Innocent man jailed for 8 months after DNA match

A basic understanding of statistics is sufficient to understand how this happens. Authorities have become obsessed with the idea of assembling giant DNA databanks of entire populations and then doing "cold hits" for unsolved crimes against the entire database.

The problem with this is that DNA testing is not 100% accurate, and when you test an entire population you're going to start getting false positives. The larger the population, the more false positives.

But then a jury may be told "there was a 100% DNA match" or that "There is a less than 1 in a billion chance that he is not the killer." And the wrong guy goes to prison.

Check out this other recent cold hit case. They took DNA samples from Occupy Wall Street protesters and then ran them against unsolved crimes files with no connection to the protesters just to see what they could get.

http://www.kens5.com/news/DNA-match-in-cold-case-murder-call...

So a protestor is arrested and charged for a 2004 murder based on this. Later found to be an error, but getting to that point assumes you can afford a good lawyer. There are lots of people getting charged for unsolved crimes they have no connection to. Not all of them can afford a lawyer and expert witnesses.

New York recently passed a law requiring DNA samples from all persons convicted of any crime, even misdemeanors such as loitering.

http://www.npr.org/2012/03/15/148692189/n-y-passes-dna-requi...

There is a nationwide push for mass DNA collections. Other jurisdictions are pushing to take DNA of people who are merely arrested, regardless of whether there is a later conviction.

http://chronicle.augusta.com/news/crime-courts/2012-09-01/so...

> South Carolina’s law enforcement agency will soon collect DNA samples from people when they’re arrested for a felony – rather than post-conviction – four years after legislators passed a law requiring the state’s DNA database to expand.

This push to states even comes from the federal level:

http://www.acluofnorthcarolina.org/?q=congress-votes-expand-...

http://thomas.loc.gov/cgi-bin/query/z?c111:H.R.4614:

> In a 357-32 vote, the House voted to offer cash incentives to states that start taking DNA upon arrest for certain crimes.

It even has it's own lobbying group: http://www.dnasaves.org/

> Every day innocent people needlessly become victims of violent crimes. Most of these are committed by repeat offenders. By passing state legislation that enables law enforcement to collect DNA from felony arrestees, at the same time as fingerprints, your state can catch criminals sooner

There you can see that 26 states now mandate collection on arrest. More than half of the 50 states.

Supreme Court Chief Justice John Roberts has ruled that DNA sampling on arrest, regardless of conviction, is permissible:

http://www.huffingtonpost.com/2012/07/18/dna-collection-mary...

The US has a nationwide database now that they do cold hits on, which is a statistically questionable process.

http://www.bioforensics.com/news/relatives_6-05.html

> Since the mid-1990s, the USA and the United Kingdom have maintained databases that use a series of such alleles to match DNA from unsolved crimes to known or suspected offenders. In the USA, states and the federal government keep DNA indexes of suspects and unsolved crimes, and share information through a computer system maintained by the FBI.

"Guilt by the Numbers: How fuzzy is the math that makes DNA evidence look so compelling to jurors?" by Edward Humes was an article that covered the problems here pretty well.

Original is gone, but here it is in cache: http://webcache.googleusercontent.com/search?q=cache:www.cal...

More information about the case discussed in that article and the problems with the cold hit methodology: http://articles.latimes.com/2008/may/04/local/me-dna4

There are also various articles by DNA cold-hit advocates who claim it is invalid and cold-hits are a good practice, for example this article:

http://lawprofessors.typepad.com/science_law/2009/04/taking-...

Doing blind searches of samples on large DNA databases in order to find cold hits gives you a very high probability of a false match. A match that the falsely accused defendant will be quite difficult to challenge since there is a general belief by the public that DNA matches on a few markers are irrefutable proof of identity.

DNA can more strongly connect a suspect who is already known through other information. Doing blind searches isn't science and it isn't justice.

http://www.bioforensics.com/articles/Legally%20Scientific%20...

> For instance, if a DNA test capable of distinguishing between 'unrelated' people with one million to one confidence was used to create a database of two million personal profiles from a population of 20 million potential suspects you could be pretty certain that most crime stain profiles run against it would produce at least one cold hit.

> You could also be more than 90% certain that it would be the wrong cold hit. You could further expect that around 80% of the personal profiles on the database would match at least one other on record from a different person.

> Problems like these had led the 1996 National Research Council publication "The Evaluation of Forensic Evidence" (NRC-II) to recommend that "When the suspect is found by a search of a DNA database, the random match probability should be multiplied by N, the number of persons in the database".

On one side we have an argument that there are severe statistical problems with this. On the other side we have those who claim that is false and that the chance of any DNA match being wrong are so small as to be virtually impossible. Yet every year the number of false accusations and convictions from cold hits grows, as we see in this latest case.

The lede paragraph in this interesting article raises the key question:

"Scientists, lawyers and politicians have raised new concerns over the quality of forensic evidence testing - so is the criminal justice system too reliant on lab tests without realising their limitations?"

I'm grateful for some of the specialized publications on a skeptical worldview, such as Skeptic magazine and Skeptical Inquirer magazine and some of the websites and books recommended in those magazines, for teaching over the last decade or more that even "scientific" approaches to questions that come up in court cases have often never been validated. I was surprised to discover that even the old technique of matching fingerprints is still the subject of a lot of recent scholarly studies

http://scholar.google.com/scholar?q=fingerprint+matching+val...

https://uchastings.edu/hlj/archive/vol59/Koehler_59-HLJ-1077...

and that there is NOT full validation of all the techniques that have long been used to show fingerprint matches between criminal case suspects and fingerprint impressions found at a crime scene. The article submitted here makes the important point even the latest gee-whiz technology, DNA typing, which is usually the key to solving crimes on television drama shows, has significant limitations. Sometimes DNA samples are poorly gathered, and sometimes DNA test results are ambiguous.

