Research On Sampling Method And Application Of Dynamic Population Based On Continuous Stratified Sampling

With the rapid development of economy and society,as the main body of sampling survey,the scale and characteristic unit of the survey object change more and more frequently.As a research hotspot,the sampling survey method of "three new" enterprises,small and micro enterprises,which are small in scale,large in number and fast in change in characteristics,has attracted the attention of all walks of life.In view of this kind of survey population,with the passage of time,the scale and structure of the survey object are constantly changing.This paper defines it as a dynamic population and classifies it.If the traditional sampling survey method is used to cover the whole population,it is necessary to update the sampling frame in time.At the same time,the cost of updating is too high,and the sampling survey will lose its timeliness due to the long preparation process.If we choose the sample rotation in continuous sampling,we can determine the rotation unit and rotation rate,but we can not guarantee that the initial samples will be in the second and later It is still retained in the rotation sampling.According to the characteristics of dynamic population,this paper summarizes the existing results at home and abroad,and deeply explores and improves the sampling method of this kind of population.The main content of this paper is divided into three parts.The first is theoretical research.Select the appropriate adjacent time points to establish the sampling frame of the previous period and the current period.In the early sampling frame,auxiliary information is used for stratification.According to the distribution of auxiliary information,fast approximation method,geometric method or nonlinear square root method are used to determine the stratification boundary.In the current sample box,the same stratification index is used to stratify the current sample.Combining the information in the previous sample frame with the current sampling frame,the total population value and its variance are estimated by using the unit information and current information retained in the previous sample.The second is random simulation.Monte Carlo method is used to simulate the random.The results show that the variance obtained by continuous stratified sampling method for dynamic population is smaller than that by simple random sampling in current period,which proves the superiority of this method.The third is empirical analysis,using the annual income data of two economic census of small and micro enterprises in Xi’an city to verify the necessity and feasibility of using this method.The registered capital index value of these enterprises is used as auxiliary variable.Because the auxiliary variable distribution is unknown,the geometric method is used to calculate the hierarchical index value.According to the value,the overall is stratified and the dynamic overall continuity is established The sampling plan of stratified sampling,combined with the above-mentioned theories,makes an empirical analysis,and provides a reference for the actual survey involving dynamic population problems.