The Research On Application Of Calibration Estimation Approach In Sampling Survey

With the development of the human society, data becomes more and more important. By analysis of the collected data, we can have a strong guiding significance in fields such as the operation of the enterprises, macro-control of national economy. The research goal of the paper is restricted to sampling survey, which is a very effective means of collecting data. The paper explores how to utilize auxiliary information in constructing calibration estimator, in the phase of sampling estimation in sampling survey, to estimate various parameters of the population. The paper mainly introduces deterministic method of calibration weights and the specific application of calibration estimation approach under different sampling designs, and further gives an overview of calibration estimation approach.The main idea of the calibration estimation is to use the auxiliary information during the adjustment of the original designed weight. Such adjustment can reduce the structural differences between the sample and the population, this will improve the accuracy of the estimation. This paper make thoroughly analysis about three main calibration approaches, and then come to some conclusions through mathematical reasoning and model comparison.After the analysis of calibration estimation theories, this paper introduces the usage of calibration approach in simple parameter estimating. This part of the paper mainly introduces and deduces the estimator of population totals under different circumstances, such circumstances include small domain estimation, non-response, two-phase sampling, two-stage sampling and dual-frame. This part of study has a significant meaning to the sampling practice in our country. In addition to a simple parameter estimation, calibration approach also has a great effect on the improvement of complex parameter estimation. Based on the theoretical deduction, this paper introduces the application of the calibration approach on the estimating of population quantiles, ratios, and population variance. This part of study widened the practical usage of the calibration approach.In order to test the advantages of calibration estimation approach, this paper create several datasets through simulation, these datasets are designed to have different original designed weight, correlation coefficient, and no answer rate. These datasets are then used to compare the accuracy of calibration approach and traditional p estimator. The result shows that the calibration estimator is generally superior to the traditional p estimator.This paper not only prove the advantages of calibration estimation approach by simulation, but also make a thorough analysis about the calibration estimating theories. The sampling survey practice and study in our country are seriously lagging behind the advanced level abroad, this paper not only provides a theoretical basis for related workers, but also make a contribution to the application and the development of calibration approach.