Publication type
Survey Futures Working Paper Series
Series Number
17
Series
Survey Futures Working Paper Series
Authors
Publication date
May 15, 2026
Summary:
Survey researchers seeking a representative sample of the UK general population do not have the advantage of access to a population register, as there is no such register in the UK. Databases of addresses are instead widely used as general population sampling frames. Royal Mail’s Postcode Address File (PAF) has been the most popular of these for around thirty years. Although it has the advantage of containing reliable and regularly updated address data, it does not enable the identification of inhabited residences.
A more recent address database product, Addressbase Plus, adds administrative data onto PAF with the aim of differentiating between residential and commercial properties. However, this classification in imperfect. Addressbase Plus was used as the initial address frame for the 2021 England and Wales Census, then council tax records and other administrative data were used to confirm private home and residential institution locations. Given the known limitations of using address databases as general population sampling frames, there has been increasing interest in the use of individual-level administrative data in survey literature and practice.
Now that alternatives to using PAF on its own are available, there is a need to set evaluation criteria for UK general population sampling frames. Outside the UK context, there is European research to draw upon, suggesting the following assessment criteria:
1. Whether the frame has multiple sources of information, for example a population register matched with additional administrative data;
2. How frequently the frame is updated;
3. Estimated under-coverage rate;
4. Which subgroups are disproportionately under-represented.
General population sampling frames cannot be evaluated in isolation and require recent population estimates for comparison.
The addition of administrative data to PAF has been trialled for various purposes, including identification of subgroups, such as children, to be oversampled and to create small area population estimates. Research to date has emphasised the challenges of combining administrative databases with PAF, due to the imprecision of the matching process. The use of health records as a general population sample frame has also been trialled in the 2023 Lothian Health Survey. Analysis of the survey sample estimated the percentage of ineligible records using comparison with Census and Office for National Statistics (ONS) mid-year estimate data. The results suggested that ineligibility levels in the frame varied by Local Authority and age group and could be high. Examination of the address data also showed that it was of lower quality than PAF, containing errors that made some survey invitations undeliverable.
In conclusion, decision-makers cannot yet have confidence that alternatives to PAF would offer benefits without risks. For this to change, a quality evaluation framework for alternative sampling frames would need to be applied, requiring access to any administrative data sources being considered for sampling purposes. In the short term, incremental benefits could be offered by Addressbase but there is limited evidence for these.
Subjects
Link
https://surveyfutures.net/working-papers/
#589052