"Statistical significance," a technique that dominates medicine, economics, psychology, and many other scientific fields, has been a huge mistake. The outcome is a case study in bad science - how it originates and how it grows. These sciences, from agronomy to zoology, the authors find, engage "testing" that doesn't test and "estimating" that doesn't estimate. Heedless of magnitude and of a genuine engagement with alternative hypotheses, they "testimate." "Null hypothesis significance testing" is in other words a scientific train-wreck, about which a small group of statisticians have been warning for a century.Ziliak and McCloskey's book shows field by field how the wreck happened, reports on the fatalities, and offers a quantitative way forward. The facts will startle the outside reader: how could a group of brilliant scientists wander so far away from scientific magnitudes? And it will inspirit the scientists who seek conscious interpretations of "oomph" rather than arbitrary columns of t-tests: how can the statistical sciences get back on track, and fulfill their quantitative promise?Ziliak and McCloskey measure the disaster in their home field of economics, and in psychology, epidemiology, and medical science.
They touch as well on law, biology, psychiatry, pharmacology, sociology, political science, education, forensics, and other fields in the grip of "significance." This book shows for the first time how wide the disaster is, and how bad for science, and it traces the problem to its historical, sociological, and philosophical roots. Many statisticians have complained about it before, but have complained science-by-science.