Association Parameters Regression for Bivariate Failure-Time Data

Copula models are often used to model the dependence structure in bivariate failure-time data. We consider a covariate effect regression method on the copula parameter for Archimedean copulas. The proposed method can handle three different data structures, namely typical bivariate data, semi-competing risks data and dependent truncation data. We derive large-sample properties of the proposed estimators, and study their finitesample performances via simulations and application to a well-known data set. Key–Words: Archimedean copulas, Censoring, Clayton model, Truncation, Local odds ratio, Semi-competing risks data.