« Matrix Door | Main | Mathematical epigraphs »

In what sense do statistical methods provide scientific evidence?

Bill Thompson's The Nature of Statistical Evidence addresses this intriguing question. Along the way, he discusses whether statistics meets the predictive and experimental verification criteria of the scientific method; critiques Bayesian inference (e.g., "Randomness Needs Explaining"); investigates various interpretations of probability and "attitudes toward chance;" offers a framework for statistical evidence as an alternative to the "true value" model; and more. Many thought-provoking points in very few pages--a practical slant on the philosophy of statistics, from a Professor Emeritus of Statistics (University of Missour-Columbia) who has also consulted for the National Bureau of Standards and the U.S. Army Air Defense Board, among others. And more diplomatically phrased than the initial statement that "the purpose of this book is to discuss whether statistical methods make sense."

The Nature of Statistical Evidence by W. A. Thompson. Springer, 2007. Mathematics Library QA276.16 .T488 2007 Link to MnCat Record