Superiority Testing Instead of Null Hypothesis Testing

Abstract

For this chapter some knowledge of power equations is required. This is given in Chapter 7 of the first part of this title. Superiority testing of a study means testing whether the study meets its a priori defined expected power. Many therapeutic studies may be able to reject their null hypotheses, and, are, thus, statistically significant, but they do not meet their expected power. Although p-values are widely reported, power is rarely given in the report. This may be a problem in practice, since lack of power indicates that the treatments are less efficacious than expected. Superiority testing assesses whether the eventual power of a study is in agreement with the power as stated in the sample size calculation of the study.