Flutterby™! : Validiity of DNA matching?

Next unread comment / Catchup all unread comments User Account Info | Logout | XML/Pilot/etc versions | Long version (with comments) | Weblog archives | Site Map | | Browse Topics

Validiity of DNA matching?

2010-01-11 18:51:07.98264+00 by Dan Lyke 4 comments

Review of Arizona CODIS database throws doubt on DNA matching, FBI isn't allowing a review of their data:

The first clue that something might be amiss came in 2005, when limited data was released from the Arizona state database, a small part of CODIS. An analyst who compared every profile with every other profile in the database found that, of 65,493 profiles, 122 pairs of profiles matched at nine out of 13 loci and 20 pairs matched at 10 loci, while one pair matched at 11 loci and one more pair matched at 12 loci. "It surprised a lot of people," says signatory Bill Thompson of UCI. "It had been common for experts to testify that a nine-locus match is tantamount to a unique identification."

Kind of like fingerprints: We have years and years of assertions that matches are precise, but the science for claiming that is pretty shakey. Via SE.

[ related topics: Law Enforcement Databases ]

comments in descending chronological order (reverse):

#Comment Re: made: 2010-01-11 23:55:43.436543+00 by: m

Medley, Thank you for the reference. It supports my biases, and is thus most gratefully received. But please note that I also very much appreciate references that refute my preconceived beliefs as well.

Laboratory error, and blunder, as well as collection and storage misfeasance and malfeasance, undoubtedly contribute far larger divergences than the interpretation protocol does. Add in political and social pressures, and I would guestimate that the total error rate climbed to the order of 1%. Such information is rarely brought to a jury, and the dazzling numbers that are used probably help convict innocents that would otherwise be freed. I base my estimates on thirty years of laboratory experience and management, including cases that involved legal chain of custody in both human and other specimens.

#Comment Re: made: 2010-01-11 21:34:33.976929+00 by: Medley

Related: http://www.pnas.org/content/94/11/5497.full

#Comment Re: made: 2010-01-11 20:01:48.60813+00 by: Dan Lyke

Oh, and I linked to a similar story on the LA Times site back in '08. That was a match between two people of different skin color.

#Comment Re: made: 2010-01-11 19:31:03.948458+00 by: m

Yeah, well that is nothing new. I was always suspicious of the initial database that was used. In 1995 it was composed of about 2000 samples. At that time it was already clear that locus matching was highly dependent on ethnic origin. A tight match for one race might be much looser for another. Two thousand samples certainly could not have possibly made a reasonable inclusion of various ethnic and family enclaves. The odds of trillions and quadrillions to one were nothing more than putrefying cow manure. As so many investigators have found to their everlasting shame (if they have a conscience), population selection biases are subtle and ubiquitous.

Look at fingerprints. Believed to be unique. But one of the issues with fingerprints is the number of points that must be congruent to be declared a match. Brazil requires 30, but the numbers vary widely. Some jurisdictions call a match on 4 points. Remember the Oregon lawyer who was matched to being on the scene of the railroad bombings in Spain, even though the FBI couldn't show he was not in the US at the time. This occurred even though the Spanish police said the fingerprints were not a match. Lower level FBI fingerprint techs said there was not a match. They techs were not allowed to disagree with their superiors. Upper management at the FBI was emphatic that they would arrest this guy, and they did in willful disregard of the evidence.

The FBI has been known to promulgate a lot of junk science with flashy pictures, numbers and hoopla. One of the favorites was promoting bullet pour matches that relied on the "well known" mathematical hypothesis that if a<>b and b<>c, if a=c, then a=b which makes for a match. You would need a mathematician, scientist or engineer on the jury to dispute that rather simple piece of nonsense. And would the other jurors believe the truth sayer int their midst?

Think about how hard it is to explain the "Birthday Paradox" to the average person that might be found on a jury.