On the comparison of procedures for testing the equality of Survival curves

Abstract

SUMMARY Lininger et al . (1979) conclude that, in small samples, a partial likelihood ratio statistict to test the equality of survival curves is not so appropriate as a Mantel–Haenszel statistic. Their results are shown to arise due to the use of a model which requires the estimation of nuisance parameters. An alternative model, which does not require such estimation, is considered. When this model is used, commonly employed tests perform similarly in small samples.