Abstract:
We expose the results of studying the power of statistical tests for checking the hypothesis of homogeneity in randomly censored data for various situations (of different censoring degrees, alternative hypotheses, and laws of distribution of censoring moments). The results of modeling show that the power of the tests depends on the distribution of the censoring times in the case when the survival functions intersect. If they do not intersect then the distribution law for the censoring moments does not have a statistically significant influence on the power of tests. If the survival functions intersect then the Bagdonavičius–Nikulin tests are the most powerful of all those considered but their power decays rapidly as the censoring degree grows. If the survival functions do not intersect then the rank tests are more powerful than the Bagdonavičius–Nikulin tests.