Duncan's multiple range test

n. A post hoc multiple comparison procedure in which several means are ranked from lowest to highest, and then the number of steps that two means are apart is used to compute a range statistic for each comparison, which is used to determine which of several means differs significantly while controlling type I error.