Estimation of Domain Mean Using General Class of Imputation Methods

Small area estimation (SAE) approach has been employed to produce realistic estimates for the variable of interest in cases when the available data are insufficient to produce reliable estimates for the domain. Missing data is a significant issue that affects sample surveys, but in case of SAE, it is particularly vulnerable. To overcome the problem of missing data in case of SAE, this study is a fundamental effort that proposes some general imputation methods for the domain mean estimation under simple random sampling. The mean square error expressions of the proposed imputation methods are determined up to first order approximation. The analytical study is carried out to establish the efficiency conditions. A simulation analysis is performed by utilizing hypothetically created data sets. Further, the simulation analysis is extended with a real data application. In addition, appropriate recommendations for practical applications have been provided to survey experts.