Riley Abel

Comparing multiple treatments separately can lead to an unacceptably highchance of falsely rejecting at least one hypothesis, creating a largefamily-wise error rate (FWER). To account for this, we use the idea ofcoherent and consonant simultaneous test procedures (STP’s) and showhow they lead to a strong control of the FWER. We then compare theresolution of various STP’s, that is, the probability of rejecting a minimalhypothesis given that the global hypothesis is rejected. Finally, we use theexample of Tukey and Dunnett’s tests to show that reducing the number ofminimal hypotheses can lead to a more resolvent STP.