(Received February 8, 1993; revised December 27, 1993)
Abstract. The large-sample frequentist property of a frequentist bootstrap for a posterior mean with respect to a Dirichlet prior of the survival function for a randomly censored data is given. The weak convergence of a bootstrap version of the Susarla-Van Ryzin estimator is established on the whole real line. An illustration of the technique and some Monte Carlo studies are also given.
Key words and phrases: Bayesian bootstrap, survival function, square-integrable martingales.