A class of tests for testing an increasing failure-rate-average distribution with randomly right-censored data

For testing exponentiality versus (nonexponential) increasing failure rate average (IFRA) alternatives (which are nonexponential) using the randomly right censored data, a class of test statistics based on a functional of the Kaplan-Meier estimator is proposed. The asymptotic relative efficiencies of tests from this class with respect to other test statistics are derived. The efficiency loss due to censoring is studied. The proposed tests are applied to published survival data.<
>