Data transformation, power, and skew: A rejoinder to Games.

The present authors (see record 1982-27158-001) concluded that if a transformation can be found that will minimize skew while tending to equate group variances, the use of that transformation will increase power of the F test. P. A. Games (see record 1983-20171-001) challenges this conclusion and presents what he considers to be counterexample. That counterexample is shown to be flawed; however, it aptly demonstrates the misconceptions that may arise with respect to the issue of data transformation, skew, and power. (11 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)