New Conventional and Calibration Estimators of Finite Population Variance Under Two-Occasions Sampling in the Presence of Random Nonresponse

This paper tackles the issue of estimating population variance in surveys conducted over two occasions (successive) sampling, specifically when dealing with random nonresponse. A traditional estimator for this scenario is proposed and then enhanced by incorporating auxiliary variable information using a calibration approach. This calibration aims to minimize the negative impact of nonresponse on the survey results. The proposed estimators aim to improve the accuracy and reliability of inferences drawn from successive sampling surveys, where non-sampling errors can significantly affect the data quality and the final population parameter estimates. The study provides expressions for the mean squared errors (MSEs) and estimated MSEs of the proposed estimators to assess their statistical properties and performance. The results of empirical studies through simulation and the use of real-life data show that some members of the proposed calibrated estimators generally outperform the proposed traditional estimator, particularly when nonresponse rates are high, with a few exceptions.