The empirical distribution function used in the classical Kolmogorov-Smirnov test is redefined in such a way that it becomes a function of a parameter called delta. Monte Carlo simulations are conducted for small sample sizes (n _ 25) in order to determine those values of delta that lead to improvements in power. A goodness-of-fit test procedure is proposed as a replacement to the classical Kolmogorov-Smirnov test. Simulation results indicate that the size of the test is not affected by delta, and that the proposed test procedure has power that is at least as high as the classical Kolmogorov-Smirnov test, and in many instances the power improvement is very substantial.
Abstract