Developing Effective Methods for Addressing Contextual Bias in Forensic Science

As submitted by the proposer: In 2009, the National Research Council criticized forensic scientists for taking inadequate measures to guard against contextual bias. The underlying concern was that forensic scientists might unwittingly base their conclusions not solely on expert evaluation of the evidence they are asked to examine, but also on task-irrelevant contextual information, i.e., other information about the case that is not pertinent to an experts judgment. Academic commentators have encouraged the use of procedures that restrict or control the flow of contextual information to analysts, in order to prevent contextual bias. But their suggestions have met with some resistance from forensic laboratories, where uptake has typically either been slow or non-existent. The resistance arises in part, we argue, from the failure of proposed procedures to take account of practical realities of forensic laboratories, where contextual information is often needed. In order to make recommendations that are acted on, we need a better understanding of how forensic analysts make decisions, the reasons that analysts might encounter, solicit, or require contextual information, and the role that this information plays in their decisions. We also need to identify the perceived barriers to context management procedures, the individual and systemic factors behind these barriers, and the obstacles faced by the few laboratories that have attempted to adopt such procedures. Our accomplished multidisciplinary team, comprising forensic practitioners as well as academics in psychological science and lawhas successfully pilot-tested several paradigms for investigating these issues. We will extend these methods by conducting comprehensive interviews with directors and section heads from a cross-section of major international forensic laboratories. We will also cast our spotlight on two distinctly different forms of forensic evidence: handwriting analysis (involving decisions about source) and bloodstain pattern analysis (involving decisions about activity). For each domain, we will conduct detailed interviews with analysts, as well as experimental studies in which analysts evaluate materials based on actual cases in order to illuminate the role of contextual information in analysts, decision-making processes. Our findings will also allow us to establish principles of sound contextual information management that will facilitate development of comprehensive yet practical protocols for mitigating contextual bias. Informed by our detailed assessment of what analysts need to know, and of the practical constraints that laboratories face in managing contextual information, we will propose model protocols for handwriting and bloodstain pattern analysis that strike the crucial, but as yet unachieved, balance between efficacy and feasibility. ca/ncf