Wenyang Zhang

       A Semiparametric Multilevel Survival Model


In this talk, I am going to present a semiparametric multilevel hazards
model for clustered survival data.  This model makes no
distributional assumptions about the cluster-level random
effects.  Another feature of the proposed model is that the
effect of a continuous covariate is represented by an unspecified
function, rather than assuming a linear effect.  Based on local linear
modeling, we propose a two-stage estimation procedure for the proposed
model.  The performance of the method is assessed via Monte-Carlo
simulations.  Finally, the method is applied in an analysis of first birth
intervals in Bangladesh.