Use of Non-response and Measurement error for Mean Estimation

Abstract

\Ve proposed an improved class estimators for the finite population mean utilizing information on two auxiliary variables, under two phase sampling technique by using simple random sampling without replacement (SRSvVOR). The proposed class estimators is modified for handling the problem of measurement error and non-response. It is assumed that the problem of measurement error and nonresponse are present in second phase sample. The expressions for bias and MSE, are drived up-to- first order of approximation. Two real data set are used to investigate the performance of the proposed estimator. The proposed estimator is compared with usual mean estimator, traditional ratio estimator, exponential ratio type estimator, traditional regression estimator, Kirgyera(19S4) type estimator, Sahoo (1930) regression type estimator, and Roy(2003) regression type estimator. It is found that proposed estimators are more efficient than all other the considered existing estimators by observing the MSE values.