Colloquium Fall 2026: Rob Cavanaugh (Mass General Brigham University of Health Professions)

Tuesday, October 27,2026
1:45-3:15 PM [In-Person]

Rob Cavanaugh (Mass General Brigham University of Health Professions)

Title: Approaching Statistical Uncertainty: A Practical Introduction to Bayesian Modeling in Communication Science and Disorders

Abstract: Despite decades of methodological critique, "p < .05" remains a dominant heuristic for interpreting scientific findings and is frequently misinterpreted as evidence for a true and important effect. In response, the American Statistical Association's ATOM principles, Accepting uncertainty, being Thoughtful, Open, and Modest, offer a broad guide to interpreting, reporting, and communicating statistical evidence. Bayesian modeling is one useful tool for putting these principles into practice, and is increasingly accessible and approachable to researchers trained in frequentist statistics. In this talk, we will work through the conceptual foundations of Bayesian data analysis, from Bayes' theorem to the specification of priors, the interpretation of posteriors and credible intervals, and the use of the region of practical equivalence. Throughout, we will draw on a tutorial comparing effect sizes common to small-N and multiple-baseline studies in aphasiology, using it to motivate the modeling decisions, priors, and interpretation behind a Bayesian multilevel approach. We will conclude by discussing how research in communication science and disorders engages with uncertainty. The broader aim, regardless of statistical framework, is to promote a deliberate practice of accepting uncertainty, being thoughtful, communicating openly, and remaining modest in the practice and communication of science.