least-significant difference test

n. In an experimental design, the computation of all possible pairwise comparisons using a t or F statistic with a pooled error term. This approach minimizes type II error but increases the probability of type I error.