Hill's classic Medical Statistics, first published in 1937, now is in its tenth edition. The revised text includes nonparametric tests for paired and unpaired observations; the section on regression has been augmented, the table of random numbers updated, and ethical guides to human research added.
The essentials of recognizing extraneous factors that may influence the outcome, yet are outside the statistician's control, are brought home effectively and lend great strength to this book. Retained are the recurring themes that have made it a living classic: sound experimental planning, considered population sampling, and cautions against observer bias. Reminders that statistical significance does not necessarily confer biologic significance are one more reason for the continued popularity of Sir Austin's text.
The well-printed, low-cost softback edition is best read as a whole rather than as a reference. Since worked samples deal thoroughly with epidemiologic problems, the book will appeal especially to workers in