基于链式方程的收入变量 缺失值的多重插补

统计研究 ›› 2009, Vol. 26 ›› Issue (1): 71-77.

基于链式方程的收入变量缺失值的多重插补

刘凤芹

收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2009-01-15 发布日期:2009-01-15

Multiple Imputation by Chained Equations of Missing Data in the Income Variables

Liu Fengqin

Received:1900-01-01 Revised:1900-01-01 Online:2009-01-15 Published:2009-01-15

摘要/Abstract

摘要： 在经济计量分析中收入变量的缺失值是一个普遍而又较难处理的问题。传统的处理方法往往导致分析结果具有系统偏差。本文提出利用基于链式方程的多重插补方法来处理收入变量的缺失值问题。文章将此方法应用到一个实际数据集，然后通过分析插补后的数据集讨论了此方法的性质，并和其他多重插补方法进行了比较。结果表明：基于链式方程的多重插补能在一定程度上纠正推断结果的系统偏差，并且给出恰当的标准差估计。

关键词: 基于链式方程的多重插补, 缺失值, 收入变量

Abstract: Missing data in the income variables is familiar and is hard to treat in econometric analysis. Traditional treats often lead systematic biases in results. It is proposed that using multiple imputation by chained equations deals with missing data in the income variables. Multiple imputation by chained equations is used by an real data set. Then the author analyzes the treated data set to discuss the character of the method and compares this method with other multiple imputation. The results show that multiple imputation by chained equations can correct systematic biases in the inference to some degree and It can give proper estimated standard error.

Key words: Multiple imputation by chained equations, missing data, income variable

刘凤芹. 基于链式方程的收入变量缺失值的多重插补 [J]. 统计研究, 2009, 26(1): 71-77.

Liu Fengqin. Multiple Imputation by Chained Equations of Missing Data in the Income Variables[J]. Statistical Research, 2009, 26(1): 71-77.