POWER EVALUATION OF SMALL DRUG AND VACCINE EXPERIMENTS WITH BINARY OUTCOMES

Power analysis based on the Fisher exact test is used to evaluate the usefulness of alternative designs for small drug and vaccine experimen ts with a binary, success or failure, outcome. The emphasis is on the comparison between designs with equal and unequal allocations of anima ls to treated and control groups having the same total sample size, an d between designs with equal and unequal allocations but with one or t wo fewer animals in the control than the treated group. Designs with u nequal allocation can sometimes be more powerful at a given level of s tatistical significance a than designs with equal allocation. Graphs a nd a table are presented to aid the choice of suitable designs for (1 - beta) = 0.80 and alpha = 0.05. (C) 1998 John Wiley & Sons, Ltd.