CLPEX Chat Board

Perhaps I'm being overly naive here, but I really do believe that if you are open and honest, you have nothing to fear -- neither by omitting potentially useful information, nor by being biased and prejudicial.

Of course if I were talking about a print that was "consistent with" a suspect's print, but not an ID, I would point out that many people could have that characteristic and it could be someone else's print. I agree that not to do so would be biased, prejudicial, and remiss in my duties to ethics, integrity, and honesty.

Nor would I allow an overbearing (and probably ill-prepared -- I agree it's frustrating that the DAs never have time for pre-trials) DA to try to pin me into an ID I didn't make. I could see it going something like this:

DA: So what you're basically saying is that this was an ID, you just didn't think there was enough information to say so.

Me: No. If it was an ID, I would have said it was an ID in my report. What I'm saying is that all the information I saw was consistent with the defendant, but that there was not enough information present to conclude that he is the only person in the world who could have made that mark. That means that it is perfectly conceivable that someone else with a similar fingerprint could have been responsible for the mark.

DA: But let's be realistic here; how many people in the world have fingerprints similar enough to have potentially made that mark? Really, with all that information in agreement, it has to be the defendant, right?

Me: Unfortunately, we don't have statistical data to make a claim answering that question. We don't know how many people have a print "close enough" to make the mark, which is why it's impossible to make an ID on this small amount of information. All I can say is that the defendant could not be excluded as the donor of this mark.

After that, if the defense got up and asked me, "So is it possible that my client did not make that mark?" I would be happy to respond, "Yes, that's quite possible."

I've had similar lines of questioning in court regarding non-IDs and handled them the same way. I had a case not long ago where I had processed a piece of paper with ninhydrin and many areas of ridge detail were developed. There were a few prints of value and many partials, smudges, etc that were not. While I was on the stand, the defense asked me if I developed any ridge detail other than the prints of value I had previously testified to. I said yes, I did, but it was not of comparison quality. He asked me if that meant that I couldn't tell who the other prints belonged to. I reponded that, yes, it meant there was not enough information to determine who the print belonged to. He followed up with, "so those prints could have been made by someone other than my client," and I said, "Yes, they certainly could have. There is no way to know one way or the other."

I'm not trying to spin it and say, "well, the other prints were his so these probably were, too" or "what difference does it make if those belonged to someone else; he's already identified three other times?" I'm not editorialising or being defensive; I'm just answering the question. Yep, someone else sure could have made those prints. Nope, I don't know who they belonged to. I truly don't have a vested interest in which side I'm "helping." I'm just telling it like it is.

I was hired to fill a Patrol Officer vacancy at Kerrville Police Department (TX) and attended Police Academy in the fall of 1973. I was taught in Academy that it is perfectly legal to lie to a suspect and sometimes that is the only way to get a confession and a guilty plea. I attended my first fingerprint training in the summer of 1976 and was promoted to ID Officer immediatly, which was a sworn position at Kerrville back then. A few years later, a senior investigator had a suspect in an interview when he (the investigator) came to me privately and asked me to walk into the interview room and say that I had made a fingerprint identification to the suspect. I balked, he reminded me he had the legal "right" to lie to the suspect, he told me the suspect was ripe and wanted to confess but was holding back, and he said if I just walked in and said I had identified the suspect's fingerprint he would confess and tell all about the crime. I still balked, but suggested a minor compromise that he accepted. I made a latent print lift of my fingerprint and, in red marker, wrote the suspect's name on the lift card, then simply walked into the interview room without saying a word and dropped it onto the table in front of the investigator where the suspect could see it. Sure enough, the suspect immediately broke down, confessed, gave a lot of extra detail to the information the investigator already knew, and a few weeks later, pled guilty.

But I felt DIRTY the instant I did it and I still regret my part in the implied lie in it. I have never repeated that performance. I have matured and here is the objection that has developed in my mind to such behavior -- not all suspects are guilty! When we, the police, lie to an innocent person, he/she KNOWS it. We have totally and completely destroyed any trust that person ever had in the police. Further, we have damaged or destroyed the trust of any close friends or relatives of that person ever had in us. Finally, if the suspect is, in fact, guilty, and he knows he was wearing gloves and could NOT have left the fingerprint we claim to have, then his behavior becomes very smug because he knows we're lying and he knows we do not have the evidence to convict him.

