Logic of Statistical Inference

Cambridge Philosophy Classics

by Ian Hacking

This book is a philosophical study of the basic principles of statistical reasoning. Professor Hacking has sought to discover the simple principles which underlie modern work in mathematical statistics and to test them, both at a philosophical level and in terms of their practical consequences fort statisticians. The ideas of modern logic are used to analyse these principles, and results are presented without the use of unfamiliar symbolism. It begins with a philosophical analysis of a few central concepts and then, using an elementary system of logic, develops most of the standard statistical theory. the analysis provides answers to many disputed questions about how to test statistical hypotheses and about how to estimate quantities in the light of statistical data. One product of the analysis is a sound and consistent rationale for R. A. Fisher's controversial concept of 'fiducial probability'.

Genres: Philosophy, Mathematics

Paperback

First published January 1, 1965

About the author

Ian Hacking is Professor Emeritus of Philosophy at the University of Toronto, specialised in the History of Science.

Wikipedia entry

Reviews

[Lucille Nguyen · Rating 4 out of 5]

Reviews the interconnection between inductive logic and statistical inference. Terribly frequentist, and owes heavily to Neyman-Pearson conceptions of statistical science. While the explanations of the Kolmogorov axioms and Fisherian fiducialism (now since outdated in most statistical practice) are excellent, the explanation of Bayesianism and the "subjective" probabilists was quite thin. Hacking acknowledges such in the introduction, mentioning only being influenced by Savage's Foundation of Statistics for the Bayesians. As such, little about Cox's postulates, Jaynes's Boolean algebra-based probability system, and similar is crammed into two short chapters at the end of the book. Which was quite disappointing, because the clear-headed logical connection Hacking gave to frequentist and "objective" statistics would have been equally enlightening for Bayesian and "subjective" statistics.

[Marks54 · Rating 3 out of 5]

This is a rerelease of an earlier book by Hacking on how to look at the philosophical assumptions behind statistical reasoning. I went through this to help in following another book on the philosophical issues regarding whether we are all in simulations or not.The book was very much as advertised, providing a background to the reasoning behind statistics, distributions, likelihood, tests, and the like. I generally knew this but it was worthwhile to hear all this from the perspective of a mathematical philosopher. As to whether it helps me work through whether we are really in simulations or not, I am less sure of that and will have to process this a bit more.