Evaluation of the reliability of the qualitative forensic method "Microscopic examination of textile fibers"

Abstract

Methods of qualitative analysis are widely used in various spheres of public activity, including forensic examination. Making the right decisions based on the results of qualitative analysis, necessitates confirmation of the reliability (uncertainty) of the methods which can be provided by the validation procedure. However, the issues related to the validation of qualitative forensic methods are debatable in the absence general regulatory requirements for qualitative analysis methods. We consider a validation procedure of the qualitative forensic technique “Microscopic examination of textile fibers,” which consists in determining a complex of characteristic external features of natural and chemical textile fibers (color, color features, morphological features) using a microscope, as well as thickness and presence/absence of a matting agent for chemical fibers. These generic characteristics are used to differentiate the fibers under study in the forensic examination of fibrous materials. The reliability of the methodology and the competence of the performers were selected as validation parameters. The parameters were determined numerically by the likelihood ratio and by the values of the share of false and correct results in the total number of tests. Ten samples of natural and chemical textile fibers from the comparative collection of the Laboratory of forensic examination of fibrous materials of the RFCSE were used for validation. Three experts participated in the experiment independently identified the presence/absence of ten external signs in each of ten samples during a week. Each expert tested a set of one hundred different external features, 39 of them were present in the samples, and 61 were absent. When comparing the test results obtained by the expert with the corresponding regulated (known) external features, a conclusion was made about the level of true or false result for each sample. A low (1.7%) level of false results was revealed in relation to the total number of tests, as well as a low (2.6%) level of false results for each of the experts indicated the reliability of the methodology and competence of the experts. The calculation of the likelihood ratio (LR) showed that the probability of true results in the assessment of a set of features is about 60 times (significantly more than one) higher than the probability of false results, which also indicates the reliability of the technique. The results of the validation experiment allowed us to conclude that the method is suitable for use in solving expert problems in the forensic examination of fibrous materials.