A Location- And Scale-free Goodness-of-fit Statistic For The Exponential Distribution Based On Maximum Correlations

摘要

We propose a goodness-of-fit statistic, Q_n, based on Hoeffding's maximum correlation (Fortiana and Grané 2003) to test the composite hypothesis that the data come from the two-parameter exponential family. We study its small and large sample properties, and we obtain tables ol'lhc critical values of Q_n and sonic power curves. We compare our statistic with the Shapiro-Wilk statistic lor exponentiality and the Gini statistic.