Tuesday, 15 March 2011
How to Weigh Evidence
Heather Douglas,
Visiting Fellow (CPS)
University of Tennessee
12:05 pm, 817R Cathedral of Learning

Abstract: This talk will address how to weigh complex sets of evidence, from multiple disciplines, that are often divergent in their implications.  Weight of evidence in this context means assessing “where the weight of evidence lies,” rather than how much one piece of evidence supports a particular hypothesis in traditional confirmation theory.  How to perform such assessments is crucial to the optimal use of science in public policy, but remains contested.  I will argue that explanations are central to weight of evidence efforts, and that explicating and then utilizing explanations (particularly to make novel predictions) can make weight of evidence efforts rigorous, while remaining sufficiently flexible.   Without explanations, it would be unclear what the set of evidence to be weighed should be.  With explanations, we can see both why competing experts weigh the evidence differently, and how we might settle disputes concerning where the weight of evidence lies.   Finally, taking explanations as the central conceptual tool in weight of evidence analysis allows us to see where more formal techniques, like causal nets or expert elicitation, fit in the process.