Census 2021 in Wales

The census in Wales and the census in Welsh 

The census of population and housing has been at the heart of population statistics for England and Wales for over two centuries, providing a detailed snapshot every 10 years. The statistics from each census have been the most detailed available on the population, its characteristics, and where and how we all live. These statistics are widely used by government, local authorities, charities, private organisations and the public. 

Census 2021 in Wales was carried out by the Office for National Statistics (ONS) working in partnership with the Welsh Government, who were also important partners throughout the process of planning and delivery. This included legislation (described in this section), communications and engagement, quality assurance and publishing outputs for Wales and relating to the Welsh language. 

In the Census 2021 white paper (published in December 2018 in English and Welsh), we set out our commitment to carrying out the census in Welsh as well as English in Wales. This included providing English and Welsh-language versions of both the online and paper questionnaires and all other important communications to the public and ensuring that the questionnaires could be completed in Welsh as easily as in English. 

We also promised a dedicated Welsh telephone helpline, Welsh language online help facilities, the recruitment of Welsh speakers for the field force, and appropriate arrangements for public engagement activities in Wales (see paragraphs 4.55 to 4.57 of the Census 2021 white paper). This article describes how that bilingual census was delivered. 

Following Census 2021, the ONS has been improving its Welsh language provision more generally, including updating the UK Statistics Authority's Welsh Language Scheme. Lessons and examples from Census 2021 have contributed to this improvement. Improving our wider Welsh language provision will be of increased importance as we move towards a population and migration statistics system that makes use of a greater range of data sources. 

As described in the relevant sections of this article, Census 2021 exceeded its response rate targets and most of its quality targets and delivered more timely outputs than ever before. The first results were released a year after the data-collection operation and all major England and Wales releases were published by the end of 2023. 

This article provides an overview of the delivery of Census 2021 in Wales. It accompanies the general report for Census 2021 in England and Wales, which provides more detail on the planning and delivery of the census. 

Following the conclusion of our Beyond 2011 programme, we delivered Census 2021 as part of a wider programme that also carried out work on meeting user needs through administrative data and other sources. Following this work alongside and after Census 2021, the UK Statistics Authority has published a recommendation on the future of population and migration statistics in England and Wales. 

Legal background and Census 2021 legislation 

The statutory authority for taking a census in England and Wales is the Census Act 1920. The Act empowers the UK Statistics Authority to undertake a census in any year that is at least five years from the year of the previous census. The ONS is the executive arm of the Authority and delivered Census 2021 in England and Wales. 

For Census 2021, new questions on sexual orientation and gender identity were made voluntary by amending the 1920 Act through a Bill in the UK Parliament. A legislative consent motion for the amending Bill was passed in the Senedd (the Welsh Parliament) in July 2019, and the Census (Return Particulars and Removal of Penalties) Act 2019 gained Royal Assent in October 2019. 

Secondary legislation is required for each census in the form of an Order in Council and regulations. The Census (England and Wales) Order 2020 was laid before the UK Parliament by UK Government ministers, following consultation with the Welsh Government. It became law on 20 May 2020, and came into force the following day. Census regulations were then made in parallel for England by UK Government ministers and for Wales by Welsh Government ministers. 

The ONS worked closely with Welsh Government officials and lawyers to ensure that the regulations for Wales delivered the same effect as those for England. The Census (Wales) Regulations 2020 were made by the Minister for Finance and Trefnydd on 28 May 2020, laid before the Senedd on 1 June 2020, and came into force on 26 June 2020.

Planning, consultation and engagement for Census 2021 began soon after the recommendation to hold a digital-first census was made in 2014. A three-year period of question development from our Topic Consultation in 2015 to the Census 2021 white paper in December 2018 established what data users needed from the census and how we would collect it.

Following this, we largely retained the population bases and definitions used in 2011, and asked most of the same questions, with some small amendments and three new questions (see "Question and questionnaire development" in this section). We also consulted on Census 2021 outputs in 2018 and 2021. Along with stakeholder engagement, this helped to shape the suite of outputs described in the Outputs section.

Among our stakeholder advisory groups was a Census Advisory Group for Wales, chaired by the Chief Statistician for Wales. Like our other Census Advisory Groups, this group ran throughout the census period from planning to outputs. Its terms of reference included requirements for information from Census 2021 and aspects of the census operation, including data-collection, questionnaire design, quality assurance, and the conduct of a bilingual census operation in Wales.

The group included members from Welsh Government, local authorities in Wales, the Welsh Language Commissioner's office and other Welsh organisations.

In addition to participating in our stakeholder groups, Welsh Government officials were also part of the internal planning structure for Census 2021. Involvement ranged from being part of the ONS's Census and Data Collection Transformation Programme Board to regular bilateral meetings and co-operation between ONS and Welsh Government officials throughout the preparation and delivery of the census.

Cross-UK co-operation

The census offices across the UK (the ONS, National Records of Scotland (NRS) and the Northern Ireland Statistics and Research Agency (NISRA)) came together in a range of forums that covered the preparation and delivery of the 2021 and 2022 censuses, many of which included officials from the Welsh Government.

These forums included a senior-level UK Census Committee (UKCC), which was chaired by the National Statistician, and included the Registrars General for Scotland and Northern Ireland, the Chief Statisticians for Scotland and Wales and senior census officials from the ONS, NISRA and NRS.

Reporting to the UKCC, a UK Harmonisation Working Group met monthly representing the three census offices and the Welsh Government. Their role was to manage relationships between census programmes of work across the UK, identify potential areas of joint working and manage risks to harmonisation of UK outputs.

Other working-level harmonisation groups included:

• Outputs and Dissemination Harmonisation Working Group*

• UK Statistical Quality Working Group

• Census Research Assurance Group*

• Data Processing Harmonisation Working Group*

• Edit and Imputation Working Group*

• UK Product Working Group

• Microdata Working Group*

• Origin-Destination Working Group*

• UK Addressing Working Group*

• UK and Ireland Geography Harmonisation Working Group*

• UK Census and Population Statistics Strategic Group*

Groups marked with an asterisk (*) included officials from the Welsh Government, as did a range of UK-wide operational level working groups that met to create and share plans and lessons around the management of the census data-collection operation.

Responding to the coronavirus (COVID-19) pandemic

Every census presents different challenges. For Census 2021, a major challenge came in the form of the coronavirus pandemic and the national lockdown restrictions during 2020 and 2021. The first national lockdown began on 23 March 2020, a year before the census data-collection operation.

Based on the latest academic thinking available at the time, we reviewed how critical functions for a successful census could be delivered in a series of scenarios. A series of "readiness assessment points" were arranged for April, June, September and December 2020, which helped to determine the focus for continuing detailed planning and prioritisation. In October 2020, we published information on the operational impact of the pandemic on the census and how we had adapted our plans.

Following our final "readiness assessment" in December 2020, we were able to announce in January 2021 that the census would go ahead. This was based on our ongoing assessment of our capacity to deliver a census that would provide high-quality data while also being safe for the public and our staff. The census in Northern Ireland also went ahead in 2021 as planned. The Scottish Government, on the recommendation of National Records of Scotland, announced in July 2020 that the census there would be moved to March 2022.

Chapter 1 of the Census 2021 general report outlines the impact of the pandemic on the census. Planning to deliver a census in England and Wales meant that we needed to be flexible enough to adapt our approach where there were differences between the responses of each government to the pandemic. For example, the rules, guidance and timing of the lockdown in place in early 2021 differed in approach between England and Wales.

Question and questionnaire development

A three-year period of question development established what data users needed from the census and how we would collect it. New voluntary questions on sexual orientation and gender identity were added, along with a new question on previous service in the UK armed forces, while the use of administrative data meant that we could drop the "number of rooms" question.

More information can be found in our question development reports. The census secondary legislation in 2020 (see Background and introduction) confirmed the questions and other details of the census.

Developing the census questions in Welsh

As for the 2011 Census, the English and Welsh census questions were developed in parallel. Between 2017 and 2018, an external agency with Welsh speaking researchers was commissioned to undertake focus groups and a series of cognitive interviews. The qualitative research tested public acceptability and comprehension of amended and newly designed census questions in Welsh. The questions were tested with people across Wales with varying dialects and Welsh language proficiencies.

To ensure questions adhered to Cymraeg Clir guidelines, some changes to the text or questions across the census questionnaires were translated by our contracted specialist Welsh language translation service provider. These changes were quality assured by the Welsh Language Census Question Assurance Group, which included Welsh language and policy experts from the Welsh Language Commissioner's office and the Welsh Government. Information about Welsh language development for specific questions is included in our question development reports.

The Welsh language skills question

A question on Welsh language skills has been included in every census in Wales since 1891. Language knowledge and skill covers whether a person can understand, speak, read or write a language, without reference to how well they can do it.

The need for data to monitor and promote the use of the Welsh language has been strengthened because of the Welsh Language (Wales) Measure 2011, which places duties on public bodies in Wales, through Welsh language standards, to not treat the Welsh language less favourably than English. Welsh language standards regulations also require Welsh Ministers and other public bodies to consider the Welsh language and the needs of its speakers when planning their workforce, developing policies, and providing services.

We tested a four-stage question in place of the single, multi-tick question used in 2011, which asked whether respondents could "understand, speak, read or write Welsh". The four-stage question design indicated no impact on respondent burden. However, the responses were inconsistent with the 2011 Census question design, suggesting it would impact data comparability with previous censuses. Therefore, we recommended keeping the 2011 Census question design of a single, multi-tick question.

More information can be found in our National identity, ethnic group, language and religion question development publication. The Outputs section of this article describes our post-census work with the Welsh Government on the topic of Welsh language ability.

Differences between England and Wales questionnaires

The main difference between the Census 2021 questionnaires in England and Wales was the inclusion of the Welsh language skills question in Wales. The other differences were:

• the ordering of response options so that Wales or Welsh appeared first in country of birth, national identity, and ethnic group questions in Wales

• the first tick-box for the main language question was "English or Welsh" in Wales (rather than "English" as in England)

• the inclusion in Wales of "Black Welsh" and "Asian Welsh" in the high-level ethnic group category descriptions in Wales (described in more detail later in this section)

• the "Christian" option in the religion question in Wales read "Christian (All denominations)", while in England it provided examples of Christian denominations

• differences in the qualifications and apprenticeship question response options (including references to the Welsh Baccalaureate)

• different examples in the "how to complete your questionnaire" section and in the relationship matrix

Following a request by the Welsh Government to review the wording of the ethnic group question to make it as inclusive as possible for the non-White Welsh population, we tested four versions of two high-level category descriptions at the first stage of the question.

The tested versions were:

• the 2011 question ("Asian or Asian British" and "Black, Black British, Caribbean or African")

• two versions incorporating "Asian Welsh" and "Black Welsh"

• a version that included English, Scottish and Northern Irish alongside Welsh

Following this testing, we recommended using "Asian, Asian Welsh or Asian British" and "Black, Black Welsh, Black British, Caribbean or African" in the high-level descriptions in the census in Wales. This change was reflected in the questionnaires and in our outputs.

For more information, see our National identity, ethnic group, language and religion question development publication.

A high-level comparison of the census questions used in England, Wales, Scotland and Northern Ireland has been published alongside the Census 2021 general report.

Large-scale tests: 2017 and 2019

The 2017 Census Test was designed to provide evidence to show what approaches in a digital-first census worked best for different population groups and to make sure our systems and services worked correctly. We also tested the impact of new questions on response rates.

We invited 100,000 households across seven local authorities in England and Wales to fill in a test questionnaire online. Powys was one of the seven local authorities. Like those in England, it was chosen for particular characteristics, which included its rural location and the proportion of Welsh language speakers.

We also sampled 100,000 randomly selected households across England and Wales, as well as 8,000 households on the Isle of Wight. This was to test different approaches, including paper-first or online-first with initial contact letters and different initiatives to maximise response (including the design of initial contact and reminder letters). In Wales, these approaches were tested in English and in Welsh.

In Autumn 2019, we undertook a data-collection rehearsal. This showed that the systems tested worked and integrated end-to-end in the way we expected. The rehearsal covered 331,359 households across four local authority areas. Ceredigion was included because of its high proportion of Welsh language speakers and large rural population. It also included a student population, a range of second homes, and some areas with digitally excluded populations.

The rehearsal enabled us to test our systems and processes for Census 2021, including:

• the quality of our preparations

• the main processes and systems for the operations for carrying out the census for households and some communal establishments

• our community engagement and communications strategies

• the Census Coverage Survey (CCS)

As for the 2017 Test, the data-collection rehearsal was conducted using both English and Welsh language materials in Wales. Reflecting the online-first and paper-first strategy used in 2021, Ceredigion had a substantially higher proportion of paper-first contact for the rehearsal and correspondingly a lower online completion rate, compared with the local authority areas in England.

Wave of contact 

Our "wave of contact" approach had a succession of timed despatches of letters and paper questionnaires before and after Census Day (21 March 2021). This encompassed our initial, pre-Census Day contacts, and the reminder and follow-up process, with field staff visiting non-responding households. The Communal Establishments and Special Population Groups operation had a different, phased approach that ran in parallel to the household operation.

Schedule for contact with households for Census 2021 

• 22 February 2021: Postcards to 25.6 million households, including 591,000 in Wales. 

• From 3 March: Initial contact letters or paper questionnaires to all households. 

• 8 March: Tranche 1 of household field force deployed. 

• 21 March: Census Day. 

• From 22 March: First despatch of reminder letters (digital-first households only). 

• 23 March: Tranche 2 of household field force deployed. 

• From 29 March: Second despatch of reminder letters (including first letters to paper-first households, and from 31 March where online responses started but not completed). 

• 30 March: Tranche 3 of household field force deployed (all of field force now deployed). 

• From 12 April: Further reminder letters. 

• From 19 April: Response-driven reminder letters. 

Our flexible approach meant that decisions on the deployment of field staff and reminders after Census Day were based on the latest management information (MI) and intelligence gathered from the field force and other parts of the operation.

Delivery of materials to households 

From 3 to 12 March 2021, all households in England and Wales received either an Initial Contact Letter (ICL) with a unique code to access the census online or a paper questionnaire (PQ). PQs also included an access code, so any household could complete the census online irrespective of the type of initial contact from us. Initial contact letters sent to addresses in Wales were also bilingual, and paper-first addresses received PQs in English and Welsh.

While in 2011 all households had been sent paper questionnaires with online access codes, being digital-first in 2021 meant that 89% of households in England and Wales (24 million) received the ICL only and 11% received the PQ (3 million). The proportion of paper-first areas was higher in Wales, with just over 50% of households receiving the paper questionnaire (748,600 compared with 746,400 digital-first) compared with just under 9% in England. 

The areas that were sent paper questionnaires as initial contact were those where we assessed that households were likely to respond to the census without follow-up but may have had difficulty in doing so online. These were areas where households responded before our follow-up period in the 2011 Census but might not have done so without paper questionnaires in 2021. The decision about which areas would get paper questionnaires was not therefore based just on estimated levels of digital exclusion in an area, but importantly also considered how likely households in an area were to respond prior to follow-up. More information can be found in our publication Designing a digital-first census.

The creation of an effective address frame was essential for the Census 2021 collection operation and in the production of high-quality statistics. The address frame was used to send ICLs or PQs to all residential addresses. While the household address frame was of good quality, addressing was one part of the bilingual delivery of Census 2021 in Wales that we were not able to deliver to the standard we would wish.

While the materials sent to households and communal establishment (CE) residents were bilingual (a bilingual ICL or both English and Welsh language PQs), we were not able to use Welsh-language addresses. Although a requirement for Welsh-language addresses was identified during the planning process, it was not incorporated into our plans early enough to implement a change for printed materials or the eQ. This provides a reminder of the need to identify and incorporate such requirements from the initial design phase to explore, test and implement them in good time. 

Website and electronic questionnaire 

During the census data-collection period, the Census 2021 website served as the host for the secure electronic questionnaire (eQ). The website also: 

• provided a wide range of information to support citizens with questions about how to complete the census 

• enabled people to request PQ or online access codes 

• helped people to locate their nearest census support centre (see "Assisted digital support" later in this section) 

• hosted a comprehensive set of materials for local authorities, community groups and others to use to explain and promote Census 2021 

The website was bilingual, with a Welsh language address (cyfrifiad.gov.uk) taking people to the Welsh-language version. It also had a toggle on the page enabling users to switch between the Welsh and English language versions. The access code sent to addresses in Wales automatically launched the version of the eQ for Wales, which was also bilingual with the option to toggle between English and Welsh. The website also hosted support pages including online help (see "Online help" later in this section). 

Requests for new questionnaires and access codes 

People could obtain new or replacement online access codes or PQs through the Census 2021 website or freephone public contact centre, as well as from field staff. People could also request an individual code or PQ to complete separately from their household. During the busiest period for the contact centre, we set up an additional phone number for PQ requests, which reduced the pressure on the main lines.

In Wales, people requesting an access code by text message (SMS) or a new PQ could specify whether it should be in Welsh or English. Like the ICLs, access code letters were bilingual. Table 1 shows that most requests where language were specified were for English-language materials.

The field operation in Wales 

The Census 2021 field force was organised into 15 operational regions, one of which was Wales. Within Wales, there were nine operational areas for the household campaign, each with a Census Area Manager and Census Area Support. We planned to recruit 116 teams to work in Wales, each with up to 12 census officers led by a team leader. Mobile teams were also recruited, to be deployed in different areas of England and Wales where needed most, determined each week. 

