Inspired by this BBC article and this thread, how are testing centres and their workers checked and audited?
The quality control processes for properly run labs involve an immense amount of effort. Gaining and maintaining accreditation is similar in principle to the way hospitals, etc get it. Regular inspections, audits, unannounced inspections, blind running of test samples, etc. There is extensive, audited documentation of everything, and double signing of important steps. Processing of samples is usually done in batches whose identity and significance is anonymised by number, and no one person runs the process from beginning to end.
No-one pretends mistakes are impossible,but typically a mistake will result in a false exclusion of a person of interest, not a false inclusion. Imagine the reference blood sample of a person suspected of a rape in town A being accidentally swapped with that of a person from town B, for example. The rapist from town A will be (albeit wrongly) excluded from being a match with the field samples taken from the victim of town A’s rape. People will notice that our two suspects are coming up as matches for cases in which they were not suspects and for which they will likely have ready alibis, and so on.
It is a badly run lab where DNA evidence can be deliberately manipulated by one person without detection.
Drug testing tends to be done differently. It is usually a less complex process, but again in a properly run lab there should be double signing, peer review, auditing and accreditation processes, etc.
Really? DNA fingerprints are stored, so you might get a false match after 20 years or more. Can you remember what you were doing 20 years ago?
And yet we see the two cases to which I linked.
OP asked about the processes. They’re pretty good when done properly, and systems are in place to multi check and error-detect. The capacity for human error can’t be entirely eliminated, but it can be radically reduced. I don’t understand what point you are making - that occasional cases of fraud or failure mean we should give up using the most powerful and well established tool in forensic science in decades? That we should improve things? Continuous improvement is built into the quality process. Any failure is one too many, but that doesn’t mean that DNA evidence is somehow unreliable, or that we should indulge paranoia.
Very true, but that reliability has got to be demonstrated. Consider the exchange below:
“So, Inspector Plod, for how long have Blandshire Constabulary been using MakeItUp DNA Testing?”
“10 years.”
“And in those 10 years, how many times has Blandshire Constabulary tested MakeItUp DNA Testing?”
“I don’t know.”
“So, basically, you’re taking their word for it, aren’t you? My Lord, I move that the case be dismissed due to the unreliability of the evidence.”
Which is why there are acreditations and standards for forensic labs… and the customers (or at least the law enforcement customers) will want the documentation to verify that the labs meet standards, for just such exchange as you imagined.
It is a common idea of those L&O soap coperas, that if a lab is shown to make a mistake, it throws all previous convictions into doubt.
Not exactly the same thing, but the same idea:
Once the confidence of the technical testimony is questionable, a lot of a case can fall apart.
As an example:
http://www.scc.ca/en/accreditation/laboratories/forensic
So plenty of others have considered this issue and come up with a way to ensure the labs are reliable.
Of course, any person can slack off, cheat, be sloppy, or bypass the system… the checks and standards suggest that pretty soon this will be caught.
(A friend of mine once worked as a student in an industrial plant for a large chemical company. One of his job was night shift, doing the batch samples every hour. he learned from the best; one night, the plant hit a problem and shut down at 2AM, and when the Engineers came in that morning, they found the shift had neatly prepared results every hour all night for samples they could not have taken. Take one sample at midnight, copy the results with slight variation, and have a nap. Busted!)