Include the Candidate List?

[RM1023]

For several weeks now I've been going back and forth in another thread about this topic. One side feels that the candidate list made of identification numbers only, should not be printed out and included in the case file since it will only lead to more problems. The other side feels like it should and that it will only improve casework. Here's a quote:

Re Do the defence have the right to ask who else you compared it with??? May I ask a question or two which will probably answer that one; when you were searching manually in the Henry collection did you supply the name on every fingerprint form you turn over in your manual search for the matching print?

I have not searched through Henry cards before, but....according to today's standard of report writing yes I would. I would not have to report the ones I flipped through to get to the correct section, but all the ones I pulled to look at. That would really be a pain the butt, good thing they are mostly all electronic now.

Do you really compare every print on the list once you have an ident on the first one (with a score of say around 6000) on the list when the other 49 on the list have scores of around 100 or less (if you do why)?

Right now, we do not search past the hit. This is still in debate in some corners. Some believe that if your SOP says you look at ##, then you really have to look at ## regardless. Some say just add the sentence, "only if there is not a hit" to the SOP. We have been asked this multiple times in court. The usual answer is, "I believe that every person's fingerprints are unique, so if I do find the correct person, there is no need to look at the others that I know cannot be the correct person. If I have 50 keys on my key ring, but only one opens the lock, why would I try the other keys once I found the correct one?"

Do you really think that stopping at the 50th candidate answers the obvious next string of questions? Why did you stop the list at 50, the true offender may have been 51 on the list, did you compare with every one on every other data base, blah blah blah.

We have been asked this multiple times in court as well for non-hits. The usual answer goes, "by the 50th candidate, the scores were so low (and we have the print out to prove it) and that the prints coming back were so off that further searching would be fruitless." It is really up to the examiner if they wish to continue. In the rare situation of when the top 50 are really close and the scores are still pretty good, then maybe querying for more candidates is warranted. The number in the SOP is for quality control to say we at least looked at this many and that number was based on past experience with the system and where the hits generally show up.

Anyway why did you or the powers to be choose 50, is there any scientific basis for the number 50. Seems to me you are making life very hard. I think ours is set at 20 maybe 25 though we would never say we compared all the other 19 candidates, what is the point in that.

This is in debate as well. I think my lab chose 50 because people have gotten hits in the 30's and 40's before. Different agencies have different numbers for different reasons. I remember one was just so they can say that they do more than the FBI. Another was because they do not have the man power to spend x number of hours on AFIS per case. There is really little scientific evidence about the probability that the correct person will show in the first ## and it may become a big issue in the future. I think Dr. Dror mentioned something about this in one of his studies (the one about moving the hit back past 20 to see if the examiner is really looking that far).

TL:DR ....

1. Pros : proof that you actually looked at AFIS and you can actually say who was compared and who was not compared (in the case of a hit right off the top).

• Cons: will only bait the defense to ask questions about AFIS such as why did you not look at all the other candidates, or why only that many.

So what do you guys think? Should a candidate list be in the case file?

[Dogs Nose]

No. As far as I am concerned it is irrelevant. AFIS doesn't go to court. You are the witness. If I am asked about AFIS, I just explain that it is a search engine much like Google and is dependant on the quality of the input at both tenprint and latent stage. Due to the pliabe nature of friction ridge skin and the variable qualities of tenprints (offender compliance/police skill or laziness etc), it can and does miss matching persons existing in the database so I would argue a list means nothing anyway. What does it bring to the table to have 50 names you have candidate checked when there are 6 million (or however many where you are)tenprint forms on the database? I don't know about you, but I don't identify anything on NAFIS but use a set of tenprints vs original crime scene images on an entirely different computer where I start the ACE-V comparison process from scratch. AFIS may just have been how I came to doing the comparison.

Our SOP's have a minimum number of candidates which must be checked on the list for volume crime. All are checked for Major Crime. The latents are stored on our AFIS in any event and can be re-searched at any time to recover a list if every needed. If I comply with my SOP's (which I do), I answer as such. I don't write the rules, I just comply with them.

Also, if you check on the candidate list past the ident you can't have any faith in the science.

