Error rates in bite mark analysis in an in vivo animal model

Abstract

Recent judicial decisions have specified that one foundation of reliability of comparative forensic disciplines is description of both scientific approach used and calculation of error rates in determining the reliability of an expert opinion. Thirty volunteers were recruited for the analysis of dermal bite marks made using a previously established in vivo porcine-skin model. Ten participants were recruited from three separate groups: dentists with no experience in forensics, dentists with an interest in forensic odontology, and board-certified diplomates of the American Board of Forensic Odontology (ABFO). Examiner demographics and measures of experience in bite mark analysis were collected for each volunteer. Each participant received 18 completely documented, simulated in vivo porcine bite mark cases and three paired sets of human dental models. The paired maxillary and mandibular models were identified as suspect A, suspect B, and suspect C. Examiners were tasked to determine, using an analytic method of their own choosing, whether each bite mark of the 18 bite mark cases provided was attributable to any of the suspect dentitions provided. Their findings were recorded on a standardized recording form. The results of the study demonstrated that the group of inexperienced examiners often performed as well as the board-certified group, and both inexperienced and board-certified groups performed better than those with an interest in forensic odontology that had not yet received board certification. Incorrect suspect attributions (possible false inculpation) were most common among this intermediate group. Error rates were calculated for each of the three observer groups for each of the three suspect dentitions. This study demonstrates that error rates can be calculated using an animal model for human dermal bite marks, and although clinical experience is useful, other factors may be responsible for accuracy in bite mark analysis. Further, this study demonstrates that even under carefully controlled conditions, albeit in a forced-decision model, errors in interpretation occur even amongst the most experienced analysts.