Two-dimensional sampling in practice

Abstract

This article explains the principles of the two-stage sampling design and presents the lessknown cross-classified sampling design. One purpose of the article is to allow the reader todifferentiate between the two survey designs and put in practice the sampling and estimationsteps. Respective variance estimators for these two designs are calculated in simple cases, andanalogies with one-way and two-way ANOVA are proposed. The comparison is motivated bythe ELFE french survey, and selections and estimations are illustrated using the softwares R,SAS and Stata.