Census 2021 outputs: content design and release phase proposals

Closes 5 Oct 2021

Section 2: Shape of the outputs and analysis release

Overview

We aim to release the first Census 2021 data on the population of England and Wales within a year of census. Following that, we intend to release all the main results within two years of census. These outputs will include a wide range of products that users can interact with to obtain data to suit their needs.

We’ve developed this timetable based on feedback received from users following the 2011 Census.

It took us nearly three years to release 2011 Census outputs. User feedback highlighted that a main area of improvement for Census 2021 outputs would be to decrease the length of time between data collection and release.

Our shorter release schedule for Census 2021 will ensure that organisations can use the data to inform decision-making and service provision while the information is most relevant. We’ve enabled this through:

  • moving away from a primarily paper-based data collection
  • making outputs production more efficient
  • automating disclosure control checks
     

For the first time, Census 2021 was primarily an online census, making it more convenient to complete for most of the population. To ensure that the census was inclusive for all, we offered a full range of support services. This included providing paper questionnaires where required. The contact centre provided guidance and help over the telephone and through digital services such as web chat and social media. We also provided census questionnaires in Braille and British Sign Language.

When we process the Census 2021 data, we will use our newly built Data Access Platform. We’ve designed this platform to meet the needs of a national statistics institute. New technologies will support the initial processing of responses in real time, allowing us to carry out analysis and adapt the data collection, if needed. We can then apply the approach we’re developing to our other data collection exercises. This has the potential to improve the statistical quality and timeliness of outputs.

Statistical disclosure control procedures will include:

  • modifying some of the data before we release the statistics, using methods such as record swapping, some form of cell perturbation or suppression
  • limiting outputs by simplifying or grouping data - for example, providing information for aggregated age groups
  • amalgamating information to ensure that data are only available in a non-disclosive way - for example, by grouping categories of data or only providing information for larger geographical areas
     

More information on data processing and statistical disclosure control is available in the Census 2021 White Paper.

We’ve also sought to reduce the time it takes us to release data for lower-layer geographies and certain topics. This is in response to feedback that highlighted this as an area for improvement following the 2011 Census. The aim is to provide data down to Output Areas (OAs) in the first phase of the release schedule when we publish topic summaries. OAs are the smallest geography we will make the standard outputs available for. These are generally around 100 households in size.

Within this first phase, we’re proposing a staggered approach to the release of data and analysis by topic. Users will have the full set of detail, from both a geographical and topic perspective, very early on in the schedule. This is down to the reduced time span for the publication of these initial releases. We will combine this data to produce area profiles.

In each release phase, the products increase in complexity. This means we will be releasing each product as early as feasible. Each phase is discussed in more detail later in this section.

In the second phase of the release schedule, we will publish multivariate data. These outputs combine variables to enable us to fully explore and understand the relationships between them. We aim to produce outputs all the way down to the smallest geographies as part of these releases.

In the third phase of the release schedule, we will publish the additional products, including data for alternative and small populations, origin-destination datasets, and microdata.

In each phase, we will publish statistical commentary or analysis alongside the data to support and explain them.

Figure 1 shows how these proposals come together to form the “shape” of the Census 2021 outputs and analysis release schedule. The exact timing of the release of outputs will depend upon the processing and quality assurance of census data and will be announced closer to the time of release.

Figure 1: Census 2021 outputs release schedule

Geography

Geographies included in releases

In late 2020, we carried out a separate consultation on Census 2021 Output Geography Policy, products and services. Earlier this year, we also published our formal response.

The geographies we’re using to support Census 2021 data is largely based on the 2011 approach. More information about that approach is available in our overview of the various geographies used to support the production of 2011 Census statistics. This includes definitions and descriptions of OAs, our smallest statistical geography.

We plan to continue using OAs as the main geographical base for Census 2021 outputs. We will produce outputs for most supported geographies by aggregating OAs, on a best-fit basis. The method we used to decide the best-fit is described in the Geography Policy for the Government Statistical Service.

We will review OAs based on Census 2021 population and household data. We will then merge or split them as needed to ensure they remain within the established population and household size thresholds. We expect around a 5% overall change between OAs compared to 2011, including some targeted re-alignment of OAs to better fit ward and parish boundaries.

We will base the main outputs for parishes on OAs. However, there will be some targeted re-alignment of OAs to better meet current ward and parish boundaries. We acknowledge that some parishes are smaller than OAs, and we’re exploring how we could produce a summary set of information for these smaller areas.

