The Moral Maze

[David Fairhurst]

Dear all, it's bee a long time since I had recourse to consult you all on a fingerprint matter, but I have a moral dilema to put to you. Don't worry, I am not on the horns of this particular dilema, it's just a "what if".

You are called out to deal with a latent print submission in the case of a child abduction.
It is less than 24 hours since the disapearance of the child.
You get a hit on AFIS which could lead police to the offender swiftly and prevent loss of life.
To get the identification verified would mean calling out another examiner and a wait of 1 hour.

Do you wait for the verification before letting the investigator have the name, knowing that every minute might be critical in saving the child's life?

[Tazman]

Hi David

Do you care more about people as individuals? Or do you consider yourself a loyal member of your agency?

Have you ever been to the website http://www.yourmorals.org ? On that website, the "Moral Foundations Questionnaire" measures five dimensions of morality. These are as follow: 1) Care for others, 2) Fairness, 3) Loyalty, 4) Respect for Authority, and 5) Purity. A person who cares for others individually generally will score higher on the first two, lower on the last three. A person who is a good member of a group will score higher on the last three. If you are a person with a very high score in the Care and Fairness categories, you would tell the investigator and consequences be damned. Your care for the child would outweigh your respect for authority. On the other hand, if the care component of your personality is lower and the respect for authority and loyalty components are higher, you would wait for verification.

That website is the work of modern normative philosopher Jonathan Haidt. His book The Righteous Mind explains those dimensions of morals and ethics with special application to political philosophy, but consideration of those dimensions can be used to understand other factors in personality, as well.

A few years ago in my hometown, an off duty cop who had been working in his garden went up to the corner convenience store to buy a six pack of beer. He just walked up there, neglecting to go into his house and get his handgun to take with him as required by department policy. He entered the store, said "Hello" to the clerk, went back to the cooler to get a pack of beer, and headed back to the front of the store. As he rounded the end of the aisle, he saw that a robber had come into the store with a gun and was holding up the clerk. The cop grabbed the closest thing off a shelf, a can of green beans, and hurled it at the robber, hitting him in the back of the head. As the robber reacted to the clobbering, the cop took him to the floor and held him while the clerk called the police. Our friend the officer got a meritorious service award for his quick actions in capturing the robber. He also got three days off without pay for violating department policy and not having his handgun with him.

So, are you willing to violate department policy to, perhaps, save a child? If so, are you willing to take your spanking? Or do you want to follow policy and risk the child's life. What if he is raped and murdered due to the hour's delay?

The bottom line is, in which case would David Fairhurst sleep better tonight? Only you can answer that.

[Bill Schade]

Well, so much for "blinding" the practitioner to case information

Your question raises three points in my mind before I can give you an answer

1) This would depend on the strength or level of association of the comparison. You make no mention of whether this is a "Helen Keller" comparison or one you have reservations about.

2) You also fail to utilize the technology available if you wait for an hour for someone to verify your work. With scanners and email, I can reach out across continents and oceans when necessary to get a second opinion on my work. In your scenario, I would be doing just that.

3) without case information, this AFIS search would probably sit a lot longer than an hour while the practitioner went to lunch, or finished a shift, or just switched to completing reports instead of doing comparisons. I think timeliness is always important in getting results back to investigators and as practitioners we have to apply ourselves efficently. I'm not in favor of with holding case information in the name of preventing errors. I think point number one above is a better method of avoiding errors, we all know when we need help with a complex comparison.

The short answer to your question for me is: I call the investigator I will risk the spanking

As my cousin Vinney said in the movie "I could use a good "spanking" sometimes

[David Fairhurst]

Tazman and Bill, thanks for your replies.
Taz: I do score high in care and fairness, even without taking the test I know that. I think that what happened to the officer in your example was correct, he got the commendation and the sanction.
I would save the life and take the spanking (without any thought or regard for the potential commendation, I would sleep well; that's enough for me.)
Bill: on your points 1)let's say its an easy ident initially, but keep reading. 2)So would I if my departmental policy allowed me to seek verification from another agency, it doesn't.

Let me modify the scenario in order to highlight a different issue:

Bill, you mentionned blinding the scenario from the examiner, let's do just that.

I'm working the same case but this time it's in office hours and there are 12 examiners available to verify the identification.
The first examiner approached cannot verify the identification and returns an inconclusive result. This is clearly now a complex comparison and we have a procedure for those.
The case is refered to a supervisor who blinds it and hands it to 3 further examiner with no knowledge of the scenario and only the latent and inked prints to work with.
Two of these blind reviewers verify the identification, the third is inconclusive.

what now?

