Abstract
The Kaplan-Meier estimator for the survival function is known to be the limit of Bayes estimators. The choice of the KM estimator is shown to imply a unique limiting posterior distribution in the range where the KM estimator is uniquely defined. Coverage properties of posterior probability intervals based on this limiting posterior distribution are examined, and posterior probabilities of classical confidence intervals are examined. A simulation shows that the limiting Bayes intervals and the classical intervals perform similarly when there is sufficient information about the survival function.

This publication has 11 references indexed in Scilit: