MODIFIED WILCOXON TEST FOR 2 DIFFERENT DISTRIBUTIONS
成果类型：
Article
署名作者：
GREEN, JR
刊物名称：
BIOMETRIKA
ISSN/ISSBN：
0006-3444
发表日期：
1979
页码：
645653
关键词：
摘要：
Given independent samples from the continuous distributions F(x) and G(y), which may be different, using the Wilcoxon or Mann-Whitney test is considered, with the normal approximation, to test between H0:pr (X > Y) = p0 and H1:Pr (X > Y) > p0, especially with p0 = 1/2. A modification of the Wilcoxon test due to Potthoff and to Zaremba is considered. Additional restrictions are introduced to give improved tests. Considerations of asymptotic relative efficiency and Monte Carlo studies show that the power of these tests compare favorably with those of the ordinary t test, the Welch-Aspin t test and a test due to Sen.