[Bill Schade]

Investigative Lead


I was trained to the "its a boy or a girl" standard, so this is difficult for me

But I'm beginning to see value in quantifying the confidence/difficulty of identifications and letting detectives investigate leads

[L.J.Steele]

I understand the temptation to cut corners, but ... if you are wrong then you would be diverting the investigation from the actual culprit to your false lead, which may delay finding the victim. Your scenario assumes that your suspect is the culprit. You didn't say whether the AFIS search was national or local. If national, what are the risks of a false positive in that big a pool.

You need to factor the risk of sending police down the wrong alley into your decision, whatever that may be. With multiple verifiers coming back with inconclusive ... that's a judgment call, but I'd expect you'd want to clearly flag this to the investigator as a lead, and be clear that there's dispute and it may not be the right answer.

[Bill Schade]

So our critics/saviors say they need to know the weight of our conclusions. We need to define the level of association. We should put things in terms of probabilities and let the "system" decide how much consideration our evidence should receive.

And when the subject comes up you refer to us "cutting corners"

I think that explains why we are resistant.

[David Fairhurst]

So, does one inconclusive examination kill the ident?
if not one, how many?
How many conclusive individualisations by competent examiners do we need?
Is the only sound identification one that everyone can see?
are all examiners created equal? Or trained equally?

[Bill Schade]

I think we are starting to realize that putting weight in the number of people who "agree/disagree/fail to commit" is not a scientific principle and instead we should put weight in the logic behind their conclusion. Perhaps the third examiners inconclusive in your example is the correct answer. Without showing the data and the logic, we can't judge the "correctness" of the conclusions.

I think one "conclusive individualisation" is enough if you can demonstrate it to others. Maybe your statement that the only sound identification is one that everyone can see is prophetic.

I know there are many gifted examiners out there, but training and experience are no longer the only measure of our conclusions. If a gifted examiner can see something, so should the rest of us.

[L.J.Steele]

So, does one inconclusive examination kill the ident?
if not one, how many?
How many conclusive individualisations by competent examiners do we need?
Is the only sound identification one that everyone can see?
are all examiners created equal? Or trained equally?

It's a good question. What happens if we turn around the example.

Latent comes in on a workday afternoon on a distinctive surface. Examiner looks at it, runs it thru national AFIS, looks at possible candidates and says inconclusive. (Examiner knows virtually nothing about the case) An hour later, media breaks the story of the child abduction with enough detail for examiner and his/her office mates to realize that the evidence likely came from that case. 1st examiner asks 2nd examiner to look at the inconclusive result. Verifier says, no this is a match. When it goes to the panel of 3, again they disagree 2 say inconclusive, 1 says match.

Does changing who has the biasing information and changing the order of the disgareement change the way you think about the problem?

I think the answer would be why do the examiners disagree -- tho that discussion may be hard to have without adding risks of peer pressure (real or perceived) on whoever is the outlier.

So our critics/saviors say they need to know the weight of our conclusions. * * * And when the subject comes up you refer to us "cutting corners".

I don't think I've ever said I want to know the weight of the conclusion, at least not in testimony. I've been following the discussions about probabilistic testimony and results of "consistent with" or "some similarities" or "cannot be excluded from" with interest and some concern. I do want to know if there's disagreement, and how that disagreement was resolved, which affects the probative value of the conclusion.

Didn't FBI just admit that its DNA population figures had errors?

http://www.post-gazette.com/news/scienc ... 1505300112

(I don't know enough about the details to know how much of a cautionary example this is.)

Investigative leads -- that's a more grey area. But if you have a procedure that says ACE, then V, and you skip V, that's significant.

[Tazman]

The truth is that the vast, vast majority of latent print identifications are correct. The number of true erroneous identifications is very, very small. And even among erroneous identifications, they are frequently verified. The number of erroneous identifications caught during verification is low.

Even for an AFIS hit, the number of close non-idents is very, very small. If you have an AFIS hit of a dozen solid points with no dissimilar points, then the changes are overwhelming that it is a correct identification.

The touchstone of science is verifiability. Not necessarily that every conclusion be verified every time before a conclusion is announced. That is not a demand of science. It may be a good quality control measure, but it is not a demand of science.

So, if you have a good identification, not a borderline consultation/inconclusive type latent, but a good, solid identification, and if a delay in telling the detective may cause harm to a victim, then I would consider violating policy for a greater good as a viable option. In the hypothetical originally proposed, we were talking not about a conviction, but about providing information to an investigator to target a specific individual. While I agree that verification is a good policy, I would suggest there may be occasions when it would be reasonable to consider violating policy in an attempt to save a life.

