(1) Department of Statistics, University of Vilnius, Lithuania;(2) Statistique Mathematique, University of Bordeaux-2,V. Steklov Mathematical Institute, RAN, Bordeaux,Saint Petersburg, France,Russia, 33076
Abstract:
We propose new two andk-sample tests for evaluating the equality of survival distributions against alternatives that include crossing of survival functions, and proportional and monotone hazard ratios. The tests allow for right censored data. The asymptotic power against local alternatives is investigated. Simulation results demonstrate that the new tests are more powerful than known tests when survival functions cross. We apply the tests to a well known study of chemo- and radio-therapy conducted by the Gastrointestinal Tumor Study Group. TheP-values for both proposed tests are much smaller than for other known tests.
