Asymptotic Bias of the Product Limit Estimator under Dependent Competing Risks

摘要

A common assumption made in analyzing competing risk experiments is that the risks are stochastically independen. Under that assumption, the product limit estimator is a consistent estimator of the marginal survival function. We show that when the risks are not independent, the product limit estimator converges, with probability one, to a survival function which may not be the same as the marginal survival function of interest.