The Washington PostDemocracy Dies in Darkness

John Lott’s shaky response to death penalty critics

The recent botched execution of convicted Oklahoma murderer Clayton Lockett has inspired a new public debate of capital punishment. Last week I wrote a response to a defense of the death penalty from conservative commentator Matt Lewis. The latest defense of capital punishment comes from John R. Lott, Jr. who criticizes what he calls “the shaky case against the death penalty.” But his arguments are quite a bit less convincing than those put forth by Lewis.

Lott begins by criticizing death penalty opponents for using cherry-picked poll results to exaggerate their claim that public support for the death penalty is in decline. He’s right that support hasn’t dropped all that much, although I think one reason why the public continues to support state executions is that the states have gone out of their way to make them look as little like executions as possible.

Lott also argues that if public support for the death penalty has dropped, it’s likely because of the dramatic drop in violent crime in the United States. I made the same point last week. There’s no question that it’s easier to drum up support for criminal justice reform when the public isn’t as worried about crime. But whether or not the public supports the death penalty has little bearing on the morality or utility of the death penalty itself, or on how it’s carried out. There are lots of issues where large majorities of the public support policies that I, or you, or John Lott find immoral, impractical or just generally unsound. If it turns out that defendants in death penalty cases are being denied their constitutional rights, that too is not something we put up for a vote.

It’s when Lott starts citing statistics that his piece begins to fall apart. First, Lott attempts to debunk the notion that the death penalty is applied in a racially discriminatory manner by noting that white people are executed at a higher rate than the rate at which white people commit murders. That’s true. It’s also a rather facile way of looking at this particular aspect of the death penalty. Prosecutors seek death only in cases where they can show aggravating circumstances — the particularly heinous crimes. If you’re looking at race and execution statistics, then, it’s important to factor in the severity of the crime. And when a black defendant and a white defendant are convicted of murders with similar aggravating circumstances, the black defendant is significantly more likely to get the death penalty. In fact, a convicted black defendant is a better predictor of a death sentence than a conviction involving multiple stab wounds, or a murder committed in conjunction with another felony.

I think the far more troubling measure of the death penalty and race is the influence of the race of the victim. In most states, defendants convicted of killing white people are quite a bit more likely to be sentenced to death than defendants convicted of killing black people. That suggests that, whether intentionally or not, the criminal justice system puts a higher value on white lives than black lives. Moreover, studies that combine the race of both defendant and victim have found that a black defendant with a nonblack victim is by far the most likely to be executed, followed by a black defendant and black victim, a nonblack defendant and nonblack victim, and, far behind, a nonblack defendant and a black victim. These are percentages, by the way, not raw numbers. So they account for the fact that some of those categories have a larger sample size than others. (The Death Penalty Information Project has a summary of all of the aforementioned research, with footnotes, here.)

Admittedly, these studies are pretty limited. When you start breaking death penalty cases down like this for a study in a particular state or city, you drastically pare down the number of cases you’re studying.

But racial bias isn’t an argument against the death penalty so much as an argument against how it is applied. As I wrote last week, the strongest argument against the death penalty is that it’s irreversible, and given that it’s run by human beings, the criminal justice system will always make mistakes. DNA testing has shown that the criminal justice system is flawed — more flawed than most of the public had probably thought.

Lott tries to dismiss these concerns, and it’s here that his statistics really get screwy. Lott writes:

Nobody wants an innocent person convicted. The Innocence Project claims that, since 1989, 34 people convicted of any type of murder have been exonerated by DNA evidence; of these, 18 had been sentenced to death. In that same time, about 260,000 Americans have been convicted of murder, with DNA evidence being used in about 12,000, or 4.5 percent. The error rate then was less than 0.3 percent, and it is actually much lower than that, since many of the exonerations came from convictions that were made before 1989.