It takes a combination of multiple lines of evidence that are consistent with one another best to prove guilt beyond a reasonable doubt. Lawyers for criminal defendants do the right thing when they argue vigorously over every factual basis for the state's claim that an accused defendant is guilty. I was an appellate clerk once upon a time for a state supreme court that reviewed all first-degree murder convictions in the state. I had to read a lot of case files with gory crime scene photographs and lots of transcripts of witness testimony to events and expert witness testimony about the findings of scientific tests. Sometimes a case is supported by so many lines of evidence that no reasonable appellate judge could doubt the sufficiency of the evidence shown at trial. Other times a conviction at trial is reversed on appeal not because of factual innocence, but because of legal errors by the prosecution. But every once in a while, even after a conviction at trial under the standard of presuming innocence beyond a reasonable doubt, a very reasonable doubt that wasn't considered fully at trial is preserved in the court record and forms the basis of reversing the trial court judgment on appeal. No one line of evidence is foolproof, not even DNA.

> his DNA sample was on record after he had willingly given it to them as part of an investigation into a burglary at his mother's home some years earlier.

This is a worrying snippet. Many people in the UK think that DNA databases should include every citizen. There are often local campaigns of "have your DNA taken, get ruled off our suspect list" when there are murders or multiple rapes.

The police have pushed to include as many DNA profiles as possible.

They don't realise that false positives happen.

Keeping a DNA database of just the criminals means that you don't have to store information from 60 million people (approx current UK population), nor process unneeded data to find matches, nor worry so much about false positives.

The good human rights stuff drops out as a result of good data practices.

DNA testing and forensics in general is a complete scam. Most of it has already been proven to be pseudoscience or else is on its way to being proven so. And of the stuff that actually is at least potentially legit, like DNA testing, virtually all of the labs in the US are so poorly run that their results wouldn't even be admissible as evidence, let alone considered reliable, if more people actually knew what was going on there. If I remember correctly the last time the US government did an audit of its accredited forensics labs every single one failed, and the vast majority aren't even accredited to begin with.

Relevant links:

http://www.alternet.org/print/story/147613/has_the_most_comm...

http://www.newyorker.com/reporting/2009/09/07/090907fa_fact_...

http://www.nytimes.com/2011/11/27/magazine/dna-evidence-lake...

http://reason.com/archives/2011/02/21/the-mind-of-a-police-d...

http://www.erowid.org/freedom/police/police_article1.shtml

http://www.pbs.org/wgbh/pages/frontline/criminal-justice/rea...

> His DNA sample was on record after he had willingly given it to them as part of an investigation into a burglary at his mother's home some years earlier.

Lesson to everybody: Don't ever voluntarily give your fingerprints or DNA. Once they get into the government databases, you can (a) never retract them (and nor can you change your fingerprints or DNA), and (b) you can fall victim to false matches.

Here's a similar case of a false match -- in this case for fingerprints:

Portland-area lawyer Brandon Mayfield was arrested in May 2004 because his fingerprint matched one found on a bag of detonators near the train station in Madrid in the March 11 2004 bombing, which killed 191 people. But Spanish authorities said the fingerprints belonged to another man, an Algerian. A US federal court later threw out the case against Mayfield, and the FBI expressed regret for the "fingerprint-identification error". As a former Army officer, Mayfield's fingerprints would be on file with the government. A law enforcement official said the fingerprints were not on file because of any crime or as part of the government's terrorism databases.

Forensics is not my field, but genetics is. Restriction enzyme digestion and gel electrophoresis seems to be the technique that is still being used for forensic DNA matching.

To my mind, it would make a lot more sense to simply genotype a few hundred well-chosen (highly variable) SNPs on a standardized array. Since there are premade arrays that give you tens to hundreds of thousands of SNPs for a few hundred dollars, you should easily be able to unambiguously identify a single individual even if (non-monozygotic twin) family members are in the haystack.

Obviously this says nothing about guilt, but from the DNA matching perspective we can and should do better.

It is of course very unfortunate that this man had to spend 8 months in jail. There was no lab error in this case (the man was a cab driver with a rare skin flaking condition and his skin flakes stuck to the nail polish of a passenger who was later the victim of murder). But of course errors do happen, someone in this thread quotes a 1% to 2% error rate for DNA testing.

Buit to put it in context, according to the Innocence Project, in 25% of the cases when DNA testing is used by the FBI during the investigation, it results in the suspect being exonerated. That 25% is a large number of innocent people who would be charged and tried and a lot of them would be convicted if not for the existence of DNA testing.

Read this interview if you have a moment: http://www.slate.com/blogs/thewrongstuff/2010/08/17/reasonab... and then tell me whether you still think DNA testing is not a force for good.

Forensics is a total sham. http://lst.law.asu.edu/FS09/pdfs/Koehler4_3.pdf

The error rate of fingerprints is at least in the single digit %, but everything else is double-digits. Moreover, while DNA is theoretically extremely reliable, the error rate of DNa testing is dominated by lab error and is on the order of 1-2%. Yet, juries are generally not told this overall error rate, but rather just the theoretical probability of a false match (1 in trillions).

I was quite surprised when reading "The Drunkards Walk" when he talked about DNA evidence. The rates of false matches is really quite low (like one in tens of million chance if memory holds) as long as there are no errors in the lab. But there are. In fact, the odds of an error on the lab are on the order of one in 200. And juries aren't allowed to be instructed on that.

sounds like someone couldn't read a gel electrophoresis ? How long did they run the gel for ? Accuracy can be increased in a read with a longer gel ... Were they using some 4" gel ? There are tons of variables. And as stated in the article some of them aren't even running the right tests.