因变量缺失下线性回归模型的估计与检验

Estimation and Testing for Linear Regression Models with Missing Response

文章研究因变量缺失下的线性回归模型,借助单点插补方法,首先给出模型的估计,研究参数估计量的渐近正态性,其次,对于模型系数的线性约束检验问题,基于Wald方法构造检验统计量并给出其渐近分布.最后,通过数值模拟验证所提方法的有效性.

This paper considered statistical inference for linear regression models when the responses were missing at random.We proposed a least-squares estimator for the parametric with imputation method and showed that the resulting estimator was asymptotically normal.To check the validity of the linear constraints on the coefficients,we constructed a Wald-test statistic and demonstrated that it followed asymptotically chi-squared distribution under the null hypothesis.Finally,some simulations were conducted to illustrate the proposed methods.