credibility doubted and dragged through the mud, probably resulting in no punishment at all for their rapist, or they can save themselves the harassment and end up with the exact same result.

The giant mess that is the handling of sexual assault on college campuses is the result of recognizing that the above situation is really terrible.

It is awful how often rape occurs and how often rapists go unpunished. America's universities are typically out front of the culture on women's rights and are also compelled by the Federal Title IX laws to pro-actively promote sex equality on campuses. Universities also have always had sort of wink-and-nod relationships with local law enforcement, even sometimes having a completely separate police department. There's a long tradition of schools handling crimes "in-house," like underage drinking and drug use. All of these factors have come together to create the shadow system of quasi-judicial proceedings within universities for handling accusations of sexual assault.

The in-house proceedings that universities are starting to adopt, whether they admit it or not, are essentially trying to lessen the standard of evidence necessary to punish somebody accused of sexual assault. If the beyond a reasonable doubt standard worked, it would be sufficient to leave sexual assault investigations in the hands of local law enforcement. The clearest difference between official legal proceedings and these in-house proceedings is that the burden of proof can be decreased or even shifted to something more akin to the preponderance of evidence standard used in civil trials, or even a more loosey-goosey "more likely than not" standard.

While this shift may actually be an appropriate change for this sort of crime, it needs to be debated, because it is a fairly radical departure from the rules of evidence and due process that have evolved over hundreds of years of jurisprudence and precedent in the official court system. In essence, our legal system has made a very conscious choice to prefer what statisticians might call a type 2 error over a so-called type 1 error. That is, the system is heavily prejudiced in favor of guilty people going free rather than innocent people being convicted. What we're seeing in the outrage over sexual assault is, at root, widespread frustration over this preference for type 2 error.

It very well may be the case that we should shift the systematic preference for one type of error over the other based on the type of crime in question and how difficult it would be to prove. I think a reasonable case could be made for a system like that.

But that's not the system we have, and I think if we're going to make that move, we need to have some really smart legal experts in charge of crafting how such a policy would work. Instead, we have university employees with no particular legal expertise forming ad-hoc committees with ad-hoc procedures and policies, with very few of the due process rights afforded to the accused that we would see in our normal courts. The result is a giant mess.

I don't really have a good answer for how universities should handle sexual assault accusations, but amateurs trying to create completely new legal principles out of whole cloth on the fly seems like a really dangerous way to go.