A separate field operation was responsible for the enumeration of communal establishments (CEs) and special population groups (SPGs). CEs are places providing managed residential accommodation, such as students halls of residence, prisons and hostels. Unlike households, CE residents received individual PQs or ICLs.

SPGs were people in residential household accommodation that required additional or specialised intervention or training, such as embassies, caravan parks and those with no fixed address (for example, boaters, continuous cruisers, fairs and circuses, and rough sleepers).

The residents of these accommodation types are often more difficult to access and enumerate. This field force had a similar structure to that of the household operation. 

Recruitment and Welsh-speaking cadre 

The Census 2021 field force numbered over 25,500 across England and Wales, the largest part of which was the household field force numbering 19,800. Recruiting the field force was challenging, resulting in a smaller field force than we had planned for, representing around 70% of the total and 67% of the household roles intended.

The difficulties in recruiting were particularly pronounced in Wales and in appointing Welsh speakers, with the 1,083 total appointed being 58% of our target and the 149 appointed to Welsh speaking roles 41% of our target figure. The high early levels of response and the flexibility built into the design of the field force meant that we were able to deliver what was needed with the field force we had. However, the shortfall (including among Welsh speaking roles) meant that this work was harder, and the risk was greater than if the full planned-for field force had been in place. 

A range of factors may have exacerbated the difficulties of recruiting in Wales; for example, the higher level of pandemic restrictions in place in Wales may have reduced potential applicants' willingness to apply for these public-facing roles. 

For the Welsh-speaking roles, feedback from candidates suggested that applicants were discouraged by the level of Welsh-language skills required, which were higher than was necessary to carry out the role. This meant that some people with sufficient Welsh-language ability applied for roles without Welsh language requirements. However, our recruitment supplier was able to redirect some of these applicants into the Welsh-speaking roles. In some areas, high Welsh-language first requirements led to significant under-recruitment, with applicants seeking roles without language requirements outside of their immediate location. 

The flexible design of the Census 2021 field force meant that mobile field force teams (who were appointed to cover England and Wales) were able to be deployed to assist where response rates were lower than expected, including in parts of Wales. We were also able to convert census field staff in high-response areas into mobile teams for deployment elsewhere. 

Following up non-responding households 

Households that had not responded to the census after Census Day were sent reminder letters and visited by field staff. Of the 11.3 million reminder letters sent in the England and Wales operation, 591,000 were sent to addresses in Wales. Of these: 

• 202,000 were sent in the first batch (from 23 March) 

• 301,000 were sent in the second batch (from 29 March) 

• 42,000 were sent in mid-April 

• 47,000 were sent in late April and May 

As with the initial contact letters, the reminder letters were bilingual.

Visits to non-responding households began after Census Day. With 76% of households in England and Wales responding by 23 March, well above the pre-campaign 70% target for households responding without any follow-up activity, the field force was able to target its resources on a smaller number of properties.

The return rate for Wales was slightly lower than in England during the field operation, standing at 70.7% on 23 March (the peak date for paper responses) and 84.5% on 1 April, compared with 76.3% and 87.8% respectively in England. In part this may reflect the higher proportion of paper responses in Wales, with a slight lag in them being registered compared with online responses, as PQs were posted back and receipted.

Section 7 in our Designing a digital-first census publication shows the peak for paper returns (on 23 March) was two days after the online peak on Census Day. The return rate gap between Wales and England narrowed during the campaign to the final return rates of 96% and 97% respectively. 

Visit schedules for census officers were updated in real-time and allocated through their Fieldwork Management Tool, based on information from our Response Management system (see chapter 4 of the Census 2021 general report) including on outcomes of previous visits and MI on census completions. 

Officers were able to offer advice on the doorstep and direct householders to support channels, to enable householders to complete their census. Areas with low response rates were additionally targeted by our mobile field force, deployed based on real-time response data. In Wales, officers were able to specify that any further visit to a property should be by Welsh-speaking field staff, enabling us to deploy those staff more effectively. 

Across England and Wales, census officers conducted over 16,685,000 visits to over 5 million households, a quarter of properties on the address frame. During these visits, of which 4.5 million resulted in contact with the household, census officers were able to offer advice on the doorstep and direct householders to support channels, to enable householders to complete their census. In Wales, the proportion visited was smaller, with 296,500 properties, or 21.2% of households, being visited, just under half of which (10.2%) received only one visit. 

While the campaign in Wales was successful and the return rates well above our targets, there were lessons to be learned from it. Many of the factors that affected the England and Wales campaign were notable in Wales, such as the shortfall in field staff recruitment and the impact of the pandemic. Other factors were specific to or particularly affected the campaign in Wales. 

The high proportion of paper-first areas in Wales may have had an impact on return rates during the campaign. As noted in the general report, while the general balance online and paper contact worked for England and Wales, for any future mixed-mode census operation, a different approach to the proportion and targeting of paper-first contact might be appropriate. This applies particularly for Wales with its higher proportion of paper-first areas.

It may also be worth considering greater operational independence in planning and delivering the field operation in Wales and incorporating local insights from existing teams and staff across the ONS at an earlier stage. This may help to better tailor the operation for the particular requirements of data-collection in Wales. This could take account of geographic, social and cultural factors, including language requirements for staff and census materials, the size and rurality of some areas of Wales, and reflecting variations across Wales. 

Greater operational independence across the census operations could also reduce issues encountered when the non-compliance operations in parts of Wales also covered a region of England. This overlap caused minor issues regarding the provision of census products, with teams in this area being issued packs containing bilingual material that was not relevant to officers working specifically within English areas. The complete independence of field operations in Wales needs to be considered to ensure consistency in approach and reduce any potential issues with logistical and administrational requirements. 

Enumerating communal establishments and special population groups in Wales 

Unlike households, active engagement with CEs and SPGs started before Census Day. Initial engagement was done though liaising with relevant stakeholder groups to establish contacts, and through the ONS engagement team reaching out to local authorities and SPG groups to discuss the strategy for their respective enumeration. 

