Improved Imputation Methods For Missing Data In Two-Occasion Successive Sampling

Missing data often complicates survey practitioners to construct reliable estimates of the required population parameters. Remembering this fact and motivated with the recent work, this article deals with some imputation methods to handle the missing data problem at the beginning of the analysis and some estimation procedures of the current population mean have been proposed in two-occasion successive sampling. The expressions for the bias, mean squared error and optimum mean squared error are derived using the concept of large sample approximations. The empirical performances are shown over the similar type of estimators designated for the complete response situation and over recently developed estimator which are defined for the situation when missing observations are observed in the sample data. Suitable recommendations have been made for survey researchers.