Using administrative data for constructing sampling frames and replacing data collected through surveys

Abstract

The UK Government agencies have an interest in making greater use of administrative data. This paper discusses the methodological challenges of using administrative data held by the Department for Work and Pensions (DWP) for constructing an alternative sampling frame for the Family Resources Survey (FRS) and for replacing income data collected in the FRS. The analysis shows that administrative data would provide a good basis for sampling the FRS by linking to socio-demographic data held administratively, which would be advantageous for sampling specific sub-populations (e.g. types of benefit recipients or households with children). Sampling using administrative data also potentially offers more representative population sampling by including some addresses currently excluded from the small users Postcode Address File (PAF) but which are important to DWP, such as people living in care homes. However, comparing FRS questions and income data held administratively, suggests little overlap. Therefore, limited gains may be realised from shortening the length of the FRS by using income data from administrative sources.

Keywords:

• administrative data
• Family Resources Survey
• sampling
• PAF
• linking

Acknowledgements

This article derives from a project investigating alternative sources of sampling frames to improve accuracy of the Family Resources Survey, jointly funded by the ESRC and the DWP, Grant reference: RES-173-27-0204. The author is very grateful to the ESRC and DWP for funding the research on which this paper reports and to John Snow, Statistical Officer at the DWP, for his statistical guidance with using DWP administrative data; also for the very helpful comments given for this article from Valerie Christian, Team Leader of FRS Developments at the DWP, and Charlie Owen, Senior Researcher, at the Thomas Coram Research Unit, Institute of Education.