Thursday, June 26, 2008

How many innocents in prison: Exonerations make up 3% of Texas DNA case resolutions

It's hard to say if the data reflect a rate or percentage of wrongful convictions, but more than 3% of criminal cases solved by DNA since Texas began using the forensic technology have resulted in overturned convictions.

Somehow I'd missed the news announced earlier this year that Texas solved its 1,000th criminal case using DNA. OTOH, Texas has witnessed 33 exonerations of innocent men (mostly in sexual assault cases) using DNA evidence.

DNA exists in only about 10% of violent crimes, so the group is a pretty random sample compared to the larger criminal class. Might the rate of exonerations to convictions based on DNA give a potential wrongful conviction rate? There are a lot of factors going into both convictions and exonerations, so I'm not sure the comparison is entirely valid - too many variables. But at least it adds another data point toward the discussion about how many Texas prisoners may be actually innocent.

I'd noted earlier that death row exonerations occurred at a rate of 1.52% in Texas since the death penalty was reinstated in 1982. The percentage of DNA cases solved resulting in exonerations, however, doubles that number.

So how many innocent people are actually in Texas prisons? If it's 1.52% (the exoneration rate from death row), that would mean more than 2,300 innocent people are locked up in Texas for various crimes. If it's 3.3% (based on the DNA exonerations), the number would top 5,000.

By contrast, the lowest estimate I've seen for the rate at which innocents are convicted - the unlikely low figure of .027% cited by Antonin Scalia - would still mean more than 400 innocent people are locked up in Texas prisons.

Increasingly I'm coming to believe the number's a lot higher than most people working in the system would feel comfortable admitting.


Anonymous said...

Innocents in prison might be better counted by the number of actual trials and appeals. How about checking Writs of Habeas Corpus?

Another approach might be to examine plea deals, especially in misdemeanors. If someone only has one misdemeanor plea in their lifetime, how likely is it that they are really a "criminal"?

Yet another approach might be to examine how many arrests actually result in a trial, conviction on dismissal. I expect prosecutors dismiss a lot of cases because they didn't get it right the first time. Then they bring new charges and get a conviction. If they can't get it right the first time, how likely are they to get it right the second time?

Basically I'm suggesting that the answer to this question might come from a bottom up approach rather than looking in the prisons. Once they're in prison, they're guilty.............The answer is in the question, "how error prone is each step of the process to get someone into prison?".

Anonymous said...

I bet we average a Nifong for every two counties.

kbp said...

Though I have never researched on the topic, I've often been curious how the rate of cases exonerated compares when you look at the term of the sentence.

Who cares about exonerating any with less than 5. or maybe even 10, years to serve?

This relates a bit with what Anon 10:02 points out.

bfrederick said...

Off the top of my head, using your statistics from this post, and not counting the many wrongful convictions resulting from misdemeanor cases where people are unrepresented or pay fines to get things over with, if DNA exists in 10% of violent crimes, if 3.3% have been exonerated by DNA evidence, if the number were 5000 based on the 3.3% exonerated by DNA evidence, then the 5000 figure does not count the 90% of the cases in which DNA does not exist, and you can multiply by 10 to arrive at an average which includes the other 90% of cases, giving you 50,000 not 5000.


Gritsforbreakfast said...

bfrederick, that would be WAAAY overstating things. The DNA exonerations were only compared to DNA based convictions, not all convictions. My estimate used DNA cases as an essentially random sample, then applied the percentage (3.3%) to the whole. You wouldn't multiply by ten, in other words; the assumption is that the number of innocents found by DNA is comparable to what you'd find if you could accurately assess the number of innocents systemwide.

bfrederick said...

I see what you mean. 3% of criminal cases solved by DNA have resulted in overturned convictions; assuming that this could be translated to 3% of all convictions, you end up with a figure of approximately 5000.

But - convictions in cases solved with DNA should be significantly more reliable than every other type of case, and the number of wrongful convictions in cases that had no DNA available (ie convictions based primarily on eyewitness identifications, jailhouse snitches, confessions, but not DNA) would be significantly higher, even if not x 10.

And the numbers don't really take into account the pleas that result from fear of trial even though a defendant is innocent.

Seriously, numbers make my head hurt. It is impossible to know, but absolutely needs to be talked about.

Gritsforbreakfast said...

"convictions in cases solved with DNA should be significantly more reliable than every other type of case"

I agree with that, and it's why I suspect the class of DNA-solved cases might be a good control group to use for assessing the innocence rate. It's the category of cases about which we can be most certain about the outcome, so the percentage is arguably closer to accurate.

In other words, if we could assess innocence in other cases as accurately as in DNA cases, the argument goes, we'd find a similar percentage of innocents locked up in cases where there is no DNA evidence. That's the assumption behind the estimate, anyway.

If you read the post about the estimate derived from the death penalty, btw, you'll see I latched onto that methodology because, as with DNA cases, you've got a) a concrete and identifiable class of cases to use for the denominator and b) more scrutiny (because of habeas appeals) and therefore arguably more accuracy in identifying innocents. The DNA cases should be even more accurate in identifying the innocent, so it's perhaps unsurprising the percentage was a little higher. Best,

Anonymous said...

It is very tempting to try to calculate a quantitative estimate of exactly what percentage of prison inmates are innocent based on DNA exonerations. However I think there are simply too many unknown factors involved to do so with any reasonable precision at this point in time.

