Traditional and Calibration-based Estimation of the Population Mean for Stratified Random Sampling under Non-Response

Objectives: To propose estimators for the population mean considering the non-response presented in the data, utilizing an auxiliary variable with the help of both traditional and calibration estimation methods. Methods: To investigate how the suggested calibration estimators perform in the case of non-response, a simulation study has been conducted using the R software. Findings: As per the percentage relative efficiencies (%PRE) using Population- I and II, the suggested estimators outperformed the existing estimators considered here, The percentage relative efficiencies (%PRE) of the proposed estimators 𝑦𝑇̄ 𝑆𝑇,𝑦𝑇̄ 𝑌 ,𝑦𝐶̄ 𝑆𝑇𝑎𝑛𝑑𝑦𝐶̄ 𝑌 vary from 200.513% to 215.641%, 200.524% to 215.654%, 217.910% to 247.667% and 217.904% to 247.662%, respectively, while 158.931% to 164.414% for the existing estimator 𝑦𝐿̄ 𝑒𝑒 with respect to the estimator 𝑦𝐻̄ 𝐻, as shown in Table 4 for Population I. Similarly, the percentage relative efficiencies (%PRE) of the estimators with respect to 𝑦𝐻̄ 𝐻 those depicted in Table 4 vary from 144.9% to 148.3%, 237.2% to 264.7 %, 237.3 % to 264.7 %, 279.6% to 324.5% and 279.6% to 324.5% for the existing estimator 𝑦𝐿̄ 𝑒𝑒, and the suggested estimators 𝑦𝑇̄ 𝑆𝑇,𝑦𝑇̄ 𝑌 ,𝑦𝐶̄ 𝑆𝑇𝑎𝑛𝑑𝑦𝐶̄ 𝑌 , respectively, for Population II. Novelty: The proposed method has efficiently improved the traditional existing estimator’s precision in estimation. Using this method, we can derive more new calibrated estimators. Keywords: Calibration approach; Stratified random sampling; Chi-square type distance function; Non-response