ABSTRACT
This chapter describes the permutation approach for solving the multivariate ranking problem, from the initial set-up phases to the computation of the final global ranking. It illustrates the steps that need to be followed to set up and solve a multivariate ranking problem. The chapter considers the situations in which the ranking process needs to take into account either confounding factors and/or of intermediate levels of aggregation of the response variables; the so-called stratified and domain analyses are very useful procedures. It addresses some practical questions: the choices of the test statistics and of the combining function and the specific pairwise permutation strategy to be used. The multiplicity issue occurs in case of simultaneous testing when a set of statistical tests is considered simultaneously. To define one-sided multivariate test statistics within the combination of dependent permutation testing methodology, a suitable combining function must be chosen.
