EVIDENCE IN CLASSICAL STATISTICS

The dominance of classical statistics raises a puzzle for epistemologists. On one hand, science is a paradigmatic source of good evidence, with quantitative experimental science often described in classical statistical terms. On the other, its hybrid of Fisherian and Neyman–Pearsonian techniques is generally rejected by philosophers, statisticians, and scientists who study the foundations of statistics. So why is the use of classical statistics in empirical science so epistemically successful? Do classical measures of evidence actually measure anything epistemically important? This chapter provides some positive answers to these questions, drawing analogies with modal principles used in sophisticated tracking theories of knowledge.