Next section: All pairwise t tests

Treatment conditions are normally included in an experiment to see how they compare to other treatment conditions. Therefore, experimenters frequently wish to compare the mean of each treatment condition with the mean of each other treatment condition. The problem with performing comparisons of all pairs of means (pairwise comparisons) is that there can be quite a large number of comparisons and therefore the danger of a high EER. The figure below shows the number of possible pairwise comparisons as a function of the number of conditions in the experiment. If "a" is the number of conditions, then (a)(a-1)/2 pairwise comparisons among condition means are possible. Therefore there are (a)(a-1)/2 chances to make a Type I error.

Next section: All pairwise t tests