A statistical assessment of clinical equivalence


  • Jochen Mau

    1. Statistics Project, Institute of Medical Biometry and SFB 175 Implantology, University of Tubingen, Hausserstr. 11, D-7400 Tübingen, F.R.G.
    Search for more papers by this author


An observed confidence distribution is proposed as a measure of strength of evidence for practically equivalent efficacies of two treatments. The concept is independent of prior opinions about relevant sizes of a difference in efficacy. It also avoids retrospective power calculations for trials with missed recruitment goals and allows combination of the confidence from independent experiments.