The four phases of the operation itself were:

• engagement – initial contact for relationship building and information gathering 

• delivery – hand delivery of ICLs and PQs to CEs and SPGs (from 23 February until Census Day), by post or by hand 

• Census Weekend – a specialised operation over Census Weekend (20 to 22 March) focused on population groups with no fixed address, such as continuous cruisers living on boats, those living on the road in a moveable residence, and rough sleepers, to ensure these population groups had the opportunity and means to complete the census 

• follow up – contacting non-responding addresses to encourage and support returns 

The CE and SPG operation in Wales was affected by the same factors as that in England, namely the shortfall in recruitment and the impact of the coronavirus pandemic. While the area manager and support posts were filled, we had 10 team leaders and 66 communal establishment officers, compared with the planned 14 team leaders and 113 officers. 

Resulting logistical problems in the distribution of kit to staff were overcome thanks to a significant additional effort by the two area managers, whose efforts and leadership helped to drive teams and CE staff to deliver what was needed, despite the issues caused by the recruitment shortfall.

Team leaders in affected areas also made a significant contribution to the success because of their willingness to cover field staff activities where recruitment was low. The pandemic and lockdown restrictions meant that much of the planned activity for the engagement phase was not possible.

Additional burdens on our partners caused by the pandemic meant that engaging with some population groups was more difficult than it would otherwise have been, including rough sleepers (a particularly vulnerable and difficult to enumerate sub-group within the homeless population). 

Support for respondents 

Public contact centre 

Our public contact centre provided help and support to complete Census 2021 and the Census Coverage Survey, through telephone advisers and through Webchat, SMS and Webmail. This included language support and a telephone capture service that enabled respondents identified as requiring assistance to complete the census over the phone with an adviser. 

People calling the freephone 0800 lines were offered three options by the Interactive Voice Response (IVR): to request a PQ, to hear frequently asked questions (FAQs), or to speak to an adviser. 

In Wales, the IVR messaging was bilingual with the Welsh language message first. Of the 4.5 million calls that reached the IVR, 238,000 were made to the Wales 0800 number. Some callers in Wales will also have used the separate multi-language number and the auto-request number for new PQs that was added during the campaign, which served both England and Wales.

Callers to the Wales line were asked the language in which they wanted to continue: 189,000 chose English and 26,000 chose Welsh, while the remainder made no choice and ended their call.

Of the callers who selected a language, 84,000 reached a contact centre adviser, while 44,000 requested PQs (in addition to those using the auto-request line) and 14,000 accessed the FAQs. Of the remainder, the majority used the pre-recorded messages in the IVR, while a minority ended their calls early, for example, while in a queue to speak with an adviser.

Webchat, SMS and Webmail were also available in Welsh. In addition to Welsh-language questionnaires being available, accessible products were provided in Welsh including braille and large print. 

Online help 

The online help part of the Census 2021 website aimed to answer any question a respondent might have had that could have affected their ability or willingness to complete the census questionnaire. This included guidance on how to obtain a PQ or a new online access code, and guidance on how to answer the census questions. 

The pages evolved to support the Census Coverage Survey and Non-Compliance phases after the main data-collection period. As with the eQ and the rest of the website, online help was bilingual, with a toggle to switch between English and Welsh-language versions of each page. 

The English language online help pages, which supported respondents in England and in Wales, had 2.1 million recorded pageviews across 1.1 million visits. The Welsh language pages had around 2,700 recorded pageviews across 2,500 visits.

The Welsh language pages with the highest recorded pageviews were similar to the most popular English language pages (see Section 4.4.4: Online help in the Census 2021 general report).

The most visited Welsh language pages were: 

• Help gyda'r cyfrifiad (Help with the census), with 450 recorded pageviews 

• Cael cod mynediad neu holiadur papur (Get an access code or paper census), with 250 pageviews 

• Ieithoedd (Languages), with 200 pageviews 

• Dechrau arno (Getting started), with 150 pageviews 

• Dechrau eich cyfrifiad ar-lein (Completing your census online), with 150 pageviews 

Assisted digital support through census support centres 

In-person assisted digital support aimed to provide face-to-face help and support for people without the digital skills, equipment or connectivity to complete the census online, or who did not feel confident in doing so. This service was provided at census support centres, primarily based in trusted or familiar locations such as libraries and community hubs. During the data-collection operation, 744 centres opened across 266 local authority areas in England and Wales.

The pandemic and lockdown restrictions had a significant impact on the operation of census support centres. Decisions made on opening centres reflected the legislation and guidance in operation at the time, including Welsh Government guidance.

In Wales, 66 centres were planned, including at least one in each local authority area, but the pandemic and lockdown restrictions meant that few were able to open. Welsh-language support was offered at all sites operating in Wales, in some cases by appointment only.

Engagement 

The primary aim of our census engagement activity was to focus on those groups that were less likely to take part in the census without additional interactions. Engagement was one of several important interactions to encourage and support participation, alongside the communications campaign and the work of the field force and public support services. 

Building on our successful approach in 2011, Census 2021 engagement activity focused on local authority (LA) and community engagement. LA engagement focused on sharing and building trust in our methods and eliciting practical support to improve the effectiveness of the census data-collection operation. 

We worked with LAs across England and Wales through the whole of the census cycle from 2017 to 2023. This included both strategic engagement across all LAs and local working between our census engagement managers and LAs. Ahead of the data-collection operation, a Local Authority Partnership Plan (LAPP) was developed between each LA and the ONS's Census Engagement Manager, setting out local engagement priorities and actions undertaken to support the census. This was used to support the partnership and to provide a clear focus on those activities and areas where LAs could provide significant added value.

The ONS asked each LA across England and Wales to nominate a census contact, and these Census Liaison Managers, or Assistant Managers, and the wealth of local knowledge and extensive networks of contacts they brought were significant in helping to deliver a successful census.

Examples of their support included providing opportunities to present to elected members of the LA and for census articles to appear in LA newsletters or social media platforms. They also helped us to use libraries' click and collect systems to reach members of local blind or partially sighted communities, visited Gypsy and Traveller sites with census teams, and supported the "light up" campaign (covered in the "Communications campaign, media and public relations" section). 

