Inference on the equality means of several two-parameter exponential distributions under progressively Type Ⅱ censoring

摘要

Two-parameter exponential distribution is one of the most widely used statistical distributions in medical sciences. However, it is difficult inference about the mean of the distribution, since it is a linear combination of both scale and location parameters. For comparing the means of several two-parameter exponential distributions, there is not provided an approach in literature. In this paper, we propose four methods for this problem under progressively Type II censoring: an approximate test, a parametric bootstrap, a generalized p-value and a fiducial approach. To compare the actual sizes and powers of these tests, a simulation study is performed. At the end, two real examples are presented to illustrate them.