A Kruskal-Wallis Type Test for Multi-treatment Effect
摘要
The problem of testing for the presence/absence of a treatment effect in case of multi-treatment is attacked here through a new test-statistic based on the same ideas as the classical Kruskal-Wallis test. Its theoretical properties are studied, under both the null and the alternative hypothesis, and an approximation based on sub-sampling is proposed. Then, the problem of testing for stochastic dominance is also dealt with.