Community engagement focused on building awareness, trust and advocacy from charities, leaders and representatives of communities from whom we may otherwise get lower response rates. These community leaders and representatives then built awareness, trust and willingness to complete the census within communities. 

For Census 2021, the activity had two core elements: national and local engagement. 

National engagement 

National engagement started years in advance of census collection and continued through to outputs and dissemination. Our approach was to form relationships in advance of the census with leaders and respected representatives of communities and population groups that faced barriers to taking part.

This approach allowed the ONS Engagement team to understand the concerns of these groups, provide up-to-date information and raise awareness of the census. This activity gained buy-in from the leaders and representatives who saw the importance of the census and their communities participating.

Local engagement 

While the national engagement began well before the data-collection campaign, local engagement began in September 2020. This engagement activity was carried out by 200 Census Engagement Managers (CEMs) and 115 Community Advisers (CAs) across England and Wales. There was an increase in both the number of engagement staff and the number that could speak Welsh compared with the 2011 Census. 

Each CEM worked with LAs and communities in their area, while CAs worked with specific population groups, mainly in urban areas. We had 12 CEMs in Wales, each mainly covering one local authority area, but some covering multiple local authority areas and two working in Cardiff. Four CAs were appointed, three in Cardiff working with Arab, Chinese and Indian communities and one across Cardiff and Newport working with the Somali community.

The CEMs and CAs were successful in using a wide range of tools to engage with their communities such as: 

• social media platforms 

• local papers and newsletters 

• radio stations 

• video conferencing 

• leaflet and poster distribution 

Welsh-language ability was a desirable skill in all local community role job descriptions; for five CEMs it was listed as an essential skill.

CEM and CA activities in Wales included interviews on Welsh-language TV magazine programmes "Heno" and "Prynhawn Da", and radio interviews on stations such as BGfm and Bro Radio. It also included Zoom-based census completion and question and answer sessions, including a "Bollywoodzoom" dance event, attended by leading figures from Sikh and Hindu communities in Wales. 

The ONS provided regular updates to the Welsh Government's Welsh Statistical Liaison Committee and Third Sector Statistics User Panel on census planning and operation. These groups included representatives from the public and third sector in Wales. 

Communications campaign, media and public relations 

Our communications campaign was integrated with the phases of contact with households. Beginning with priming audiences for the upcoming contact, it moved on to encouraging completion before and after Census Day. This included targeted advertising for areas and populations with lower response rates.

The campaign made use of a range of communications channels to reach a large proportion of the population. This ranged from TV and radio to social media and digital, including social media influencers and paid media partnerships. The communications campaign also focused on messaging for groups facing barriers to taking part. With a common style to the whole campaign, and our "it's about us" message, for these targeted communications we used different languages and bespoke photography, representing different population groups.

In keeping with the bilingual census, all campaign materials and content used in Wales were produced bilingually, ranging from adverts and videos to posters and leaflets. 

We undertook quantitative and qualitative research with Welsh speaking audiences to ensure that the campaign proposition and messaging resonated with Welsh audiences. 

In addition to the Welsh-language census website, we had Welsh-language social media channels promoting Census 2021. We worked closely with Facebook and Twitter to develop search prompts, which created a huge amount of free promotion. These were bilingual, the first time that Twitter had produced a prompt in Welsh.

The campaign used Welsh media channels, both English and Welsh-language, optimising content for Welsh speaking audiences with targeted advertising. Our press releases, including 53 in Welsh, achieved coverage through a range of outlets before and during the data-collection campaign, including Wales Online, BBC Wales, West Wales Chronicle and South Wales Argus, and Welsh-language interviews on S4C and BBC Radio Cymru. 

We worked closely with the Welsh Government on many aspects of the campaign, sharing the campaign proposition and messaging to obtain feedback. This included working with a Welsh language officer on the development of the Welsh language TV advert and social media videos and animations. We also collaborated on the paid for media plan for Welsh audiences. 

The public relations campaign included a search for 22 "community heroes". Fronted by "Gavin and Stacey" star Joanna Page, this campaign searched for ordinary people doing extraordinary things to help their communities, whom we recognised with a "purple plaque" in Census 2021 brand colours. The heroes were announced on 11 March 2021 and included three people from Wales. 

Alongside this was the illumination of over 200 buildings across England and Wales in purple lights over Census Weekend to create a sense of occasion. These included the Wales Millennium Centre in Cardiff and the National Library of Wales in Aberystwyth.

School programmes 

Primary and Secondary Education programmes were developed to raise awareness of the census and develop statistical literacy. These were the Census Secondary Education Programme (CSEP) and a creative primary school education programme called "Let's Count!" (LC). 

These programmes were also designed to reach parents, particularly those in areas where high proportions faced other barriers to completion. Across LC and CSEP, a series of free education resources were developed, informed by the national curriculum. These resources were tailored to focus on educating pupils on the census, the use of census data to shape services locally and nationally, and the importance of everyone taking part. 

The websites and materials were available in English and Welsh. 

More than half of primary schools in Wales registered for LC, giving the programme a reach of over 122,000 children. The CSEP saw 38.7% of secondary schools register across England and Wales, including 81 schools in Wales.

Response rates and mode of completion 

The overall household response rate for Wales was 96%, compared with 97% in England and Wales as a whole. The figures for the 2011 Census were 93% and 94% respectively.

Table 2 shows response rate figures for each local authority area in Wales, Wales as a whole, and England and Wales. While the overall response rate was slightly lower in Wales, the rates were high and the variability was small, with all areas seeing response rates between 94% and 98%.

More information for both England and Wales can be found in the Local authority comparison tool, available through our Measures showing the quality of Census 2021 estimates web page. 

The online share of household returns in Wales was 67.6%. While lower than the 90.1% in England, this largely reflects the proportion of households in "paper-first" areas in Wales (see "Delivery of materials" in Data-collection operations).

