The effect of task irrelevant contextual information on judgment, decision making, and opinion in forensic odontology identification.

Abstract

Non task relevant contextual information can give rise to context bias which has been shown to cognitively influence all comparative forensic science decisions that rely on human judgement. However, whether forensic odontology identification opinions which require human judgement and evaluation are affected by contextual bias is, to date, empirically unverified. This project aims to explore and provide sound empirical evidence regarding the influence of non task relevant contextual information on forensic odontology identification casework outcomes. The opinion formation process in forensic odontology was found to be an under researched area at the start of this thesis. This deficiency could explain the commonly held assumptions that categories on standardised opinion scales represent decisional confidence in identification and that dental radiographs alone provide sufficient information for definitive identification. A scoping review of the available validation studies using dental radiographs for identification reveals support for sufficient inter individual discriminability in dental radiographs to allow matching and definitive identification decisions. While an analysis of the relationship between category levels and confidence found correlations between the identification categories and decision confidence, it remained uncertain whether this confidence was derived only from probabilistic weight estimates of the evidence or if contextual effects contribute as well. These early foundational findings informed the design of the main experiment used to address the central research question of whether non task relevant case information contextually influences identification conclusions and opinions. In the main experiment in this project, forensic odontologists and dentists participated in an online web based survey where they formed identification opinions. Participants were required to read contextual case information that either supported or contradicted the true match status of pairs of matching or non matching radiographs which they then compared. Subsequently, they were asked to provide probabilistic estimates of whether the pairs of radiographs were a match or non match, assign a category of identification and state their confidence in their decisions. The overall findings suggest that strong contextual non relevant case information affected the judgement and evaluation process and concluding category decision. Additionally, training and experience appear to affect the interpretation of the categories used on the identification scale. The tenet and value of scientific expert opinion require the evaluation to be based only on relevant information. The finding that contextual information biases the opinion provides an added reason and a strong argument for the management of non relevant contextual information. Concomitantly, the finding that the interpretation and assignment decisions are affected by the connotation, granularity, and positions of the terms in the scale implies that different scales cannot be compared directly. Although more research is required, it does suggest that familiarity with the scale is an important factor for its efficient and correct application. Finally, the finding that the expert participants appear to understand the implications of the identification levels in an opinion better than the comparison group despite the different geographical forensic odontology training and practice backgrounds, is pleasing and provides support for the value of training and calibration.