敏感问题随机应答技术模型分层整群抽样下参数的估计  

Estimation of parameters in stratified cluster sampling on randomized response technique for sensitive question survey

作　　者：高歌[1] 范玉波[1] 王冕[1] 

机构地区：[1]苏州大学公共卫生学院卫生统计学教研室,苏州215123

出　　处：《第二军医大学学报》2009年第2期170-177,共8页Academic Journal of Second Military Medical University

基　　金：国家自然科学基金(30571620)~~

摘　　要：目的:为敏感性问题提供科学的较复杂抽样调查方法及其统计量的计算公式。方法:使用CochranW.G.的经典抽样理论,2种随机应答技术(RRT)模型,全概率公式,均数、方差的性质等理论与方法进行公式推导。结果:推导出二分类敏感问题双无关问题RRT模型分别在整群抽样、分层整群抽样下总体比例的估计量及其估计方差的计算公式;推导出数量特征敏感问题加法RRT模型分别在整群抽样、分层整群抽样下总体均数的估计量及其估计方差的计算公式;并在苏州大学学生婚前性行为、考试作弊次数的调查中取得了成功的应用效果。结论:本研究提供的敏感性问题2种RRT模型的整群抽样、分层整群抽样调查方法与统计量的计算公式信度较高。Objective： To explore a scientific, complicated sampling method and corresponding formulas for sensitive question survey. Methods： Cochran W. G. ＇s classic sampling theories, two randomized response technique（RRT） models, total probability formulas, and properties of mean and variance were used in this study. Results： Formulas for the estimation of the population proportion and its variance for dichotomous sensitive questions on Greenberg model in cluster sampling and stratified cluster sampling were deduced. Formulas for the estimation of the population mean and its variance for quantitative sensitive question on the additive constant model in cluster sampling and stratified cluster sampling were deduced. Our survey methods and formulas were successfully applied in surveying the pre-marriage sex and cheating in examination in Soochow University. Conclusion：Our survey methods and formulas on two RRT models for sensitive question survey in cluster sampling and stratified cluster sampling have high reliablility.

关 键 词：敏感问题 RRT模型 分层整群抽样 参数估计 估计方差 

分 类 号：R195[医药卫生—卫生统计学]