Enhanced Robust Estimators for Estimating Population Means When Confronted with Non-Response and Measurement Error

Abstract

The use of estimators in statistics, quality assurance, and survey methodology can never be over flogged just as the use of sampling. Two of the major challenges of statisticians or surveyors due encounter at the course of data collection in the field of medical and social sciences is non-response and measurement errors. This poses serious problem during data compilation, computation and estimation stages. In this paper, a robust-based classes of estimators are proposed in the presence of non-response and measurement errors through the use of imputation scheme incorporated with measurement errors parameters. The properties of the proposed estimators (Biases & MSES) were derived up to the second degree approximation using Taylors’s series approach. The conditions for the efficiencies of the proposed estimators over the existing estimators was also considered and established in this research. The empirical study conducted using simulated data from normal distribution, exponential distribution, chi-square distribution, uniform distribution, gamma distribution and poison distribution revealed that the modified classes of estimators of the proposed imputation schemes are more efficient and satisfactory than the compared existing estimators. Thus, the proposed modified classes of estimators under imputation scheme were recommended for use in the real life situation especially in the presence of non-response and measurement errors during data analysis and estimation stages.