Comparing Calibration Estimators in Labor Force Survey of Iran
by
Arman Bidarbakhtnia
1414663111,Statistical Center of Iran(SCI),Tehran, Iran
Coauthors: Hamid Reza Navvabpour

Using suitable auxiliary variables in calculating weighted estimators can improve efficiency of estimators. This type of estimations are using for compensating unequal selection probabilities, nonresponse, noncoverage or departures from known population counts. Simplest estimators that are using auxiliary variables are regression and ratio estimators. Devill and Sarndal (1992) shows that regression estimators can be shown as a weighted estimators that their weights have produced with minimizing a distance function such that calibration equations are satistied. The weights are producing in this manner are known as calibration weights and the resulting estimators are calibration estimators. In this paper we have compare several calibration estimators according to several distance functions in Labor force Servey of Iran and it has seen that D&S method has less impact on the sample distribution and leads to a more precise estimation. All programes are written in SAS/IML.

Key Words: Calibration Weighting; Model Assisted Estimators; Nonsampling Errors.

Date received: July 23, 2006