But the bottom line is that lying is wrong. Now, I will do this: When I am taking inked major case prints from a suspect, ahead of time I will ask the officer to leave the room for a few minutes and leave me alone with the suspect. When I am alone with the suspect, I will quietly say something like, "You know I'm not an officer. I'm just a civilian fingerprint examiner. Now, I don't know if you left any fingerprints or not. Only you know that. If you didn't do this crime, you don't have anything to worry about because you and I both know I won't find your fingerprints anywhere. But if you did do this crime and left your fingerprints anywhere, then I WILL find them. And when I tell the officer we've identified your fingerprints, he will have no further reason to bargain with you. So, if you did this crime, the best thing you can do right now is to admit it to the officer and cut a deal before he gets my fingerprint report. Make it easy on yourself. But I repeat, if you didn't do this crime, you've got nothing to worry about."

That little monologue is not a lie, but accomplishes about the same thing with a guilty person without destroying the faith an innocent person has in the police. It preserves my integrity while, at the same time, helping the investigator.

In regard to reporting "probable" identifications, that never appears in my formal report. But I might mention to the investigator, if I know him and trust him, that I "think he's on the right track" without telling him I have a "probable identification." If the investigator has ever given me reason to doubt his trust, then all he gets from me is "inconclusive," exactly as reported in the formal, written report.

But I know, too, that neither the prosecutor nor the defense attorney want the literal truth from me on the witness stand. Both will try to "spin" my testimony to suit their needs. So I have to walk that tightrope across the courtroom to the jury without falling off either side --- true, also, when I work and testify as a defense expert in a private case. But occasionally, as a police examiner, I have been faced with the dilemma of weak evidence (perfectly acceptable legitimate access issues, for example) and a defense attorney who is going to attack me in an inappropriate way (for example, when there is a serious "legitimate access" issue and the defense attorney if focusting on "how many points does it take?"). In such a case, the defense attorney is so far off base it is a terrible disservice to the defendant. So in some cases, I have been known to initiate a contact the defense attorney and coach him/her on how to cross examine me in order to avoid allowing the prosecutor to put too much spin on my testimony. In one case when I was testifying on a case I had worked for the police and was testifying for the prosecution, I even arranged for an examiner from another department to coach the defense attorney ahead of time and sit with her during the trial.

Dave Grieve told me one time that ethical behavior is like throwing darts. The more you concentrate and the harder you try to be ethical, the more you hit the bullseye. But still, once in a while, a dart goes wide and misses the mark. You don't quit the game, you pick up your darts and keep practicing. I try to be ethical, but I know I'm not perfect at it. I keep practicing. And I hope I'm getting better.

Should we tell an officer we think a latent was "probably" made by the suspect? I think so, but I also think we need to make sure he doesn't misunderstand and treat it like a positive identification. Should we put it in our written reports? Only when it is accepted in the science and only then to the degree that it is allowed in our policies and proceedures. Should we say something about it in court? Only if asked, and then only with strong caveats in our testimony.

A friend who read my last post reminded me that anything I say to a suspect, and any response, may be required to be revealed in disclosure. I know that and keep the investigator advised of the exact conversation, including responses and replies between me and the suspect. But I believe what I am doing is honest, truthful, ethical, etc., and would not be embarrassing in court if it came out during testimony. On the other hand, if I were to participate in a lie about a fingerprint identification in order to induce (coerce?) a confession, I would expect the jury or the judge to hold that against me, the police, and the prosecution. I would hope the judge and jury would frown on the police lying to trick a suspect into confessing. I know it's legal and I know it sometimes "serves a greater purpose," but I still find it distasteful.

