C. Ceccarelli, G.M. Giorgi, A. Guandalin
The use of auxiliary information considerably improves the estimates and allows their production for unplanned domains, as well. Different kinds of auxiliary information can be used to improve estimates. Demographic information from administrative sources is frequently used, but sampling information can also be used. In this context we present an extension of calibrated estimator () of a total, originally proposed by Deville and Särndal in 1992. Our proposal of estimator is able to take into account both demographic and sampling information. It also agrees with the needs of the Italian Labour Force Survey. The joint use of these kinds of auxiliary information in estimators raises some questions. How their sampling error can affect the efficiency of estimates and whether introducing exogenous elements of sampling error is justified by the increased efficiency or at least improvement in estimates? The aim of this article is: (i) to derive the expression for the estimate of variance of the estimator with sampling information, (ii) to present a new tool, based on sampling error, quick and able to support the introduction of additional auxiliary variables in the constraints system of the estimator. Some interesting results are achieved both analytically and by applying the proposal methodology to the monthly estimates of the Italian Labour Force Survey.
calibrated estimator; composite estimator; sample constraints; exogenous variance; constraints effect; constraints selection
Tipo di pubblicazione: