Confidence intervals are a more useful complement to nonsignificant tests than are power calculations

Open Access
Publisher Website
Google Scholar
Abstract
Many leading journals, including Behavioral Ecology, emphasize the importance of considering the power of statistical tests in the light of nonsignificant results. However, there is considerable scope for misinterpretation of what this advice actually implies. The common conception among biologists is that a nonsignificant result with low power is not to be relied on, but a nonsignificant result with high power is strong support for the null hypothesis. Here we will draw on a recent paper by Hoenig and Heisey (2001) to explain why use of (post hoc) power analysis actually provides no more information than does the p value itself.

This publication has 0 references indexed in Scilit: