Nonparametric Goodness-of-Fit Tests for Discrete, Grouped or Censored Data

The problems of application of nonparametric Kolmogorov, Cramer-von MisesSmirnov, Anderson-Darling goodness-of-fit tests for discrete, grouped and censored data have been considered in this paper. The use of these tests for grouped and censored data as well as samples of discrete random variables is based on Smirnov transformation. The convergence of statistic distributions to the corresponding limiting distribution laws has been investigated under true null hypothesis by means of statistical simulation methods, as well as the test power against close competing hypotheses. For discrete and grouped data the criteria have been compared by power with Pearson chi-squire test. The criteria have been also compared by power with the modified nonparametric tests for censored samples.