ABSTRACT
A narrowly focused “one hypothesis, one experiment, one decision” approach is often inadequate for the advancement of medical science and practice. Armitage (1991) wondered how statistics might have developed if R. A. Fisher had been employed in medical or industrial research. He suspected that especially experimental design might have developed differently, for agricultural experiments are essentially non-sequential in contrast to the sequential nature of most medical experiments. This difference is amplified by the very special nature of the “units of observations” in clinical trials, the patients, in comparison to those in other scientific investigations, e.g., with respect to blinding, informed consent and compliance in clinical trials. The nature of biomedical research requires stepwise procedures with respect to the sequence of questions to be addressed by separate trials during the four traditional phases of clinical research.
