Every now and then, CNN and other newsmedia run puff pieces on people who are exonerated of previously convicted crimes, usually through DNA evidence and the work of groups like Innocence Project.

In this one, however, I found something that disturbed me.

Sometimes an Innocence Project client is confirmed to be guilty by DNA evidence, but the group doesn’t make the number of those cases available. Theoretically, If key DNA material in a case is properly preserved, there’s no time limit on revisiting old cases, according to the Innocence Project.

This worries me slightly – I understand that their goal is a reform of the justice system. In many ways I sympathize with their cause, since psychological science has proven time and again that certain longstanding identification techniques (books of “known criminals”, badly arranged lineups) can easily be abused and give false information, and that memory fades and changes over time.

At the same time, the question this passage raises is, is Innocence Project wilfully exaggerating the extent of “wrongful” convictions for their purposes? What other purpose does hiding the record and ratio of guilty/innocent determinations by their DNA testing serve, except that it may come out that most of the people they test are in fact guilty, and that the justice system may be mostly working as it should?

And if they were getting a whole lot of exonerations, wouldn’t they be willing to say, perhaps, that “over 50%” or “over 75%” or even “over 25%” of the people they tested were innocent? Heck, 10% or even 1% (1 in 100) would be a not-inconsequential figure and better evidence for their cause. Instead, they only list “208 exonerated.”

There’s an old rhetorical fallacy from the baloney detection kit known as “Observational Selection” (aka “counting the hits but forgetting the misses”) and I submit today that Innocence Project, in presenting the statistic of “208 exonerated” on their webpage while refusing to tell us how many of their subjects are confirmed guilty, are very guilty of this.


Category: Courthouse

About the Author


2 Responses to Innocent or Not?

  1. Abel says:

    What’s the old saying? There are lies, damn lies, and statistics.

    I agree with you 100 percent on this.

  2. trumwill says:

    I guess I don’t see the harm. It would be helpful to know how often we get it right, but their mission focuses on when the authorities get it wrong. Say that they only exonerate 3% of the convicted; that’s very important for the 3%. In the case of the other 97%*, no harm done except some time and money. If it were a case that we were weighing two evils (guilty going free and innocents convicted) I’d be a lot more concerned about it. In this case, though, unless there is reason to believe that the DNA is exonerating guilty individuals, we’re weighing one evil (guilty going free) and one neutral (guilty staying in prison).

    I really think that the important statistic here is the 208 exonerations. Their goal is to prove innocence when applicable and prevent future cases of that happening with proposed legislation. It’s a worthy goal by any measure. So too are organizations that seek to make sure that police and prosecutors are given the lattitude to arrest and convict the guilty.

    In the case of the DNA tests, even if we only get it wrong a fraction of the time, I’m glad that there are organizations that seek the evidence to discover when that happens.

    * – I suspect the statistics are more muddled than that. There are probably three categories: innocence confirmed, guilt confirmed, and inconclusive, with probably the last of the three being quite common. It’s difficult to prove a negative, and there are likely more than a few cases where there is no DNA link to the crime but the conviction is upheld anyway as it is very difficult to overturn a conviction in many jurisdictions.

Leave a Reply

Your email address will not be published. Required fields are marked *

If you are interested in subscribing to new post notifications,
please enter your email address on this page.