Evidence and Evolution: A Review

TL;DR

The paper reviews Elliott Sober's 2008 book on the philosophical and statistical foundations of evolutionary biology, discussing hypothesis testing, model comparison, and debates like intelligent design.

Contribution

It critically analyzes the book's approach to statistical reasoning in evolution, highlighting its accessible philosophical insights and critique of model selection methods.

Findings

Explains statistical approaches to testing hypotheses in evolution.

Critiques the book's advocacy for AIC over Bayesian methods.

Highlights philosophical arguments against intelligent design.

Abstract

"Evidence and Evolution: the Logic behind the Science" was published in 2008 by Elliott Sober. It examines the philosophical foundations of the statistical arguments used to evaluate hypotheses in evolutionary biology, based on simple examples and likelihood ratios. The difficulty with reading the book from a statistician's perspective is the reluctance of the author to engage into model building and even less into parameter estimation. The first chapter nonetheless constitutes a splendid coverage of the most common statistical approaches to testing and model comparison, even though the advocation of the Akaike information criterion against Bayesian alternatives is rather forceful. The book also covers an examination of the "intelligent design" arguments against the Darwinian evolution theory, predictably if unnecessarily resorting to Popperian arguments to correctly argue that the creationist perspective fails to predict anything. The following chapters cover the more relevant issues of assessing selection versus drift and of testing for the presence of a common ancestor. While remaining a philosophy treatise, Evidence and Evolution is written in a way that is accessible to laymen, if rather unusual from a statistician viewpoint, and the insight about testing issues gained from Evidence and Evolution makes it a worthwhile read.