-
Views
-
Cite
Cite
Handbook of Parametric and Nonparametric Statistical Procedures, 4th Edition by D. J. Sheskin, Biometrics, Volume 63, Issue 4, December 2007, Page 1306, https://doi.org/10.1111/j.1541-0420.2007.00905_13.x
Close - Share Icon Share
Extract
How does one write a brief review of a book that is 1736 pages long? Sheskin strives to write a “comprehensive reference in the areas of parametric and nonparametric statistics,” and achieves this goal to a considerable respect. The book certainly doesn't cover every conceivable statistical procedure, but what it does cover, it addresses thoroughly and in tremendous detail.
The book is laid out in 40 chapters, each about a particular type of statistical procedure. The emphasis of the book is really on testing. Estimation, when it is covered, is included as an adjunct to the tests discussed. The exception to this is a series of chapters on measures of association, but even these are laid out in a way that emphasizes the tests. Refer to an online listing of the contents for full details on which procedures are covered.
Every chapter follows the same progression: (1) a text statement of the hypotheses, assumptions, and related background information, including previous chapters that might be considered as prerequisites; (2) an example consisting of a story built around some (presumably artificial) integer data; (3) a detailed formulation of the null and all possible alternative hypotheses, in the context of the example; (4) the hand calculation of the test statistic; (5) interpretation of the results in the context of each possible alternative hypothesis; (6) additional analytic procedures (such as follow-ups to an ANOVA, confidence intervals for parameters, power calculations, and so forth); (7) additional discussion, including theoretical details and practical issues; (8) additional examples, frequently consisting of new stories invented around the original data set; and (9) references and endnotes. The book has a lengthy introduction, covering much of the same material that an introductory class might (except for the actual tests and confidence intervals). There also occasionally appears some SPSS output, especially during the new (to this edition) chapters on multivariate analyses.
