Advantages of the Closed Testing Method in Multiple Comparisons Procedures

Abstract

This article deals with the use of the Closed Testing approach in Multiple Comparison Procedures (MCPs). MCPs occur when after rejection of a global hypothesis of no effect of a given treatment, a set of pairwise comparisons between levels of that treatment are performed in order to find out significant differences between levels. Given a set of partial hypotheses, such as the pairwise comparisons of an MCP, the Closed Testing approach concentrates on testing the family of all the non empty intersections of these partial hypotheses. The results of our simulation study highlight the advantages of closed testing methods and prove that they are more powerful than other classic MCPs controlling the FWER.

Keywords:

• Closed testing
• Familywise error rate
• Multiple comparisons

Mathematics Subject Classification:

• 62J15
• 62L10