Table 2 shows the proportions of households in each local authority area that were "paper-first" (receiving a paper questionnaire), along with the online completion rates for those households and for "digital-first" households (receiving an ICL). The figures for Wales by mode of initial contact were similar to those for England and Wales combined.

At the local authority level, the online completion percentages were between 35.9% and 49.6% for paper-first households and between 90.8% and 95.2% for digital-first. There was greater variability between smaller areas, with online responses between 28.6% and 64.4% for paper-first areas, and between 85.9% and 99.3% for digital-first.

More information on the online share of returns can be found in our Designing a digital-first census.

Language of completion 

Respondents in Wales were able to complete the census in Welsh or in English. While measuring the Welsh-language response rate for paper questionnaires (PQs) is simple as people completed either a Welsh or an English version of the questionnaire, the online response is more complex.

The capacity to toggle between English and Welsh throughout the electronic questionnaire (eQ) means that people could view and complete questions in different languages within a single response, and members of a household could complete the census in different languages. To provide a figure comparable with that for the PQ, Table 2 includes as a "Welsh language questionnaire" responses for those households that used the Welsh language pages to launch or submit their census response, or both. 

In total, 23,500 households completed the census using a Welsh language questionnaire, equating to 1.7% of household responses in Wales. There was a broadly even split between modes of completion with 12,350 doing so online and 11,150 on paper. However, as paper responses made up only a third of census responses in Wales, this means that the proportion of Welsh responses was higher on paper, with 2.6% of household PQs completed in Welsh compared with 1.4% of online responses.

The proportion of Welsh language responses was lower than for the 2011 Census. In our Taking the Census in Wales report (PDF, 267KB), we reported that by June 2011, an estimated 3.4% of household returns from Wales had been completed in Welsh. In 2011, the proportion of Welsh-language responses was higher for online responses than for paper: 5% compared with 3%.

Table 2 shows that the proportion of Welsh language responses in Census 2021 varied considerably between local authority areas. The local authorities with the largest proportion of Welsh language responses were those with the highest proportions of usual residents aged 3 years and over who could write Welsh. These were Gwynedd (13.6% Welsh language responses); Isle of Anglesey (6.1%) and Ceredigion (7.1%). Meanwhile in 14 local authority areas, less than 1% of households responded using the Welsh questionnaire.

It is worth noting that the figures used here record a household's response languages, not those of individual respondents.

Differing Welsh-language ability within households is demonstrated in the Census 2021 estimates for Welsh language household composition: while 23.7% of households included at least one person able to speak Welsh, less than half that number comprised people who could all speak Welsh (9.0%). Welsh-language census responses recorded here would not include a household where, for example, the householder launched the response in English, other members of the household completed it in Welsh, and the householder or final respondent submitted the response in English.

The Census 2021 data went through a rigorous quality assurance (QA) process to ensure that: 

• the data were being processed correctly 

• unexpected features of the data were dealt with appropriately 

• the final estimates were plausible in the context of the other information available to us 

Quality assurance covered both the population estimates and the census results for each topic included on the census questionnaire. 

For the first time, we invited local authorities to contribute to our quality assurance of the population estimates for their area, giving all local authorities and other relevant bodies, including the Welsh Government, the chance to see indicative census estimates before the data processing was complete. This meant that they could tell us about any aspect of those estimates where other evidence suggested we needed to look at it again within our quality assurance process. 

In addition to taking part in the Local Authority Insight Initiative, the Welsh Government was also involved in the QA of estimates for Wales, including Welsh language ability. An executive panel consisting of the National Statistician, the Deputy National Statistician responsible for the census, and the Chief Statistician for Wales signed off the census results for publication as part of the QA process. 

More on processing and quality assurance can be found in chapter 6 of the Census 2021 general report.

Main statistical releases 

The strategic aims of Census 2021 outputs were flexibility, timeliness, accessibility, and relevance. The first release on 28 June 2022 provided rounded population estimates of usual residents, along with estimates by sex and five-year age bands, down to local authority area level. The rounded population estimate for Wales was announced at Ysgol Gynradd Gymraeg Tonyrefail in Rhondda Cynon Taf, the runner up in the Census 2021 Let's Count primary school campaign. They were also laid before the Senedd on the same day. 

The first unrounded population estimates and characteristics data from Census 2021 were published in topic summaries. A topic summary is a set of data supported by statistical bulletins, interactive products, and quality information, grouped by a similar theme. Alongside the first topic summary release, we also published measures and interactive tools showing the quality and quality assurance of the census data. 

The first topic summary, published on 2 November 2022, included the unrounded population estimates. These estimated that the size of the usual resident population in Wales was 3,107,494. This was the largest population figure ever recorded for Wales and represented an increase of approximately 44,000 (1.4%) since 2011. 

Further topic summaries covering most census topics and questions were published from November 2022 to January 2023. The bulletins for these were published in html (or web) format in English and initially as a downloadable PDF in Welsh, with Welsh-language html pages added later. Most topic summary bulletins included a summary of the data as they related to Wales. They also linked to the dataset pages on the ONS website, which allowed the data to be downloaded for Wales or for smaller geographies within Wales (such as local authorities and local health boards). 

For the Welsh language topic release on 2 December 2022, the ONS published a short release (in English and Welsh), with Welsh Government publishing the main Welsh language in Wales bulletin on the same day. For other topics, the Welsh Government published statistical bulletins on their Census of Population webpage, summarising the main points for Wales alongside our releases. Following the release of the Welsh language data, the ONS and Welsh Government published a joint workplan of coherence of Welsh language statistics (see Quality section). 

A major innovation for Census 2021 was the Create a custom dataset tool, which opened up census data for users to explore and find the data they need. This was part of a blended offer that also included ready-made multivariate tables. These products, along with the interactive tools and analysis published alongside them, provided users with access to a wide range of census data. The complexity of creating bilingual products meant that the create a custom dataset tool was produced in English only. 