I remain concerned about "consistent with", if one can't tell the significance of the common factors. If you only had level 1 detail, that might be consistent with the suspect, and thousands of other people. If you can't articulate that to the investigator -- he or she may give the information more weight than it deserves. For many other factors, shoes, tire treads, maybe even blood types (can you get a blood type without enough material for DNA?), the investigator and/or the jury may have a better gut sense of the frequency issue. If a shoe mark guy says it is consistent with Size 10 women's TEVAs, one may be able to get a gut sense of how common or uncommon that shoe is among the pool of potential suspects. But will an investigator have any sense of the frequency of an Arch with two Galton points?

As to lying to suspects -- yes, it is legal in the U.S. (heck, I found references to it in a 13th century manual for papal inquisitors), but it can lead to false confessions. Professor Kassin's done a great deal of work on this.

http://www.williams.edu/Psychology/Facu ... ssions.htm

One of the more interesting problems is here .. behavioral confirmation
http://www.williams.edu/Psychology/Facu ... y_2003.pdf

As I understand it, if the police officer believes the suspect is guilty, he or she will attribute various ambiguous non-verbal cues as signs of guilt, and, the investigator's body language will often cause the suspect to react in ways that a third-party observer may view as consistent with guilt. I've got serious concerns about the effects of providing information that you, the print expert, can't put into context because you don't have the underlying data to explain the significance of your observations.

[One of the keys according to various confession folks is whether the suspect's confession provides verifiable information _not known to the police_ when the confession is made. There's all sorts of interesting things that can happen in an interrogation with body language, questioning methods, etc. that can cause a suspect to echo back to the investigator information the police already know.]

Ah, I hear several of you say, but I've got experience to tell me this is significant. I can't prove it, but I know it. My first employer warned me about gut insincts that I couldn't prove -- they can sometimes get one into big trouble.

The hunch raises the yellow caution flag of various heuristics:
Representativeness heuristic -- you don't know the probabilities of the occurance of these features, and thus you may mis-estimate their significance.
Availability heuristic -- the more easily you can recall a similar case, the more likely you will think it is, regardless of the actual probability.
Anchoring -- giving the information may cause the investigator to anchor his or her belief in the suspect's guilt, and ignore information that might exonerate or lead to other suspects.
Tunnel vision -- discussed in the Stacy report on Mayfield -- the investigator may overvalue the information and focus on that suspect, to the exclusion of other, better, suspects.

Yes, you say, but let's let the investigator or prosecutor or jury decide what to do with the information, I'm not responsible for his/her/their reasoning errors. By mentioning "consistent with", you are saying that the information has significance, but you can't say how much. Essentially you are vouching for the significance of the information with your reputation. The investigator is depending on you to help him or her save time and pursue the best suspects. Do you want to do that if you can't back up your gut feeling with some frequency information?

Many of you seem to be assuming that what you are seeing would be a match if you had a better latent so it is ok to give the investigator help so he stays focused on the right guy. What about the converse -- if you had a better latent, you would be able to clearly exclude. By saying "consistent with" to the investigator, based on incomplete info, you've led him or her to focus on the wrong suspect, which means the right culprit stays out there longer (and perhaps remains uncaught), and able to commit more crimes.

Hi Lisa

I appreciate your perspective and I cannot disagree with most of what you write. But let's look at that rape case that RL Tavernaro and I were discussing earlier in this thread. The entire family of the victim had loops, but on the bedroom lamp the rapist touched was found a whorl pattern latent. And not just any whorl, but one with a clear pattern and three or four clear "points." As it happened, when a suspect was developed through other means and the fingerprint from the lamp was compared to the suspect's fingerprints, lo and behold, there was a perfect "match" of the available features in the latent print with one of his fingers that would have been consistant with the positioning of that finger by a person turning off the lamp, as the rapist had done.

You are absolutely correct -- we cannot calculate the probability of a random person having that exact pattern and those several points in the exact position relative to the pattern. One in ten thousand people? One in a million? Maybe only one in a thousand? Can't say with any scientific degree of certainty. But based on my "training and experience," the odds against any given person in the population having that exact pattern (not just any old generic whorl, but an exact overlay of the one from the lamp) and the several points that are visible would be phenomenally great. Steve Horn may want to comment here because we are not talking about an AFIS hit, but about a suspect whose name came up in the investigation independently of the fingerprint.

