Some goodness of fit tests for exponential distributions

Summary This paper considers tests of exponentiality against unspecified alternatives. The test functions are modifications of those given by K ac , K iefer and W olfowitz [Ann. Math. Stat., Vol. 26 (1955) 189–211]. The distributions of these statistics are obtained with the aid of the Monte Carlo method, and the tables may serve to apply a test of exponentiality for the usual levels of significance (sample sizes n ≥ 10). A Monte Carlo comparison is also made with competing tests for exponentiality, namely with criteria arising from probability plotting and due to S hapiro and W ilk [(1964) in an unpublished manuscript, but recently described in “Statistical Models in Engineering” by H ahn and S hapiro (Wiley, 1967)], and with tests based on the conditional distribution of total lives and on a likelihood ratio.