FREDRIC C. GENTER, PH.D.
To the editor: Young, Bresnitz, and Strom (1) present two nomograms for sample size determination when the response variable is either dichotomous or approximately normal. In the instructions for use of the nomogram for dichotomous variables, the authors make an erroneous assertion on page 250, stating "The symmetrical design of the nomogram allows an arbitrary designation of control group." This is not true, and can have dire consequences.
It is true that the assignment of the "control" and "treatment" labels can be arbitrary. Once assignment is made, however, one must be careful in the calculation of the percentage change in
GENTER FC. Nomograms for Sample Size. Ann Intern Med. 1983;99:737–738. doi: 10.7326/0003-4819-99-5-737_2
Download citation file:
Published: Ann Intern Med. 1983;99(5):737-738.
Copyright © 2018 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use