Dear members of Allstat, I am currently working on a study which includes 3 treatment arms (equal n) and for which three pairwise comparisons are planned. I looked at the stats section of the protocol and noticed that there are two adjustments for multiple comparisons: Bonferonni correction in the sample size calulation and Tukeys method in the analysis. I have two questions: Do you think that the adjustment at sample size calculation was neccessary if the Tukey method will be used in the analysis? Does anyone know of a formula for estimating the sample size using Tukeys method? Kind regards, Des