This book gives a clear comprehensive explanation and defense of the Bayesian account of scientific reasoning. It will be read not only by philosophers and theorists of scientific method but also by working scientists, uneasy about the justification of the statistical methods now in use. Since the book is designed to explain to the uninitiated the controversial theories it discusses, it can serve as an introduction to the role of statistics and probability in science. Confronting the problems of induction and the ...
Read More
This book gives a clear comprehensive explanation and defense of the Bayesian account of scientific reasoning. It will be read not only by philosophers and theorists of scientific method but also by working scientists, uneasy about the justification of the statistical methods now in use. Since the book is designed to explain to the uninitiated the controversial theories it discusses, it can serve as an introduction to the role of statistics and probability in science. Confronting the problems of induction and the confirmation of scientific theories, Howson and Urbach reject the "objectivist ideal" and the fashionable non-probabilistic standard of scientific worth (Popper, Lakatos, Fisher, Neyman and Pearson). The authors contend that "scientific reasoning is reasoning in accordance with the calculus of probabilities", and (using nothing more advanced that elementary algebra) they give a concise introduction to this calculus. Howson and Urbach examine the way in which scientists actually appeal to probability arguments, and explain the "classical" approach to statistical inference, which they demonstrate to be full of flaws. They then present the Bayesian approach, showing that it avoids the difficulties of the classical system. Finally, they reply to all the major criticisms levelled against the Bayesian method, especially the charge that it is "too subjective".
Read Less
Applications of Bayesian statistics are just now elbowing their way into service in many fields, yet statistical education for users of statistics who are not professional statisticians is still mired in "orthodoxy." This somewhat older book is an excellent introduction to the field and explains the arguments and counter-arguments clearly and in a studied, logical fashion. There is a certain amount of polemic (par for the course), but polite in a British kind of way (nothing like the rough-and-tumble anti-frequentist zingers of ET Jaynes). The math is accessible to anyone who took algebra, and is present in just the right quantity.
