Calibration Estimator of Regression Coefficient using Two Auxiliary Variables

Abstract

Surveys are often based on complex sample designs where sampling units frequently have different probabilities of being selected. In the survey data analysis, sampling weights must be used to incorporate the sample designs. Regression coefficients are estimated to find the relationship between the study and auxiliary variables. Kish and Frankel (1974) deliberated the use of sampling weights in the estimation of regression coefficients. This paper describes calibration based approach to estimate the regression coefficient using two auxiliary variables. The variance estimation of proposed estimator is also developed. The empirical results based on synthetic and real population show that the proposed estimator, in terms of percent relative bias and percent relative root mean square error, performs better than the existing estimator. The proposed variance estimator shows a satisfactory performance in empirical evaluation.