SUMMARY
This paper proposes a class of estimators based on information of two auxiliary variables. The expressions of mean square errors of the proposed class of estimators are derived in a general form. It is shown that the proposed class of estimators is always more efficient than regression estimator based on two variable, estimators proposed by Abu-Dayeh (2003) and estimators recently proposed by and Lu and Yan (2014). In addition, we support this theoretical result by an empirical study using original data to show the superiority of the constructed estimators over others.