Logarithmic-type Estimators of Population Variance in Stratified Random Successive Sampling in the Presence of Random Non-response and Measurement Errors

Mansur Rabiu *

Department of Statistics, Zamfara State University, Talata Mafara, Nigeria.

Ahmed Audu

Department of Statistics, Usmanu Danfodiyo University, Sokoto, Nigeria.

Isah Muhammad

Department of Statistics, Binyaminu Usman Polytechnic, Hadejia, Nigeria.

*Author to whom correspondence should be addressed.


Abstract

Aims: To investigate logarithmic-type estimators of finite population variance that combine auxiliary information, stratification, and two-occasion successive sampling in the presence of random non-response and measurement errors.
Study Design: A mathematical formulation accompanied by a simulation-based comparison of conventional and logarithmic-type variance estimators.
Methodology: A finite population was divided into three strata, with a matched component retained from the first occasion and a fresh component selected at the second. The study considered artificial multivariate-normal populations of sizes 10,000, 9,000, 7,000, 5,000, and 3,000, a second-occasion sample size of 1,000, and 1,000 simulation repetitions. Performance was summarised using absolute relative bias (ARB), mean squared error (MSE), and percentage relative efficiency (PRE) relative to a conventional estimator. The numerical comparison was organised by matched and fresh sample allocation.
Results: In the displayed balanced allocation (500 matched and 500 fresh units), T_{2}^{\ast \ast } and T_{4}^{\ast \ast } achieved PRE values of 186.40 and 123.58, respectively; T_{1}^{\ast \ast } and T_{3}^{\ast \ast } had PRE values below 100. At an allocation of 900 matched and 100 fresh units, T_{1}^{\ast \ast } and T_{2}^{\ast \ast } had PRE values of 2528.81 and 2237.17. All four proposed estimators had PRE below 100 for the displayed allocations of (700, 300), (400, 600), and (300, 700). Absolute relative bias also differed considerably across settings.
Conclusion: The numerical results show substantial but allocation-dependent differences in MSE between the logarithmic-type estimators and the conventional estimator. Auxiliary-information estimators may offer efficiency gains in selected matched-sample configurations, while joint consideration of bias and MSE remains essential.

Keywords: Stratified successive sampling, population variance, logarithmic-type estimators, auxiliary information, non-response, measurement error, mean squared error, relative efficiency


How to Cite

Rabiu, Mansur, Ahmed Audu, and Isah Muhammad. 2026. “Logarithmic-Type Estimators of Population Variance in Stratified Random Successive Sampling in the Presence of Random Non-Response and Measurement Errors”. Asian Journal of Probability and Statistics 28 (12):53-59. https://doi.org/10.9734/ajpas/2026/v28i12986.

Downloads

Download data is not yet available.