What to Believe and How Much¶
What to Believe and How Much is a short guide to statistical thinking written by José Blanca.
Statistics is central to modern science, but most of us learned it as a collection of recipes: if you have two groups, you run a t-test; if the p-value is below 0.05, you have found something. This book does not explain how to run any test. It tries to explain what we are doing when we do statistics, and why we do it the way we do. What does a p-value really mean, and why does almost nobody interpret it correctly? What does it have to do with the replication crisis? Can a probability represent a degree of belief? The book covers maximum likelihood, inverse probability and Bayesian inference, and presents the century-long fight between Bayesians and frequentists as a dialogue between the two schools. It ends by defending a pluralist position and a simple idea: good tools need a good craftsman.
It is written for anyone who has used statistical tools (biologists, psychologists, physicians...) without stopping to think about the philosophy behind them.
The book is also available in Spanish, as Qué creer y cuánto: Una breve introducción al pensamiento estadístico: epub or pdf.