Minimum points for an ID

[bficken]

Hello out there,

One of my coworkers is pushing for us to establish a minimum point standard for identifications. Specifically, she wants to say no print that has less than 8 level 2 minutia details will be considered an ID.

Last I'd checked, some labs out there are creating standards that say "anything more than X number of level 2 details is DEFINITELY an ID," however I don't know of anyone that is officially saying "We won't ID anything with less than Y level 2 details." My understanding was that we were still going off of the IAI study from the 70s that stated there was no scientific reason for establishing such a minimum. But I could be wrong, and my concept could be outdated. Thought I'd check in.

[bficken]

As a follow-up note - this would be an administrative policy, for quality assurance purposes only. And all manuals would make that clear.

[LPE123]

Specifically, she wants to say no print that has less than 8 level 2 minutia details will be considered an ID.

this would be an administrative policy, for quality assurance purposes only.

Eight points as an absolute minimum to declare an identification, but that is an administrative policy for quality assurance purposes only? What does that mean? Does it mean that I can decided an identification is correct on seven points, but if I tell anybody I have violated department policy and will be fired?

Where did the number "eight" come from? Is there a valid scientific study establishing a higher degree of accuracy at that number? Or did she select "eight" arbitrarily because it sounds good?

To the best of my knowledge, all of the peer reviewed, published literature including the IAI Resolution you mention, David Ashbaugh, the Ne'urim Declaration, SWGFAST, The Fingerprint Sourcebook, JFI articles, etc., are clear on the point that there is no scientific basis for any predetermined number of level 2 detail.

So . . . . . where did she come up with the number "eight?" Is her source better than those I listed above?

[ER]

Unless your agency has a history of erroneous ID's with 7 or fewer 2nd level minutiae, this seems to be a solution in search of a problem.

[bficken]

but that is an administrative policy for quality assurance purposes only? What does that mean? Does it mean that I can decided an identification is correct on seven points, but if I tell anybody I have violated department policy and will be fired?

Sorry, let me clarify.

The co-worker that is requesting the change recognizes all of the above-referenced research that states there is no SCIENTIFIC reason to have a point standard. I think we all agree on this part.

Her request comes from a quality assurance angle, and trying to ensure that we aren't all saying dramatically different things on the stand when we testify. I think we all know that examiners can disagree. I think we also all know that there is a gray area of disagreement that is acceptable, and an area that is not. For example, if someone is not identifying prints that have no distortion, and have 22 clear level 2 details, then that person probably needs additional training. His/her decision falls out of the range of acceptable disagreement.

Laboratories seem to have various ways of handling disagreements between examiners, and what is acceptable and what is not. I believe it is from that topic that my co-worker is coming from. Essentially she wants to define our "bottom" end at 8 points, saying we do not ID anything with less than that. It is in the same vein as many laboratories who have the top end defined, by saying "We absolutely ID everything that has 12 clear points or more," it's just the opposite end of the spectrum.

We have had no problems before. My co-worker is mostly getting this from her former agency, and from the fact that our state agency has the same policy. I'm just wondering how many other labs out there are doing this.

[LPE123]

Did I neglect to mention the courts? I have not tried to verify the numbers, but I'm pretty sure thousands of courts have accepted the view that there is no minimum number of points required for an identification. I don't believe I have heard of a single court in my career that has accepted a number and refused to admit into evidence anything below "eight points."

If you had a policy that eight points or more requires only single verification, whereas anything seven points or fewer requires blind verification, or double verification, or verification by a supervisor, then I could accept the "quality assurance" explanation. But would you want to declare a clear print with seven solid points and even some good level three detail of no value in a rape murderer case because your policy says eight points?

[Bill Schade]

If you are going to set a number, better make it 10.
The attached image comes from a training class I've taken and there are 9 points of similarity shown in two prints from different sources.
CLPEX also has some close non matches posted that would be a problem for your proposed SOP

[ER]

But would you want to declare a clear print with seven solid points and even some good level three detail...

Or an entire arch with only six points?

Further administrative policies are a great idea. Drawing a line in the sand is too rigid and will result in strong disagreements, discord in the unit, and examiners teasing out further points that don't really exist. See the Evett and Williams paper.

It is known.

[LPE123]

The images Bill Schade posted, I think, must be from an AFIS high score non-ident. So, to take this discussion into a slightly different direction from that originally launched by bficken, should the "minimum points for an ID" be greater for an AFIS hit than for a named suspect?

I remember the first time I heard that proposition. I was adamantly opposed. But as friends have shown me close non-ident hits off AFIS (and I have made one or two of my own), and having taken Glenn Langenburg's class on application of statistics to latent print exams, I think now that there is a valid reason for applying a higher threshold for an AFIS developed candidate than for a suspect with other reasons to suspect him before the start of the comparison.

