题 目: Pseudo semi-empirical likelihood for missing covariate data

报告人：王启华研究员（中科院数学与系统科学研究院）

时 间：2009年3月30日下昼2：00-3：00

地 点：伟易博新楼217

摘要: We consider a semiparametric population model which parameterizes the conditional density of a response Y given covariates X,Z but allows the conditional and marginal distribution of the covariates to be completely arbitrary. Covariate vector X may be missing. We first extract model information and then incorporate the informatiion by developing a pseudo semi-empirical likelihood to make improved inference. A pseudo semi-empirical likelihood estimator for the parameter vector describing the conditional density is defined, which is proved to be asymptotically normal and is more asymptotically efficient than the pseudo-score estimator. The semi-empirical loglikelihood and the semiparametric log-likelihood functions are proved to be distributed asymptotically as two weighted chi-squares, respectively. A simulation is conducted to compare the proposed methods and the existing methods.