Abstract

Abstract: The auxiliaryinformation is used frequently by the survey sampling researchers at theestimation stage in order to develop efficient estimators from long decades.Besides the classical ratio, product and regression methods; exponentialestimators are useful for estimating the finite population mean when there islow degree of correlation between study variable and auxiliary variable. Inthis regard, Singh and Vishwakarma [8] suggested a ratio type exponentialestimator in two-phase sampling. Again, Searls [17] used coefficient ofvariation for improving mean per unit estimator. Srivastava [18] developed anestimator of population mean using estimated coefficient of variation of y fromthe sample value when population coefficient variation is not known. Further,Upadhyaya and Srivastava [19,20] have suggested an improved estimator ofpopulation mean Y ̅ in symmetrical population considering the estimatedcoefficient of variation of  y. Motivatedby these, the present paper proposes some improved ratio type exponential estimatorsof finite population mean under simple random sampling without replacement intwo-phase sampling using auxiliary variable. The proposed estimators have beencompared with the mean per unit estimator, the conventional two-phase ratioestimator and the two-phase ratio type exponential estimator suggested by Singhand Vishwakarma [8] both theoretically and empirically.

Keywords: Two-phase sampling,Ratio type exponential estimators, Auxiliary variable, Bias, Efficiency.