Daubert case?

[Outsider]

There is only a problem if there is a discrepancy between expectations and reality. The service that you, as a profession, say that you provide is reporting individualisation, exclusion and inconclusive. This means that an LPE would report as an exclusion the inked prints of anyone in the world who did not deposit the latent. This includes the person who has the fingerprint which is most like the person who did. You also say that a competent examiner will do this without error. You have put the “service specification” in the public domain so this creates the expectations of the police, prosecutor, judges and, possibly, juries.

It seems reasonable that a court, particularly in a capital case, will want to put the onus on the experts and their profession to demonstrate that they deliver what they say they do. If we cannot be sure that a match represents true individualisation then some sort of statistical thinking has to be done by somebody (not a lot, just a bit of fully informed common sense based on realistic expectations).

This was not a single identification to the defendant. This was not a suspect developed by AFIS, rather through police intelligence AND THEN several id's were made. There is other evidence in the case to associate the defendant. There even was a level of blind testing done later by an external agency as well....

It seems to me that this puts this case right at the opposite end of the continuum from the case that has taken my interest.

[Thomas Taylor]

It seems to me that the big fuss lately has been "examiner bias." We are supposed by the courts to be biased in favor of the police or other fingerprint examiners who have made an identification that we are asked to verify. If "bias" is a human failing, surely we are potentially influenced by it.

Is the judge human? Is she subject to bias, too? I think maybe she has a defense bias and has succumbed to the mantra of Cole, Huber, Starrs, and the rest, who wouldn't recognize a fingerprint if one poked them in the eye. Should it surprise us if this judge has a defense bias and some other judge has a prosecution bias? After all, judges are human, too.

But what I really think is that it is too early to be making broad assumptions about the future of fingerprint identification based on this one errant ruling. Even if her decision cannot be appealed in Maryland, it has not become the "Law of the Land."

[g.]

Lisa,

Speaking as a defense attorney, I have deep concerns about admitting evidence "for what its worth" and trusting to cross-examination, closing, and maybe jury instructions to guide the jurors.

I understand your point, I didn't want to intrude on the strategies of those whose case this is. I was referring to the fact that as a compromise, one that I am NOT a fan and think that there would be greater error and harm in the long run, but at least puts the evidence in front of the jury, is to do what I affectionately refer to as "neutered testimony". This is what Pollack first suggested in Plaza I and has been done in a handful of Forensic Document cases. This is where the examiner can present the images, the matching features, perhaps even say it's a match, but not allowed to opine as to what that signifies and perhaps for this judge would not allow "individualization opinions to the exclusion of all others" and discussion of certainty, error, etc. Simply put them before the jury and discuss characterstics of similarity and dissimilarity. Again I can't stress how dangerous I believe this to be, compared to the much smaller probability of an error (especially in a case where multiple ids have been made to the individual). It also seems to run counter to Rule 702 and the entire reason for the expert in the first place... But at least it's viewed by the jury. I didn't actually want to suggest this openly, since I do not support it, but that's what I was alluding to...but since you called me out....

Outsider,

It seems to me that this puts this case right at the opposite end of the continuum from the case that has taken my interest.

Well, I don't think it's on the 'opposite' end. Mayfield is on the opposite end to me, in that it is similar to the DNA debate of a "TRAWL" case v. "CONFIRMATION" case. Developing a suspect from a database with no other evidence to tie to the case is a trawl. This has a very different prior probability (everyone in the database has equal chances of being a suspect) until the evidence is examined. Whereas in the confirmation case, a suspect is developed hopefully due to good intelligence. The examiner then compares the information and a match occurs. The prior is much higher than in the latter confirmation case. [PS-the examiner does NOT have to know about the intelligence, just be given a name to compare...so no need to jump to the whole bias argument...it is irrelevant to the probabilities]. Some statiticians have argued that these are different cases, and the confirmation case would have a stronger posterior probability for the prosecution's hypothesis of guilt. There are other statiticians who might disagree on some nuances and interpretation. In any event, yes Outsider, you hit the nail on the head with that distinction in that there are different prior probs, which was what I was alluding too...[if anyone wants the papers and discussions w/ respect to DNA CODIS database cases and those discussions, I also have those...]

Outsider, if you wish to discuss more offline about 'that case', I would be happy to...



g.

[L.J.Steele]

Is there anyone who has access to the case materials who can post good quality images of the latent, exemplar, and any charted exhibits? (I'm assuming these were court exhibits and are now part of the public record.)

This case raises a few of my yellow flags:

It's a death penalty case. I've worked on a couple at the appellate stage and have gotten a strong sense that there are potential problems with tunnel vision, public and internal pressure to arrest somebody, and the kind of long hours that can lead to fatigue and mistakes. These were all issues mentioned in the Stacey report.

The suspect came from the investigators. Anyone know if the witnesses addressed what information was given to them by the investigators about the suspect when they were making their examination and/or before they wrote their reports? Depending on what they knew and when, there may be serious confirmation bias issues.

