Fiona Fidler: Misinterpretations of evidence, and worse misinterpretations of evidence
P-values are frequently misinterpreted. Confidence intervals are too. So are Bayesian statistics. Sometimes this simple equivalence is used as an argument that statistical cognition shouldn’t play a role in deciding which analysis approach to adopt in practice, or to teach to students. But are misinterpretations of these different displays of statistical evidence equally severe?
Do they have the same consequences in practice? In this talk I’ll present the limited empirical evidence related to these questions that we have so far, and suggest that, at the very least, we don’t know enough to assume Abelson’s law yet, i.e., “Under the law of the diffusion of idiocy, every foolish application of significance testing is sooner or later going to be translated into a corresponding foolish practice for confidence limits” (Abelson, 1997, p. 130). There may be other sound reasons – technical or philosophical reasons —to reject one approach or another, but we shouldn’t (yet) consider them cognitively equivalent.
Associate Professor, University of Melbourne
Fiona Fidler is Associate Professor at the University of Melbourne, with a joint appointment in the Schools of BioSciences and History and Philosophy of Science. She is broadly interested in how experts, including scientists, make decisions and change their minds. Her past research has examined how methodological change occurs in different disciplines, including psychology, medicine and ecology, and developed methods for eliciting reliable expert judgements to improve decision making. She originally trained as a psychologist, and maintains a strong interest in psychological methods. She also has an abiding interest is statistical controversies, for example, the ongoing debate over Null Hypothesis Significance Testing. She is a current Australian Research Council Future Fellow, and leads the University of Melbourne’s Interdisciplinary MetaResearch Group (IMeRG).