click for paper.

Summary of Murdock et. al

###Introduction With the work some researchers are doing in the forensic field there is bound to be criticism as to the methods currently used. Criticism is usually helpful and wanted, but “just as substandard work deserves derision, so does substandard criticism.” While there is currently no firm statistical backing in firearm and toolmark identification it is not reasonable to say that the probative value of evidence can be assessed. Firearm and toolmark identification works well enough that the evidence presented in court is mostly reliable with little error. While it would be nice to incorporate random match probability in firearm and toolmark analysis, there is currently no model that can reliably do so.

Firearm and toolmark examiners have been trying to find such a statistical model to evaluate their work in an effort to make more accurate analyses. While there are some models that provide some useful information they are each limited in their own way. There is also currently no accepted model to assess other forensic evidence like fingerprints, bloodstain patterns, or shoeprints. Creating a model to give a probative value for evidence is like trying to predict longevity or the future weather. While there are some ways to give an estimate, but there is no certainty in it.

###Review of the Current Literature in Statistical Applications for Firearms and Toolmarks The need to create a model to assess random match probability was recognized as early as 1932 with Gunther and Gunther. However, no reliable data was presented until 1959 when Biasotti recorded the number of groups of consecutive matching strations with bulled fired from the same and different guns. In 2007, Neel and Wells extended Biasotti’s work when they compared 2D and 3D striated toolmarks. They used Z-tests at the 99% confidence level to compare empirical frequencies. There were statistically significant differences found to which would back the rejection of the null hypothesis. Neel and Wells also used Bayes’s theorem to determine a likelyhood ratio for the matching of their striae. In the more recent Hamby study 1623 cartridge cases of the same caliber were collected for both manual and algorithmic comparison. The RMP estimate was found to be 0.0001% with a very small error rate in the manual calulation. The algorim proved to have a similarlly small error rate.

###Current State of the Use of RMPs in Firearm and Toolmark Identification Currently there is a lack of a database to base RMPs in firearm and toolmark identification. Incorporation of RMPs in firearm and toolmark identification would be much more difficult than incorporating into DNA analysis. “RMPs can be used in DNA profiling because analysts base their estimates of association on arrays of genotypes, which are indentification (50). Firearm and toolmark identifications are based on arrays of individual, not subclass, characteristics.” DNA is also less likely to change than firearms and tools. Where it takes a significant amount of mutation for DNA to change it would only take a bit of sandpaper to change the face of a tool. The amount of change is also much harder to predict than DNA. Where DNA usually mutates in a few different processes, that are ultimately infintesimal, a tool can change in many more. The pressure of application, grit, and wear of sandpaper can affect the tool in nearly unpredictable ways.

###Absolute Versus Practical Identification and Subjectivity Another criticism in firearm and toolmark identification is that the examiners deal in absolutes. In present day analysis of evidence examiners use practical certainty. Practical certainty is the assumption based on past experiences that a prediction will be correct. “There is a practical certainty that our car will start in the morning (assuming it is in good mechanical condition), or that our (normally obedient) dog will come when called.” The examiners base their identifications on years of examinations and training.

The subjectivity of the identification is another popular criticism in this field. Some claim that since an examination is subjective it has no scientific basis and is not valid. People use subjective analysis everyday in their lives. “We must remind ourselves that we may be diagnosed with influenza by a doctor after 60 seconds of a very subjective examination based on the doctor’s training and experience, or that we pay a good deal of money to an automodile mechanic who has made a vary subjective determination of some mechanical problem,” Subjective analysis will always play a role in analysis of evidence.

Title & Abstract Discussion

A good title for a scientific paper should be clear, consise, and explain exactly what the research is about. This paper’s title does a horrible job at doing these things. At first glance it may seem that this paper is about the development and application of random match probabilities to firearm and toolmark identification. After reading the abstract it’s clear that the paper has another purpose. Instead of providing useful information about the research in this field all that’s provided is criticism with a heavy dose of shade. A more suitable title could have been “Criticism of the Development and Application of Random Match Probabilities to Firearm and Tool of Identification.