Meghan Prusinowski

The current standard test method, ASTM E1967-19, for refractive index (RI) measurements of forensic glass samples recommends one comparison criterion to determine whether the known glass is a potential source of each questioned fragment. However, in practice, forensics laboratories currently calculate comparison intervals in a variety of manners. Using alternative comparison criteria to interpret glass without current data to support the chosen criteria poses a significant risk to the forensic community, as this could lead to potential issues in interpretation, testimony, and admissibility of glass casework.In this study, various comparison criteria are evaluated to determine rates of false exclusion and false inclusion utilizing RI data from 950 different sources of tempered and laminated vehicle windows and two additional sets of densely sampled single source glass. The results of the study indicate that the current recommended criterion performs well, with a false inclusion rate of less than 5% for both types of glass. However, the false exclusion rate is higher on several criteria. Alternate criteria, such as using a 3σ range above and below the mean of the known glass sample provide more equivalent false inclusion and exclusion rates, around 5%. In addition, including more measurements for a known or questioned fragment than currently recommended in ASTM E1967-19 provides greater discrimination power.Overall, this study demonstrates that the previous sampling strategies and comparison criteria described in ASTM E1967-19 may need to be updated in accordance with contemporary samples and laboratory practices.

Several organizations have outlined the need for standardized methods for conducting physical fit comparisons. This study answers this call by developing and evaluating a systematic and transparent approach for examining, documenting, and interpreting textile physical fits, using qualitative feature descriptors and a quantitative metric (Edge Similarity Score, ESS) for the physical fit examination of textile materials. Here, the results from 1027 textile physical fit comparisons are reported. This includes the evaluation of inter and intraanalyst variation when using this method for hand‐torn and stabbed fabrics. ESS higher than 80% and ESS lower than 20%, respectively, support fit and nonfit conclusions. The results show that analyst accuracy ranges from 88% to 100% when using this criterion. The estimated false‐positive rate for this dataset (2% false positives, 10 of 477 true nonfit pairs) demonstrates the importance of assessing the quality of a physical fit during an examination and reveals that potential errors are low, but possible in textile physical fit examinations. The risk of error must be accounted for in the interpretation and verification processes. Further analysis shows that factors such as the separation method, construction, and design of the samples do not substantially influence the ESS values. Additionally, the proposed method is independently evaluated by 15 practitioners in an interlaboratory exercise that demonstrates satisfactory reproducibility between participants. The standardized terminology and documentation criteria are the first steps toward validating approaches to streamline the peer review process, minimize bias and subjectivity, and convey the probative value of the evidence.

A physical fit is an important observation that can result from the forensic analysis of trace evidence as it conveys a high degree of association between two items. However, physical fit examinations can be time-consuming, and potential bias from analysts may affect judgment. To overcome these shortcomings, a data analysis algorithm using mutual information and a decision tree has been developed to support practitioners in interpreting the evidence. We created these tools using data obtained from physical fit examinations of duct tape and textiles analyzed in previous studies, along with the reasoning behind the analysts' decisions. The relative feature importance is described by material type, enhancing the knowledge base in this field. Compared with the human analysis, the algorithms provided accuracies above 90%, with an improved rate of true positives for most duct tape subsets. Conversely, false positives were observed in high-quality scissor cut (HQ-HT-S) duct tape and textiles. As such, it is advised to use these algorithms in tandem with human analysis. Furthermore, the study evaluated the accuracy of physical fits when only partial sample lengths are available. The results of this investigation indicated that acceptable accuracies for correctly identifying true fits and non-fits occurred when at least 35% of a sample length was present. However, lower accuracies were observed for samples prone to stretching or distortion. Therefore, the models described here can provide a valuable supplementary tool but should not be the sole means of evaluating samples.

An ongoing advancement in forensic trace evidence has driven the development of new and objective methods for comparing various materials. While many standard guides have been published for use in trace laboratories, different areas require a more comprehensive understanding of error rates and an urgent need for harmonizing methods of examination and interpretation. Two critical areas are the forensic examination of physical fits and the comparison of spectral data, which depend highly on the examiner’s judgment. The long-term goal of this study is to advance and modernize the comparative process of physical fit examinations and spectral interpretation. This goal is fulfilled through several avenues: 1) improvement of quantitative-based methods for various trace materials, 2) scrutiny of the methods through interlaboratory exercises, and 3) addressing fundamental aspects of the discipline using large experimental datasets, computational algorithms, and statistical analysis. A substantial new body of knowledge has been established by analyzing population sets of nearly 4,000 items representative of casework evidence. First, this research identifies material-specific relevant features for duct tapes and automotive polymers. Then, this study develops reporting templates to facilitate thorough and systematic documentation of an analyst’s decision-making process and minimize risks of bias. It also establishes criteria for utilizing a quantitative edge similarity score (ESS) for tapes and automotive polymers that yield relatively high accuracy (85% to 100%) and, notably, no false positives. Finally, the practicality and performance of the ESS method for duct tape physical fits are evaluated by forensic practitioners through two interlaboratory exercises. Across these studies, accuracy using the ESS method ranges between 95-99%, and again no false positives are reported. The practitioners’ feedback demonstrates the method’s potential to assist in training and improve peer verifications. This research also develops and trains computational algorithms to support analysts making decisions on sample comparisons. The automated algorithms in this research show the potential to provide objective and probabilistic support for determining a physical fit and demonstrate comparative accuracy to the analyst. Furthermore, additional models are developed to extract feature edge information from the systematic comparison templates of tapes and textiles to provide insight into the relative importance of each comparison feature. A decision tree model is developed to assist physical fit examinations of duct tapes and textiles and demonstrates comparative performance to the trained analysts. The computational tools also evaluate the suitability of partial sample comparisons that simulate situations where portions of the item are lost or damaged.