Portal de Periódicos da CAPES

Commentary on: Monson KL , Smith ED , Peters EM . Accuracy of comparison decisions by forensic firearms examiners. J forensic sci. 2022; 68(1):86–100. https://doi.org/10.1111/1556‐4029.15152

2023; Wiley; Volume: 68; Issue: 3 Linguagem: Inglês

10.1111/1556-4029.15258

1556-4029

Nicholas Scurich, Hal S. Stern,

Forensic Fingerprint Detection Methods

See Original Article here See Authors' response to Scurich et al Commentary on here Editor, The article entitled "Accuracy of comparison decisions by forensic firearms examiners" reports the results of an empirical study of firearm examiners [1]. To their credit, the authors posted some of the raw data underlying the results online. However, reviewing the raw data reveals an anomaly that could have serious implications—both for the interpretation of the study results and for the practice of firearm examination. The Association of Firearm and Tool Mark Examiners (AFTE) defines class characteristics as "measurable features of a specimen which indicate a restricted group source. They result from design factors and are determined prior to manufacture [2]." "The class characteristics for a fired cartridge case include the cartridge caliber, the shape of the firing pin, and the tooling marks on the breech [3]." Eliminations based on class characteristics are recognized to be "very easy" because they involve comparing objectively measurable features as opposed to subjectively comparing patterns [4]. The study in question purportedly did not have any different-class comparisons. As stated in the companion article, which provides the "full details of the planning, design, and logistics" of the study in question, "all items in an individual comparison set shall be fired from the same make and model firearm, precluding elimination based on class characteristics" (p. 4, emphasis added) [5]. However, the raw data contradict this claim. The table below reports the raw responses for cartridge case comparisons; these labels and numbers come directly from the datafile posted online. Notice the second and third rows of the table: 48 eliminations were based on individual and class characteristics and 401 eliminations were based on class characteristics only. Roughly one-third (449/1400) of all cartridge case eliminations in the study were reportedly based on class characteristics. This fact is never mentioned in the article, and it presents an interpretive challenge for the reader. Some possible explanations for this anomaly include: (A) The authors were incorrect in stating that the study was designed in such a way to "preclud[e] elimination based on class characteristics." In fact, the study did include comparison sets that allow for eliminations based on class characteristics. If this is correct, however, it undermines the reported false positive error rates (i.e., false identifications for known nonmatching cases) because the false positive error rates are based on a set of comparisons that include a number of "very easy" different-class comparisons. In other words, the reported false positive error rate is artificially low because there is virtually no risk of a different-class elimination being called an identification. (B) The authors were correct that the study was designed in such a way to "preclud[e] elimination based on class characteristics." If this is correct, however, nearly one-third of the responses are wrong in claiming to make an elimination based on class characteristics only. These participants ultimately got the "correct" answer (elimination) but for a wrong reason. (C) The authors and examiners have different notions of what constitutes a "class characteristic." If this is correct, however, it suggests that "qualified" practicing firearm examiners have considerably different interpretations of AFTE's definition of class characteristics (cited above). Two additional observations must be made. First, it appears that certain guns led to more eliminations reportedly based on class characteristics than others (e.g., Jimenez gun #7 led to about 25% of all the eliminations based on class characteristics), suggesting that these are not random or reckless responses. Second, it is not possible to know how many eliminations based on class characteristics were actually included in the study by examining the reported class eliminations; the number of true class eliminations may be greater or less than the number of reported class eliminations. Thus, the study was designed to be a rigorous trial of examiner ability; as a result, error rates derived from this study may provide an upper bound to a possible error in operational casework, as evidentiary specimens may generally be assumed to be less challenging than those used in this study. (p. 87) This assertion, along with the reported error rates, should not be accepted unless or until the results from the raw data are reconciled with the study design that purportedly "preclud[ed] elimination based on class characteristics."