因变量缺失下线性回归模型的随机约束估计

Stochastic Restricted Estimation of Linear Regression Models with Missing Responses

本文研究了线性回归模型在因变量存在缺失的同时回归系数附加有随机约束条件时的估计问题,基于完整数据方法和单点插补方法,给出了模型系数的两种约束估计,并研究了这两类估计量的渐近正态性。最后通过数值模拟验证了所提估计方法的有效性。

This paper discusses estimation of linear regression models in the presence of multicollinearity and there are stochastic linear restrictions on the regression coefficients. Based on the complete-case method and single imputation technique, two stochastic restricted estimators are proposed. Asymptotically, properties of the proposed estimators are shown. Finally, some simulations are conducted to illustrate the proposed methods.