[David Fairhurst]

We have, on occasion, been asked for a candidate list in cases where the latent is unidentified and the investigators are stumped.
Why do you want the candidate list? We ask. They say it might provide a lead.
When we explain that the candidate list contains the names/numbers of those people we know definitely DIDN'T leave the latent, while the offender may still be somewhere in the system but wasn't on the list, they shut up.

[David Johnson]

When we explain that the candidate list contains the names/numbers of those people we know definitely DIDN'T leave the latent, while the offender may still be somewhere in the system but wasn't on the list, they shut up.

I don't think that this statement is true. If you look at the list then you eliminated that finger of that person but you did not look at all ten of the fingers so how can you have eliminated that person. (Unless of course you know it was a #3 finger before your search).

That is one reason why we do not report the list of candidates because even though they were on the list it could have been that AFIS just hit the wrong finger of the right person. I am sure that is a very small possibility but since we did not actually check the other 9 fingers of the person we would not eliminate them.

[ER]

There are a few other reasons to maintain a copy of the candidate list in the case file or, at least, to be able to access an electronic version of the list from the AFIS system.

QA/QC - If an examiner is discovered to have missed a couple of AFIS hits, QA may want a review done on a number of AFIS cases to see if others were missed. A candidate list would allow them to check for past errors.

CYA - Let's say a latent doesn't hit but is saved to the unsolved file for ongoing search. Later, it hits to someone who's been in the system since before the original search. A candidate list will easily show that the person wasn't one of the original candidates. Wouldn't you want to be able to prove to your boss that you hadn't missed it the first time?

Keeping the list is required for ASCLD-LAB/ISO labs (as far as I know).

It is reasonable to assume that that a defense attorney will eventually demand to have the candidate list as part of discovery. We all know that it's pretty useless, but someone will eventually ask for it. It is likely that a judge will eventually side with the defense and force you to turn it over. You can probably convince most judges about how useless the list is, but eventually one will tell you to find the list and turn it over. Isn't it just easier to print out the list in the first place?


As for searching past the hit...

Also, if you check on the candidate list past the ident you can't have any faith in the science.

Um, no.

Checking further down on the list after a hit accomplishes three pretty important things for your investment of a couple of minutes.

First, it lets you check to see if your bad guy is listed twice in AFIS under an assumed name. Of course, this almost never happens, but it could be pretty important when it does. I have faith in the science. I don't have faith in AFIS.

Second, it lets me look for close non-matches, which I collect. It is extremely important for our science to be actively looking for the closest non-match that we can find. This will constantly improve our understanding of the limits of how close two fingers can look, and help with training our new folks.

Third, it lets me say in court. "This is an ID. I checked every other candidate on the list which is comprised of the closest fingers out of the X million fingers in the database, and none of the them were even close. Except the one, which is an ID."


That's why you keep the list, and you check them all.

[David Johnson]

Keeping the list is required for ASCLD-LAB/ISO labs (as far as I know).

What ISO requirement are you talking about? I believe there is something in ISO stating that you have to notify if you search a print or if you register it in a database but you do not have to document the candidate list of your negative searches. Now if your own SOPs state that you have to save the list then I would say that you need to change your SOPs.

This is an ID. I checked every other candidate on the list which is comprised of the closest fingers out of the X million fingers in the database, and none of the them were even close. Except the one, which is an ID.

Again are they really the closest fingers in the database. I know that when I search a loop I will get returns in the top 25 that are not always loops. The AFIS triangles or score may be the closest out of the database but I don’t think that is how an examiner would rate similarity.

I agree that there may be some small benefit with the checking to see if the same person is on the list a second time but it seems like a waste of time.

Just like if I searched multiple databases at the same time I wouldn't then check the additional searches after I had made an ID.

[ER]

I'm not familiar enough with ISO requirements to know for sure. I made the assumption that they require the candidate list after talking with someone whose lab was required to demonstrate that they could retrieve electronic versions of the candidate list during their ASCLD-LAB/ISO audit.

Are they the closest fingers in the database? Well, yes, if you're measuring by the AFIS score. No, if you're measuring by a different standard.


So you've still got benefits to the QA department to check examiners' cases for a pattern of errors, to the examiner to prove that they didn't miss an ID when the latent hits on an unsolved search, to the trainees who get to see a collection of close non-matches, and to the AFIS system by cleaning up duplicate records.

