Scarier Numbers

Over at the Agitator, Radley Balko has a post he’s titled “Scary Numbers.”


In September 2004, Mark Warner, then Virginia’s governor, ordered a random audit of 31 old criminal cases after a vast trove of biological evidence was discovered lying around in old case files saved by state forensic serologists. The testing of those 31 samples led to the exonerations of two convicted rapists. Warner, embarrassed by the revelations, then ordered in late 2005 that every sample obtained between 1973 and 1988 be rechecked. It amounted to thousands of files . . .

It goes on to suggest something truly horrible:


UVA’s Garrett suspects that the error rate may actually be as high as 17 percent. As he discovered in his own research, Barbour’s conviction, based on the testimony of a single eyewitness, reflects the reality that of the first 250 people exonerated by DNA testing, a whopping 76 percent were misidentified by eyewitnesses.


Thank God for DNA, right?  Well, of course, because without DNA for use in subsequent exoneration, the number of innocent people convicted under this study would be zero.  There could be a stray somewhere along the line, where someone else confessed or some smoking gun piece of evidence came out of nowhere, but once a person is convicted, they are guilty.

The presumption of innocence is gone. Their every denial is jaundiced and suspect.  Guilty people obviously deny they did it. Happens all the time. But we know they’re guilty.

What DNA exonerations have done is exposed the legal system’s worst nightmare; that it fails to distinguish the guilty from the innocent with a frequency never before believed.   Justice Scalia once suggested that there was a wrongful conviction rate of .027%.  This, he thought, was acceptable, as no system is perfect.  The old “take one for the team” argument never persuaded me; it’s a great unless you or your loved one happens to be the innocent asked to suffer for the system’s flaws.

But this assumed rate was nonsense from the outset.  It was based on the handful of proven innocents, the beneficiaries of DNA exonerations by the Innocence Project, as a percentage of all people convicted.  In other words, Justice Scalia failed to extrapolate the percentage of wrongful convictions in that tiny class of cases where DNA evidence is involved to the general universe of convictions.

The reason is clear. If they aren’t proven innocent, they’re guilty. And if they’re guilty, the system works. Not perfectly, but with a wrongful conviction rate so minimal as to barely be worth a yawn from Justice Scalia.

Or, the same influences that produce wrongful convictions in this tiny sphere of DNA cases exist in almost every case that comes before the court. They are:



  • Mistaken eyewitness identification
  • Perjurious testimony (testilying)
  • False confession
  • Racial influence
  • Guilty pleas of convenience

The only difference is that there is no DNA evidence to be used to subsequently exonerate the defendants. They get convicted. They stay convicted. Everybody (except the defendants and their families) sleeps well at night. Job well done.

But this is a delusion. Not to blame the Innocence Project, or the miracle of DNA exonerations, but this has put an enormous emphasis on innocence as the primary concern. Certainly, we do not want to convict the innocent.  But when the only question asked is whether the defendant was innocent, we continue to feed the monster of wrongful convictions, overcharged convictions and the faults of the system.

If we knew in advance who was “really guilty,” this might make sense. Perhaps we would demand that prosecutors and police work harder, spend more time, look deeper, be more honest, when it came to those for whom everybody felt there was some doubt as to their guilt.  And there are a few defendants for whom such doubt exists, but prosecutors feel compelled to press the charges nonetheless.

Most of the time, however, prosecutors and police are absolutely certain they’ve got the bad guy. We see this even after the DNA exonerates the defendant, and they still insist he was guilty.  We see this every day when prosecutors inform us of how they’re absolutely certain the defendant is guilty, when we know otherwise. They take comfort in their certainty. There’s no comfort for the defendant against whom the evidence is overwhelming, except for the fact that he didn’t do it. This happens too.

The point is that the flaws of the system remain, whether the defendant is guilty or innocent, and doom both without discrimination.  The “tried and true” means of conviction are every bit as dangerous to the guilty as the innocent, because the system can’t tell them apart.

We know they are flawed. We know they produce the desired result of conviction, even though they are untrustworthy. They’ve been used forever, and the law is nothing without stability and precedent, even if we merely perpetuate the system’s flaws.

The problems are known.  DNA has produced the proof to support the issues we’ve long argued, but the next step, the extrapolation to the general population of the convicted, seems to never really make it into the discussion.  I would argue that it’s because we need to be able to sleep well at night, and judges, cops and prosecutors wouldn’t be able to do so knowing that they were responsible for sending innocent people to prison 17% of the time.

But the truth is, that would never happen, because they remain absolutely certain that they are never that wrong, and that all the people, save the .027%, are guilty.  And for the rest of us, those of us who aren’t being asked to take one for the team, we don’t seem sufficiently bothered by these flaws in our system to lose much sleep either.



3 thoughts on “Scarier Numbers

  1. REvers

    The only way to make this a socially popular issue is to somehow spin it into a “Do It For The Children” campaign. Or maybe link convictions of innocents to some sort of Muslim plot.

  2. Fabio Dias Moreira

    You have a point, but Scalia’s .027% is not *necessarily* inconsistent with Garrett’s 17%. The .027% rate is measured relative to the general population (whether innocent or guilty); the 17% is measured relative to convicts (whether true or false). Since the latter is a much smaller set than the former, the false convicts show up as a much, much larger proportion of the latter.

    This is a very common statistical fallacy, and other Supremes have fallen victim to it; see, for instance, Souter’s dissent to Illinois v. Caballes, where he claims that “the dog that alerts hundreds of times will be wrong dozens of times”, which would only be true if almost all cars carried drugs around, given the circa 10% false positive rates quoted beforehand in the same paragraph.

  3. SHG

    I thought I had made that point with regard to Scalia’s stat, but obviously not clearly enough. Thanks.

Comments are closed.