In the field of ranking and selection, the quality engineer would like to compare the mean lifetimes of several treatment populations with the control for exponential distribution. In most cases, the scale parameters are unknown and unequal. We proposed a one-stage multiple comparison procedure for comparing k-1 treatment exponential mean lifetimes with the control based on doubly censored samples under heteroscedasticity. The tables of critical values for the multiple comparison procedures are listed for the practical use. At last, one example of duration of remission for 20 patients each taking one of four drugs for the treatment of leukemia is given to demonstrate the application of our proposed method.
Communications in Statistics: Simulation and Computation 50(5), p.1473-1483