So, I have this latent print that I honestly believe was "probably" left by the suspect, but it lacks sufficient detail to individualize. Should that fingerprint from the lamp alone be sufficient to convict? Of course not!

But now let's consider hypothetically what other evidence may exist. Perhaps the suspect's name came up through an "identification" of a photo lineup of known sex offenders living within a couple of miles of the victim's house. Can we calculate the probability of this particular eyewitness/victim making a correct identification from a photo lineup of six men? No. Should that "identification" alone be sufficient to convict? No. I think we would all agree that there are a lot of innocent people serving time based on no evidence but an eyewitness identification that later proved to be erroneous. So should the "identification" made from a photo lineup of six know sex offenders be enough to convict in the absence of any other evidence? Of course not!

Let's say we have a footwear impression that was found in the soft dirt outside the window through which the rapist entered the house. Because of the coarsness of the dirt, there are no clear individualizing details, but we can determine the brand, model, and size of the shoe that left the print. When located, the suspect is wearing a pair of shoes of the same brand, model, and size. Now, here we can roughly calculate the probability of any random person wearing that exact brand, model, and size of shoe, although we cannot individualize the impression in this case. Should this footwear impression alone be sufficient to convict? Of course not!

Let's also say that there was no semen and no DNA, but the victim is adamant about the rapist climaxing. Then we might assume he wore a condom. And perhaps coincidentally, an empty condom wrapper was found outside the window, next to the footwear impression. When the suspect was detained, amazingly, in his wallet was another condom with a wrapper identical to that found outside the point of entry. Can we calculate the probability of a random male having one of those exact packages of condoms in his wallet? No. Should that alone be sufficient to convict? Of course not!

You know by now where I'm going with this. I could add more hypothetical evidence, but let's stop here. No single "coincidence" by itself should be sufficient to convict, but by the time you add them all up, even though you cannot calculate the probabilities individually of any of the specific "coincidences," there may be enough for the jury to decide, beyond a reasonable doubt, that this guy is the rapist.

So, Lisa, my question is this: Should we withhold the photo lineup because we cannot calculate a precise probability, or "error rate," for this specific lineup? Should we withhold the information about the condom because we cannot calculate an exact probabilty of how many men carry these exact condoms in their wallets? And if those tidbits of information are admissible in court, why should the "inconclusive" fingerprint that matches in all respects, but lacks sufficient detail to individualize, be withheld from the jury?

After all, it is up to the defense attorney to expose in court the lack of certainty in ALL of the areas of evidence. My "probable" identification could only assume disproportionate importance if the defense attorney chooses not to attack me on the weaknesses of that evidence. And, of course, that same is true of all of the other coincidental evidence, as well.

Of course, at this point in time I would never officially report nor testify to a "probable" identification. But does that mean we should bottle up our opinions totally? That is the topic of this debate, and I believe that a qualified expert SHOULD be able to state an opinion such as that. I believe such a statement of opinion should contain a strong caveat that the "probable identification" is NOT positive and that some number of other people in the world probably have the same combination of pattern and points, but we cannot calculate those probabilities. To deny the investigator, the judge, and the jury that information about the latent fingerprint from the lamp, weak as it is, would be the same as denying them the information about the photo lineup, the footwear impression, and the condom. And all of those probable but uncertain bits of evidence, each weak when standing alone, combine to form a case stronger than any of them alone would justify. But all of those bits of information are important for the "finder of fact."

Pat,

I see your point, considering ALL the information may be ‘valuable’ but it may be valuable to the investigation and not the identification. If the identification is a scientific determination then the person arriving at the conclusion needs to use scientific principles and protocols, this would include trying to make the process as objective as possible and make your conclusion based on the object (the data in the latent) and not on outside information. This extraneous information may be statistically useful but it could also just be biasing information. Before we start using this kind of information, I think we need to set up some valid principles to insure reliable quality conclusions.

Yes, this information could be useful to the fact finder but it could also be biasing information that leads the investigator in the wrong direction. The fact finder may also be convinced that your conclusion is scientific when in reality it’s just your subjective opinion without any scientific merit behind it.

