Abstract: | A method is proposed for calculating the small sample powers of rank tests which are based on the method of n rankings. A class of normal shift alternative hypotheses is considered, and Hodges–Lehmann efficiencies are calculated for the Friedman test. |