Scholars Award: Explanatory Weight of Evidence Analysis

University Of Tennessee Knoxville, Knoxville TN

Investigators

Abstract

Introduction This project will develop a new approach to the weight of evidence problem. That problem has to do with how best to combine, in a meaningful and reliable way, complex sets of evidence including evidence from multiple disciplines, disparate methodologies, and often apparently conflicting results Intellectual Merit In this project, an approach to the weight of evidence problem using explanations as conceptual tools to organize the evidence and test hypotheses will be developed. The explanatory approach will then be applied to current scientific cases where the weight of evidence challenge is pressing. Finally, this approach will be compared with other approaches being developed, such as Bayesian Belief Networks, meta-analyses, evidence ranking, and expert elicitation techniques. The various approaches will be compared with respect to their rigor, their ability to consider a wide range of evidential sources, their ability to grapple with a range of contexts of application, the transparency of their method, and their ease of use. Potential Broader Impacts The need for an approach to the weight of evidence problem, particularly in cases where direct human subject research would be unethical, has grown in recent years. Federal agencies have been called upon to utilize the best available science in setting policy, but when the science comes from multiple disciplines using different methods and points in various directions, how to do this cogently and transparently has become a serious challenge. This project will help address that challenge, by developing an approach and comparing it to the other approaches noted above. The PI hopes to be able to hone the explanatory approach such that it is both rigorous and transparent, while still remaining qualitative. Indeed, it may be that the explanatory approach is a necessary step for any of the quantitative approaches currently being proposed.