Some Results on Generalized Difference Estimation and Generalized Regression Estimation for Finite Populations
Claes M. Cassel, Carl‐Erik Särndal, Jan Hakan Wretman · Biometrika · 1976
A superpopulation model is formulated under which the generalized difference estimator is optimal in an expected variance sense. The result is applied in the situation where an auxiliary variable is available and linear regression through the origin is assumed. A generalized regression estimator is proposed and shown to be better than a range of traditional competitors.