国家卫生统计与专题调查样本地区的实验设计——社会经济多变量综合因子作为分层抽样的标识

本次研究旨在通过分层整群随机抽样抽取一个经济而有效的国家卫生统计和专题调查的地区样本。采用人口普查分地区社会经济、文化教育和健康等多个指标的资料,用主成分因子分析法确定抽样分层的标识,在此基础上用K-MEANS聚类分析法对总体聚类分层;采用分层整群随机抽样技术抽取多组不同样本容量的样本,分别计算出每个样本各变量的统计量;通过各样本变量的统计量对代表总体参数精确度的比较和各样本变量的统计量与总体参数分布拟合度检验确定最佳样本容量和样本地区。结果表明,抽取60个县(市)可以代表全国,至少90个县(市)可以代表全国不同类型地区。

Selection of a National Areal Sample for Acquisition of National Health Information in china

The objective of the study was to select a most economic and efficient area sample for collection of national health survey. Data was obtained from national population census and the principle component factor was been as the stratification criterion using the principle com- ponent analysis. All primary units (counties/ districts) were stratified into 5 strata by the score of principle component factor using the K-MEANS cluster analysis. Systematic random sampling was used to select a smple allowing each unit having egual probability of being se- lected regardless of its size. Comparing sample statistics with population parameters (mean va- lues and tests of the goodness of fit for frequ- ency) for five samples of 30,45,60, 90,120 counties/districts, we found that 60 sample co- unties/districts may be adequate to represent the whole nation and more then so sample un- its for 5 strata.