[timbo]

Just as important as ridge detail in agreement is ridge detail in disagreement - and in Bill's example, there is some very clear ridge detail in disagreement. And, clearly these prints have been specifically cropped - I wonder what appears just outside the areas we can see in the images.....

Throwing out an arbitrary number of points also fails to take into account spatial relationships. Which is another thing that Bill's example has issues with.

I think most people have seen the famed 'no point' ident - how would a SOP like this account for that print?

To ER's point (pun intended), in my opinion you will see your examiners tease out extra points that aren't there just to hit that magic number of 8. This phenomenon is backed by studies.

A quality assurance procedure like that being implemented could be an indicator to other latent print examiners around that the agency has likely had some internal proficiency issues, hence the need for such an administrative policy. From your other comments below, this does not appear to be the case - but it is a QA procedure that is not in line with the generally accepted 1973 IAI or 1995 Ne'urim standards......

From an examiner's perspective, I would see a procedure like this as being indicative of my management not trusting the capabilities, training, experience and expertise of their examiners.

[Peter Griffin]

Our lab deals with this issue as a "Non-routine Verification".
"When the original examiner has completed a case in which less than eight (8) Level 2 details were used to effect the identification, the examiner will notify the Unit supervisor...to invoke the non-routine verification procedure.

The Unit supervisor will assign the case to an examiner for the purpose of verification. If the initial verifier agrees with the examiner, then blind examination by a third qualified examiner will be required."

The above procedure will allow you to ID with less than 8 Level 2, but at the same time enforces a tougher standard for that ID.

I think you should write your manuals to be specific, but not too specific. You want to a good balance between covering things in detail, but being flexible enough to deal with the odd case that you are sure to get every now and then. This applies to both comparison and processing work.
PG

[josher89]

I think a policy like this would make you wind up with paint on the bottoms of your shoes. Being stuck in a corner with nowhere to go is the worst place to be.

Administratively, having a point standard may solve the issue of whether or not a print gets looked at for comparison purposes, but not for a decision of ID or exclusion.

Would this policy prevent you from ID'ing this print (I think it would)?

2015-10-05 10_23_08-Zero Point ID JFI 2003.pdf - Foxit Reader.jpg

[Ducky]

LPE123 asks
"Where did the number "eight" come from? Is there a valid scientific study establishing a higher degree of accuracy at that number? Or did she select "eight" arbitrarily because it sounds good?"


This number came from Dr. Locard's triparte rule which was the basis for the FBI 12 point standard. Many agencies adopted the 8 point minimum from this rule. The 12 point beyond debate was statistically the Galton number for absoluteness. Dr. Locard realized that clarity and level 3 is also a factor to build upon a statistically significant number of points in agreement without discrepancies.

In 1914, Dr. Locard published his conclusions of the fingerprint identification and the criteria that should be used to assure reliability based upon statistical analysis study. His study showed the following tripartite rule summarized as follows [2] :

If more than 12 concurring points are present and the fingerprint is sharp, the certainty of identity is beyond debate.
If 8 to 12 concurring points are involved, then the case is borderline and the certainty of identity will depend on:
the sharpness of the fingerprints;
the rarity of its type;
the presence of the center of the figure [core] and the triangle [delta] in the exploitable part of the print;
the presence of pores [poreoscopy];
the perfect and obvious identity regarding the width of the papillary ridges and valleys, the direction of the lines, and the angular value of the bifurcations [ridgeology / edgeoscopy].

[Neville]

Hi Timbo
I agree, well said.

When we changed from the 12 point rule I made a personal decision to never id anything less than 12 points. But it became clear that my 12 points was another person’s 8 points or another person’s 16 points. Trainers over the years had changed plus we got experts in from overseas, so although we were, in our own ways, well trained we saw things differently. I just did not see a dot as a point.

So to say 8 points is the minimum in my view is very difficult. I would say we look at nothing under 8 points. Even if it had 10 points, I would not say we can ID this one, only it is worth looking at.

But then there was that Russian expert who Ided a print on 3 or 4 points wasn't there, and their courts accepted it.

Ah but that is all in the past and as we were told in my latest district staff training we must make allowances for the new young cops as they see things differently, I think they should re-invent the wheel.

[sandra wiese]

In another recent thread regarding a wrongful conviction based solely on an erroneous identification, I posted this petition by the defense:

http://media.al.com/news_impact/other/d ... tition.pdf

Which will go a very long way, IMHO, to explaining the practical reasons for abandoning this 8-pt idea. In the petition/trial it was 9 points, but the premise remains the same.

Go to that thread to read the news article regarding the basic details of the case, which I believe will make the petition much more understandable. (Links to both in that thread.)