Key points are not available for this paper at this time.
ARTICLES in the medical literature increasingly present statistical tests of significance. Too frequently one has the feeling that these have been added only after bitter correspondence between the author, the editors and a consulting statistician. Used in this way, tests of significance add little to the report of an investigation. However, if early in the design of the study, the investigator considers, with the help of a statistician if he plans eventually to consult one, the methods that will be used in the analysis of the data, the result is apt to be an improved investigation focused on a specific . . .
Jane Worcester (1966) studied this question.