On Variance Estimation for the French Master Sample
It is common practice to take advantage of auxiliary information collected from a large survey to improve the estimation for a smaller related survey. A comparable technique was used to draw the French Master Sample conditionally on the rotation groups of the New Census chiefly for practical reasons, that is, to benefit from a recent sampling frame. In this article, we are interested in the case where a sampling design, including the sampling units themselves and their selection probabilities, is defined conditionally on an external survey. More specifically, we focus on variance estimation for estimators arising from the French Master Sample.
Auxiliary information, balanced sampling, conditional sampling design, cube method, expansion estimator, primary sampling unit, rotation group