If you are talking about a complex latent in which no unanimity is possible among examiners, then perhaps you should back off anyway, even if it is just a baggie of marijuana for a first time offender.

[David Fairhurst]

I think we are starting to realize that putting weight in the number of people who "agree/disagree/fail to commit" is not a scientific principle and instead we should put weight in the logic behind their conclusion. Perhaps the third examiners inconclusive in your example is the correct answer. Without showing the data and the logic, we can't judge the "correctness" of the conclusions.
I think one "conclusive individualisation" is enough if you can demonstrate it to others.

The touchstone of science is verifiability. Not necessarily that every conclusion be verified every time before a conclusion is announced. That is not a demand of science.

So let's look at the bench notes of the five examiners.
The three who identify the latent have comprehensive notes following an established protocol. They record agreement in 1st, 2nd and 3rd level detail as well as some minor flexion creases. Each makes note of the same distorted area which was problematic. Using sound reasoning they all provide similar explanations of the cause of the distortion and conclude that the detail observed in the undistorted areas is sufficient to identify the latent.
The notes of the first verifier, who had contextual knowledge of the case, are sketchy. They remark that due to differences in a distorted area of the latent they are not comfortable in identifying the latent and so deem it no value for identification.
The blind reviewer makes no mention of a comparison of 3rd level detail or flexion creases. While not put off by the distorted area, they conclude that there is insufficient detail in the rest of the latent to identify it.

Do we require unanimity in the bottom lines to declare an identification?
Should the supervisor take account of the reasoning and logic expressed in the examiners' notes?
Should SOPs limit the supervisor to reporting the latent as unidentified or no value where the reviewers vary in their conclusions?
What would have happened if one of the identifying reviewers had been asked to do the first verification?

[David Fairhurst]

What happens if we turn around the example.
Latent comes in on a workday afternoon on a distinctive surface. Examiner looks at it, runs it thru national AFIS, looks at possible candidates and says inconclusive. (Examiner knows virtually nothing about the case) An hour later, media breaks the story of the child abduction with enough detail for examiner and his/her office mates to realize that the evidence likely came from that case. 1st examiner asks 2nd examiner to look at the inconclusive result. Verifier says, no this is a match. When it goes to the panel of 3, again they disagree 2 say inconclusive, 1 says match.
Does changing who has the biasing information and changing the order of the disgareement change the way you think about the problem?

My way around you have 1 biased identification, 2 unbiased idents, 1 biased inconclusive and 1 unbiased inconclusive.
Switching the scenario as you describe. If the review panel does not know about the case then, with the first examiner, you have 4 unbiased conclusions: 2 identifications and 2 inconclusive. Add in the biased verifier and you have 1 biased ident, 2 unbiased idents and 2 unbiased inconclusives. Given that the bias is supposed to push the examiner to make the identification, I'd say there is no effect and you may proceed as before.
If, however, you have been unable to find 3 examiners with no knowledge of the source of the latent then yes, you have a different scenario altogether. All your Idents are biased and your only unbiased examiner was inconclusive. Perhaps it's time to pick up the phone or email another agency to seek blind verification as Bill suggested earlier.

[L.J.Steele]

The truth is that the vast, vast majority of latent print identifications are correct. The number of true erroneous identifications is very, very small. And even among erroneous identifications, they are frequently verified. The number of erroneous identifications caught during verification is low.

Even for an AFIS hit, the number of close non-idents is very, very small. If you have an AFIS hit of a dozen solid points with no dissimilar points, then the changes are overwhelming that it is a correct identification.

The touchstone of science is verifiability. Not necessarily that every conclusion be verified every time before a conclusion is announced. That is not a demand of science. It may be a good quality control measure, but it is not a demand of science.

Agreed. A genuine erroneous ID is rare, tho one needs to be aware of the possibility. Fabrication/fraud is more common, and also worth pondering particularly in a high-stakes investigation where someone may feel pressure for evidence to arrest or search a suspect. The sloppiest crime scene work I see in cases is frequently in what become capital cases where emotions are high during the investigation.

I would posit that the reason for the excellent track record is consistent, rigorously high standards and good QA. I would say use a high-quality inconclusive or unverified ID as an investigative lead with great caution, consciously considering the risk that an error may send investigators down the wrong track, and document the heck out of what you are doing and why.

[Tazman]

Dear Ms Steele,

Wise advice, indeed. And I truly appreciate your flexibility in recognizing that there may be a rare situation in which, if the examiner is well aware of the risk and is willing to take it, there might be a valid excuse for doing so, but with extensive documentation to show why.

Thank you!