Key points are not available for this paper at this time.
ABSTRACT Wilson (1967, 1974) indicated that when successive measurements are obtained in psychophysiological investigations, analysis of the data with repeated measures F tests will very likely result in positively biased tests of the repeated factor null hypothesis. The present paper demonstrates how the Greenhouse and Geisser (1959) three‐step approach to significance testing, a procedure intended to control the probability of a false positive, can be applied to experimental designs containing any number of repeated measures variables. In addition, the paper contains a short computer program for obtaining a numerical solution.
Keselman et al. (1980) studied this question.