Uncertainty, Information, and Narrative: A Statistical Perspective on Scientific Storytelling

Robert Kass
Maurice Falk Professor of Statistics and Computational Neuroscience, Carnegie Mellon University

Abstract
Like many other human endeavors, science is built on story telling: the aim of science is to explain how things work. But when science is reported, either in the popular press or in scientific journals, the stories almost always blend results with speculation. Too often the distinction between scientific findings and speculation gets blurred, and the reporting of speculative theories as if they are necessary consequences of results helps erode public confidence in science. One reason authors de-emphasize such subtleties is widespread discomfort with uncertainty and the procedural nature of scientific knowledge, but these are the everyday business of statistics. Using examples from neuroscience, history, and the "bible codes" controversy, this non-technical lecture (aimed at a broad audience of humanities and science students and faculty) summarizes several of the most basic statistical lessons for our citizenry, and indicates how these lessons could inform narratives. It also modernizes the definition of scientific method, incorporating widely accepted features that are too frequently ignored in interpretive summaries of scientific discoveries.

This lecture was given March 28, 2019.