[Pair differences-U-test. A distribution-free comparative study of two therapies].
Distribution-free comparison of the effects of two therapies The Wilcoxon-Mann-Whitney test forms the basis for the following common test-problem according to which the U-test should be performed. On two rows of independent -- connected in parallell -- matched paired samples a continuous variable is determined. Observations on a single pair are contributed to the identical individuum. The question that arises is formulated as follows. Can the differences of a pair of the two rows of samples -- connect in parallel -- be distinguished from each other with regard to their location? To answer this question a modified application of the U-test is proposed. This test can be used to compare the efficacy of two different treatments. Other uses of this test are also possible.