Sequential Tests of Hypotheses in Consecutive Trials

Summary Methods of testing the global null hypothesis in planned consecutive trials are proposed. One decision rule based on the inverse sampling principle stops further experimentation with the rejection of the global null hypothesis, if in already k of the m planned experiments the individual null hypotheses have been rejected at a certain level. This method for k ≥ 1 also allows early stopping with the “acceptance” of the global null hypothesis, if already m − k + 1 of the individual null hypotheses could not be rejected in their respective experiments. The second stopping rule is a repeated‐significance‐test‐version of Fisher's method for combining test results from different experiments.