The Methodology Study Of Sampling Design And Sampling Estimation Based On Overlapping Sub-population

With the change and development of today's economy and society,the objects in sampling survey and the structure of its overall goal are getting more complicated.The traditional random sampling with a single sampling frame may cause many problems,such as incomplete coverage and high cost of maintaining and updating of sampling frame.This paper holds that the sampling design and estimation based on overlapping sub-population have the advantages of complete coverage,low cost,high efficiency and simple operation,which is an effective method to solve these problems.In order to improve the estimation accuracy of overlapping sub-population sampling,combining the generalized regression estimation method on one-stage sampling,we propose a new estimator named Hartley-GREG based on Hartley's estimation method and introduced auxiliary variable and superpopulation model.Considering incomplete auxiliary variables,estimators based on both-side and single-side auxiliary information are proposed respectively,named Hartley-GREG-Bs and Hartley-GREG-Ss.In addition,for the two situations of the two-stage sampling design,the estimator forms of the Hartley-GREG estimator under the two-stage sampling are gradually derived.Through numerical simulation,the new estimator is compared with the traditional estimators Hartley and F-B.The results show that: the accuracy of new estimators is higher than that of traditional ones in most situations,and the estimation effect of the new estimator is greatly improved under the linear model.To better reflect the popularization and application value of overlapping sub-population sampling method,this paper gives the specific sampling design as well as the new generalized regression estimation program based on the overlapping sub-population of industrial enterprises under the scale.We hope that it can provide some reference for the application in China's statistical survey.