Until you can establish how much weight the fact finder should give this evidence, you don’t have valuable information, you may just have an active imagination and you’re making the evidence fit your preconceived idea. Maybe that should be the caveat that examiners tell the investigator

Pat A. Wertheim wrote:As it happened, when a suspect was developed through other means and the fingerprint from the lamp was compared to the suspect's fingerprints, lo and behold, there was a perfect "match" of the available features in the latent print with one of his fingers that would have been consistant with the positioning of that finger by a person turning off the lamp, as the rapist had done.

Would you feel confident enough about the hunch to take the same consequences if subsequent evidence clearly excluded the suspect as if you had made a mis-ID? Let's say your condom is found, analyzed, and clearly excludes the suspect based on the DNA -- what then?

Pat A. Wertheim wrote:But based on my "training and experience," the odds against any given person in the population having that exact pattern (not just any old generic whorl, but an exact overlay of the one from the lamp) and the several points that are visible would be phenomenally great.

Let me ask a couple questions here as we're moving beyond my knowledge of the field?

Would the suspect's family members be more likely to share the pattern and a few common points?

If you were to put the data you had in AFIS, how many similar matches might come up in the same geographic area?

Pat A. Wertheim wrote:Perhaps the suspect's name came up through an "identification" of a photo lineup of known sex offenders living within a couple of miles of the victim's house. * * * So should the "identification" made from a photo lineup of six know sex offenders be enough to convict in the absence of any other evidence? Of course not!.

No, but for a different reason. A lineup in which all the people present are potential suspects isn't a test of the witness' ability to identify -- there's no wrong answer. A properly constituted double-blind, sequential array, with the right cautions and record keeping on the other hand, is about as good as we can to protect the memory and get the right answer. One of the reasons we see so many exonerations is that the methods used for arrays, lineups, and showups are demonstrably flawed, but there is huge resistance in the courts and in some departments to even testing new methods.

Pat A. Wertheim wrote:So, Lisa, my question is this: Should we withhold the photo lineup because we cannot calculate a precise probability, or "error rate," for this specific lineup?

The one you describe ought to be supressed, but because it is flawed in its design. You've also told us nothing about the incident that bears on the reliability of the ID -- lighting, event duration, stress, weapon presence, cross-racial, etc. But, if you can get past those issues, it goes to the jury for what it's worth. In an increasing number of states, it goes in with a jury instruction on the possibility of good-faith error and the problems of eyewitness ID.

I suggest, tho, that shoeprints and condom ownership are things that a lay trier of fact or an investigator has a better gut sense of than fingerprint features. You are testifying, in effect, that this information is significant, but you can't testify to how significant it is or back up your intuition with data. That makes me darned nervous.

Pat A. Wertheim wrote:That is the topic of this debate, and I believe that a qualified expert SHOULD be able to state an opinion such as that.

Hasn't the problem been stated as investigators not understanding the result you are giving is, in effect, a "maybe" -- you can't include or exclude, so police should neither give up on the suspect nor proceed as if they've got proof. And the concern that investigators aren't understanding the meaning of this result? If the investigators are having trouble with the terminology being used, is giving a "consistent with" answer, no matter how hedged it is, going to make things more clear?

What I hear is a desire for a grey scale...
No value ... can't include or exclude as there isn't enough information.
No..exclusion
Inconclusive...there's some information, but not enough to include or exclude.
Probably...I've got a gut sense that the correspondence I'm seeing is significant but not enough for an identification.
Yes...identification.

Is that a fair summary of the debate so far?

In over 20 years of examinations of fingerprints, I have occasionally seen a pattern type which appeared the same as the suspect's and found 2 or 3 points which appeared to be in the same unit relationship, and yet, upon further examination, the prints were not the same. With the advent of AFIS, we have seen more of this type of occurrence. I would be very cautious of saying something like this was "probable". I try to think what I would want done if it was a member of my family who was being investigated. I would want the examiner to be darn sure and not going on some gut feeling.

