A rank-based contingency table approach to testing for equality of treatment-effect patterns for several groups

Appropriate rank procedures are not presently available for the problem of testing for equality of treatment-effect patterns in several two-way layout s. In this paper, we consider a rank-based contingency table approach to ad dress this problem. We propose an asymptotically distribution-free test pro cedure that is based on position vectors, using the concept of grouping sim ilar block rank permutations to obtain an effective number of pattern categ ories. We illustrate the application of the proposed procedure to two data sets and provide the results of a substantial Monte Carlo simulation study of the power properties of the procedure.