The Importance of Critically Examining the Level of Propositions When Evaluating Forensic DNA Results

The proposal of a discussion about the use of software to help assign likelihood ratios for forensic DNA profiling results, and the use of their output in the legal process, is both timely and important (see also related contributions elsewhere in Frontiers, e.g., Biedermann et al., 2014). Ever since their introduction in forensic science, DNA profiling analyses have been accompanied with the results of calculations of various sorts. Their scope is well illustrated and documented in several reference monographs (e.g., Evett andWeir, 1998; Buckleton et al., 2005; Balding and Steele, 2015). This solid body of scholarly research and established practice has contributed to the widely held view among scientists and recipients of expert information that eliciting the probative strength of forensic DNA profiling results is per se a numerical task. In this commentary, we intend—in a first part—to make the point that although calculations are, by virtue, an integral part of the quantification of probative strength, it is equally important at the outset to be clear about the question “Why are we doing a calculation?” (Buckleton et al., 2005, p. 151). We will argue that this is not a question that statistics can answer. Stated otherwise, we will contend that, as much it is important to be clear in any instance about what a particular computation exactly purports to do, it is essential to define the questions that are of interest in a particular case at hand. In a second part, we will emphasize on the extent to which, why and how recently issued guidelines (e.g., ENFSI, 2015) encourage such thinking about cases prior to conducting calculations, if any.