Testing for New Better than Used in Expectation with Incomplete Data
Hira Lal Koul, V. Susarla · Journal of the American Statistical Association · 1980
This article proposes a test for the problem of testing that a life distribution is an exponential distribution against the alternative that it is new better than used in expectation, not exponential, on the basis of randomly right censored data. The test statistic is an analog of the “total time on test” test statistic.