One-sided sign-type non-parametric procedures for comparing treatments with a control in a randomized complete block design

SUMMARY Non-parametric procedures are presented for comparing several treatments with a control when the data are collected in a randomized complete block design with no interaction. The procedures are generalizations of some well-known sign-type tests, and include both overall tests and multiple comparisons procedures. A numerical example is used to motivate the problem and illustrate the proposed methods. Some concluding remarks are offered.