ABSTRACT
The necessary restrictions imposed on articles by the scarcity of publication resources does result in greatly reducing the confidence we can put in statistical tests of significance as reported in the literature. The same number of results of a given level of significance would be expected from either “design.” The tradition of independent repetition of the experiments should be transferred from physics and chemistry to the areas where it is now a rarity. Journals should make space for brief reports of such repetitions, and foundations should undertake their support. If a hypothesis had been tested by a number of independent investigators, all but one of whom had obtained negative results, the publication of the positive results of the single investigator who had had the good fortune to obtain a “significant” result would normally bring the others out in the open.
