摘要
目的介绍一种基于设计的复杂抽样调查数据总体特征量及其方差的估计方法。方法采用抽样权重估计有关的总体特征量,用泰勒级数线性法估计其方差,并给出具体的分析实例。结果加权估计量是总体特征量的近似无偏估计,所构建的95%CI可用于有关估计量准确度的评价及假设检验。结论基于设计的复杂抽样调查数据分析方法说明了调查设计的特征,在大规模调查情况下都能得到可靠的结论。
Objective To Introduce a design-based approach to estimate population quantities of interest and their variances. Methods Sampling weights are utilized to calculate population quantities of interest, and their variances are estimated by Taylor series linearization, and a practical example is also presented in this article. Results Weighted estimators are approximately unbiased for population quantities of interest, and their can then be used for the evaluation of the accuracy of point estimators and for hypothesis tests. Conclusion A design-based approach to analyze data from complex sample surveys takes account of the characteristics of survey design, and usually yields reliable results in large scale surveys.
出处
《中国卫生统计》
CSCD
北大核心
2008年第4期377-379,共3页
Chinese Journal of Health Statistics
关键词
复杂抽样调查
抽样权重
泰勒级数线性法
方差估计
Complex sample survey
Sampling weights
Taylor series linearization
Variance estimation
