CLPEX Chat Board

One more thing.

My entire reasoning is in the belief that correlation means that one thing always moves in sequence with another, like y=2x. As x goes up, y always goes up (that doesn’t seem right, you never hear people say x is correlated to y, they say x is a function of y). I may be wrong about how I’m thinking of correlation. It may mean that ‘generally speaking’ things are related somehow but no exact function exists to demonstrate the relationship? It’s been a long time since I’ve had a math class so I really don’t know the complete definition of correlation. . Maybe since Glenn works with this stuff everyday, he can tell me if I’m even using the word correlation correctly? Maybe dependent and independent is what I’m thinking of?? Maybe correlated means that the events aren’t mutually exclusive?

Have I made everyone as confused as I’ve made myself? You probably think I’ve been drinking (too much spiced wine, hot buttered rums, and doctored up eggnog). Actually, I haven’t had any, just imagine the thoughts I’d have after a few drinks (scary isn’t it).

I’d better go make cookies or something (something less mentally challenging).

Happy Holidays,

L.J.Steele wrote:At the ABFDE Daubert conference in Vegas, there was some discussion about ballistics and the counterpart to that phrase -- matched to that firearm and no other in the world.

A minor suggestion in use of terminology: Although by no means a firearms examination expert, my training indicated the term "ballistics" refers to the science of the travel of a projectile in flight. Matching a bullet or cartridge to an individual firearm would be the conclusion of a forensic firearms examination of physical markings and properties, using comparitive analysis methodology similar to a forensic toolmark examination.

A minor suggestion in use of terminology: Although by no means a firearms examination expert, my training indicated the term "ballistics" refers to the science of the travel of a projectile in flight. Matching a bullet or cartridge to an individual firearm would be the conclusion of a forensic firearms examination of physical markings and properties, using comparitive analysis methodology similar to a forensic toolmark examination.

"Firearms identification" is the more technically precise term, but ballistics seems to be an accepted shorthand. Calvin Goddard, the father of the field, started using ballistics as the term for what he did, and later said he regretted it.

Kasey,

I feel remiss for not citing your study in the first place. After you mentioned it I recall the details in general. I’m sure that previous knowledge helped bolster my opinion. I should have been able to give credit there. That study was a significant step in understanding the nature of our work.

Ms. Steele,

I just want to reiterate that I believe we should testify to precisely how we came to our decision. For most of us, the decision to individualize is rooted in the fact that FRS forms uniquely, and is triggered by our complete confidence that the unknown source matches the known exemplar. It is not derived from seeing all the prints in the world, for it would be impossible for a human to analyze that much information even if all that information were accessible. It would also be impossible for a person to accurately recall every fingerprint they compared, or even the ones from the day before for that matter.

The fact that some tool mark (or ballistic) experts do not testify exactly the way we do is not a point of contention or a cause for concern for me. The nature of their science varies from ours, and thus guides their testimony to be slightly different.

You also seem to be a bit contradictory. You claim that slight variations in our testimony are probably meaningless when influencing a jury--Yet you seem to argue that slight changes in our testimony are vitally important… hmm. To this I would like to offer an anecdote. I recall testifying in a homicide trial where the accused left an object at the scene with his fingerprint on it. I was peeved that it was a high profile case, the firearms expert was being hammered via cross examination, and I, the last prosecution witness was not even given a pretrial conference. The senior prosecutor finally spoke to me minutes prior to me taking the stand.

He only wanted to know if I was completely certain in my findings. Out of spite I smirkingly told him I was “pretty sure” but don’t hold me to it. That statement triggered synchronized panic. My second conference came in the witness box. After I was sworn in the senior prosecutor asked for a recess. The junior prosecutor politely went over my notes with me and verified that I had complete confidence in the match, all the while the senior prosecutor squirmed, fussed, and wiped sweat from him brow at the prosecution table. When everyone was satisfied the trial went on.

I don’t believe that all lawyers share your opinion that slight variations that back away from complete certainty are probably meaningless when conveying our opinion to a jury. I also don’t believe backing away from complete certainty while we are actually doing our job is advisable either. As I said earlier I think it affords the highest level of scrutiny.

Strict Scrutiny wrote:I feel remiss for not citing your study in the first place. After you mentioned it I recall the details in general. I’m sure that previous knowledge helped bolster my opinion. I should have been able to give credit there. That study was a significant step in understanding the nature of our work.