These benefits may not be worth it to some to save the list or to compare everyone on it, but they are still benefits.

[RM1023]

As for ISO.....as far as I know, it states that every examination be recorded/reported and that all examination documentation be kept. So you "examined" the print on the screen and came to the "conclusion" that it is or it is not a hit. So somewhere there needs to be a report stating what you did and whatever you used as the "exam document" must be kept. Since it is impractical to print out every ten print card or get screen captures for every candidate, a copy of the candidate list can serve as a substitute since the cards can still be obtained with that information.

Now then there's the argument that looking on the screen is not really an "examination" since the real comparison is done off the ten print card manually so AFIS is just an admin document, a procedural step that within the real comparison. If that's the case, then just a simple summary would suffice and no candidate list needed unless there is no hit.

Then there is the argument that the whole purpose of ISO standards is to be able to account for and replicate every action, and come to the same conclusion as the examiner. So you will need some way of retrieving all the candidate's you "examined" in AFIS to account for your actions. Printing the candidate list is the simplest way to do that.

Then there is the argument that the AFIS transaction number should be enough instead of the candidate list itself......and it goes on and on till either the QA Manager or the Latent Print Supervisor gives in.

[bficken]

I feel as though this is one of those issues that everyone will go back and forth indefinitely, because there is a logical argument to both sides. But for anyone out there who is taking a poll on who does what and why...

We do include the candidate list. I don't believe our ASCLD-LAB/ISO accreditation specifically states "the candidate list of an AFIS search must be included..." but I didn't read every line of their requirements. I believe it instead falls under the ISO requirement of all work done by an examiner must be presented in such a way that every step can be duplicated/replicated by someone else, without the original examiner ever needing to be present. The only way for someone to repeat my AFIS search/examination is to know which candidates I looked at. We choose to cover this by putting the candidate list into the case file.

It also does help with QC. There have been plenty of 'no hit' AFIS cases whose latents were later identified to suspects that detectives came up with on their own. It is always helpful to be able to go back to the original AFIS search candidate list, and prove whether or not the examiner made an error. In addition, our QC policy states that a certain number of our cases will be reworked annually, to ensure our work is up to standard. Providing the candidate list enables the process of reworking our AFIS searches.

As for the court issue - this probably depends on where you are and how well the defense attorneys are trained. We get asked questions such as "How many candidates did you choose for your search...Why did you pick that number and not more...Did you compare it to every candidate or did you stop with the one you identified it to..." etc. even when the attorney has never actually looked at our case file. Personally, with our attorneys I feel that the more thorough I am in presenting my work, and the more information I provide to back it up, the less opportunity they have to poke holes in anything and the less questions they ask me when I testify. However, I can also see logic in the concept of "the more information you supply, the more you open yourself up for questioning." This part of the issue is the one I feel everyone can argue endlessly without a solid conclusion, because we will all draw that "how much information do you volunteer" line in a different place.

[josher89]

It is always helpful to be able to go back to the original AFIS search candidate list, and prove whether or not the examiner made an error.

I would hope that isn't a primary reason to keep the candidate list in the case file; I don't think it's going to be very productive to go back and "prove" if an error occurred. There could be many reasons why a candidate was missed; poor quality exemplars being the number one reason. We always include a copy of the fingerprint card we used for the comparison in our case file so that if the suspect does ultimately get identified, and that suspect was eliminated by an examiner, the hope is that the card used for the comparison was poor quality or it just didn't contain the ridge detail present on the latent lift. We all know that 10-print cards that are obtained aren't always collected by trained personnel or aren't always the best quality. I know it is our responsiblity as examiners to do a thorough job but if someone has 15-20 arrest cards in AFIS, are we to look at each card to find the 'perfect' one? I don't think so. I think we need to make sure we include what we looked at and what we used for the comparison in the case file but we can't possibly look at every card on file for a suspect. When reporting a non-ident, it should be reported "not identified to DOE, John, DOB: 05/06/1969 at this time". '...at this time' gives you options because it may still be possible that they left the print but you couldn't identify it based on the card you looked at. This still doesn't mean it couldn't be used for disciplinary reasons, I just hope it isn't the primary one.

