蒙特卡洛模拟对分层整群抽样下样本轮换的评价  

Evaluation based on Monte Carlo simulation for sample rotation under stratified cluster sampling

作　　者：傅颖[1] 高歌[1] 刘升学[1] 周云华[1] 施佳琛[1] 

机构地区：[1]苏州大学医学部公共卫生学院流行病与卫生统计学教研室,苏州215123

出　　处：《安徽医科大学学报》2014年第11期1618-1621,共4页Acta Universitatis Medicinalis Anhui

基　　金：国家自然科学基金(编号:81273188)

摘　　要：目的采用蒙特卡洛模拟(SAS编程)对分层整群抽样下样本轮换连续调查的方法和公式进行信度、效度和精度评价。方法以嘉兴市中小学生肺活量的调查结果为模拟总体参数,各在抽样比为10%和40%时应用SAS编程进行分层整群抽样下样本轮换的蒙特卡洛模拟,分别模拟100个样本并按公式计算获得100个总体均数的95%CI,采用95%CI法比较模拟计算出的总体均值估计量与总体均数,以评价该调查方法及其公式的信度、效度。另外,分别在两种抽样比下模拟分层整群抽样下样本轮换连续调查和固定样本连续调查各100个样本并比较两种方法下100个样本均值的标准差,以评价抽样精度。结果两种抽样比下模拟总体均数几乎均被包含在总体均值的95%CI(仅在10%的抽样比下有两次不包含);两种抽样比分层整群抽样下样本轮换连续调查的样本均值的标准差较固定样本的均值标准差小。结论分层整群抽样下样本轮换的方法及其统计公式有较高的信度、效度和精度,今后可应用于医学领域的连续性调查。Objective To evaluate the reliability, validity and precision of the successive survey using sample rota- tion under stratified cluster sampling and its formulae based on Monte Carlo simulation. Methods Survey results of some students in Jiaxing about vital capacity were taken as simulative population parameters. Monte Carlo simula- tion was conducted with SAS programming when sampling ratios were 10% and 40% respectively. 100 samples in each sampling ratio were simulated and 95 % CI （ confidence interval） could be acquired. In order to evaluate relia- bility and validity of the survey and its formulae, simulated overall mean estimators were compared with population mean using 95% CI. In order to assess the sampling precision, 100 samples were respectively simulated for rotation survey and panel survey at each ratio（ 10% and 40% ） and their standard errors were represented. Results Al- most all the 95% CI of 100 estimated overall means at each sampling ratio （ 10% and 40% ） included the popula- tion mean except for two at 10%. The standard error of rotation survey was lower than panel survey. Conclusion Sample rotation under stratified cluster sampling and its formulae are of high reliability, validity and precision, which can be applied to the future medical successive survey.

关 键 词：蒙特卡洛模拟 样本轮换 信度 分层整群抽样 

分 类 号：R181.2[医药卫生—流行病学]