You are testifying, in effect, that this information is significant, but you can't testify to how significant it is or back up your intuition with data. That makes me darned nervous.

As well it should. This is commonly called "circumstantial evidence" is it not? Put enough "coincidences" together and you run a trail of sucessive approximations toward truth, if not quite there. The defense can be expected to call it "intuition." What I see are facts -- a whorl with 4 ridge details in the same spatial relationship with one (and only one) finger of one (and only one) of the suspects. What I report are facts based on professional education, training and experience. I do not testify to the significance. In fact all my reports are left to the agency, the P.A. and (if he so chooses) the court to determine the significance, if any.

How many cases are convicted with no scientific evidence whatsoever? Would it be fair to say the majority of them? We are again crossing the line here in assuming that such information ends up in court. The prosecutor is the gatekeeper here, not the lab. If I see a latent print that has ridge details insufficient for positive identification (and I so state in my report) but is such that a specific finger of a possible suspect cannot be eliminated as a potential donor of that print, am I not withholding potentially valuable (and it's up to the submitting agency and the P.A. to determine how valuable to their investigation) information? Am I the sole determinant of what is valuable to the submitting agency and the P.A.'s case? Am I limited to reporting only the "slam dunks?"

The agency gives me a lamp, asks for latent print development and asks me what I found. I tell them. I also tell them that I cannot ID the print. Can I not tell them that I cannot exclude one of their suspects? I hope not. We have to look at these cases from both sides of the fence, do we not?

I have been watching this thread with some interest now and wanted to see how the discussion went before jumping in. There was quite a bit of discussion about how to express the “weight” of inconclusive opinion (probably him, cannot exclude, use in court v. investigative tool, etc.).

This is perfect timing.

The discussion about statistics, models, and the need for a tool to calculate how rare (or specific) is a minutiae arrangement is not a futuristic concept. In fact for 100 years, always at arms length, endeavors have been made to develop these tools. In the last 30 years they have progressed significantly. These tools exist, the statiticians that have developed the various models have offered them to us to take the mantle and validate them on the massive databases/repositories that we have control and domain over. We have failed and dropped the ball time after time, rejecting the notion of statistics to evaluate fingerprint evidence. Some excellent models have been proposed by Roxburgh (1933), Amy (1946), Kingston (60s) Osterburg & Sclove (70’s), Stoney (80s) and Champod (90’s). I have many of these papers (in pdf) if people wish to read them. I believe that this should be taught as standard FRS ID science. We have rejected them, for various reasons, but the most common I hear is “because they don’t encompass everything an examiner takes into account”. The short answer is: they don’t have to.

For the mark described in the case example, the whorl with 4 minutiae, this could be calculated by the models. And a probability of that arrangement could be calculated, not unlike a DNA model. We as FP examiners could then go to court and simply testify to the value of the evidence. And leave it at that. Heidi’s comments about “what a jury does with it then is out of our hands”…I agree. That’s not our job as scientists.

More recently, building upon research of Champod at the University of Lausanne in Switzerland, others are taking up the torch and building statistical tools that evaluate a “match”, producing instead of a probability (like .0000000001) but rather it spits out a Likelihood Ratio (which is a ratio of probabilities). In fact simply put, it is the probability that a crime scene fingermark comes from the “suspect”/exemplar divided by the probability it comes from someone else in the population. When the ratio is greater than 1, it suggests that it came from the suspect. When the ratio is in the millions, it [emphasis] “REALLY” suggests that it came from the suspect.

This software, developed by Cedric Neumann and his team in the Forensic Science Service of the UK (Cedric is a student at U Lausanne, i.e. UNIL) exists. It is real and is currently being validated on a large fingerprint database. (I should also mention the improvements to it by a brilliant mathematician from UNIL, Nicole Egli, who likes to hide outside the spotlight, but needs to be recognized for her genius!).

Point being, this hypothetical whorl could be run in the software, a likelihood ratio (LR) could be produced (maybe in this example it might be around 100,000 (10 ^ 5). So now, the scientist could go to court and present the image to a jury, show the 4 points in relationship and say the LR is 100,000. Which means, the print is 100,000 times more likely from the suspect, than someone else.

