A post hoc technique that sequentially removes the smallest group means and then performs t-tests between the remaining groups, requiring adjustments on each step to control overall Type I error across the sequence of removals, especially when group sizes differ and heteroscedasticity is suspected — used after a significant ANOVA.