A simple, general, and efficient method for sequential hypothesis testing: The independent segments procedure.

Abstract

We propose a new sequential hypothesis testing procedure in which data are collected and analyzed in a series of independent segments. As in fixed-sample hypothesis testing and in previous sequential procedures, the overall α level can be set to any desired value. Like other sequential procedures, the independent segments procedure generally requires smaller samples than fixed-sample procedures-often approximately 30% smaller-to achieve the same α level and statistical power. Relative to other sequential procedures, the new method has the advantages that it is simpler to use, requires fewer assumptions, and can be used with a wider array of statistical tests. Thus, in some circumstances the independent segments procedure may provide an attractive option for increasing the efficiency of statistical testing. (PsycInfo Database Record (c) 2021 APA, all rights reserved).