Wednesday, July 22, 2015

Resources on Texas and federal hair microscopy reviews

Today I compiled these resources for a colleague on Texas and federal reviews of flawed forensics in hair microscopy cases and thought there may be other Grits readers who're interested:
There's also a good discussion of the limitations of microscopic hair analysis with useful footnotes in this fat but useful reference book, which sits on a shelf above me but is also available online. And the problems with hair microscopy share themes with other non-scientific comparative forensic disciplines critiqued in Chapter 5 of this 2009 National Academies of Science treatise; here's a link directly to the subset of that discussion on hair analysis.


Anonymous said...

Wasn't ASCLD and TxDPS responsible for accrediting labs practicing this voodoo?

Chalk up another home-run for taxpayer-sponsored "accreditation"!

Where's the accountability? Do we get our money back, or do we get to shell out more cash for re-trials and wrongful conviction payouts?

No accountability. Right, Ralph "Bud" Keaton?

Gritsforbreakfast said...

To be fair, both those, particularly DPS, were charged with accreditation many years after the discipline's practice developed. I'm not a great fan of ASCLD-LAB, though I think the lab folks at DPS do a good job with what they've got. Regardless, it's unfair to criticize them harshly without simultaneously acknowledging that the move toward professionalization of forensics that they represent is part of why we've eventually arrived at a place where the most error-ridden disciplines are being re-evaluated.

Also, I don't think microscopic hair analysis is voodoo, per se, it's just not science (see this rule). More tradecraft. I grant it's a real skill, as is ballistics/toolmarks analysis (bite marks, not so much), but in the end it's based on a subjective judgment, albeit an informed one, whose error rate was frequently ignored or understated by its practitioners. The technique is fine for excluding people but the error rate is too high to "match" them the way DNA can.

Anonymous said...

I'm curious if anyone has performed a post-hoc DNA testing of alleged hair sample matches via mtDNA testing, either to test accurancy generally or challenge specific convictions.

The problem is that you can't get nuclear DNA from a hair shaft; nuclear DNA is what is commonly tested to match individuals with a high degree of certainty. mtDNA is much less useful because it is passed down the maternal bloodline unchanged, so a lot of people will match. In that sense, it's not good for positively identifying someone, but it can still be used to exclude a match.

There's a good explanation here:

According to the article, mtDNA testing can still exclude 99% or more of the population, so it seems like a worthwhile angle for convicted persons to request if forensic hair identification was a factor in a conviction.

Anonymous said...

I should have read the whole article ... mtDNA studies show a false positive rate of about 11% where an expert determined that hair matched. Also notes that it's been used as a basis for post-conviction relief where the defendant is excluded from being the source of hair, yet an expert testified that defendant matched hair at the crime scene.

Anonymous said...

I'm not sure I understand the flow chart question, "Was there a positive, probative association?"
Doesn't all hair comparison analysis connect the hair to either the victim or the defendant?
Or are they only looking for association to the defendant/convicted?

Gritsforbreakfast said...

9:18, I interpret that as saying "was the hair evidence used to help accuse the defendant?"

@1:11, also keep in mind hair evidence is usually an add on, something developed at the crime lab months after the fact when detectives have already decided who to accuse. Before DNA, it was almost never the primary accusing evidence. So some of the 89% "correct" ones could just happen because the cops accused the right person based on other evidence and they got lucky that reality matched their confirmation biases.

Harry Berry said...

Networking problems easier to solve and him I quote Open Flow give network managers programmatic control their networks using industry standards using the same distributed systems libraries and packaging’s be used PMP to orchestrate the rest that infrastructure both Martinique Oliver sharp guys so what is open flow did not show up-close a protocol for server software AK the control or to send instructions the Open Flow enable switches these.

Harry Berry said...

Positive commoditization automation support very difficult to get industry colleagues enterprise firms to disclose what their networking OPP X percentage is cap ex percentages are there VCP550 total atonal IT budget and not much in the public domain from places like Gartner forced her and others so in order to have a conversation around this made some conservative assumptions it's not hard to imagine some enterprises.