부산대학교 도서관

In Reply.—Most of the issues raised by Drs Miller and Schumaker have been addressed in separate letters, three of which appeared in The Journal (240:2005; 2142; 2143, 1978) and one that will appear in Science. We welcome the opportunity to summarize our responses here.Miller and Schumaker state that the assumptions for "critical values" have been criticized by a number of prominent statisticians. They support this statement by reference to a single letter by two statisticians, which Science declined to publish. However, a subsequent letter by these same statisticians is to be published in Science, with a rejoinder that refutes each criticism made. Basically, the rejoinder describes fundamental epidemiologic realities of which the critics may not have been aware. Since the rejoinder is to be published in its entirety, it is not repeated here.In any case, our article in The Journal uses only one statistic from the