Data for National Parks will use an exact geography in line with 2011. We also intend to produce simple population counts for postcodes, as we did following the 2011 Census.

We’re aiming to make census outputs for key geographies available early in the release schedule, following the initial releases of Census 2021 data. This contrasts with the 2011 approach, which released geographical detail and type in stages. This is an ambitious aim, and we will make final decisions on this prior to publishing a final release schedule in February 2022.

If we do need to produce outputs for some of the key geographies later than others, we propose to prioritise estimates for administrative and statistical geographies down to OA level. This will include, for example, Local Authorities, Lower Layer Super Output Areas (LSOA) and Middle Layer Super Output Areas (MSOA). Following these releases, we will produce estimates for additional geographical areas, such as health areas and Parliamentary Constituencies.

Boundaries

All Census 2021 outputs will reflect the geographies in place in May 2022, rather than as of Census Day, 21 March 2021. As a result, we expect that census outputs will reflect any ward changes that become effective for the planned elections in May 2022. This is consistent with the approach we took in the 2011 Census outputs.

To prevent disclosure of confidential information about people and workplaces, we will build upon the approach of the 2011 Census. This means that we will provide users with look-up tables that map each OA to a range of other geographies, on a best-fit basis. This will create best approximations of data for new geographical areas, such as changed wards.

For Census 2021, the guidance will suggest using higher levels of geography to build other geographies where possible, rather than adding all individual OAs together. This is because the small cell key perturbation method that we will apply to Census 2021 data has less impact on larger areas. These areas have larger counts and are likely to be less disclosive. This will minimise the slight differences in totals that statistical disclosure control creates. This extra layer of protection is needed to allow the flexibility for users to create their own tables.

We recognise that some geographical boundaries may change over time. Although the overall rate of change for geographies affected is relatively low, it can have a big impact in specific areas. Where there are significant boundary changes following the release of census outputs, we will consider providing some revised outputs as ready-made tables if users demonstrate sufficient need. For example, we may do this if there are significant changes made following the Boundary Commission for England’s (BCE) 2023 review of electoral boundaries. The BCE is due to publish its final report in June 2023.

Users may also wish to request commissioned tables for these new areas, if preferred. We will assess these on a case-by-case basis and charges for the cost of their production will apply.

  •  

We’d like to know your views on the proposed approach for providing outputs to reflect boundary changes made after May 2022. In addition, we’d like to know if you have a need for data produced for different geographic boundaries.

Phase 1: Initial findings and topic summaries

Initial findings

We aim to release initial census findings within 12 months of the census. These are likely to be rounded national population estimates for England and Wales.

We plan for these to be followed as soon as possible by more detailed unrounded population and household estimates. These estimates will be broken down by age and sex, for England and Wales at local authority level. We will separately produce mid-year population estimates (MYEs) for the UK. These will be broken down by local authority, sex, and age.

Topic summaries

We then plan to move into a staggered release of topic summaries. We intend to provide these largely as univariate tables. This means that they only include data from a single variable, for example, religion or general health. These are similar in concept to the Quick Statistics from 2011. We will be able to make the tables produced on a single variable available at the most detailed level of the classification.

During this phase, we aim to produce these tables down to OA level and for all geographies defined using the OA best-fit approach. If this is not possible because of operational issues, we will release this data as soon as possible after this phase.

We’re proposing to cover topic summaries, in the order provided, for the following areas:

  1. demography and migration
  2. ethnic group, national identity, language and religion
  3. sexual orientation and gender identity
  4. health, disability and unpaid care
  5. housing
  6. labour market and travel to work
  7. education
  8. armed forces veterans


All tables in this phase will be populated based on the usually resident population of England and Wales. Tables will not include totals or subtotals.

Commentary, which will draw out insights from the data and provide supporting contextual information, will accompany each of these releases. This information could provide context to any significant changes in those topics since 21 March 2021, when people gave their answers to the census questions. However, it’s too soon to know what additional contextual information we could provide this early in the release schedule.

We're currently exploring the products that we could use to provide the data and commentary, and we're conducting user research to inform their design. Those products are likely to include interactive data visualisations, which present the data in innovative and accessible ways for different types of user. If you’d like to be involved in the development of these products, please email us at census.outputs@ons.gov.uk for more information.

We’ve provided the proposed table specifications for the phase one releases in the “Topic summaries” tab in Draft proposals for outputs data content (XLS, 610KB).

We’d like to know if the recommended ordering of the topic summaries and the proposed table specifications meet your needs.

Area profiles

We will use the data from the topic summaries to create dynamic area profiles. These will pull together selected census information across a wide range of topics for a specific area. They will be available for geographies where we’ve published data. Area profiles will allow users to compare the local and national picture and will provide similar data to those contained in the Key Statistics in 2011.

A range of geographies will be available. We’re aiming to make these area profiles available for a range of geographical levels from the whole of England and Wales right down to OAs from the outset. If this is not possible because of operational issues, we will add additional levels of geography after they become available.

These area profiles are similar in concept to those available in Nomis Area Reports.

We’ve provided the proposed area profile specifications in the “Area profiles” tab in Draft proposals for outputs data content (XLS, 610KB).

We’d like to know if the area profile specifications meet your data user needs.


We will base the main outputs for parishes, or communities in Wales, on OAs. In some cases, parishes are smaller than OAs. A result, we cannot produce them using the OA best-fit policy. To provide information for these specific small parishes, we will consider producing an additional set of parish and community profiles alongside the main outputs. These would consist of a summary set of information for these smaller areas. We would produce these for all parishes and communities with a population of at least 50 people, similar to those produced from the 2001 Census. We could also provide additional, very basic information on the number of residents and households for any parishes below that threshold.

We’d like to know if you need an additional set of profiles to give summary data for all parishes and communities.

Phase 2: Multivariate data for the usual resident population base

Multivariate data

In this phase of the release schedule, we will publish the richer, multivariate data. Data users can use these outputs to combine variables and, through this, can fully explore and understand relationships between them.

We aim to produce these outputs, in the form of tables, down to OA level. During this phase, we will do this for all geographies defined using the Output Area best-fit approach.

In 2011, we provided this in two main sets of tables that were designed for different levels of geography. Generally, we made the Local Characteristics (LC) tables available for OAs and above, but these tables contained less detailed classifications of the characteristics they covered. Similarly, we generally made the Detailed Characteristics (DC) tables available for MSOAs and above. These tables contained more detailed classifications of the characteristics they covered.

The model for 2021 will be somewhat different than for 2011. We’re introducing functionality for users to build their own tables, alongside the availability of a set of prescribed ready-made tables.

This functionality gives users flexibility in three main ways, which include:

  • letting them choose whether to select a ready-made table or to build their own
  • putting them in control of what census variables to combine when they’re building their own tables
  • giving them the ability to get more detail for smaller areas than was possible in 2011 where that does not risk the confidentiality of respondents
     

All tables in this phase will be populated based on the usual resident population of England and Wales.

Tables will not include totals or subtotals. We will apply the small cell key perturbation method to Census 2021 data as part of the disclosure control methodology. Data users should note that this may lead to slightly different estimates depending on how the tables were defined.

Ready-made tables

In general, we will make ready-made tables in phase two available down to OAs. They will be available for all areas within the geographic level they’re released for. This means that users can be sure that, when they request a table, we will provide it for all requested areas.

Many of the ready-made tables we’re proposing are similar and comparable to the LC tables published following the 2011 Census. We are also proposing some additional tables, subject to assessment of the statistical disclosure risk of the data.

Firstly, we’re proposing additional tables on multi-language and multi-religion households. In each case, we’re proposing three new tables on:

  • tenure
  • ethnic group of the household reference person
  • occupation of the household reference person


The household reference person is an individual within a household who acts as a reference point for producing further derived statistics. They can also characterise a whole household according to the characteristics of the chosen reference person. The 2011 Census glossary explains how we choose the household reference person for a range of household types. We’ve included these tables to meet the need for more information on cultural diversity within households.

Secondly, we’re proposing tables for the new questions on sexual orientation, gender identity and armed forces veterans.

The “Ready made tables” tab in the Draft proposals for outputs data content (XLS, 610KB) contains the proposed specifications for all the ready-made tables. These specifications include the geographical levels the data will be available for.

We’d like to know if the ready-made tables’ specifications meet your needs.
 

Build-your-own tables using flexible datasets

Being able to produce build-your-own tables from flexible datasets means that data users have the chance to combine variables in new ways. The build-your-own tables system takes users’ requests, creates the outputs and applies disclosure control methods on the data. Finally, the system runs checks to determine whether the resulting outputs are potentially disclosive and, therefore, whether it can provide the data to the user or not.

