Deals with methods for the design and analysis of surveys, studies and experiments when the data are qualitative or categorical. Estimation of the degree of association or difference assumes equal importance with assessment of statistical significance. Presents the standard error for most measures of association or difference mentioned in the text. New topics include: construction of confidence intervals; planning of comparative studies with unequal sample sizes; the Fisher-Irwin "exact" test for a fourfold table; attributable risk; the Cornfield method for making inferences about the odds ratio; and measuring inter-rater agreement. Includes new and revised problems and worked examples.
This book has a very narrow focus, specifically statistical tests for rates and proportions (odds ratios). It is probably not suitable for a classroom setting, lacking exercises for students, but it does contain worked problems and derivations for all statistical testings making it an excellent reference guide for practicing statisticians. It is well indexed with handy topical subdivisions in the table of contents, enhancing its utility.