Statistical power of goodness-of-fit tests based on the empirical distribution function for type-I right-censored data

摘要

In this study, the power of common goodness-of-fit (GoF) statistics based on the empirical distributionfunction (EDF) was simulated for single type-I right-censored data. The statistical power of theKolmogorov–Smirnov, Cramér–von Mises and Anderson–Darling statistics was investigated by varyingthe null and the alternative distributions, the sample size, the degree of censoring and the significancelevel. The exponential,Weibull, log-logistic and log-normal lifetime distributions were considered as theyare among the most frequently distributions used when modelling censored data. We conclude by givingsome general recommendations for testing the distributional assumption of parametric survival models inhomogeneous populations when using EDF-based GoF statistics.