A BIVARIATE WEIBULL AND ITS COMPETING RISKS MODELS

Competing risks analysis with two causes of failure was considered. Since independence of causes are not realistic in many situations, bivariate dependent model was considered for the underlying causes of failure. The absolutely continuous bivariate Weibull model proposed by Ryu (1993) was modified in order to derive a Weibull competing risks model with crude and net hazards equals. This condition allow identifiability of the marginals corresponding to each cause of failure. Identifiability and estimation of its parameters by maximum likelihood method are investigated. Also, tests for some hypotheses of interest were studied. Simulation studies for comparison of the proposed, Ryu's and independent Weibulls bivariate models and of proposed with independent competing risks models were performed. Applications to real data also are presented.