This means, for any specific request of data, the result may be that the system can provide the data for some but not all areas requested. This is more likely to occur where users have:

  • combined multiple variables
  • requested the most detailed classifications
  • requested smaller geographical areas

 

The creation of data using this capability will be an iterative process for users. It’s expected that users will experiment with what the system can provide for each area. As such, the specifications for the build-your-own tables, which we will make available for users to choose from, are an indicative set of variables and classifications. It does not mean that users will be able to obtain the data that they request.

As a guide, if one aspect of a table increases in complexity, another aspect is likely to need to decrease in complexity. This is similar to the relationship between the 2011 Census LC and DC tables.

The “Build your own tables” tab of the Draft proposals for outputs data content (XLS, 610KB) contains the proposed specifications for the flexible datasets.

We’d like to know if the build-your-own tables from flexible datasets specifications meet your needs.
 

Census analysis

We will publish commentary, interpretation and more detailed analysis of the new census data on the ONS website. These will complement new data releases and interactive data visualisations will accompany them. This will provide users with impartial and insightful commentary to better understand our population and their local community needs. It will also inform public policy and public service delivery.

In line with the data releases, census analysis will develop in complexity as we progress through the programme and new census datasets become available. As the level of detail in the published data increases, the depth of the analyses will increase, and the insights will become greater.

In earlier stages, we plan to provide analysis for the main population bases, including usual residents, households and dwellings. In later stages, the plans will also include analysis using alternative population bases, such as the workday and out-of-term populations. These population bases are discussed in detail in “Section 5: Population-base specifications”.

Planned analyses in phase two aim to meet many of the analytical commitments made in the Census 2021 White Paper in December 2018 that do not require linkage with other data sources. These analysis proposals are subject to review following user feedback and viability testing.

These analysis plans are discussed in more detail in “Section 3: Main changes to variables compared to the 2011 Census”. That section also includes questions on the extent to which these plans meet user needs. It’s too early to specify exactly when we will release different analysis products, so all analysis for phases one to three are discussed together.

We’d also like to know your needs for analysis on all census topics.

Phase 3: Alternative population bases, origin-destination data and microdata

Prior to phase three, all data that we release will be based on the usually resident population. In the third phase of the release schedule, we will produce other data products.

This includes data for alternative and small populations and origin-destination flow datasets looking at migration and commuting patterns. We will also make microdata samples available for research purposes and continue to provide analysis about the published data.

This period will also include the introduction of the commissioned table service for Census 2021. Where it’s not possible for users to create the outputs they need themselves, users can commission us to produce and supply them. We will do this if the requested data passes statistical disclosure checks. There is normally a charge for this service.

Alternative population bases

Alternative population bases allocate those counted in the census to different geographical locations where individuals could be counted. Examples of alternative population bases include:

  • workplace
  • workday
  • out-of-term
  • short-term resident
     

We discuss these products, and the impact of the coronavirus pandemic on the size of these populations, in “Alternative population bases” within “Section 5: Population-base specifications”. We’ve included questions on the extent to which proposals for alternative populations meet user needs.

Small populations

Small population tables aim to provide information on identities about which users have specific needs. This may be to better understand that small population or analyse potential inequalities between that small population and the wider population.

These tables will be bespoke outputs, where the size of the population in that group means confidentiality constraints will limit the release of detail on the population in standard outputs. The tables will only include members of the chosen population, not the whole usually resident population. Examples of proposed small populations include Sikh, Cornish and Jain. This is in line with the approach taken for small populations in 2011.

One of the important aspects of these outputs is that we produce them only when the number of people in that small population meets an agreed threshold. This minimises the risk of disclosing personal information, and often suits these small populations quite well as they’re usually clustered geographically. We discuss these products in “Small populations” within “Section 5: Population-base specifications”. In that section, we’ve included questions on the extent to which proposals for alternative populations data meet user needs.

Origin-destination data

Origin-destination data are sometimes known as “flow” data. These data describe the movement of people from one location to another. In line with the 2011 Census outputs, we plan to release four different categories of origin-destination data in a combination of data visualisation products and data tables.

  • Migration flow data: national and international migration during the year prior to Census Day.
  • Workplace flow data: travel to work in the week before Census Day and method of transport.
  • Second address flow data: location of second address in relation to an individual’s usual residence or workplace.
  • Student flow data: migration patterns of individuals living at student addresses one year prior to Census Day.

 

As per our approach in 2011, we will classify the origin-destination data in three ways: public, safeguarded and secure.

