CLPEX Chat Board

Verification may be blind to the initial examiner’s decision, in
which case all types of decisions would need to be verified.

I believe that this just means that while (non-blind) verification can be applied to individualization decisions alone, as is done is many agencies, you can't have blind verification if it's only applied to individualization decisions.

Rah wrote: On ACE returning 4 decisions - ACE *does* only return 4 terminal decisions. If Analysis returns No Value, we're done; if it returns a Value decison, we aren't done, and we proceed to one of the 3 Comparison/Evaluation decisions. A value decision isn't an end state. There can be a variety of paths through ACE, but there are only 4 distinct results that come out at the end.

Two decisions, five possible results. In the analysis portion, you make one decision that has two possible results - value/no value. In the comparison portion you make one decision that has three possible results. Final score: Decisions - 2; Possible Results - 5.

It's really a semantics issue, kind of like, "it depends on what "is" is," to paraphrase a past President. But, in this field, it's good to keep our semantics in order, I think.

Hi all,

I agree this has become a discussion about semantics which underlines the importance of how to explain our thinking and decisions relating to the evaluations of finger marks. Many moons ago I've posted about the examination of latents and the semantics of Individualisation as a step preceding Identification. Adding to this current topic about decisions leading to a conclusion I think it is valid to include our ability to discern that the 'mark' "is" a latent fingerprint, as opposed to other types of marks. Following from this decision then the ACE part and V. The reliability of a decision of fingerprint individuality is dependent on the principles of biological uniqueness and the detection of sufficient objective features to support the conclusion. Understanding the principles of biological uniqueness leads to the ability of explaining how the fingerprint is formed and why. Here we dont have to rely upon semantics as researchers such as Babler and Munger have done most of the heavy lifting. Recording the objective data used in the examination requires modern computer technology with a validated procedure. Combine these things with our competency to make accurate observations and our confidence to defend the conclusions we reach then reliability is guaranteed. Regards from oz. Les

Hopefully getting away from counting the number of possible outcomes, which is obviously not the point...

The paper measures a number of different rates including:
TPR - true positive rate (calling the ID)
TNR - true negative rate (calling the exclusion)
FNR - false negative rate (bad exclusion)

So if the TPR drops too low, you're using Inconclusive too much instead of ID.
If the FNR goes too high, you're using Exclusion too much instead of ID.

It seems that an agency would want (in general) to raise TPR and lower FNR. TNR could be allowed to fluctuate with little effect on the system. TPR and FNR are obviously related and this data shows that examiners with high FNR tend to have low TPR.

How can we identify these examiners and give them training to reverse both of these rates? (While still keeping bad ID's super low.) This is one of the biggest question that this study raises.