The Application Of Doubly Robust Inverse Probability Weighting Methods In Missing Data Of Income And Expenditure

As we all know,income and expenditure data play an important role in reflecting the lives of residents.Government and scholars through the analysis of income and expenditure of residents to draw the corresponding conclusions,to provide a scientific basis for the formulation of relevant national policies.Want to get accurate analysis of the results,you need a complete and accurate income and expenditure data to do support.In the case of household income and expenditure data,there will often be no answer,making the data incomplete,affecting further analysis.When data is missing,the usual way is to delete the missing part of the data,only the integrity of the data analysis,in this way,although you can continue to the next analysis,but delete the missing data,only Analysis of the data with complete observations,so that the data used to analyze the overall representation of the deterioration,the corresponding analysis results will be unreliable.Therefore,it is necessary to make some adjustments to the missing data,to a certain extent reduce the error due to no answer,weighted adjustment method is to deal with missing data a class of methods.In the many methods of adjusting the missing data,the weighting adjustment method is mainly to solve the unit without answer a class of methods.At present,the weighted adjustment methods are Politz-Simmons adjustment method,weighted group adjustment method,post-stratification method and re-sampling weighting method.In this paper,based on the existing weighted adjustment method,the theory of bistable inverse probability weighted adjustment method and the application of income and expenditure data are discussed.On the basis of the relevant literature to sort out summary on the first of the missing data mechanism it has been assumed,and illustrates the concept of weights;secondly,on the theory of double stable inverse probability weighting adjustment methods are described,and introduced already some four kinds of weighting adjustment method:Politz-Simmons adjustment method,the weighted group adjustment Act,poststratification adjustment Act and re-sampling the weighted adjustment Act;again,the sound of the double inverse probability weighting adjustment method with the conventional method of adjusting the weighting group,poststratification adjustment methods and adjustment associated with the resampling four methods difference between depth analysis method,from the assumptions and methods of use of information,the applicable case,the error is reduced as well as the advantages of these live areas are compared,on the basis of these four methods o of adjusting the weighting deep analysis,but also for adjusting the effect of the four methods were analyzed and summarizd,obtained through simulation analysis of the operation,in the absence of rate gradually increased,compared to other Method,the bistable inverse probability weighted adjustment method has better adjustment effect;finally,combined with the income and expenditure data(data to Chinese Academy of income distribution CHIP database 2013),a robust double inverse probability weighting adjustment method has been applied,and to adjust the effect were analyzed and sununarized.Analysis concluded that due to the income and expenditure variables residents have good correlation itself,and there is a significant regression model fits dual sound premise applications inverse probability weighting adjustment method,therefore,the inverse probability of double stable population balance of payments data applications Weight adjustment method to adjust,can be a good reduction in data due to missing estimates caused by the deviation,making the analysis more reliable.