As a result of the impact of the coronavirus pandemic on population flows, public and safeguarded origin-destination data may contain less detailed characteristics than in 2011. This would be necessary to protect confidentiality. We mention impacts of the coronavirus pandemic on products in “Section 6: Taking a census during a period of change”.

We will place the most detailed origin-destination data, classified as secure, in our Secure Research Service (SRS). We will make this available to approved or accredited researchers. We’re still considering the most appropriate means of making the safeguarded data available. However, the UK Data Service is likely to hold these. This is in line with previous censuses.

As a result of their complexity, we may produce these data after the public data tables. We will include more information in the full outputs release schedule, which we’re planning to publish in early 2022. A fuller understanding of the scope and likely timescales will inform that schedule.

Microdata

Microdata products are samples of anonymised records for individuals and households, which include a selection of associated census characteristics.

To make data available as widely as possible and to maximise benefits from the census, we plan to classify microdata in three ways: public, safeguarded and secure. This is the same approach we’re using for origin-destination data. This approach strikes a balance between protecting confidentiality and ensuring data can be available for a range of users, from inquiring citizens through to expert analysts.

In 2011, we made a “Microdata Teaching File” available for public access. This contained anonymised records on a limited set of variables for a random sample of 1% of people in the 2011 Census output database. It’s freely available for anyone to download under the terms of the Open Government Licence.

We intend to produce a similar, downloadable public access teaching file for Census 2021. This aims to meet the needs of users for an accessible, non-disclosive microdata file to improve the use of census microdata. It will also provide an educational tool of real data to assist with the teaching of statistics and social sciences.

The secure microdata samples have the highest level of detail and the largest sample sizes. We will also store them in our SRS. As a result, they will be available only to approved or accredited researchers to protect the confidentiality of personal information. We’re still considering the most appropriate means of making the safeguarded data available, but the UK Data Service is likely to hold them.

Following user feedback, we also plan to produce a microdata file that will contribute to the University of Minnesota’s Integrated Public-Use Microdata Series (IPUMS) project. This is an international project that brings census microdata together from over 100 countries.

The proposed safeguarded and secure Census 2021 microdata products are described in the next subsections.

 Public-access file

•            Sample size: up to 1% of individuals

•            Statistical unit: persons

•            Lowest geography: region

Safeguarded individual region file

•            Sample size: up to 5% of individuals

•            Statistical unit: persons

•            Lowest geography: region

Safeguarded individual grouped LA file

•            Sample size: up to 5% of individuals

•            Statistical unit: persons

•            Lowest geography: grouped local authority

Safeguarded household file

•            Sample size: up to 1% of households

•            Statistical unit: households

•            Lowest geography: region

IPUMS file

•            Sample size: up to 1% of households

•            Statistical unit: households

•            Lowest geography: region

Secure individual file

•            Sample size: up to 10% of individuals

•            Statistical unit: persons

•            Lowest geography: local authority

Secure household file

•            Sample size: up to 10% of households

•            Statistical unit: households

•            Lowest geography: local authority

As a result of the complexity of these products, we may produce these data after the public data tables. We will include more information, based on a fuller understanding of the scope and likely timescales, in the full outputs release schedule. We’re planning to publish this schedule in early 2022.

Census analysis

During phase three, we propose to continue producing analysis of the multivariate tables. We’re also proposing to include detailed analytical topic and policy reports on small population and alternative population bases in our releases.

We will publish these reports on the ONS website. They will aim to provide users with the most relevant and insightful commentary for their communities. They will also aim to help users better understand their demographic make-up and inform policy and service delivery.

Analysis proposals are detailed in the supplementary document Census 2021 analysis programme proposals (PDF, 504.8KB)These analysis proposals are subject to review following user feedback and viability testing.

Beyond March 2023: UK data and further analysis

UK data

In addition to carrying out the census and producing outputs for England and Wales, we are responsible for collating data from across the four nations of the UK to produce UK outputs. These outputs provide totals for the UK, as well as comparable data for small areas across the UK.

Historically, the census has been taken on the same day in all four nations of the UK. We’d then combine data from each census, where the questions were harmonised, to produce UK statistics. We then used these data as the baseline for other UK-level statistics we produce.

However, each nation has its own operational, user and respondent needs. The coronavirus pandemic led to each country reviewing whether to conduct their census in March 2021. Following this review, we concluded that we should conduct the census on 21 March 2021 as planned. The Northern Ireland Statistics and Research Agency (NISRA) came to the same conclusion for Northern Ireland. This date allows us to use Census 2021 data as a baseline from which to monitor recovery from the pandemic.