It is safe to say however that the continuing cavalcade of DNA exonerations is showing us that what used to be cynical estimates regarding the percentage of wrongful convictions may not be so unreasonable.

One question I don't hear being asked is "are all crimes equally prone to wrongful conviction?" Most of the DNA exonerations we have seen to date involved sexual assaults, often particularly violent assaults. My guess is that sex crimes in general, and sex crimes against children in particular, might be more likely to result in wrongful convictions.

Juries are supposed to weigh evidence in an objective manner, giving the presumption of innocence to the accused. Sex offences tend to produce a strong emotional and visceral response in jury members. Neil Vidmar at Duke University Law School has done some interesting research into the impartiality of juries in sex offence cases. He is not the first to note that it is often very difficult, or even impossible, to get an impartial jury , especially when the alleged assault is against a child.

Prosecutors know this well and often play to the fear of jurors with "put this monster away so he can't do this again". That is well and fine if there is sufficient evidence to support a conviction. But it can also sway jury members into making an emotional decision rather than a rational one, and convict and innocent person.

dudleysharp said...

“To be sure, 30 or 40 categorically innocent people have been released from death row . . . “. ‘ (2)

This is about 0.4% of those sent to death row.

This when death penalty opponents were claiming that 119 inmates had been released from death row because they were exonerated with evidence of innocene.

Based upon a standard of inmates released from death row with actual evidence of innocents, the number may be as low as 25.

Death penalty opponents never required actual innocence in order for cases to be added to their “exonerated” or “innocents” list.

They simply invented their own definitions for exonerated and innocent and deceptively shoe horned large numbers of inmates into those definitions - something easily discovered with fact checking.

There is no proof of an innocent executed in the US, at least since 1900.

If we accept that the best predictor of future performance is past performance, we can reasonable conclude that the DNA cases will be excluded prior to trial, and that for the next 7700 death sentences, that we will experience a 99.8% accuracy rate in actual guilt convictions, or 0.2% wrongful convictions of actually innocent people.

This improved accuracy rate does not include the many additional safeguards that have been added to the system, over and above DNA testing.

Gritsforbreakfast said...

Dudley, for a minute there I thought your comments were satire!

Re: Your statement that there's no definitive proof of an innocent executed, true enough. But the flip side is there's strong evidence that's PROBABLY happened, but officials responsible won't admit their mistakes. See e.g. the Cameron Willingham case, the Texas man who IMO is the most clearcut example of an executed innocent. (See also this professional peer review of the bad science in the case.) Before DNA exonerations, the same people who say conclusively there have been no executed innocents made similar claims about other types of cases. But the same techniques are used to convict death row residents as rape defendants. Why wouldn't the same errors that caused the conviction of DNA exonerees translate to capital crimes? I think they clearly do.

Either the number of innocents sent to death row is zero, or you don't get to argue that innocents couldn't be executed. Neither you nor I know for sure whether an innocent has been executed, but we know to a 100% absolute certainty that innocents get sent to death row from time to time.

Finally, DNA is unavailable in 90% of violent crimes, so why do you say it will reduce error rates to .2%? That's absurd. And what are these other improvements? We still use scientifically invalid lineup procedures, don't record interrogations, tolerate forensic science with 10+% error rates ... what in the world are you talking about?

@5:27 - the reason most DNA exoneration have been in rape cases is that there isn't DNA evidence in most cases, and it's not even collected for most non-violent crimes. If one lab hadn't (almost accidentally) kept a bunch of old rape kits in Dallas, none of those cases would have made.

To me the key is to look at the CAUSE of wrongful convictions proven by DNA - e.g., most stem from bad eyewitness IDs. That means all cases relying on that ID technique risk errors convicting an innocent (from non-violent theft to capital murder), and contrary to Dudley's claim, those problems have not been fixed.

dudleysharp said...

I already conceded that from 25-40 actual innocents may have been sentenced to death row since 1973, or from 0.3-.05% of those so sentenced.

I no of no one who says actual innocents are not found legally guilty for all sentences.

However, there is both an appeals and executiveclemency/commutations process after conviction.

I am aware of the Wiilingham case and have read the report. Damned worrisone. We'll have to wait a while for a full and final review.

I have never denied the risk of an actual innocent being executed, but do state there is no proof of such occurring within the US, at least since 1900.

However, as the due process protections are at a totally different level since 1973, that is normally were the current debate resides.

As you are very knowledgeable about criminal justice issues, I will ask you to consider.

Let's say the current evidence supports that we have sentenced 25-40 actual innocents to death row since 1973 and that the current evidence does not prove any actual innocent executed or that, so far, the evidence finds a 100% success rate in appeals/clemency/commutation in preventing actual innocents from being executed.

I specifically no not state that no actual innocents have not been executed. I won't.

But, from an accuracy standpoint, is there any other sanction anywhere in the world, that has a better record than 99.5-99.3% accuracy in a guilty finding for the actually guilty with a 100% record in releasing the actually innocent? Probably not.

In the future, it will be more accurate, because those numbers of inmates released from death row because of DNA exclusion, will not make it past the pre trial process in the future.

Furthermore, I think the evidence is solid that innocents are more at risk without the death penalty.

Please review the next post.