What does that mean? Well in a city of millions of people….if that’s the only evidence in the case and no other evidence is produced, I certainly would hope a jury would not convict (based on beyond a reasonable doubt criterion). In a town of 1000 people though, and this individual lived next door, and he had a history of stalking the victim, and he was found in possession of something of the victim’s and there was a size 8 shoe print in blood at the scene, but only size could be determined…then all of these circumstantial things begin to make the case look pretty solid. And that’s it folks, the jury looks solely at an objective metric of the match and weighs it along with everything else in the case. But that’s the jury’s job. Not ours.

The weight is off of us. We simply testify to the strength of the match. That’s it. No longer is there a “single shade of grade” (inconclusive) [which to me is the most meaningless opinion we can provide in court, especially when it was almost a match or almost an exclusion…two obviously different things….yet we must report: “one single shade of grey…it’s the same = inconclusive”. Statistics/Likelihood ratios suddenly give a lot more value to those shades of grey and gives the court something that reflects the true “VALUE” of the evidence.

Let me finish with 2 things.

I have some great papers that discuss this. One by Ian Evett about categorizing likelihood ratios (using expressions like “strong evidence”, “extremely strong”, etc.) b/c of his disdain for a somewhat useless “consistent with” opinion (I have to agree again). I have a second paper by Taroni, et. al (a prof at UNIL) regarding evaluating evidence
with LRs.

The other thing, most importantly, is that Christophe Champod and Cedric Neumann, (plus myself and Paul Chamberlain) will be presenting quite a bit on this at IAI conference in San Diego. Also, there will be 2 workshops that will allow people the chance to see the software and use it a bit. From what I hear, the workshops are currently full, but I would encourage people to attend the lecture block to learn more about it. This
stuff is NOT that complicated for us to understand and has real value [how many of you have ever had to testify to the mathematics of an AFIS algorithms in court; yet you use AFIS everyday to develop suspects? I would suspect you would find the same thing here, you don’t need to be a statistician to use these tools, merely understand a few fundamental concepts and have basic training to use the tool. If its novelty/scientific basis is challenged in court, then the big guns will come out and testify to it….]

This stuff is a reality. It’s here and it can be an exceptionally valuable tool in our arsenal, providing us with objective guidance for when to call a match, measuring the strength of the match, providing weight of our evidence to a jury, etc. To me, this is the future of forensic science and the time is now to ready ourselves for it.

[PS- thanks Pat for the mention in the thread way back there, but it's Christophe who has done so much with this; my statistical work deals with measuring variations among examiners, comparing experts to non-experts, and trying to learn more about the decisions we make during ACEV. I can't take any credit for the statistical models that people like Christophe, Cedric, Nicole, and Didier have developed. Those people just amaze me! (and they still find time to go clubbing =) ]

g.

We have rejected them, for various reasons, but the most common I hear is “because they don’t encompass everything an examiner takes into account”. The short answer is: they don’t have to.

I hope you're on to something here, but I remain skeptical without seeing the program. Here's why:

Let's take Pat's whorl (from this thread?) as a hypothetical example. A whorl with 4 "points" (whatever THAT means--lets say 2 dots and a short ridge). Let's further assume that we have excellent control detail consisting of ridge outline contours, pore locations and outlines and such, but our latent print contains only level I pattern and ridge flow detail and 4 other Level II "details." I look at the latent and see that the whorl is a counter-clockwise spiral surrounding a "happy face" consisting of two eye "dots" and a mouth "short ridge". If I run a horizontal line across the dots, I see that the short ridge curves slightly upward forming the "mouth" below the eye dots. One eye dot is nearly perfectly round, the other is round with a slight ridge concavity at the top right corner. I also note that the right side of the "mouth" is longer by approx. 2mm than the left side. The left "short" side of the mouth is 3mm closer to the left dot than the right side is to the right dot. The right side of the short ridge extends farther beyond a perpendicular to the eye line than the right side, and there is a bulge midway along the short ridge slightly closer to the longer right side that corresponds to a pore opening visible on the control, and the short ridge is rounded on one end, tapered on the other. (And I sure hope this isn't what some courts expect as "contemporary" note detail or we'll all spend the rest of our careers writing instead of comparing!)