I recall mention of other evidence. Does anyone know if it is genuinely independent, or has similar potential confirmation bias issues?

Was the verification blind?

Was the outside review truly bind? Or did the outside investigators know the original results? (Ken Moses' problem in making a review knowing the FBI had made a match.)



It would likely be helpful if we weren't speculating in a vacuum, but could look at the evidence itself.

[Norberto Rivera]

So the saga continues....

This is my third attempt at articulating a post on this. I agree with most of the posts here. I don't agree with the ruling and the decision to exclude the evidence, but I understand how the judge wasn't persuaded with zero error rate on the methodology. She probably has some personal agenda against the death penalty, the defense attorney might be over-zealous in the defense of his client and I could go on and on with "might's" and "might be's". The reality, at least in my humble perspective, is that nothing is ever 100% certain when it involves human interaction except death. Sorry folks, but we're all gonna end up pushing daisies eventually.
The ruling says that the methodology is "subjective, untested, unverifiable" and "purports to be infallible". Subjective it is. Considering all the variables that affect a comparison analysis I don't see a way around that one. Untested?... 100 years of history, millions of prints on computer databases around the world, a plethora of literature and studies....how exactly should it be tested??? I must have missed something. Unverifiable? This one is significant, not because it's true, but because we haven't done a good enough job to prove it false. The high-profile misidentification cases like Mayfield and McKie didn't help either. This is where the universal standards and the standardized training come in. I'm sorry if I'm offending anyone, but I'm a pro-point standard kind of guy. (mind you, still taking into account the entirety of the impression's characteristics, not JUST the number of points) The reason is simple, you can't argue with numbers. Exactly how many numbers are sufficient? I can't answer that one at this point, but it's a good research opportunity I may explore at a later time. Which leaves me with the last one: "purports to be infallible". This methodology is used by humans. It doesn't make sense to separate the human aspect of the methodology because in the end, it's still a human's subjective conclusion. (two humans if you count the verifier) To say that the methodology has a zero error rate just doesn't make sense to me unless someone can produce a machine that can do ACE-V correctly every time. I'd love for AFIS to be at that point, but...uhhhh...not yet. So yes, somehow we need to figure out how to involve that human factor into the error rate, and you have to include non-IAI-certified examiners in that calculation because even that requires two years of full time experience with a Bachelor's Degree to even qualify. How many cases do you work in a two year time frame? So much for competency standards!
My head hurts now. I think I've stirred the pot up just enough for today.

[Brian Moline]

Judge Souder's decision on ACE-V appears to be primarily based on the DOJ Inspector General Oversight and Review Division report on the Mayfield case. That is what the defendant submitted at the Frye hearing without objection from the State of Maryland. I did not notice what if any documentation was submitted by the State of Maryland for review and consideration by Judge Souder. Does the State of Maryland also have the right to submit documents for consideration and review at a Frye hearing?

[Charles Parker]

With Mr. Meagher testifying I would suspect he had a plethora of information available.

He has probably testified in over 40 Daubert Hearings and I personally cannot see him letting this one go very easily or dropping his guard.

From my POV--more at play here than just fingerprints.

I would like to know since several latent prints were identified if those LPE that made the identifications would characterize them as:

Simple-----Moderate-----or Complex

Mayfield was Complex---Cowans was Complex---McKie was Complex.

Just curious.

[Strict Scrutiny]

I fully encourage review and scrutiny. I believe the fingerprint science will prosper from honest criticism. Although I believe the Judge in this case makes important points, especially in relation to blind testing, I think the totality of her decision is flawed and will eventually be reconciled with reality.

Knowing how important it is to “get it right” we must be careful about lending credibility to those who do not deserve it. Defense experts like the Habers of Human Factors Consultants have been blatantly dishonest with some of their “forensic” arguments, and contribute to flawed reasoning. Every Judge, jury, prosecutor, and defense attorney should be aware of this.

Dr. Haber is a highly trained scientist who understands statistics well. He has argued that LPEs might have an error rate of about 50% based on his extrapolation of statistics, and that a coin flip could be more accurate. At the risk of sounding harsh, this is a fraud.

Several years ago I attended an IAI conference where Mr. and Mrs. Haber were supposed to address fingerprint examiners with “their side”. The most persuasive thing they had to offer that day was a video of someone in a gorilla suit. They did no tests, presented no facts, but just boldly stood in front of the audience and argued for the sake of argument.

“Experts” such as this have shown their intentions, and in the past those intentions have not been to enlighten, but to mislead. As a field it is our duty to consider honest criticism, but it would also behoove us to not lend credence to charlatans. Out with the Habers and let in those who demonstrate they care about getting it right.

