To the Editor.—

We read with interest Lantos' article1 about statistics, sample size, and clinical trials, but the thrust of his argument, which questions the role of clinical trials in determining good clinical practice, cannot be allowed to pass without challenge.

Lantos is right to question the widespread use of the P value as arbiter of the validity of the result of a study. He argues that mathematical proofs do not require statistical justification, but mathematical proofs are not science but tests of logic.

This content is only available via PDF.
You do not currently have access to this content.