It seems that starting with the lie detector some time ago that the validity of many CSI style forensic methodologies like fiber analysis, bite analysis etc. look ever more questionable, and are being increasingly debunked when looked at with a critical scientific eye. Many people have been sentenced to long prison terms or the death chamber in some cases, using extremely questionable forensic assumptions.
The scientific method has been around for some time. Why are so many of the forensic methodologies used by law enforcement over the past few decades and assumed to be powerful and explanatory, actually so full of crap? People’s lives were at stake, why was no one looking into the validity of these claims, many of which amounted to litle more than law enforcement folklore?
Can you provide any examples other than the polygraph where a forensic technique that was widely accepted was shown not to work?
There have been some serious problems with scientific evidence in various jurisdictions, but these have more to do with incompetent and/or immoral personnel as opposed to the science itself being on a shaky foundation.
Also, are you sure the problem has been in the science, rather than in the public perception of the science?
For example, fingerprints have always had the possibility of a false positive, about one in a million. DNA tests are 100 times better (about 1 in a hundred million) but the perception of the average Joe on the streets is that fingerprints are an absolute match and DNA tests are kinda iffy.
As another example, most forensic tests such as blood type, fibers, dental imprints, etc. can rule out a suspect but cannot positively identify a suspect. The news won’t report it that way, though. The forensic scientists will say something like “Semen samples indicate a type AB perpetrator. Suspect 1 is type AB. Therefore, the evidence is consistent with the hypothesis that Suspect 1 is the killer.” The press will report it as “Blood type proves killer’s identity! Film at 11.”
In 2005, the FBI announced that they would stop conducting examinations of bullet lead in cases where a ballistic match couldn’t be obtained. The theory behind testing the recovered bullet to determine the composition of trace elements within it was that if they could determine the composition, they could match the bullet to a production batch from the manufacturer. Thus, if a suspect had a box of ammo from that batch, you could claim that the bullet recovered at the crime scene was consistent with ammunition that was in a defendant’s possession.
It should be obvious why this type of evidence isn’t very useful to a prosecution and could be unduly prejudicial to a defendant. Any number of people could been possession of a box of ammo from a particular batch, and there’s no guarantee that the dispersion of trace elements would be uniform throughout every individual bullet. The FBI spent a lot of money acquiring the equipment and personnel to make this type of analysis before abandoning it.
Note that the FBI asked the National Academy of Science, a nonprofit, to evaluate that type of analysis. There is no federal government organization that oversees or sets standards for forensic evidence use.
First off, many of those things were controversial in their heyday. Especially the lie detector.
Secondly, acceptance is just as social as it is scientific and law enforcement has a huge bias for things that will give them the results they want. Its probably not difficult for a department or a government to introduce some controversial method and fight off opposition by getting enough experts to testify that it works. Eventually, the abuses lead to allegations and lawsuits and those at the top pull the method.
This is also how people get away with things like the twinkie defense and other nonsense. Experts who will testify the way you want are picked by law enforcement, lawyers, etc for their own needs. Non-experts like juries, policy makers, judges, lawmakers, etc in many cases have no interest in being skeptical and may even be biased to accept some conclusions to look tough on crime or whatever.
I wasn’t aware that technique was ever used with law enforcement. All of the applications I’ve seen for it have been in archaeology (identifying bullets as being from the Lewis and Clark Expedition, for instance).
And bite analysis has at least some usefulness. Once when I was a kid, my sister bit herself and tried to claim that I did it. Mom was able to very quickly put the lie to that claim by comparing the marks to our mouth sizes.
Just a nit-pick on the so-called “Twinkie defense.” Most people do not understand the argument the defense made. They never said “My client killed someone because he ate a Twinkie.” What they said was “My client killed someone because he was mentally disturbed, as evidenced by his lack of sleep, mood changes and behavior changes. As evidence of behavior changes we note that he used to eat healthy foods and then started eating Twinkies.”
(This is why it’s taking the SD so long to fight ignorance… the news media are so busy creating new ignorance.)
http://www.news-medical.net/news/20090818/DNA-evidence-found-at-crime-scenes-can-easily-be-falsified.aspx
This information was recently made public. Even DNA isn’t unassailable, and is apparently easy to falsify if you’ve got access to a decent lab and some training. I’d be more concerned about police labs “enhancing” evidence to “help” the case. I don’t think the average person could use this to frame someone, but there are always those mad scientists.
I was immediately reminded of this article in the current New Yorker. The relevant information is on page 13 with the description of the Lime Street Fire (pdf), an experiment that showed that the assumptions of what evidence constitutes arson are seriously flawed.
As for why, in the case of arson, certain patterns of flame were assumed to indicate an accelerant had been used. The logic behind it sounded reasonable, so they were accepted by juries and by prosecutors.
The problem about Forensics is not the science, but what measures they set up. As you said, the chance of a FP with fingerprints is about one in a million. But if you accept only 5 markers as a “match” then it gets much higher.
Here ya go, this has been on my mind recently; no you don’t want to know the details, sometimes being an ER nurse sucks. My google-fu is weak, but I remember about 10-12 years back an article that the UK was conducting some kind of review of prior convictions that relied heavily on retinal hemorrhage evidence.
One thing to remember with many types of forensic evidence (especially the more “observational” forms of evidence, such as bite mark analysis, tire and shoe print analysis, and fingerprints) is that the evidence can give investigators useful information, even if it’s never used at trial. The problem arises when conclusions are drawn from the evidence that are either incorrect or are not very probative of a defendant’s guilt but end up being very impressive to a jury because the limitations of the evidence aren’t explained sufficiently to them and they trust in the expertise of an expert witness.
Because the police and the courts are not scientists. You have many forensics developed first in a science lab, then people who are not necessarily that well trained doing the tests.
These people soon become experts because of their constant exposure to this type of analysis. “I’ve done over ten thousands fiber analysis in my career, and to my expert opinion, that fiber matches the one from the defendant’s sports coat.” Who are you going to find to debunk someone like that?
Even fingerprints, the best established and most trustworthy type of forensic analysis has come under criticism. Many people are tied to a crime scene from partial or poor fingerprints. Usually an expert (aka the guy who looks at all the finger prints back in the police headquarters) will testify how, in his extensive experience, he can absolutely guarantee that this set of finger smudges came from the defendant.
I’ve also seen cases where certain experts are used by the prosecutors over and over again. For example, when I was in Texas a couple of decades ago, the same two psychiatrists were the experts in almost all death penalty cases because they were sure to testify that the defendant would murder again. Since they had so much experience testing suspects in death penalty cases, it was hard for the defense to show that the science behind this diagnosis was faulty. (Especially if the defense attorney was drunk or asleep at the time which seemed to be quite a common occurrence).
Over zealous prosecutors fishing for experts to testify against the accused combined with poor forensic training allow poor evidence in the courtroom.
As many people have noted in this thread, evidence doesn’t prove anything, even when it’s good evidence. It simply establishes a certain amount of likelihood. If there’s only a partial finger print such that any 1 of 100 people could be the guilty person, but only 2 people who had any particular reason to commit the crime, that’s pretty decent evidence. If the guilty party could very well be just about anyone in town, it’s not very good evidence even though it’s still marking out 99 out of every 100 people.
Criminal evidence is an odds game. You’re trying to get the cumulative odds of everything so low as to make it unreasonable for it to be anyone else. Any one piece of evidence doesn’t need to cut the odds by all that significant an amount to be useful.