论文信息 - k-Sample tests based on the likelihood ratio

k-Sample tests based on the likelihood ratio

General k-sample tests are developed, including the classical Kolmogorov-Smirnov, Cramer-von Mises and Anderson-Darling k-sample tests, as well as new powerful omnibus tests based on the likelihood ratio. Conventional tests are sensitive to location difference among distributions, but are dull to detect the variation in shape. The new tests are sensitive to both location and shape. In fact, if the distributions of k-sampled populations are different in location only, the new tests are as powerful as the old ones. Otherwise, they are much more powerful.

Jin Zhang | Yuehua Wu

[1] Joseph Naus,et al. Tables of Critical Values for a k-Sample Kolmogorov-Smirnov Test Statistic , 1973 .

[2] W. J. Conover,et al. Several $k$-Sample Kolmogorov-Smirnov Tests , 1965 .

[3] I. McKeague,et al. Empirical likelihood based hypothesis testing , 2003 .

[4] Jin Zhang. Powerful goodness‐of‐fit tests based on the likelihood ratio , 2002 .

[5] W. J. Conover,et al. Practical Nonparametric Statistics , 1972 .

[6] Jin Zhang,et al. Powerful Two-Sample Tests Based on the Likelihood Ratio , 2006, Technometrics.

[7] M. Stephens,et al. K-Sample Anderson–Darling Tests , 1987 .

[8] W. Kruskal,et al. Use of Ranks in One-Criterion Variance Analysis , 1952 .

[9] J. Kiefer,et al. K-Sample Analogues of the Kolmogorov-Smirnov and Cramer-V. Mises Tests , 1959 .