I don’t know where Lott gets the number 34. I can’t find it anywhere at the Innocence Project link he provides. The actual number of people convicted of murder who were later exonerated by DNA testing is 104, which you can find by doing a search of the Innocence Project database. (I also called the Innocence Project to confirm this number. They did. They also don’t know where Lott got the number 34.) Use the correct number, and the error rate approaches 0.9 percent. Which means that about 2,300 of those 260,000 people convicted of murder were innocent.

But even if Lott had used the correct number, there are problems with his formula. Lott links to this article, which states that DNA evidence was “a factor” in about 4.5 percent of murder cases. But that doesn’t mean it was the determinant factor in those cases. Isolate the cases in which DNA is determinative, and you’ll start to shrink that sample size of 12,000. Moreover, the error rate in cases for which DNA is available, determinant and testable will naturally be much lower than the error rate in cases for which DNA isn’t a factor. Consider, for example, the error rate in cases in which the prosecution can show that the semen collected from a raped and murdered victim is a DNA match to the defendant. Barring mishandled evidence or a straight up frame job, wrongful convictions in these cases will be close to nonexistent. Now consider cases in which the defendant was convicted based on, say, eyewitness testimony, ballistics testing or bitemark matching. That rate will be much higher. About 90 percent of murder cases don’t involve DNA, and thus governed by this higher rate of error. But prosecutors still seek the death penalty in many of those cases.

Lott also mistakenly assumes that we’ve already reviewed all of those 12,000 murder cases for which DNA evidence was relevant. We haven’t. We continue to find new exonorations from years past. There have been 84 new cases in the last five years (that figure includes all crimes, not just murder). Basically, the more resources groups like the Innocence Project get to review these cases, the more exonorations they’re able to produce.

So let’s review: In arriving at his proposed error rate of 0.3 percent, Lott divided the 12,000 murder convictions for which DNA was in some way a factor by the 34 cases for which he claims DNA testing produced an exoneration. But both his numerator and denominator are wrong, and both are wrong for a variety of reasons. His 12,000 is likely considerably less than 12,000. His 34 is actually 104, and even that figure continues to rise. He then mistakenly presumes the error rate in DNA-related cases will be similar to the error rate in cases for which DNA isn’t relevant.

So what is the actual error rate? A recent study by a group of researchers led by Samuel Gross, a law professor at the University of Michigan Law School conservatively estimated the figure at 4 percent for death penalty cases. They estimate that it’s higher for non-capital murder cases. As of last year, there were 3,088 people on death row. Even we go with the low-ball 0.9 estimate of innocence using Lott’s formula (with the correct 104 figure), that’s about 30 innocent people on death row. If we go with the 4 percent estimated by Gross and his colleagues (which, again, they say is a conservative estimate), we’re looking at over 120 innocent people condemned to death. Even Lott’s own 0.3 percent would represent 10 people.  Maybe he’s comfortable with 10 innocent people getting executed. I’m not. Especially with so many (mostly) GOP-led legislatures trying to limit post-conviction petitions and passing laws to speed up executions. (If we kill them quickly, we’ll be less likely to discover that they were innocent!)

As I wrote last week, it’s a mistake to look at DNA testing as a panacea. It isn’t relevant in the vast majority of criminal cases, so we can’t rely on it to catch our mistakes. It’s really more of a wake-up call. Death penalty supporters who say we can just sit back and rely on DNA testing to save us are putting forth a dangerous proposition. At some point, we’ll either have found all of the wrongful convictions that can be exposed by DNA testing, or the remaining wrongly convicted in those cases will all have died. From that point forward, we won’t hear about exonorations nearly as often. DNA testing will go a long way toward preventing wrongful convictions, but — and this is important — only in the small set of cases for which DNA testing is dispositive. Here’s the scary part:  If we haven’t fixed the problems with the criminal justice system by then, we’ll continue to have the same rate of wrongful convictions in non-DNA cases that we have today. But at that point, we’ll be much more likely to plod along with a false sense of confidence, because we’ll no longer have a transcendent technology to remind us that we sometimes get it wrong.