On enhanced estimation of population variance using unconventional measures of an auxiliary variable

Most of the research work on ratio, product, and regression estimators are usually based on conventional measures such as mean, quartiles, semi-interquartile range, semi-interquartile average, coefficient of skewness, coefficient of kurtosis, etc. The efficiency of these conventional measures is doubtful in the presence of extreme values in the data. In this paper, we propose an enhanced family of estimators for estimating the population variance using unconventional location measures such as tri-mean, Hodges-Lehmann, and decile mean of an auxiliary variable. The performance of the proposed family of estimators is compared with the existing estimators using a simulation study and two real populations. Also, the robustness of the proposed estimators was examined using an environment protection data with extreme values. The results showed that the proposed family performs better than its competitors not only in simple conditions but is also robust in the presence of extreme values.