论统计调查中的域估计及其实证研究

The Theory and Example of Domain Estimation for Sample Survey

抽样调查是获得统计数据的一项重要途径,而对需要进行研究的层次作为子总体进行分层抽样实际上就是域估计问题。首先介绍了域估计的两种方法,包括直接估计和间接估计,然后对域估计进行了案例分析,最后对间接估计和直接估计的特点进行了对比分析,发现在一般情况下,如果样本量很充足,也无缺失数据,则使用传统的直接估计方法最为便利。当直接估计方法不能很好地满足我们的需求时,使用模型来进行间接估计就是一种必要的选择。

Sample survey is an important way to find statistics; the problem of sample survey by dividing layer is domain estimation in fact. At first two ways were introducecl for domain estimation in this paper, then the example of domain estimation were showed, the character of directly estimation and indirectly estimation were compared at last. The results showed that in general cases, if the sample amount is very sufficient, there are no data of lacking, it is most convenient to use the traditional directly estimation method. When directly estimation can＇t well meet our demands, use models to estimate indirectly is a kind of essential choice.