Publications and Preprints

Competing risks data consist of failure times and causes of failure. In practice, causes may not be known for all individuals of a study due to experimental constraints. We propose tests for testing equality of two competing risks against the alternatives of dominance of sub-distribution functions, sub-survival functions and cause-specific hazard rates under the assumption that causes are missing at random. The proposed tests are applied to simulated data to evaluate empirical powers to assess their performances.