[mdavis]

I see three potential problems in this thread:

First, it is becoming more and more difficult to convice the courts that a latent print exam is unbiased. I've spoken before about the problems involved when the XYZ Police Crime Lab examiners perform a latent print individualization on evidence submitted by the XYZ Police Department. We all know that a true scientist is not intentionally influenced by the source of the evidence, but convincing the court of this is very problematic. Many of these examiners are not civilians, but commisioned officers of the XYZ Police Department. It may be OK, but it doesn't look good. I think we may see some changes here in the future to other agencies.

Second, vast AFIS databases are a source of dozens of "look alike" candidates that can pass muster to a sloppy or overconfident examiner. What used to be "safe" when individualizing from a small list of likely offenders is now fraught with danger when the edge of the envelope looms ever closer to our increasing experience and resulting (over)confidence. We cannot afford to make some of these borderline idents from blind searches that used to be "slam dunks" in the pre-AFIS world.

Third, we have given lip service to providing error rates. How on earth can this be done? No one truly knows (except the perps, as stated above) if the latent is theirs. With no absolute known answer, it is virtually impossible to determine if an ident is valid or not, and hence impossible to determine true error rates in actual case work. Proficiency tests are either put together using actual case examples (with true accuracy in question), or they are artificially manipulated from known examples in accordance with the collective bias of the proficiency committee.

There will always be judges with bias who come from backgrounds in defense or prosecution. Attorneys are notorious for bias toward both their clients and personal victory. We must strive to leave the bias with the courts and not allow it to influence our own work, lest we find ourselves in the same un-defensible positions as those who seek to destroy the strongest of forensic evidence coming from our laboratory system. There is no need for any examiner to be defensive if they have done everything right.

[Pat A. Wertheim]

Okay, Folks, the prosecutor has not given up yet:

http://www.examiner.com/a-1017710~_Radi ... -Baltimore

[L.J.Steele]

A motion for reconsideration worked for the prosecution in Plaza. I'll be curious to see what happens here.

[Charles Parker]

From the article just posted.

Rose’s attorneys have praised Souder’s ruling and suggested other judges follow suit. They say fingerprints are a pseudoscience that can’t be trusted and lack “serious scrutiny,” according to Patrick Kent, the head of the Maryland Public Defender’s Forensics Division.

For those attorneys that might be reading. While I agree that an argument can be made about the science. I do not accept the argument but I can see why some people believe that.

Taking that into consideration even if you did not belive that fingerprints is science, then I would think that you would consider it technical.

But to come out and say that something that has a 100 year history and with the documentation of its use and continued use----calling it a pseudoscience is wrong and from my POV that attorney has violated the canons of their office.

What is the more incorrect statement: Those in the business, the courts and a large number of learned people saying that fingerprints is based upon a science OR Those that have never learned it, are not scientists making a public statement that it is Pesudoscience. Is that attorney really representing his profession and his client by calling fingerprints a pesudoscience.

Just my 2 cents from an old pesudoscientist whose only goal in life is to put the wrong people in prison for crimes they did not commit.

Even lawyers are biased and make mistakes. I wonder who peer reviewed his statement before he made it.

[Ann Horsman]

I have always thought our profession to be both art and science, much like the practice of medicine. Yes, as many professions would be classified as such. While we are not brain surgeons we do, like surgeons, develop our skill (the art) over time with experience and training. Yet somehow, the 'art part', is looked down upon.

Obviously the art part of our work is very important just as it would be for a surgeon. I know machines could never do this job, AFIS misses many , many ID's. I know this because I test the system all the time with ID's I make and it misses (often these are Helen Keller latents!) quite a few of them.

I have spoken to my friends who are MD's and asked them if medicine is an art or all science. Universally the answer is something such as, "it's an arty science"!

I will next have to ask my friends if they take contemporanious notes during surgery and if they were biased by previous brain surgeries....

[Les Bush]

Hi,
The science of individualisation by means of fingerprint pattern recognition has both technical and scientific bases. The reliance on experience/competency from the first 100 years is heavily weighted toward the mechanics of fingerprint technicians observing sufficient evidence and for a long time that evidence was quantified on the basis of poorly understood probability variables. By taking away the safety net of numerical sufficiency the new standard for reliability is emerging as a combination of scientific intelligence and proveable data. A good analogy from history would be the Will West case, the question is would Bertillion have been able to differentiate the two men since he designed the system and tolerances for each measurement, but the technicians performed their task within the limits of their understanding of the technique? The current court case under discussion is a continuation of a trend to ensure the reliability and sustainability of expertise in the science of fingerprints.

[Justice Pie]

Someone mentioned this earlier...

“Countless doctors misread X-rays, yet these errors would never be seen as a reason to prevent doctors from testifying about broken bones in court,” he wrote. “The isolated errors of the Mayfield case should likewise not be the reason for the exclusion of fingerprint evidence. In fact, it was another fingerprint examiner using the exact methods used in this case who exonerated Mayfield and identified the true bomber.”

But, clearly Souder is making this decision based off 1 side of the coin.