Exercise:

Either a two-factor study with fixed effects. The postman admits=5 levels and factor B admits b=4 levels. The study is balanced

with n = 5 observations per cell. We want to determine a set confidence intervals to compare the effects in pairs of the levels of

A by setting level j = 1 of B. So we want to estimate _i1 _i1 for i > i. So there are 10 differences to estimate. Is it better to use the method of Tukey or the Bonferroni method? Justify your answer.