Abstract:
In this work we study the Archimedean copula model generated by the Laplace transform of the Power Variance Function (PVF) frailty distribution to model the dependence structure of multivariate lifetime data. The PVF copula family includes, among others, the Clayton, Gumbel (Positive Stable) and Inverse Gaussian copulas as special or limiting cases. Dependence properties of the copula models are described. From a Bayesian framework, parameters of the marginal distributions and the PVF copula are simultaneously estimated using piecewise exponential distributions. We illustrate the usefulness of the methodology using data from the Australian NH&MRC Twin registry.