Post date: Apr 15, 2015 4:59:14 PM
Estimation and inference with semiparametric models
Zhezhen Jin, Department of Biostatistics, Columbia University
Abstract: It is challenging to obtain variance estimates if estimating functions are nonregular, such as non-smooth and non-monotone, in particular for censored data. We develop self-induced objective functions and Monte Carlo smoothing method for the estimation and inference for semiparametric models. The approach does not require any explicit form, not even the estimating function itself. We present general convergence theory and demonstrate the methods with examples.