Minimizing the total sample size when multiple 1-shot systems are compared against a common baseline

摘要

A test design is determined for simultaneously comparing the individual reliabilities of each of multiple alternative one-shot systems against the reliability of a single baseline, with a common specification for the size and power of each alternative-vs.-baseline comparison. Using the procedure of Y.J. Lee (1984), the total sample size (baseline plus alternatives) can be minimized by setting the ratio of the number of baseline system replicates to the number of replicates for each alternative system equal to the root of the number of alternative systems.