SURVEYMEANS过程在抽样调查资料分析中的应用  

Application of the SURVEYMEANS procedure in the analysis of sampling survey data

作　　者：李长平[1,2] 胡良平 Li Changping;Hu Liangping(Department of Health Statistics, School of Public Health, Tianjin Medical University, Tianjin300070, China;Specialty Committee of Clinical Scientific Research Statistics of World Federation of Chinese Medicine Societies, Beijing100029, China;Consulting Center of Biomedical Statistics, Academy of Military Medical Sciences, Beijing100850, China)

机构地区：[1]天津医科大学公共卫生学院卫生统计学教研室,天津300070 [2]世界中医药学会联合会临床科研统计学专业委员会,北京100029 [3]军事医学科学院生物医学统计学咨询中心,北京100850

出　　处：《四川精神卫生》2017年第5期415-421,共7页Sichuan Mental Health

基　　金：国家高技术研究发展计划课题资助(2015AA020102)

摘　　要：传统的统计分析方法在进行差异性分析、线性与广义线性回归分析时,基本上都是基于样本来自无限总体、完全随机抽样的基础上估计抽样误差。而调查数据往往来自于分层、整群、多阶段或不等概率等复杂随机抽样方法,此时若采用前述提及的经典统计分析方法,则不能准确估计抽样误差。本文通过具体实例,介绍如何应用SAS软件中的SURVEYMEANS过程,更好地实现对通过各种抽样方法获得的数据进行统计描述和简单的统计分析,以便达到准确估计抽样误差、对总体参数描述和估计的目的。When performing a difference analysis or a linear and generalized linear regression analysis,traditional statistical methods are basically based on the sample from the infinite population or completely random sampling to estimate the sampling error.However,the survey data are usually collected from complex random sampling methods,such as stratified,cluster,multi-stage or unequal probability.At this point,the sampling error cannot be accurately estimated if the classical statistical analysis methods mentioned above are adopted.Through specific examples,this article aimed to apply the SURVEYMEANS procedure in SAS software which can better implement the statistical description and analysis of the data obtained by various sampling methods,in order to estimate the sampling error and population parameters accurately.

关 键 词：SAS软件 SUVEYMEANS过程 简单随机抽样 分层抽样 分层整群抽样 

分 类 号：R195.1[医药卫生—卫生统计学]