The Scottish Government’s review concluded that, because of the impact of the coronavirus pandemic, they should move Scotland’s census to March 2022, to secure high-quality outputs. This means that there will be a one-year difference in reference dates and that the first results from Scotland will not be published before March 2023. 

All the UK census offices are working closely together to understand how this difference in reference dates will impact UK-wide population and housing statistics. This is both in terms of the timing and scope.

Every year, we also produce mid-year population estimates (MYEs) for the UK, broken down by local authority, sex and age. Every 10 years, these mid-year population estimates are rebased using new census data. We will deliver UK population estimates for mid-2021, based on Census 2021 in England, Wales and Northern Ireland, and on the 2011 Census in Scotland. In the following year, the Scotland mid-year population estimates will also be rebased using Scotland’s 2022 Census data.

While we’re confident that we will be able to produce some UK outputs, it will be more difficult in other areas. These include the four areas outlined here.

Workplace populations an out-of-term populations

There are likely to be some minor impacts on England and Wales outputs for workplace population and out-of-term populations. In both cases, flows into areas from Scotland will be missing.  

UK origin-destination outputs

We’ve explored the options for producing UK origin-destination outputs. We’ve concluded that it’s unlikely that we can combine the data in a meaningful way to produce a single reliable set of UK origin-destination outputs. As a result, we will make origin-destination outputs for 2021 available for England, Wales and Northern Ireland combined. The National Records of Scotland will publish data for Scotland separately.

Second address flow data

Second address flow data will continue to be available for England and Wales only, as Scotland and Northern Ireland do not collect this information. 

Areas with a difference in reference dates


There are other examples where there’s a potential for a high rate of change between reference dates, including health and labour-market data.

To ensure the UK-wide population statistics we publish meet data needs and are of a high enough quality, we’re engaging with UK census data users through our working groups. We’re considering factors such as the level of harmonisation between nations and the impacts of the difference in reference dates.

For example, we’ve been exploring the need for UK population estimates as at March 2021 or March 2022. We’ve progressed this work outside of this consultation, as a result of the short timescales. So far, we’ve not identified a need for these products in addition to the mid-year population estimates. Please contact census.outputs@ons.gov.uk if you want to be involved in this process.

The full Census 2021 Outputs Prospectus will include more detail on our plans for UK data outputs. We plan to publish this by February 2022. We will provide guidance to users on comparing data from different nations. We’re also considering if and how to present data for England, Wales and Northern Ireland together.

We’d like to know if you need data for England, Wales and Northern Ireland to be published alongside each other, irrespective of the feasibility of producing UK outputs.

Census analysis

Census analysis after March 2023 will generally focus on more complex projects that require linking Census 2021 data with other data sources. Other sources include previous censuses, the ONS Longitudinal Study, and other government surveys and administrative sources. Through this we can undertake new, innovative and deeper analysis of census data to inform policy and service delivery.

These analytical projects will complete any commitments made in the Census 2021 White Paper in December 2018 that we’ve not met in previous phases.

For example, we acknowledge the great interest in income data and the value that this can provide in planning, policy development and evaluation. We’ve been working with tax and benefits data from the Department for Work and Pensions (DWP) and Her Majesty’s Revenue and Customs (HMRC) to develop small-area income data that can be linked with the data collected in Census 2021. We’ve already produced some research outputs, alongside other research on Population Characteristics to demonstrate the potential of this approach.

We’re continuing to develop these research outputs and published our latest set of experimental statistics on small area income distributions in June 2021. These statistics were the first to incorporate self-employment income. We currently plan to publish a further update in 2022. We’re still working towards the aim of directly linking this data to census data in the future.

We include details of our analysis proposals in the supplementary document Census 2021 analysis programme proposals (PDF, 504.8KB). We will continue to develop and review these plans in light of evolving user needs, particularly with respect to the pandemic and feasibility testing. User engagement, including the response to this consultation, will steer the development of these plans. We discuss analysis planned for beyond March 2023 separately from the phase one to three analysis. This is because it generally requires linkage with other data sources, including from Scotland’s census.

As we progress the transformation of the UK population and statistics system, we’re working to put administrative data at the core of our population, migration and social statistics systems. Our aspiration is to be as radical, ambitious and inclusive as possible to improve the quality, frequency, representativeness, relevance, coherence, accessibility and timeliness of these core statistics.