671
Bayesian Model Selection: Analysis of a Survival Model with a
Surviving Fraction
Howard Seltman, Joel Greenhouse, and Larry Wasserman
Abstract:
We describe a model comparison problem approached from a
a Bayesian perspective. This is illustrated with a case study of a
randomized and controlled clinical trial investigating
recurrence of depression. The time until recurrence
is modeled as a survival model with a surviving fraction. Posterior
distributions are simulated using Metropolis-within-Gibbs Markov chain
methods. Sixteen versions of linear combinations of two covariates
in the log odds of being in the surviving fraction and the log of the hazard
rate are modeled. Bayes factors for comparing the models are
obtained by using the bridge sampling method of calculating normalizing
constants.
Here is the full postscript text for this
technical report. It is 219812 bytes long.