requires only crude estimates (upper bounds) of relevant tail probabilities under the
alternative hypothesis. Using this approach, the (optimal) Hodges-Lehmann efficacies are
obtained for certain tests arising in multivariate analysis and regression analysis.