METİN, Cenker Burak Türkiye İstatistik Kurumu (TÜİK) - Örnekleme ve Analiz Teknikleri Daire Başkanlığı, Turkey , ÖZDEMİR, Yaprak Arzu Gazi Üniversitesi - Fen Fakültesi - İstatistik Bölümü, Turkey

Nonresponse Error and Examining the Effects on The Estimation of Population Mean: Two Subpopulation Study

25376

Nonresponse error appears due to various reasons as a result of lack of measurement for taken units in survey. Nonresponse error which takes place in some groups has common characteristics with respect to target variable causes nonresponse bias for estimators. Bethlehem (2002), Särndal and Lundström (2006) defined nonresponse bias subject to response propensities. The purpose of this study is to examine the effect of nonresponse error on the estimation of population mean with respect to mean square error (MSE) and nonresponse bias by using response propensities. Monte Carlo simulation was conducted since in theoretical way it is impossible to evaluate mean square error of the estimator when nonresponse occurs. By simulation study, mean square error and nonresponse bias values observed for the estimation of population mean at different population structures, sample sizes and response propensities. It is recognized by simulation results indicated that when response rate is high and the difference of propensities between two sub groups is large in a population, nonresponse bias for estimation of population mean is greater when it is compared with the case in which the response rate is low and the difference of propensities between two groups is small.

54

Nonresponse error , Nonresponse bias , Random response model , Mean square error , Relative bias

Journal Of Natural an‎d Applied Sciences

63