Using auxiliary variables in sampling methods and an application with regression estimator

The aim of this thesis is to study the effect of using auxiliary variable to obtain more accurate and more efficient results in estimation stages of survey sampling. In this sense; various ratio and regression estimators, which use auxiliary variables, their biases and mean square errors are investigated for the estimation of population mean in simple random sampling and stratified random sampling. Moreover, Generalized Regression (GREG) Estimator, which can be used in different sampling design, is considered. For comparable estimators, in which conditions which of them should be chosen is analysed.In application, using auxiliary variables taken from 2005 Annual Industry and Service Statistics average turnover of enterprises which work in Manifacturing Industry and have 20+ employees at 2006 are estimated by various ratio, regression estimators and GREG Estimator. In simple random sampling GREG Estimator gives the best result. In stratified random sampling Seperate Regression Estimator gives the best result through the estimators that mean square errors can be evaluated. It is shown that Monte Carlo simulation results may be taken when the mean square errors can not be evaluated exactly.Key Words:1.Auxiliary Variable2.Ratio Estimator3.Regression Estimator4.Generalized Regression (GREG) Estimator5.Monte Carlo Simulation