Research On Total Error Estimation Methods Based On Two Different Random Samples Random Sampling

In the process of sampling survey,the data source is becoming more and more complicated,and the sampling environment has also changed greatly.For the population with auxiliary information known,the population is unknown or the sampling frame is difficult to compile,the traditional sampling method is to first randomly select samples.However,the investigator has doubts about whether the randomly sampled sample is a probability sample,so a new method is needed to test whether the randomly sampled sample is a probability sample;for the population where all information is unknown and only a part of the information is known,traditional sampling methods will first Making assumptions may cause the population to be inestimable or have large deviations,and the size of the population cannot be accurately estimated.Therefore,new sampling designs and methods are needed.The integration of response-driven sampling and simple random sampling is a new field,which can not only take advantage of the characteristics that non-probability sampling does not require a sampling frame,but also can take advantage of the possibility that probabilistic sampling can evaluate and infer sampling plans.In this paper,through combing the development process and basic principles of using auxiliary information sampling and response-driven sampling,it is proposed that for randomly selected samples from populations with unknown populations or difficult to compile sampling frames,auxiliary information can be used to test whether they are probability samples.,Provide a basis for the random sample for the probability sample,and then use the auxiliary information to solve the non-response phenomenon as a supplement to the application of auxiliary information in the sampling theory;for the total unknown information,only one part of the total information,first use responsedriven sampling survey The whole population,then find the sample from the clear sampling frame in the sample of response-driven sampling,and then use simple random sampling to retake the same sample size from the subpopulation with the sampling frame to replace the original sample,and then use two methods Fusion can solve the problem of sampling and estimation of mixed samples,reflecting the feasibility and effectiveness of the new sampling method.Finally,for the proposed response-driven and simple random fusion methods,a random simulation comparison analysis was performed,and the sampling average error related indicators were used to judge the sampling efficiency,and a conclusion was drawn that the sample fusion method is more effective than response-driven sampling.