Wednesday, April 26, 2006

Science of Elimination Utilizing ACE-V

The Science of Elimination Utilizing ACE-V

    The debate over dactyloscopy’s conclusions of individualization and exclusion using ACE-V has provided us with a new appreciation for the detailed demands of science.  In essence, science is not accommodating nor is it polite. Science is simply a method that allows theories to logically develop and evolve.  A difficulty in defining the science of fingerprint identification, within ACE-V and the Scientific Method, is in understanding the related supporting statistics and the uncertainties.  One of the areas of friction skin comparison that abruptly encounters the threshold of invalid statistical support is in the hypothesis of elimination.  Due to uncertainties in many elimination type comparisons, it may not be judicious to routinely offer hypotheses of exclusion relating to friction skin comparisons.  In fact, it may be unscientific.  

    On average, examiner provisional (non-verified) hypotheses of exclusion may have an unacceptable rate of error.  This is mainly due to the fact that examiner’s conclusions cannot always be proven on a practical basis, and that many particular search attempts can be considered less than a scientific endeavor.  Generally speaking, the search can be considered and investigation, an art.  Here is an analogy; How many times do you comb a large grassy yard looking for a small 9mm case, before you can call the elimination of the yard, as containing the evidence, scientific?  First, we would not know for sure the evidence was ever there, and if it is there, it may have been overlooked due to the vast area and perspectives needing to be covered for such a small object.  The amount of information needed to "accurately process" such a hypothesis generally exceeds our ability to quantify it.  A search of a basketball court for the same object is easier to search, and the elimination of the basketball court is much easier to prove on a practical basis as the needed perspective angles to be covered is very small.  Our proper hypothesis of the grassy yard search for this 9mm case should be "inconclusive".  However, this does not convey the great effort made by the investigator(s) crisscrossing the yard multiple times in a grid search.  Regardless, we can't eliminate the yard as not containing the 9mm case.  We simply don't have that information.  The 9mm case may still be there, hidden from our view, perspective, or other ability to locate it with relatively reasonable effort and within a reasonable time.  In additional, elimination is a scientific conclusion regarding a comparison that has satisfactorily passed a formal comparison methodology.  It is important to note that a formal scientific based result must be derived from a scientifically aligned comparison methodology and that result must be provable on a practical level.  In our case, elimination must also be verified by a peer.  Both "individualization" and "elimination" are formal products of ACE-V, not just ACE.   The courts are expecting the application of the full methodology, not a truncated version of it.

    The failure to find an existing print identification is a provisional (unverified) elimination error.  If the second examiner utilizing the accepted comparison methodology also fails to find the existing match, thus duplicating the original type of error, a formal elimination error has occurred.  This failure is due to an incomplete evaluation of available information.  This “available information” must also include an assessment and understanding of the value of the information being analyzed. A proper analysis of the value of the information set, or rather, an understanding of how particular information can support a conclusion, is imperative to the correct application of the comparison methodology. With the common difficulty of comparing and finding low quality print matches, within a single or multiple set of exemplars, it is recommended that a second qualified examiner verify all print comparisons.  Early, SWGFAST guidelines did not require verification of an examiner’s provisional hypothesis regarding elimination.[1]  Should ACE elimination errors occur with frequencies of an estimated 8-10%; we could safely say that verification is indeed necessary.  Should we want to align the process with the scientific method, we would also need to say elimination verification should be mandatory.  In many complex cases, and those with many friction print comparisons, the correct conclusion is; Inconclusive.  Some agencies appropriately use the term "No Match Found" to indicate that search efforts have been made, yet a reasonable likelihood exists that a fingerprint match is yet undiscovered.  Better exemplars or more complete exemplars may assist in process.  Most persons are not good with intuitive probabilities.  It is not reasonable or scientific to make an educated guess that there is no match.  If better exemplars are needed, such a note should accompany the conclusion of "inconclusive".

    With friction skin comparison elimination, the goal is to show that the friction skin of a particular person did not make a print impression.  The difficulty is that, elimination must represent all the palmar friction skin of a specific person, and in some cases, include plantar prints.  A conclusion of elimination states that; the subject was not the source of the friction skin print in question. An error of elimination can happen when the print impression, made by the subject, was overlooked by the examination team and/or there was an error in understanding the supporting statistics /uncertainties involved.  It’s this second point that is the cause for concern here.  The examiner may not understand that their conclusion of elimination is not scientifically supportable.  Essentially, the scientific method was not followed in its proper format or in its intent.