Now in order for a computer program to assess this image, it would be necessary to either ignore about half of the unique ridge detail as seen by the examiner (as stated above), or find a way to add an exhaustive list of "choices" from which the computer operator (a latent print examiner turned computer data input expert) must pick and choose (in context of this latent print) in order to come up with a theoretical percentage of probability of, what, consistency? What constitutes a "detail?" What would this program do with the no-point (no level II detail) identification made in (if I recall correctly) Illinois of the thumb print on a rifle scope identified using only Level I and III?

I suspect that the very few bad idents that are made are made because the examiner invoked "the eye of faith." The question before us is still, exactly WHAT do we see, not a numerical accounting of what we have. Is this not back to counting "points?"

I really hope the math guys are on to something here and prove me off-base. But as a scientist, forgive me for being a skeptic when we try to over-simplify and then digitize an analog process. If we limit our data input to two dots, a short ridge and counter-clockwise spiral whorl, we can find hundred of thousands of images that will "fit". Add in the other details, and we can effectively eliminate perhaps all but a very small handful. I still wouldn't call that an ident if my potential database came from IAFIS. But I wouldn't refuse to tell investigators that it looked pretty good, and I think that information with appropriate disclaimer should be acceptable in a written report.

For once I find myself agreeing with a defense attorney. L.J.Steele is right -- if it is not absolute, it is prejudicial and may well be completely wrong. Therefore, you either have "Identification, Elimination, or No Conclusion." To say "Probably" or "Probably Not" only confuses the issue and cannot possibly yield an accurate verdict, except by happy accident. Let the footwear examiners and the document examiners say "Probably," but I hope I never see fingerprint identification lessen the impact of our evidence that way.

No one has suggested using "Probably" or "Probably Not" as a reported result. I suggest that disregarding the "happy face whorl" in this example as an absolute "No Conclusion" is to withhold potentially valuable evidence from submitting agencies. Is that not the same as saying "I have no idea if it could have come from one of the suspects fingers" which, in and of itself, is not true.

"Point Standards" served our business well in the 1890's and early Twentieth Century because no one in the field had the experience or knowledge to know differently. By the 1930's, many forward thinking examiners had abandoned point minimums. In 1973, the IAI officially rejected them.

Likewise, the idea that there is no statistical basis for anything less than an absolute identification has served the science well. Examiners cannot help but feel occasionally that a latent print was "probably" made by the suspect, but we have never had a tool to quantify that feeling or determine numerical probability. In the absence of such a tool to calculate a mathematical probability, the dogma developed that an unknown print can either be individualized or excluded, or else we can say or think nothing of its origin. The result is "inconclusive."

Well, it is not truly "inconclusive" if you have formed a secret conclusion, subjective as it is, of some kind. You might call that an "opinion." So, if you are truly an "expert," should you not be able to tell an investigator, write a report, or testify to your "opinion?" There is nothing in law nor in rules of evidence to prevent that, but there is every reason to believe such a report and testimony would be supported in law. The jury should be cautioned to consider what weight to give your evidence, but they are given that instruction under any circumstances.

I am not promoting the promiscuous use of SWAG reports ("Scientific Wild A-- Guess"). But occasionally, there are latent prints in all of our labs that give us strong, unshakable feelings of identification, but we can't take that final step to "individualization." All I'm saying is that, in those cases, we should be able to say something. Like mdavis' happy face, or the perfect overlay whorl with three or four points that Bob and I looked at years ago.

Glenn's post encourages me. I thoroughly enjoyed Paul & Cedric's talk last year in Boston, although I heard a few impolite groans and grunts from traditionalists in the audience who wanted to make their rejection of this research audible. As for me, I was cheering inside. The more research in any area of our discipline, the greater the advancement we will make as a science. The more the dinosaurs defend dogma with the "if it ain't broke, don't fix it" mentality, the more we become archaic and obsolete. And eventually, like the real dinosaurs, extinct.