Results from a simulation study of the power of eight statistics for testing that a sample is form a uniform distribution on the unit interval are reported. Power is given for each statistic against four classes if alternatives. The statistics studied include the discrete Pearson chi-square with ten and twenty cells, X210and X220; Kolmogorov-smirov, D; Cramer-Von Mises, W2; Watson, U2; Anderson-Darling, A; Greenwood. G;and a new statistic called O A modified form of each of these statistic is also studied by first transforming the sample using a transformation given by Durbin. On the basis of the results observed in this study, the Watson U2statistic is recommended as a general test for uniformity.
展开▼
