Title: A two-sample nonparametric statistic based on the power divergence
Authors: Hidetoshi Murakami - Tokyo University of Science (Japan) [presenting]
Abstract: A testing hypothesis for the shifted location parameter in two-sample is considered. In two-sample problem, the Wilcoxon rank sum test is often used to test the location parameter. We propose a new nonparametric statistic based on the power divergence measure between the two distribution functions. In addition, we derive the exact critical value of suggested statistic for small sample sizes. Deriving the exact critical value of the statistic can be difficult when the sample sizes are increasing. Since an approximation method to the distribution function of the test statistic can be useful, we consider the moment-based approximation. Simulations are used to investigate the power of the proposed statistic with various distributions for small sample sizes. We compare the robustness to the actual significance level and the power of suggested statistic with various nonparametric statistics.