Improvement on estimating median for finite population using auxiliary variables in double-sampling

Abstract

The aim of the study is to suggest a difference-cum-ratio type of median estimator for finite population median using two-auxiliary variables in double sampling. Using simple random sampling without-replacement scheme (SRSWOR) the estimated mean square error (MSE) and BIAS are computed for the new suggested median estimator. The suggested median estimator has a smaller MSE than all other median estimators currently in practice, showing a valid contribution to the literature. In addition, some members of the suggested estimator and theoretical comparison of MSE are also computed. Finally, the numerical and graphical comparison of percent relative efficiency (PRE) is also computed for five different real data sets. The PRE of the suggested estimator were 13610.69%, 177.59%, 17626.95%, 204.13% and 181.29% for dataset I, II, III, IV and V respectively which reveals the importance of the new estimator.