统计研究

• 论文 • 上一篇    

复杂抽样设计下的域估计问题研究

吕萍   

  • 出版日期:2017-07-15 发布日期:2017-07-18

The Research of Domain Estimation in Complex Survey Data

Lv Ping   

  • Online:2017-07-15 Published:2017-07-18

摘要: 随着国内定量研究方法的开展和大型调查数据的免费公布,研究者不仅使用抽样调查数据对总体分析,还需要对域总体进行分析。本文对调查数据满足域精度推断的域估计问题进行研究。首先,根据实际调查中的域估计问题,指出解决域估计问题最好的方法是事先确定好需要估计的域,并在抽样设计时兼顾域的估计精度。但是,在实际调查中还包含计划外的域,通过对简单随机抽样下的域估计问题的研究,说明非计划域的估计问题的最大难点是域样本量的随机性。然后,针对实际中的抽样调查数据多来源于分层、多阶段、整群和不等概等复杂抽样设计的问题,指出需要结合复杂抽样设计信息、域样本量的随机性、域样本在总体的误差层和误差群中的分布,对复杂抽样设计下的域估计问题进行研究。最后,以中国家庭追踪调查(China Family Panel Studies, CFPS)为例,对复杂抽样设计下的域估计问题进行案例研究。

关键词: 域估计, 方差估计量, 复杂抽样设计, 层, 群, 权数

Abstract: With the development in quantitative research and survey data of academic institutions open freely in China, researchers not only use the sample survey data to analyze the population, but also analyze the domains. In this paper, we study the problem of domain estimation which satisfies the precision estimation of the domain. According to the actual survey experience, it is pointed out that we need to make sure the domain and consider domain estimation in the sampling design. Then, we study the domain problem in the simple random sampling and find the problem is the random sample size in domain. In practice, survey data refers to sample design in which samples have been sampled in a way that is multi-stage, stratified, clustered and unequal probability sampling design. We need to consider the complex sampling design information, domain sample size and the domain samples distribution in population, especially the stratum and clusters. This paper uses the data of China family panel studies to study the domain estimation problems.

Key words: Domain Estimation, Variance Estimator, Complex Survey Design, Stratum, Cluster, Weight