Eric Rasmusen’s Weblog

A comment to an earlier post about stigma worried about false conviction, and the stigma that those who are convicted falsely would bear. It referred to the Innocence Project and cases like that of Michael Williams.

16-year-old Michael Williams was convicted of rape. A girl had been raped in her home by an intruder whom she identified as Michael Williams. Williams had just gotten out of jail, where he’d been sent after a fight with the same girl in which he’d gotten violent and broken a window. He was in love with her, and she had no interest in him. Other, lesser evidence supported him– for example, he did not own the kind of clothes described by the victim. He was convicted by a jury in less than an hour and sentenced to life in prison. In 2005, however, he was exonerated. The Innocence Project does DNA studies to look for false convictions, and in this case enough bodily fluids were preserved on cloth to show that Williams could not have been the intruder.

What cases like this show is that sometimes people are falsely convicted even when the evidence against them is very strong. The Innocence Project has exonerated 159 people this way– all by DNA evidence, suggesting that there must be many more innocents behind bars for other offenses, for which DNA testing isn’t applicable. What does this imply for how we should deal with possible criminals?

First, we have to remember that anecdotes, though powerful and good to have, can be very misleading. Suppose there are 1,000 men falsely in prison for rape. A site devoted to false convictions for rape says that in 1992 there were 158,000 men in prison for rape in the U.S. Thus, our current standards would lead to 1 false conviction for every 157 true convictions, not a bad rate. Blackstone famously said that it is worth letting ten guilty men go free to avoid punishing one innocent man. We have a much stricter standrd. In fact, the Statistical Abstract, table 289, says that in 2002 there were 95,000 rapes reported, and that number has stayed pretty level since 1980 varying from 91 to 107 thousand). Thus, over the past 20 years there have been about 2,000,000 rapes. The rate of false conviction per rape would then be 1 in 2,000, not a bad rate. Put differently: 1,999 rapists go free for every 1 falsely convicted.

Thus, despite cases like that of Williams, our big problem is not false convictions, but false acquittals (including people who are never brought to trial because the evidence is so weak).

Indeed, it might be that increasing the conviction rate would reduce the number of false convictions. Here is how that paradox would work. Suppose that for every rape, there is a 1 in 2000 chance that we will falsely convict someone. Then, we reduce our standard of proof drastically, so the false conviction rate rises to 1 in 1000. As a result, however, we raise the true conviction rate enough that rape falls by more than 50%– whether by deterrence, or, more likely, because the rapists are incapacitated by being put in prison. If under the old system, the number of rapes was 100,000 per year, now it falls to 40,000. Under the old system, the number of false convictions was 50 per year, but now it is 40 per per year. Requiring less evidence for conviction thereby reduces the number of false convictions.

It is nonetheless interesting to think about crimes for which it is hard to get enough evidence to avoid a large number of false convictions if we are to get more than a trivial number of true convictions. “Date rape” might be an example. If we require a unanimous jury to find guilty “beyond a reasonable doubt”, it would be very difficult to get any convictions at all. How often could a jury say that the act in question was involuntary beyond a reasonable doubt, with just the evidence of one woman against the evidence of one man? Thus, the current standard decriminalizes date rape, in effect, unless juries decide on their own that the reasonable doubt standard is too strong.

So what do we do? To use a lower standard would result in numerous false convictions.

I’ll have to think about this (and I welcome comments), but here are some immediate thoughts:

1. The victim can currently bring a civil suit for damages. She just has to prove the defendant’s culpability by “preponderance of evidence” (”more likely than not”) and by a bare majority of a jury, not a unanimous jury. The problem, however, is that it is costly to bring suit and if the defendant has no assets, he gets off free.

2. A defendant who loses a civil case for damages based on a criminal act should not have any bankruptcy protection for his house, car and so forth. I don’t know what current law says on this.

3. If the defendant is under 21, his parents should be liable for damages. I don’t know if this is the current rule.

4. Should we make his other relatives liable also, regardless of his age? Better that they bear the cost than that the victim be uncompensated.

5. When a criminal case is brought, the jury should unanimously find the defendant guilty beyond a reasonable doubt before he is sentenced. But if they do not, they should next see if a bare majority of the jury thinks he is guilty more likely than not. If they do, civil liability should result automatically, without the need for a separate trial. Also, their finding should be publicized, so stigma can operate.

6. Stigma is a very good penalty in cases like this. Stigma adjusts to the accuracy of the court system. If a court system has low standards for conviction, there will be less stigma from a conviction. If courts convict randomly, there would be no stigma at all.

7. If a defendant is liable for civil damages for criminal acts and cannot pay, the plaintiff should have the option of sending him to prison, at a rate of so many days for so many dollars (e.g., 1 day in prison for $1,000 in damages). If the plaintiff is subsequently exonerated, the defendant must repay him in dollars at the same rate.

8. In a civil case, punitive damages are currently awarded by a bare majority of jurors. I’m not sure of the evidence standard. Punitive damages should require the same standard as criminal punishment– unanimity, and liability beyond a reasonable doubt.

This entry was posted on Monday, May 16th, 2005 at 7:14 am and is filed under ResearchOld, Thoughts-New, Social Reg, Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.