By David Martin, University of Southampton
The census comprises a detailed questionnaire about household and individual characteristics, sent to every residential address in the country. Unlike other surveys, there is a legal requirement to complete the census and this leads to very high levels of population coverage (around 94% in 2011). Statistical modelling is then applied to provide estimates for the entire population, from which a rich series of data products are published.
Individual census responses are confidential and kept securely for 100 years, but the very high level of population coverage enables the census to provide our most geographically detailed understanding of demographic, social and economic conditions, while preserving confidentiality. The huge range of users spans central and local government, businesses, charities and academic researchers. Applications include allocation of public funding from central to local government, the planning of education, healthcare and housing, and research in areas such as inequalities in health, ethnic differences in migration and broader understanding of the spatial population structure.
The high geographical resolution enables investigation of relationships between human and physical environments, for example analyses of environmental justice. The census data provide our most complete picture of ethnicity and are a key reference point for re-basing official estimates of international migration and population projections. In addition to providing basic demographic and economic information, census topics reflect a tension between covering the same topics as before (so we can measure change) and new topics (reflecting current demands).
The 2021 census will include new questions about gender and past service in the armed forces (the latter in England and Wales only). A further innovation for 2021 will be the integration of some data from administrative sources into the census outputs, including number of rooms and income estimates. These reflect a longer-term ambition to steadily transform the census process with data that can be continually updated through the 2020s. 2021 will also be the first time that UK censuses have been designed ‘digital first’. Most addresses will receive a unique access code with an invitation to complete the census online (with a target of 75%). Areas where online response is expected to be low will be sent paper questionnaires from the outset, although everyone still has the option to complete either online or on paper.
The organization of the census itself reflects a huge amount of geographical analysis, including preparation of the best possible address list, optimal deployment of a large temporary workforce, detailed modelling of expected responses. Differential allocation of effort between areas is essential to achieve comparable levels of response overall – essential for fair resource allocation and reliable comparison between local authorities.
Census outputs are produced for a hierarchy of geographical areas, with the most detailed counts being available for larger areas. These range from a few simple counts for the most detailed postcodes (typically around 15 addresses) through an extensive range of cross-tabulated statistics for Output Areas (mean population 309 residents) to the most detailed statistics (the finest breakdowns of age, employment, ethnicity, etc.) for larger areas. Census Output Areas are the foundation of this statistical geography and were first developed for the 2001 census using a Geographic Information System (GIS) and automated zone design process to generate consistent small areas which ensure the confidentiality of the data. They were updated for the 2011 census and will again be updated for 2021, to take account of major residential developments.
The high level of boundary stability makes possible the mapping of change over time. Output Areas are aggregated to form larger areas known as Lower Layer Super Output Areas (LSOAs, mean population 1614) and Middle Layer Super Output Areas (MSOAs, mean population 7,800). These census areas are the basis for many important non-census data products including annual small area population estimates, the government’s Indices of Deprivation (LSOA) and detailed mapping of Covid-19 prevalence rates (MSOA). An innovation in 2011 was the creation of a separate set of small area, known as Workplace Zones, optimized for the publication of data about workers and workplaces.
The census and the pandemic
It is reasonable to ask why the census is going ahead under the current pandemic restrictions. The answer is complex, but reflects both a pressing need for updated data in light of changed social realities (not least the need for Covid-19 denominator populations, and international migration data in light of Brexit), but also acknowledgement that developing and delivering a once-in-10-years census has been a very long process which cannot be easily rescheduled. The need for migration data, for example, is made more acute by the suspension in 2020 of other key data sources such as the International Passenger Survey.
The census regulations for England, Wales and Northern Ireland came into force in 2020 and set out the detailed arrangements, including the questions to be asked, based on several years of prior planning, including large-scale census testing and rehearsal. The digital-first approach means that the need for face-to-face interaction is much reduced compared to previous decades. The censuses in Northern Ireland and Scotland are coordinated with, but separate from, that in England and Wales – being led by the devolved administrations and statistical agencies. Scotland took the decision to defer its census process to 2022, and that will cause new challenges for everyone needing consistent statistics for the whole of the UK.
Some of the characteristics of greatest interest will themselves be affected by Covid-19 restrictions. Census questions about employment, place of work and mode of travel to work will be interpreted differently in 2021, and the unique residential distribution of university students on 21 March will undoubtedly present enumeration challenges. Although the census questions themselves cannot be altered, the online guidance has been updated to reflect Covid-19 circumstances.
2021 seems likely to be one of the most remarkable censuses of modern times: taken at a moment of pandemic lockdown and huge social change, committed to a new mode of data collection and to integrate administrative data into the outputs. The design is informed in many ways by geographical analysis and the results will deserve very close attention from geographers. Whatever the detail of your vision for how we should build back better, fairer and more sustainably, 2021 census data will be an essential part of the evidence base. Expect results to become available from spring 2022.
About the author: David Martin is a Professor of Geography at the University of Southampton. He is a Deputy Director of both the UK Data Service and the National Centre for Research Methods. You can follow him on Twitter @GeogDave
Suggested further reading
Martin, D., Dorling, D. and Mitchell, R. (2002), Linking censuses through time: problems and solutions. Area, 34: 82-91. https://doi.org/10.1111/1475-4762.00059
Martin, D. (2006), Last of the censuses? The future of small area population data. Transactions of the Institute of British Geographers, 31: 6-18. https://doi.org/10.1111/j.1475-5661.2006.00189.x
Catney, G. (2017), Towards an enhanced understanding of ethnic group geographies. The Geographical Journal, 183: 71-83. https://doi.org/10.1111/geoj.12162