Research On The Application Of Spatial Sampling Method In Resident Income Survey

When the traditional sampling design is applied to the scene with strong spatial correlation,it is difficult to fully capture the spatial information and ignore the influence of the spatial attribute of the target population.In fact,many literatures show that there is spatial correlation among the target variable units of any survey population.As an extension of traditional classical sampling techniques,spatially balanced sampling has become a hot research issue in the field of geostatistics.In particular,BAS sampling technique is famous for its simple sampling process,high sampling efficiency,good sample representation and good adaptability to unexpected situations in the sampling process.This paper takes the household income survey in Shanxi Province as the background,and residents’ disposable income as the survey object to carry out classical sampling design and spatial sampling design respectively,and evaluates the feasibility of BAS sampling in household survey through comparative analysis of sampling process and effect.This paper firstly expounds the basic theories and methods of sampling survey,compares and analyzes the advantages and disadvantages of traditional probability sampling and spatial sampling in theory,and summarizes the advantages of spatial sampling method and BAS sampling technique on this basis.Then,the sampling process is simulated to study the difference between spatial sampling design and traditional sampling design and the sampling effect when the sampling population obeys the ideal Gaussian distribution and uniform distribution in space.In the empirical part,firstly,the spatial bubble map and trend surface analysis are used to demonstrate whether the population has spatial correlation.Secondly,simple random sampling and PPS sampling in traditional sampling methods are used to carry out sampling design and statistical inference for the per capita disposable income of Shanxi residents.In the spatial sampling design,regional GDP with high correlation with target variables was selected as spatial auxiliary information to improve the representativeness of samples.Then,balanced samples of BAS sampling and spatially correlated Poisson sampling were obtained by combining the longitude and latitude coordinates of counties in Shanxi Province.Finally,the results obtained by the spatial sampling method and the traditional sampling method are compared respectively with the true value of per capita disposable income in Shanxi Province in 2020.Multiple indicators such as sampling error,estimated variance and survey cost are selected as the evaluation criteria to measure the sampling efficiency.The performance evaluation and comparative analysis of different sampling methods are carried out to obtain the evaluation of BAS sampling effect.This paper draws the following conclusions:(1)The simulation results show that,in the ideal population distribution,the spatial balance sampling method is superior to the traditional sampling method in terms of sampling accuracy and error,among which BAS has the best simulation effect,and with the increase of sample size,the simulation sampling effect becomes better.(2)Empirical results show that the disposable income of Shanxi residents has obvious spatial characteristics.In addition,under the condition of the same sample size,the BAS sampling designed in this paper is more accurate in the estimation of sampling relative error and variance,and the obtained samples are significantly more representative,achieving a good spatial distribution of sample points.(3)The application of spatial balanced sampling technique in the survey population with strong spatial autocorrelation can not only improve the representativeness of the samples,but also save the survey cost to a certain extent,which is more favorable for improving the estimation quality of the overall parameters.This paper proves that the application of BAS sampling design and spatial sampling method in household survey is feasible to a certain extent,thus providing some reference for the government and the Bureau of Statistics in the household survey related work design.