Bootstrapping the Kaplan-Meier estimator


Stuart Barber & Christopher Jennison.
We describe several existing bootstrap methods of inference for the Kaplan-Meier estimator of survival probability, and propose a new alternative. Unlike many existing analytic methods, these bootstrap methods do not depend on the asymptotic normality of the Kaplan-Meier estimator, which is dubious in small to moderate sized samples.

Some key words:
Bootstrap, censored data, Kaplan-Meier estimator.


Back to publications list