Statistics in Transition

This paper provides a comprehensive description of an integrated system of cross-sectional and longitudinal weighting for rotational household panel surveys. To be concrete and detailed, it develops the weighting procedures with reference to the EU-SILC integrated design. EU-SILC (Statistics on Income and Living Conditions) covers data and data sources of various types: cross-sectional and longitudinal; household and personal; economic and social; and from registers and interview surveys. The standard integrated design involves a rotational panel in which a new panel is introduced each year to replace one quarter of the existing sample; persons enumerated in each new panel are followed-up in the survey for four years (Verma and Betti, 2006). A common rotational sample of this type yields each year a cross-sectional sample as well as longitudinal samples of various durations. These sample data have to be weighted to make them more representative of the target populations they represent. The paper begins with a summary of the main features of EU-SILC and an overview of the integrated weighting system for the different types of data coming out of the rotational panel annually. It describes a step-by-step procedure for construction of initial weights to be applied to each new sample as it is introduced into the survey. An innovative feature of the weighting procedure is the concept of base weights (Verma and Clemenceau, 1996). Starting from the initial weight of each individual in the original sample, the person’s base weight is constructed for each subsequent year to compensate for panel attrition. The final objective is of course to construct cross-sectional weights and longitudinal weights for use in data analysis. Procedures are described for constructing these two types of weights from the same base weights. This makes the whole system of weights internally consistent and integrated.