Review

This would serve as a great adjunct to an introductory or second semester course in statistics. You won’t learn much (if anything) about specific statistical tests, and you won’t learn any of the mathematics underlying modern statistics from this book. What you will learn is a set of ways to think about what you (should) hope to achieve from applying statistical methods to a problem. Vickers clearly believes, with reason, that statistics can be used to help us understand the world, and by understanding, improve it. So he wants to root out silly, misapplied statistics, and get statisticians and the rest of us, to do statistics right. “One hypothesis, one p-value” could almost serve as the dedication of this book. He wants us to think deeply about the problems we are working on, and distill the solutions to those problems down to specific testable hypotheses, and to then conduct the surveys, studies, or experiments that will let us evaluate the hypotheses. This is not earth-shatteringly new, but Vickers provides lots of examples where this isn’t done, even (or maybe especially) in peer reviewed publications.

Metadata