Cross-sectional variance estimation for the French "Labor Force Survey"

Pascal Ardilly, Guillaume Osier


This paper describes the method that was implemented by the Statistical Office of France (INSEE)
to calculate cross-sectional variance estimates for the French “Labour Force Survey”
(LFS). A home-made SAS application, named POULPE, was used for the calculations. After
outlining the LFS sample design in the first part, the paper presents the software POULPE,
particularly its technical capabilities and the theoretical principles underlying the variance estimation
methods it implements. The third part develops how POULPE was utilized to compute
variance estimates for the LFS, focusing on the statistical problems which were met and how
they were solved. Finally, we provide estimated standard errors for totals and ratios and we
make comments about those values.


Multi-stage sampling; software POULPE; standard error; confidence interval; design effect

Full Text:



Copyright (c) 2015 Survey Research Methods

Hosted by the Library of the University of Konstanz