    With many comparisons, proving elimination can be extremely difficult. Yet, it must be practically provable according to the established guidelines for uniqueness and of sufficiency of information. Thus, it follows, that the certainty of elimination may remain unsubstantiated.  As with a conclusion of individualization, elimination is also a formal scientific conclusion.  Such a conclusion would need to be illustrated via prediction and comparison protocols coupled with thorough evaluation.  In many cases, low quality and incomplete exemplars or comparison prints themselves do not allow for thorough comparison of predicted characteristics. Thus, these uncertainties would scientifically prevent a convincing conclusion of elimination. With regards to proof of elimination, the key word is practical, whereas the term practical applies to the limitations of a science-based application. “…There will never be a formal method of determining for every mathematical proposition (in our case prediction and comparison) whether it is absolutely true, anymore than there is a way to determine whether a theory in physics is absolutely true.” [2]

    Assuming the examiner is qualified, we must understand the dynamics of the search difficulties to understand how accurate examiners can be.  Remember the familiar adage of “A needle in the haystack?” This has a meaning that a search would prove very difficult, likely unsuccessful, or in some cases near impossible.  This is the equivalent to searching a low quality print impression, against a large and technically incomplete exemplar set or exemplar haystack.  Technically, for a scientific exclusion of the needle from the haystack, one must show that each piece of hay does not conceal the needle or that the needle is too big to be concealed by the areas searched.  With fingerprints, one may be able to show that the search print has a whorl pattern and the exemplars can be shown to contain no whorl patterns.  Exclusion can then be proven, as it would not be possible for the print to have been made by the exemplar prints.  If our search print lacks a pattern type or other level one feature, the difficulty of the search increases significantly and the reliability of conclusions resulting from the search diminishes.  This creates an increased level of uncertainty that can for our practical application, render a conclusion of elimination non-provable.  A non-provable conclusion would also be an error of elimination.  The correct hypothesis would be that the comparison is inconclusive. An additional aspect of uncertainty regarding exclusions is that the examiner’s mental search efficiency is reduced with lower quality print impressions.  This compounds the issue.  

    The key to maintaining a reliable science is to understand what conditions of a comparison scientifically support exclusion.  Examiners cannot presuppose that their "intuitive assumptions" would scientifically support exclusion.  They need to understand that increased uncertainties regarding relevant information culminate in a need to consider an assessment of inconclusive.  To routinely venture into a gray area such as attempting to prove exclusions with low quality information is not be the best approach scientifically.  To maximize accuracy regarding exclusions, it may be appropriate to offer conclusions of exclusion only on a special case-by-case basis.  One solution would be to state in some fashion that; "no identification was found," and when necessary state that the comparison is inconclusive regarding the fact that exclusion could not be proven.

    It is imperative to avoid non-supportable conclusions as fact.  The main point here is to offer information that qualified experts have searched and did not find a match, and that exclusion is not provable with the information currently available.  In most criminal cases, the fact of exclusion offers little informational value regarding an investigation.  Examiners understand that all surfaces touched by friction skin do not necessarily offer latent print impressions of identifiable quality.  Therefore, the information regarding a conclusion of elimination may not actually offer information that could further an investigation.  In fact, in some cases, investigators and jurors may be mislead by assuming that a person did not touch an item.  Fingerprint information by itself does not provide proof of guilt or of innocence.  While properly formulated hypotheses of inconclusive are impartial, false information can be unintentionally injurious to the investigative process as a whole. 

    The collection and application of accurate information helps further an investigation while inaccurate information detours investigations.  Examiners must remain cognizant of their science’s strengths as well as its limitations.  This includes the limitations of the both the application and the examiner.

Craig A. Coppock 20051203
Updated 20181202

Reference:
1.  SWGFAST; A federally funded Scientific Working Group on Friction Ridge Analysis, Study and Technology, to improve related forensic science.  
2. Livio, Mario: 2003, The Golden Ratio (1.61803…), Broadway Books, NY p. 240

Also see the related post "Scientific Method; Scientific Method and Information Theory" for comprehensive insights on the introduction of error in the ACE-V / Scientific Methodology in the June 2017 post of this blog.  ACE-V is a product of the Scientific Method.

This article posted to the blog "Fingerprint Individualization | ACE-V | Scientific Method" at:  http://fingerprintindividualization.blogspot.com

No comments: