蒙特卡洛模拟对分层整群抽样下样本轮换的评价

发布时间：2018-10-18 20:08

【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。
[Abstract]:Objective to evaluate the reliability, validity and accuracy of the method and formula of sampling rotation continuous investigation under stratified cluster sampling by Monte Carlo simulation (SAS). Methods the results of investigation on vital capacity of primary and middle school students in Jiaxing City were taken as the simulation parameters. Monte Carlo simulation of sampling rotation under stratified cluster sampling was carried out with SAS programming when the sampling ratio was 10% and 40% respectively. In order to evaluate the reliability and validity of the method, 100 samples were simulated and 95 CI were calculated according to the formula. The 95%CI method was used to compare the estimated total mean value and the total mean value calculated by the 95%CI method in order to evaluate the reliability and validity of the survey method and its formula. In addition, the standard deviations of the mean values of 100 samples under two sampling ratios were compared to evaluate the sampling accuracy by simulating the continuous investigation of each sample under stratified cluster sampling and the continuous survey of 100 samples with fixed samples. The standard deviations of the mean values of 100 samples under the two methods were compared to evaluate the sampling accuracy. Results almost all of the simulated total mean under two sampling ratios were included in the 95%CI of the population mean (except twice at the sampling ratio of 10%). Standard for sample mean of continuous sample rotation under stratified Cluster sampling with two sampling ratios The difference is smaller than the standard deviation of the mean value of the fixed sample. Conclusion the method of sample rotation under stratified cluster sampling and its statistical formula have high reliability, validity and accuracy, which can be applied to the continuous investigation in the field of medicine in the future.
【作者单位】：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;
【基金】：国家自然科学基金(编号:81273188)
【分类号】：R181.2

【参考文献】