Alongside these products, we published a greater range of quality information and other supporting information than previous censuses. Two of the most important of these publications were published bilingually: the Census 2021 dictionary and Quality and methodology information (QMI) for Census 2021. 

Data-visualisation 

Much greater use was made of data visualisation in the topic summary statistical bulletins and analysis articles than in 2011, including interactive maps, charts, population pyramids and heatmaps.

We also published a range of stand-alone interactive products, ranging from games and semi-automated releases, to enable users to find census data for bespoke areas. The complexity of creating bilingual products meant that these were produced in English only. 

The interactive Census 2021 products included:

• Census maps

• localised How your area has changed in 10 years articles

• Build a custom area profile and Create a population profile tools 

• Origin-destination data visualisation

Analysis 

The census analysis programme produced a comprehensive set of research that shone a light on public-policy topics. By the end of 2023, we had published over 50 analysis articles.

These publications were based on multivariate data, ranging from breakdowns of populations covered by the topic summaries by age and sex to a profile of centenarians in England and Wales, and variables calculated from census data, such as overcrowding and underoccupancy. This wider range of analysis included articles on different population groups such as unpaid carers, UK armed forces veterans or groups defined by ethnic group, religion or both (such those identifying as Sikh, Jewish and Somali). 

Working with the Welsh Government, we prioritised for publication in Welsh those articles relating to Welsh language, second homes, national identity and other specific topics. 

Publications available bilingually included: 

• Overcrowding and under-occupancy by household characteristics 

• More than 1 in 10 addresses used as holiday homes in some areas of England and Wales 

• People in England and Wales with a different address in the UK a year before the census 

• Number of vacant and second homes 

• Towns and cities, characteristics of built-up areas 

• Household characteristics by tenure 

In addition to the ONS publishing bulletins on Wales alongside our topic summaries, the Welsh Government also produced analysis publications, working closely with the ONS. 

Published on the Welsh Government's Census of Population webpage, these included:

• Welsh language by population characteristics (including ethnic group, national identity, religion, sexual orientation, gender identity, labour market and health) 

• Welsh language composition of households in Wales 

• analysis of outcomes in health, housing, education and economic status across ethnic groups, for disabled people, and by sexual orientation and gender identity 

• analysis of second addresses used as holiday homes and characteristics of the population by area deprivation 

Microdata for education and research 

In addition to publishing estimates from Census 2021, the ONS makes samples of census data available for research and educational purposes, while ensuring that personal data are kept secure, including through microdata samples and the longstanding Longitudinal Study. We also make de-identified record-level census data available to accredited or approved researchers through the UK and Welsh Governments' preferred secure data-linking facilities. Both are accredited under the Digital Economy Act 2017 (DEA) and only allow access to accredited researchers. 

The UK Government's preferred facility is the Integrated Data Service (IDS), a new cross-government initiative with ONS acting as the lead delivery partner, described in Section 8.6 of the Census 2021 general report. The SAIL (Secure Anonymised Information Linkage) Databank is a research resource focused on improving health, well-being and services funded by the Welsh Government's Health and Care Research Wales.

Census 2021 data for Wales were made available in the SAIL Databank in February 2024 for researchers, enabling the linkage of Census 2021 data to other de-identified data in the trusted research environment. 

The ONS's Secure Research Service (SRS) currently operates in parallel to the IDS. The SRS gives accredited or approved researchers secure access to de-identified, unpublished data to work on research projects for the public good.

Chapter 7 of the Census 2021 general report describes the measures of quality for Census 2021 data, including response rates and confidence intervals. As Table 2 shows, the response rate for Wales was 96%, higher than the 94% target rate, and all local authorities (LAs) in Wales had a response rate over 94%, again exceeding the target of 80% for all LAs. 

Confidence intervals are a statistical measure of uncertainty. Our 95% confidence interval had a lower limit of 0.16% below the England and Wales population estimate and an upper limit of 0.07% above. This was well within our target of plus or minus 0.2% and a narrower overall width of the interval than in 2011.  

For Wales, the confidence interval width was 0.82%, slightly wider than for most English regions. All LAs in Wales had confidence intervals narrower than the target width of 6%. For 15 of the 22 Welsh LAs, the width was less than 3%. Confidence intervals for individual LAs can be found in Measures showing the quality of Census 2021 estimates. 

Welsh language ability joint workplan 

The Census 2021 estimate that 17.8% of people aged three years or older living in Wales were able to speak Welsh was a decrease of 1.2 percentage points since Census 2011.

In contrast, estimates on Welsh speakers aged three years and older in Wales produced from the Annual Population Survey (APS) for the year ending 31 March 2021 were 3.4 percentage points higher than in the year ending 31 March 2011. The APS also estimated a much larger proportion of the population aged three years and older in Wales were able to speak Welsh (29.2% in the year ending 31 March 2021). 

Differences in the estimates of Welsh language ability between the census and household surveys such as the APS are longstanding, and both the Office for National Statistics (ONS) and the Welsh Government have explored possible reasons for some of these differences in the past. While household surveys typically provide us with higher estimates of Welsh-speaking ability, this is the first time that the census has estimated declining numbers of Welsh speakers, and the APS has estimated increasing numbers of Welsh speakers. 

The Welsh Government considers the census to be the authoritative source for information about the Welsh language ability of the population in Wales. It uses census data to monitor progress against its aim of a million Welsh speakers by 2050, and wanted to understand more about these longstanding differences between the census and other data sources.

In April 2023, the Welsh Government and the ONS published a joint workplan of coherence of Welsh language statistics, outlining a programme of research to analyse these differences and potentially make recommendations on the future production of statistics on the Welsh language. This workplan involves several work projects, including quantitative and qualitative research. 

The first findings from the workplan were published in October 2023, and considered linked, de-identified responses of individuals who had completed the Labour Force Survey (LFS) around the time of Census 2021. Work on the outstanding research projects from the joint workplan has continued in 2024 and 2025, alongside joint research on the provision of Welsh language statistics as part of our work on the future of population and migration statistics (see chapter 11 in the Census 2021 general report).