ABSTRACT
We shall consider only semiparametric estimation, because under any specified GPH1 model a unique parametric family of survival distributions such that for any constant explanatory variable the survival distribution belongs to this family can be found. For example, for the GPHGW model this family is the family of generalized Weibull distributions, for theGPHGLL model this family is the family of generalized loglogistic distributions, etc. But under a specified GPH1 model and the corresponding class of survival distributions this model coincides with the AFT model. Parametric estimation for the AFT model is considered in Chapter 5.
