Are flexible designs sound

摘要: Flexible designs allow large modifications of a design during an experiment. In particular, the sample size can be modified in response to interim data or external information. A standard flexible methodology combines such with weighted test, which guarantees type I error level. However, this inference violates basic principles. example independent N(mu, 1) observations, test rejects null hypothesis mu < = 0 while average observations is negative. We conclude that its most general form corresponding not valid. Several possible are discussed focus on alternative tests.