摘要
对于缺失数据,本文根据目标变量和辅助变量的无回答者总体总量的无偏估计,利用再抽样(复制)技术,构造了使用回答概率的回归插补;进而,利用再抽样(复制)技术,得到了该插补估计的方差估计;并进行了大量模拟,模拟结果表明使用回答概率的回归插补估计及其方差估计具有良好的性质。
For missing data, based on the unbiased estimation of the population quantity of the nonresponser of the interest variables and the auxiliary variables, regression imputation using response probability and its variance estimate are formulated by resampling technologies such as Jackknife and Bootstrap. Simulation results show that this imputation estimation and its variance estimation have good performance.
出处
《工程数学学报》
CSCD
北大核心
2008年第4期616-622,共7页
Chinese Journal of Engineering Mathematics
基金
武器装备预研基金(9140A19030106HK0108)
关键词
无回答
抽样调
方差估计
回归插补
回答概率
再抽样技术
nonresponse
survey sampling
variance estimate
regression imputation
response probability
resampling
