Forensic Science

[BrilliantBeast]

I was watching previews of 60 minutes Sunday and one story was about the fingerprint matching and how the process was very unscientific. There was not clear agreement by experts as what to look for or process to apply matching fingerprints. I didn't get to see the piece, so I am not sure of the conclusion.

My questions is how much else of forensics is based on "junk" science that has not been clearly tested to determine chances of getting a false positive or false negative? Most TV shows or even investigative shows on DiscoveryChanel portray all this as 100% reliable.

We already know the lie detector is very questionable. What about things like ballistics and weapons matching is based upon real science rather than some guy in a white coat speculating theories about carpet fibers matching and such.

[Phaycops]

Oh my goodness, I was just wondering something similar myself with respect to Luminol. In fact, I came in here to start a thread on it! Luminol, for those who don't know, is a chemcical you spray around and it bonds to leftover blood proteins (even after cleaning), making them visible under a black light. Now my question is, other stuff, notably detergents and probably other household goods, luminesce under black light. Won't they continue to do so after the Luminol is sprayed, causing some confusion about splatter of bloodstains? Let's say, for instance, that you dropped your container of liquid detergent and a bit splashed out on your way to the laundrette, and you just wipe it up without water. Then you kill someone on the other side of the room and Luminol is used to find the cleaned blood splatters. Won't both splashes show up under the light? Isn't this, um, bad?

Other than that, I would have thought, though now I'm not so sure, that forensic science is pretty scientific. There is always a problem in trying to re-create something that happened in the past (believe me! I'm a geologist!). You can always find something to support your pet theory, but unfortunately, forensic science isn't exactly peer-reveiwed in the same way geology is (which catches mistakes and biased results, etc.). An interesting topic, that I'd really like to discuss.

[Yahweh]

[meaningless fact]Cat urine also glows under a ultraviolet light.[/meaningly fact]

DNA evidence can be destroyed if the clothes are drycleaned (for all future criminals out there, never dryclean your clothes... it gets awfully expensive).

[davefoc]

This is a big topic and I would like to write a detailed post with links but I've already exceeded my procrastination limit for the day and must go do real work. It would really be nice if we had a real expert that could add some comments.

Howerver, off the cuff list

1. Polygraphs - close to complete BS. Might have occasional value if target is convinced they work
2. Hand writing comparisons - highly suspect, recent requirements for real testing to validate have actually indicated that field is far less reliable than was assumed.
3. chemical analysis of bullets to determine which boxes they came from - long established technique promoted by FBI which is probably completely bogus

It also seems like there are other techniques that might have some validity to them but biased testimony by experts makes it difficult to sort out the truth.

The maggot testimony in the Westerfield trial comes to mind for this. My favorite moment was when defense expert didn't know the difference between a mean and a median. This was a guy who was paid about $30K by the defense to come in and spout BS for a couple of days.

Bight mark analysis seems to be subject to gross overuse. I think there was a case a few years ago where many convictions were thrown out because a state "expert" was wildly overstating the relibility of his findings.

As to your example of finger print matching, I have always been a little skeptical of this, but I think the actual results suggest that it's pretty good. The fact that it is possible to digitize a print and find a possible suspect out of millions of people seems like pretty good evidence to me that it works well. If those searches came up with a lot of hits for people that were absolutely innocent that would be good evidence that the technique was not as reliable as was commonly thought, but I don't think that happens. Maybe at the margins where it is necessary to make comparisons of prints in bad condition there is room to be a little skeptical, but I don't have a feel for how often that is the case.

[repairman]

I work in a crime lab and have first hand experience with alot of this.

Anybody that is testifying that the spots that they found with Luminol is blood based on the Luminol is giving false testimony. Luminol can give you an indication of where to look to find the blood but as has been pointed out luminol will react to almost any oxidizer. After finding spots that may be blood then they are tested with a presumptive test (phenothalein). This still is only an indication that the substance is blood. This will be collected and submitted to the Serology section where they will determin if the substance is blood (with DNA and other tests). As far as fingerprints go there has been lots of studies to try and find different fingers with the same fingerprint over the last 100 or so years (So far none have been found). It is true that there are no studies to determin the frequency of fingerprints, but how do you go about that if you cannot find any that match? As far as AFIS (Automated Fingerprint Identification System) the fingerprints are digitized and when a fingerprint is searched the system brings back several of the bes matches and the examiner then makes the determination if any of them are a match or not. so Identifing fingerprints is very dificult work and it gets very hard when the prints are only partial prints and when those partial prints are smudged or on an uneven surface the job becomes almost imposible. I certainly have alot of respect to the analysts here in my lab.
If the analyst has any doubts about the match they will not call the print a match. After they have worked a case it then goes to another analyst to be verified independently by them. If at this point the two analysts do not agree then the print will not be identified (It will be considered of no value) the case is then sent to the supervisor for one last review before any reports are sent out.

The rule here in the lab I work in is check, check, and recheck! The worst thing that could happen is not that the guilty go free but that an inocent person is convicted of a crime that they did not comit.