LAUSR

Sir, To address the potential for extraneous contextual information to bias forensic examinations (1,2), forensic disciplines must consider ways to effectively manage this information. At the heart of developing contextual information management (CIM), procedures is the identification of types of information that are task-relevant and task-irrelevant. Effective CIM will enable forensic examiners to have access to the task-relevant information, while at the same time shielding them from the task-irrelevant information. How a CIM approach might transpire will largely depend on the method of forensic analysis. Many of the CIM approaches that commentators recommend are designed with pattern comparison disciplines (e.g., latent print examination, shoeprint comparison, and DNA analysis) in mind. For example, in (linear) sequential unmasking (3,4), the expert must examine the latent (crime scene) sample before being exposed to the known (e.g., suspect/elimination) sample. In this way, knowledge of the features in the known sample cannot influence (or bias) the observation and interpretation of the latent sample. This approach, however, is less useful when we consider bloodstain pattern analysis (BPA), a discipline that requires the recognition rather than comparison of patterns. We assert here that the difficulty in making recommendations for CIM in BPA goes beyond the type of analysis that is required (i.e., pattern recognition), and is instead inherent in the BPA classification method itself. Bloodstain pattern classification involves two overlapping components: one component is the analyst’s assessment of the observable characteristics of the pattern; and the other component is the analyst’s opinion about the event(s) that gave rise to those observable characteristics. The contention is that these two components are not well separated and are generally conducted and reported simultaneously. The resulting classification becomes not so much a classification of the pattern, but a classification of the underlying mechanistic event. A method such as this, that blurs the boundary between observation and interpretation, only serves to increase vulnerability to contextual bias. While most bloodstain pattern analysts would contend that they do make observations about the size, shape, and distribution of bloodstains as the first step in their analysis, the current method for pattern classification does not provide analysts with a platform to clearly report observation-only based findings. These concerns were raised by Bernstein in 2005 (5), but have not yet translated into any significant evolution of BPA methodology. A well-defined process that distinctly separates the currently overlapping components into a Two-Step Method for BPA would serve two purposes. First, it would bring more objectivity and transparency to an otherwise subjective decision-making process. Second, it would allow for a clear and effective way to recognize and manage task-relevant and task-irrelevant contextual information at each step. We propose that Step One is limited exclusively to the characterization of the pattern (i.e., observing and measuring its features). Here, contextual information would have a limited role to play, and efforts to remove all task-irrelevant contextual information should occur. We propose that Step One be completed and fully documented before Step Two begins. Step Two is where analysts can form opinions about the nature of the event(s) that gave rise to the pattern. Here, the examiner needs to develop hypotheses about the possible event (s). To ensure that these hypotheses are relevant, it will be valuable, perhaps even essential, to incorporate relevant contextual information (e.g., victim’s injuries, contradictory witness statements) into the assessment. The inclusion of this information, and how it has (or has not) extended an analyst’s opinion, however, must be transparent and clearly documented. The basis for any opinions proffered in Step Two should be primarily based on the context-limited pattern characterization completed in Step One. To connect these two steps, more work is required to better understand and quantify the relationship between pattern characteristics and their causal event(s). Furthermore, to introduce this Two-Step Method for BPA, we need to consider how a change to current terminology could best serve characteristic-based observations (Step One), as well as the mechanistic-based opinions (Step Two) (6). We invite other bloodstain pattern analysts, and the wider forensic community, to comment on and contribute to the development of such a method of bloodstain pattern analysis.