Statistical experimentation in therapeutics.

The issue I propose to examine is whether, or in what circumstances, the statistical approach to the assessment of curative measures is legitimate. This issue is straightforward, since, unless the mere enumeration of cases is regarded as a statistical procedure, the distinction between the statistical and the non-statistical approach to a problem is absolute; an investigator must inevitably be aware that he is making a clearcut and significant decision when he choses to adopt one method rather than the other.