Multiple Statistical Inferences

Abstract

Clinical trials often assess the efficacy of more than one new treatment and often use mquestions about subgroups differences or about what variables do or do not contribute to the efficacy results, remain. Assessment of such questions introduces the statistical problem of multiple comparison and multiple testing, which increases the risk of false positive statistical results, and thus increases the type-I error risk. In the previous chapter six commonly-used methods for controlling the risk of this problem have been addressed. This chapter gives a more mathematical approach of the problem, and gives examples in which different methods are compared with one another.