Common Errors in Statistics (and How to Avoid Them), 4th Edition

TO LEARN MORE

Potential flaws in the bootstrap approach are considered by Schenker [1985], Wu [1986], Diciccio and Romano [1988], Efron [1988, 1992], Knight [1989], and Gine and Zinn [1989]. Some improvements are suggested by Fisher and Hall [1990, 1991]. Canty et al. [2006] provide a set of diagnostics for detecting and dealing with potential error sources.

Berry and Stangl [1996] include a collection of case studies in Bayesian biostatistics. Clemen, Jones, and Winkler [1996] subject Bayesian methods to an empirical evaluation. Kass and Raferty [1995] discuss the problem of establishing priors along with a set of practical examples. The Bayes’ factor can be used as a test statistic; see Good [1992].

For more on the strengths and limitations of meta-analysis, see Teagarden [1989], Gurevitch and Hedges [1993], Horwitz [1995, Egger, Smith, and Phillips [1997], Smith, Egger, and Phillips [1997], Smith and Egger [1998], Smeeth, Haines, and Ebrahim [1999], and Gillett [2001]. To learn about the appropriate statistical procedures, see Adams, Gurevitch, and Rosenberg [1997], Berlin et al. [1989], and Hedges and Olkin [1985]. On the topics of power, number of studies, and sample size per study, see Sterne, Gavaghan and Egger [2000]. Sharp and Thompson [1996, 2000] analyze the relationship between treatment benefit and underlying risk. Smith, Spiegelhalter, and Parmar [1996] describe a Bayesian meta-analysis.

For practical, worked-through examples of hierarchical Bayesian analysis, see ...