Posterior analysis of latent competing risk models by parallel tempering

Latent competing risk models are examined from a Bayesian point of view. The parallel tempering algorithm is applied for posterior inference and compared with di4erent Markov chain algorithms such as the Gibbs sampler in terms of mixing. A simple remedy is suggested for reducing the computational cost of the parallel tempering, and posterior estimates are obtained from the relabeling algorithm. The methodology is illustrated by both simulated and real data. c 2003 Elsevier B.V. All rights reserved