Imputation of missing data for domain mean estimation using simple random sampling

Abstract

The estimate of domain mean is a significant issue in sample surveys. However, if the data is missing, it becomes very necessary. In the case of missing data, this paper proposes some direct and synthetic domain mean estimators using simple random sampling. To evaluate the performance of the suggested estimators against existing estimators, the algebraic formula of mean square errors is deduced. Additionally, a thorough, extensive simulation study was conducted utilizing a normally distributed population. Certain applications that contain actual data are also made available. The results of the simulation show the superiority of the suggested direct and synthetic Searls power ratio imputation approaches over the direct and synthetic mean imputation approaches, direct and synthetic ratio imputation approaches, and direct and synthetic power ratio imputation approaches by minimum mean square error and maximum percent relative efficiency. Furthermore, the proposed direct and synthetic imputation approaches are demonstrated using a real data based on the crop production from Agra district, located in the Indian state of Uttar Pradesh.