Statistical Inference: A Commentary for the Social and Behavioural Sciences

by Michael Oakes

Critical analysis of the use of statistical inference in social and behavioural research, its proper role, logic, and its abuse. Argues that ignorance and misunderstanding of the role of statistical inference has had a detrimental effect upon research in the field. Points the way to an appropriate appreciation of the part played by quantification in these disciplines. Examines the significance test versus interval estimation and evaluates Neyman-Pearson, Fisherian, Bayesian, and Likelihood inference. Presents arguments in a plainly written, easily understood manner.

196 pages, Hardcover

First published March 19, 1986

Reviews

[Christophe Van · Rating 5 out of 5]

Thirty years after its publication, this book is still very worthwhile reading. It is not a textbook focusing on mathematical statistical theoretical foundations or on how-to-do-it technique. Rather, as its subtitle implies, it discusses the substantive question of what-do-the-numbers-mean.

That discussion is especially valuable given the current "crisis of reproducibility" in psychology and biomedicine and the related concerns about "p-hacking."

The book does not require advanced training to understand its content. It is acccessible to anyone who has learned how to compute a p-value to test a statistical hypothesis or how to compute a confidence interval.

Strongly recommended to social scientists, psychologists, biomedical researchers, policy analysts, practicing statisticians and data analysts who report or rely on confidence intervals and significance levels. And to anyone teaching statistics, econometrics and quantitative research methods generally.