No worries, SS... I just wanted to make sure folks interested in the topic who may not have seen it were provided the reference. Obviously it's a topic of interest to me, but my post wasn't looking for credit... just relaying the info. Thanks for bringing up the important topic. There has also been some recent look into the use of automated latent quality measures. This will be a future Detail, but for those who frequent the board, you can get a preview. The applicable section relates "success rate" to quality measures:

http://fingerprint.nist.gov/latent/ir_7377.pdf

3.7 Latent Quality Measures
Among the topics were given emphasis in the workshop program were “Latent Quality Measures.” The purpose of a quality measures is to assess the suitability of a fingerprint for some specific purpose, for example automated searching [23][24][25]. The quality measure algorithm might, for example, assign a value between zero and unity to an input fingerprint image, depending upon the algorithm’s assessment of how suitable the particular fingerprint is for automated searching. Assuming that 1.0 indicated the highest possible value, a computed value of 0.8 to 1.0 would indicate “very good;” a value between 0.6 to 0.8, “good;” 0.4 to 0.6, “fair;” 0.2 to 0.4, “poor;” and a value less than 0.2, “unusable.” The computed score can then be used to predict the success rate of a search. It can also be used to screen out fingerprints of very low value, as these would have a negligible success rate, and would waste system resources.

For most of us, the decision to individualize is rooted in the fact that FRS forms uniquely, and is triggered by our complete confidence that the unknown source matches the known exemplar.

I fear I am going round in circles here, or perhaps just being insufficiently clear. I do not disagree that an examiner has, and should have complete subjective confidence in an identification before he or she reports it as a match. However, I think that one must understand that confidence can be manipulated by external factors --- I would expect that every one of Dr. Dror's subjects was completely confident in his or her match when testifying to it in court. Yet, but manipulating the information given to them, Dror was able to, years later, get 3 of 5 experienced examiners to say that the prints were not a match and explain why -- in effect, showing a possibile doubt. I assume that the Mayfield examiners, and the folks in the 20-ish other cases that Cole writes about were confident in their results as well.
It is good, and necessary to be confident that one has reached the right result by the careful application of the right method. The question is how does none express that view in court or in a report.

He only wanted to know if I was completely certain in my findings. Out of spite I smirkingly told him I was “pretty sure” but don’t hold me to it. That statement triggered synchronized panic.

I agree that there are prosecutors who would react to any change in the usual way of expressing one's results with horror. And those in the defense bar might mistake a change in terminology for a change in methodology. Attorneys don't like change -- we like to be able to cite to 18th and 19th century cases, too. Juries -- hard to say, I'm not a jury consultant, but I think the exact words used are probably less important than whether the expert explains himself or herself clearly and whether the testimony makes sense to a layman.

At the risk of flogging the dead horse, few prosecutors would go forward with an eyewitness ID case if the key witnesses expressed anything less than 100% confidence in their identifications. By a growing number of states will not permit the witness to testify as to a level of confidence. Some do not allow the judge to expressly factor in the witness' expressed confidence into their decision about whether the identification is reliable enough to be admissible. Does that mean that the witness's confidence is irrelevant? No. It comes through in other ways -- it comes through in tone, body language, and the witness' steadfastness on the stand.

Similarly, a witness is not allowed to vouch for another witness at trial. A witness can't say "I believed X". Nor, in many jurisdictions, can a witness be expressly asked if Y lied. Again, the concept will come through in tone, body language, conduct, and the witness' demeanor on the stand. The jury is likely smart enough to figure out whether a doctor believed a victim's complaint of injury, or a police officer believed a defendant explainatin of his wereabouts during an interrogation.

So I think there is a false dicotomy presented here. You don't have to chose between 100% confident and some nebulous, but lesser standard, in chosing "reasonable scientific certaintly". You, as a profession, can insist that an examiner stake his or her career on each an every ID testified to (which will effectively require the examiner to be personally 100% confident), and still conclude that the words to be used in reports refer to the facts, the method and the science and not use the examiner's subjective feeling about his or her results as an implicit shorthand for that idea.

I won't beat the dead horse any longer either. I just want to express that I strongly agree that outside influences can adversely effect our decision making. We as a profession need to be more diligent in keeping those influences out of the ACE-V process. This includes opinions transferred from the original examiner to the verifyer.