A Modified Anderson-Darling Test for Uniformity

Abstract. The subject of assessing whether a data set is from a specific distribution is crucial in statistical inference. This topic is critically important for uniform distributions as in generating random samples, a random sample is taken from the standard uniform distribution and then converted to the sample as required. Two different modifications of the Anderson-Darling A2 test are presented. The critical values for the modifications and the usual A2 test are also re-computed for different sample sizes. This is followed by a power comparison of the various tests.