Mean estimation with calibration techniques in presence of missing data

The problem of estimating the population mean using calibration estimators when some observations on the study and auxiliary characteristics are missing from the sample, is considered. Some new classes of estimators are proposed for any sampling design. These new classes employ to all observation (incomplete cases too) in the estimation without using any imputation techniques. On the basis of properties derived and some simulation results, the proposed estimators are compared with other complete case estimators.