Scientific reasoning

This second edition of a Bayesian study of scientific method has been thoroughly reworked and considerably expanded. It serves two purposes at once: it teaches the basics of probability theory and offers a philosophical examination of how scientists draw conclusions from evidence. New chapter exercises accompany expanded treatments of regression analysis, distributions and densities, randomization, and conditionalization.

The authors take on the hard questions of induction and how scientific theories gain support from evidence. They argue against the pursuit of an objective ideal and against the non-probabilistic yardsticks for judging good science favored by Neyman and Pearson, Fisher, Popper, and Lakatos.

At the center of the book is the claim that sound scientific thinking amounts to applying the rules of probability. Readers need little beyond elementary algebra to follow the concise introduction to that calculus.

The book looks at how working scientists actually make probabilistic arguments. It then explains the classical approach to statistical inference, pointing out a number of its flaws. The Bayesian alternative follows, shown to sidestep those problems. The closing chapters respond to the main objections raised against Bayesian methods, most notably the accusation that the approach is too subjective.