Statistical Rules of Thumb, Paperback (2nd Ed.)
Descriere
Contributor(s):Author: Gerald Van Belle Praise for the First Edition: ""For a beginner this book] is a treasure trove; for anexperienced person it can provide new ideas on how better to pursuethe subject of applied statistics.""--Journal of Quality Technology Sensibly organized for quick reference, Statistical Rules of Thumb, Second Edition compiles simple rules that arewidely applicable, robust, and elegant, and each captures keystatistical concepts. This unique guide to the use of statisticsfor designing, conducting, and analyzing research studiesillustrates real-world statistical applications through examplesfrom fields such as public health and environmental studies. Alongwith an insightful discussion of the reasoning behind everytechnique, this easy-to-use handbook also conveys the variouspossibilities statisticians must think of when designing andconducting a study or analyzing its data. Each chapter presents clearly defined rules related toinference, covariation, experimental design, consultation, and datarepresentation, and each rule is organized and discussed under fivesuccinct headings: introduction; statement and illustration of therule; the derivation of the rule; a concluding discussion; andexploration of the concept's extensions. The author also introducesnew rules of thumb for topics such as sample size for ratioanalysis, absolute and relative risk, ANCOVA cautions, anddichotomization of continuous variables. Additional features of the Second Edition include: Additional rules on Bayesian topics New chapters on observational studies and Evidence-Based Medicine (EBM) Additional emphasis on variation and causation Updated material with new references, examples, and sources A related Web site provides a rich learning environment andcontains additional rules, presentations by the author, and amessage board where readers can share their own strategies anddiscoveries. Statistical Rules of Thumb, Second Edition is an ideal supplementary book for courses inexperimental des
