在不可忽略抽样机制下目标变量的估计问题

抽样调查是通过对有限总体的重复抽样,用样本数据对总体的目标变量进行估计,但是若样本的抽样过程与目标变量有关,则样本分布不能代表总体分布,此时用样本数据来估计总体会产生很大的偏差。针对这种在不可忽略的抽样机制下如何进行目标变量的估计问题展开讨论,详细介绍了三种处理该问题的方法并对这三种方法进行了比较,得出第三种概率密度函数的方法是处理该问题比较好的一种方法。

A Study on the Estimation Problem of Response Variable under Non-ignorable Selection Mechanism

Sampling survey make use of sarnple data to estimate population variable by repeated sampling for finite population. But when these selection process are related to the response variable, sample distribution cannot represent population distribution. Ignoring the effects of the sampling process on the distribution of the observed outcomes in such cases may bias the inference very severely. This paper deals with this problem of Non - ignorable selection mechanism and proposes three methods and compares those methods, concludes that the third method is a better one.