Comparison of estimators in finite population sampling on successive occasions

Three estimators of the population mean on the current occasion, which can be applied in finite population sampling on successive occasions, are given and compared. One of these estimators is an unbiased estimator which uses only the data contained in the sample on the current occasion. Other two estimators use also the data contained in the sample on the previous occasion. One of them is an estimator in the form related to the conventional estimator, and the other is a new estimator. The new estimator is always consistent, whereas the others are consistent for some special weights used in the estimation. The bias of the new estimator is less than that of the modified conventional estimator. The variance of the unbiased estimator is larger than those of the others. For some weights used in the estimation, the variance of the new estimator is less than that of the modified conventional estimator.