New Memory Type Estimators for Systematic Sampling

This study aims to enhance estimation accuracy in systematic sampling by proposing a set of novel Exponentially Weighted Moving Average (EWMA)-based memory-type estimators. While memory-type estimators have been explored in other sampling frameworks, they have not yet been adapted to systematic sampling, which is known for its uniform population coverage and greater efficiency compared to simple random sampling. To address this gap, we develop three new estimators: An EWMA-based ratio estimator, an exponential ratio estimator, and a regression estimator. Through comprehensive simulation studies using both synthetic and real-world datasets, we demonstrate that the proposed estimators consistently outperform traditional methods in terms of efficiency. Notably, the ratio and regression-type estimators exhibit superior performance in different distributional settings, particularly when the weight parameter ϑ is set to 0.3 for symmetric distributions. These results offer a practical and robust alternative for survey statisticians and practitioners working with structured populations. The proposed methodology contributes both theoretically and empirically to the field of finite population estimation under complex designs.