Applying principal component analysis (PCA) to the selection of forensic analysis methodologies

Abstract

Booker et al. [3] stated that “in many engineering defect investigations, the choice of forensic analysis methodology is rarely questioned, and the forensic analysis methodology is chosen by the analyst without due diligence, resulting in a methodology that may not be the most accurate or expedient. This in turn has the potential to increase the risk and cost of the decision-making process.” Therefore, a need was identified to identify and analyse the variables that influence practitioners’ selection of an appropriate forensic analysis methodology. A review of current literature highlighted the fact that there was no definitive correlation identifying the primary variables that affect the selection of forensic analysis methodology. Noting that the determination of the correct methodology for a specific instance is not within scope of this paper, the first stage of the analysis was to apply previously collated methodology taxonomy, and then apply Principal Component Analysis (PCA) to determine which variables had the highest significance of influence on the selection of Forensic Analysis Methodology from 132 case studies selected from Case Studies in Engineering Failure Analysis [2] (Appendix A.). The analysis has determined that the methodology selected to conduct forensic analysis is primarily influenced by the geographical culture and industry of the analyst. Therefore, there is a requirement to acknowledge and understand the role of cultural and industry biases when choosing an appropriate methodology, and subsequently, when evaluating a forensic analysis case study.Although the mere recognition of the existence of cultural and industry bias does not guarantee that the right decisions will be made, it does, however, facilitate an objective rigour to the process of the selection of the forensic analysis methodology to be applied, which in turn, will mitigate subjective reasoning and potentially reduce the risk and cost of the decision-making process. Without an international standard for forensic analysis, or universally acknowledged guidelines for the selection of forensic analysis methodology, analysts will continue to be susceptible to methodology selection bias, which will increase the probability of choosing a sub-optimal methodology, which, in turn, can prolong the investigation and adversely affect the accuracy of the analysis.