Aligning Estimates for Common Variables in Two or More Sample Surveys

Abstract

In practice, for many sample surveys the estimates of several population totals are based on one set of weights, which reproduces the known population totals of auxiliary variables. Such a set can always be obtained by using the general regression estimator. If some variables, not necessarily with known population totals, are jointly collected in two sample surveys, then it may be desirable that the weights of both surveys produce the same estimates for the unknown population totals of the common variables. In this article we propose adjusting the general regression estimator to meet this consistency requirement by considering the common variables as additional auxiliary variables. It turns out that the adjusted general regression estimator generalizes Zieschang's method.