I've heard about an agency that wanted to use a non-trained to competency examiner (the only one in their department) to perform AFIS searches so they could get a candidate list to use as an investigative tool. Basically, it sounded like they would locate each person on the candidate list and interview them about their involvement in that particular case. With a bit of a *gasp* and a couple of phone calls, the matter was quickly resolved--they realized that the candidate list returns what computer thinks is similar, not what is similar. I don't know how the candidate list actually serves a global purpose for a case file, other than being transparent. I think every AFIS database can, or should, be able to go back and print out the candidate list if necessary but it's (usually) only one page so I don't see the big issue.

That's my two cents.

[bficken]

Josher -

Hypothetically an examiner runs a latent in AFIS, and decides that the latent does not identify to any of the candidates. Some time later, detectives develop a new suspect. Comparison of the latent to the new suspect results in an identification. Then it is discovered that the suspect was in fact on the original AFIS candidate list. In that situation - the examiner would be pulled off of casework until an investigation into how the 'missed ID' occurred can be conducted. Some level of remedial training and possible supervised casework would follow, before the examiner would be allowed to return to independent casework. The theory is that even though tenprint cards can be of poor quality, an examiner should recognize that possible distortion is occurring. Therefore the examiner should try to pull up an alternate record for that person that has a clearer version of the known, before exacting a conclusion. In other words, (except in very rare occasions)there is no justified reason for a missed ID, even in AFIS searches.

This is all part of our quality control. Please keep in mind that my comments do not reflect my personal opinion. I'm just reiterating the way things are managed here.

BUT, errors and how to handle them were not the topic of this thread. Regarding the keeping of the candidate list - the attempt to catch an error is not the 'primary' reason they are kept, no. But the lists are and can be used for that reason.

[Boyd Baumgartner]

Then it is discovered that the suspect was in fact on the original AFIS candidate list. In that situation - the examiner would be pulled off of casework until an investigation into how the 'missed ID' occurred can be conducted.

What about the Verifier? Wouldn't he/she be just as responsible as the Examiner?

[bficken]

Boyd - We verify only identifications, not exclusions. So the original examiner is the only one who looks at the images of the candidate standards in a 'no hit' AFIS search.

[Tazman]

What about the Verifier? Wouldn't he/she be just as responsible as the Examiner?

On AFIS candidate lists? Do you actually declare "exclusions" on the candidates on an AFIS list and have them all verified?

If that's your policy, I guess that's fine. But it just seems to me like a lot of wasted time going overboard for very little gain. I mean, if you figure how to spend your time with a sort of "cost/benefit" analysis, the cost of verifying non-idents to report out as verified exclusions for every AFIS search seems very high in relation to the rare benefit of catching a missed ident.

[Boyd Baumgartner]

Tazman, we do declare exclusions on AFIS runs and have them verified.

Obviously, at the end of the day it's an organizational policy decision and an interesting discussion to say the least because there's a balance to be struck. I think the balancing act which stands out to me with regards to bficken's mentioning of the Examiner being taken off of case work is that if it is determined to be that important to remove someone from casework, shouldn't that be an indication that it is equally important to then verify the work?

If strict labor intensity/pragmatic philosophy is what drives the organization you could:

• Only process fixed items from the interior of a scene because exterior items do not prove the person entered or committed any crime found to have occured on the interior of the car/house/etc and transient items could have been touched prior to or after the commission of a crime.
  
  Only lift AFIS quality prints.
  
  Only ID prints with a high administrative point standard.
  
  Only run impressions with known orientation and anatomical source (finger/palm)
  
  Only Verify Individualizations
  
  Choose the single most successful chemical processing technique by item
  
  etc.

Obviously at some point, the quality of the service you provide begins to diminish due to lack of thoroughness and I think you run into ethical dilemmas. Where is that point, I for sure couldn't pin point it, but I'm sure you would agree?

I guess the two biggest questions I would ask you in response would be:

• Are you really using ACE-V if you aren't verifying exclusions?
  
  Is your understanding of scientific methodology strictly positivist in that the aim of science is to positively establish truth, or is it to weed out error with the residual conclusion being the best explanation?

I believe for sure it's the second of the two understandings. If that is the case, you would expect to find more error in organizations using 100% ACE-V with the understanding that erring on the side of transparently weeding out error is best case scenario.

Thoughts?