AN HEWMA-BASED ESTIMATOR FOR POPULATION MEAN UNDER SIMULTANEOUS NON-RESPONSE AND MEASUREMENT ERRORS IN TIME-BASED SURVEYS

Estimating key unknown population parameters, such as population mean and the population total, has been a fundamental objective in sample survey theory. However, this task becomes challenging particularly in time-based surveys when the observed data are contaminated by non-response and measurement errors. The presence of non-response and measurement errors can substantially reduce the accuracy and reliability of conventional estimators. Motivated by these practical challenges, this study proposes a novel estimator for the population mean based on a hybrid exponentially weighted moving average (HEWMA) framework. The HEWMA framework effectively integrates information from both current and past surveys. The proposed estimator optimally combines time-based information while mitigating the adverse impact of missing and mis-specified observations. Analytical expressions for the bias and variance of the proposed estimator are derived, and explicit efficiency conditions are established to compare it with the existing classical estimator. The performance of the estimator is further evaluated through extensive Monte Carlo simulations across measurement error and varying levels of non-response. The practical applicability of the proposed estimator is also demonstrated using a real survey dataset. The results indicate that the proposed HEWMA-based estimator consistently achieves lower variance and higher efficiency than the existing estimators cross a wide range of practical scenario. Therefore, the proposed HEWMA-based estimator offers a reliable and efficient alternative for estimating the population mean in time-based surveys affected by non-response and measurement errors.