in a factorial analysis of variance, a set of nonredundant comparisons among the mean values for different samples. In other words, the set contains no overlapping information, so that the results from one comparison provide no information about the results of the second comparison. Also called orthogonal comparison.