| Abstract: | Comments on the use of Bonferroni procedures for avoiding increases in Type I errors as outlined by R. Rosenthal and D. B. Rubin (see record 1985-15969-001). It is contended that these procedures conserve power for more important tests while sacrificing power for less important tests. It is suggested that with judicious choices of parameters (assessing both costs and benefits when choosing weights for individual tests and overall level of protection against Type I errors) the insurance cost is not large. (4 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved) |
