A quantile goodness-of-fit test for Cauchy distribution, based on extreme order statistics. (Q1771834)

The author presents a test statistic for testing goodness-of-fit of the Cauchy distribution. This statistic has the shape of a quadratic form of the first and the last order statistic and its distribution does not depend on the real parameter of the sampled Cauchy distribution. Critical constants for this test statistic are evaluated on the basis of 50 000 simulations for each considered sample size. A simulation study is done to compare this test statistic with the Anderson-Darling, Kolmogorov-Smirnov and the von Mises test statistics in the sense of their powers under several alternatives. Results for the considered alternatives show that the presented test statistic is more powerful for testing the goodness-of-fit of the Cauchy distribution than the other mentioned statistics.