Abstract
A modification of the small sample up-and-down method is studied for the normal, logistic, and extreme value distributions. Two specimens are tested “in series” (i.e., simultaneously) at each stimulus level, and the decision to increase or decrease the stimulus level for the next test is made as soon as the first specimen responds. This procedure can markedly reduce the total test time in certain situations. Overall, the efficiency of the new procedure, as judged using Dixon's E f criterion, is sufficiently high to warrant its consideration when overall test time is much more critical than specimen cost.