本文在响应变量随机缺失情形下讨论了半参数部分线性变系数模型的估计问题.首先采用局部线性方法估计系数函数,然后进一步估计常数系数.最后利用回归方法借补缺失的响应值,再用全部数据估计常数系数.本文进一步讨论了利用完整个体方法及借补方法求得的参数估计的渐近性质,并进行了模拟比较.
We develop regression imputation estimators of semiparameter varying-coefficient model with response variables missing at random. It is shown that the proposed estimators are asymptoyically normal with given asymptoyically bias and variance. A simulation example is used to compare behavior between the complete case estimation and the imputation estimation.